Top Banner
329 Syst. Biol. 48(2):329–351, 1999 De nitions in Phylogenetic Taxonomy: Critique and Rationale PAUL C. SERENO Department of Organismal Biology and Anatomy, University of Chicago, 1027 E. 57th Street, Chicago, Illinois 60637, USA; E-mail: [email protected] Abstract.— A general rationale for the formulation and placement of taxonomic de nitions in phy- logenetic taxonomy is proposed, and commonly used terms such as “crown taxon” or “node-based de nition” are more precisely de ned. In the formulation of phylogenetic de nitions, nested refer- ence taxa stabilize taxonomic content. A de nitional con guration termed a node-stem triplet also stabilizes the relationship between the trio of taxa at a branchpoint, in the face of local change in phylogenetic relationships or addition/deletion of taxa. Crown-total taxonomies use survivorship as a criterion for placement of node-stem triplets within a taxonomic hierarchy. Diversity, morphol- ogy, and tradition also constitute heuristic criteria for placement of node-stem triplets. [Content; crown; de nition; node; phylogeny; stability; stem; taxonomy.] Does one type of phylogenetic de nition (apomorphy, node, stem) stabilize the taxo- nomic content of a taxon more than another in the face of local change of relationships? Is one type of phylogenetic de nition more suitable for clades with unresolved basal re- lationships or uncertain outgroups? Which type of phylogenetic de nition is preferable for clades whose members are entirely living or extinct? Questions like these have not been rigor- ously addressed in phylogenetic taxonomy. Rather, attention has been focused on (1) the adaptation of traditional rules governing synonymy and redundancy for use within the phylogenetic system, and (2) the recom- mendation that “widely recognized” names are better restricted to crown taxa than to more-inclusive taxa with extinct basal mem- bers (de Queiroz and Gauthier, 1990, 1992; Rowe and Gauthier, 1992; Bryant, 1994, 1996; Lee, 1996). Despite increasing use of phy- logenetic de nitions in systematics, a gen- eral rationale has yet to be proposed for the formulation and placement of phyloge- netic de nitions, and many commonly used terms such as “crown taxon” and “node- based de nition” have yet to be speci cally de ned. LACK OF A GENERAL RATIONALE De Queiroz and Gauthier (1990, 1992) rst articulated the general structure of phylo- genetic de nitions, outlining apomorphy-, node-, and stem-based de nitions (Fig. 1). Most phylogenetic de nitions have been constructed in the systematic literature since then without explanation or justi cation for the particular type of de nition used. The justi cation given for preferential use of node- and stem-based de nitions for crown and total taxa, respectively, is incomplete or inaccurate, as reviewed below. Total taxa and crown taxa require stem- based and node-based de nitions, respectively.Given that a total taxon includes all cur- rently known and potential extinct out- groups that are most closely related to a particular crown taxon (Table 1), the total taxon must have a stem-based de nition (de Queiroz and Gauthier, 1992). The stem- based structure of the de nition ensures the inclusion of all taxa up to, but excluding, the common ancestor shared with its sister to- tal taxon. This justi cation is suf cient, be- cause neither an apomorphy- nor a node- based de nition would include all potential extinct outgroups. A crown taxon, in contrast, does not require a node-based de nition, although one is commonly assumed. De Queiroz and Gauthier (1992:469) simply remarked, “Names can be associated unambiguously with crown clades using node-based def- initions.” Likewise, Rowe and Gauthier (1992) and McKenna and Bell (1997) pro- posed node-based de nitions for Mammalia without explaining why de nitional type is preferable. Lee (1996:1103) remarked, “A crown-clade de nition results when both taxa implicated in a node-based de nition
23

Sereno, 1999b

Mar 10, 2016

Download

Documents

Felipe Elias

PAUL C. SERENO 329 Department of Organismal Biology and Anatomy, University of Chicago, 1027 E. 57th Street, Chicago, Illinois 60637, USA; E-mail: [email protected] Syst. Biol. 48(2):329–351, 1999 VOL. 48 SYSTEMATIC BIOLOGY FIGURE 1. Node-based and stem-based phyloge- netic denitions, which usually have been shown graphically by encircling portions of a cladogram (de Queiroz and Gauthier, 1992), are indicated here by a dot (node-based) or arrow (stem-based). 330
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Sereno, 1999b

329

Syst Biol 48(2)329ndash351 1999

Denitions in Phylogenetic Taxonomy Critique and Rationale

PAUL C SERENO

Department of Organismal Biology and Anatomy University of Chicago1027 E 57th Street Chicago Illinois 60637 USA E-mail dinosauruchicagoedu

Abstractmdash A general rationale for the formulation and placement of taxonomic denitions in phy-logenetic taxonomy is proposed and commonly used terms such as ldquocrown taxonrdquo or ldquonode-baseddenitionrdquo are more precisely dened In the formulation of phylogenetic denitions nested refer-ence taxa stabilize taxonomic content A denitional conguration termed a node-stem triplet alsostabilizes the relationship between the trio of taxa at a branchpoint in the face of local change inphylogenetic relationships or additiondeletion of taxa Crown-total taxonomies use survivorshipas a criterion for placement of node-stem triplets within a taxonomic hierarchy Diversity morphol-ogy and tradition also constitute heuristic criteria for placement of node-stem triplets [Contentcrown denition node phylogeny stability stem taxonomy]

Does one type of phylogenetic denition(apomorphy node stem) stabilize the taxo-nomic content of a taxon more than anotherin the face of local change of relationshipsIs one type of phylogenetic denition moresuitable for clades with unresolved basal re-lationships or uncertain outgroups Whichtype of phylogenetic denition is preferablefor clades whosemembers are entirely livingor extinct

Questions like these have not been rigor-ously addressed in phylogenetic taxonomyRather attention has been focused on (1) theadaptation of traditional rules governingsynonymy and redundancy for use withinthe phylogenetic system and (2) the recom-mendation that ldquowidely recognizedrdquo namesare better restricted to crown taxa than tomore-inclusive taxa with extinct basal mem-bers (de Queiroz and Gauthier 1990 1992Rowe and Gauthier 1992 Bryant 1994 1996Lee 1996) Despite increasing use of phy-logenetic denitions in systematics a gen-eral rationale has yet to be proposed forthe formulation and placement of phyloge-netic denitions and many commonly usedterms such as ldquocrown taxonrdquo and ldquonode-based denitionrdquo have yet to be specicallydened

LACK OF A GENERAL RATIONALE

De Queiroz and Gauthier (1990 1992) rstarticulated the general structure of phylo-genetic denitions outlining apomorphy-node- and stem-based denitions (Fig 1)

Most phylogenetic denitions have beenconstructed in the systematic literature sincethen without explanation or justication forthe particular type of denition used Thejustication given for preferential use ofnode- and stem-based denitions for crownand total taxa respectively is incomplete orinaccurate as reviewed below

Total taxa and crown taxa require stem-based and node-based denitions respectivelymdashGiven that a total taxon includes all cur-rently known and potential extinct out-groups that are most closely related to aparticular crown taxon (Table 1) the totaltaxon must have a stem-based denition(de Queiroz and Gauthier 1992) The stem-based structure of the denition ensures theinclusion of all taxa up to but excluding thecommon ancestor shared with its sister to-tal taxon This justication is sufcient be-cause neither an apomorphy- nor a node-based denition would include all potentialextinct outgroups

A crown taxon in contrast does notrequire a node-based denition althoughone is commonly assumed De Queirozand Gauthier (1992469) simply remarkedldquoNames can be associated unambiguouslywith crown clades using node-based def-initionsrdquo Likewise Rowe and Gauthier(1992) and McKenna and Bell (1997) pro-posed node-based denitions for Mammaliawithout explaining why denitional typeis preferable Lee (19961103) remarked ldquoAcrown-clade denition results when bothtaxa implicated in a node-based denition

330 SYSTEMATIC BIOLOGY VOL 48

FIGURE 1 Node-based and stem-based phyloge-netic denitions which usually have been showngraphically by encircling portions of a cladogram (deQueiroz and Gauthier 1992) are indicated here by a dot(node-based) or arrow (stem-based)

are represented by extant formsrdquo And Wyssand Meng (1996559) stated that the deni-tions of crown taxa ldquoare thus node-based intheir formulationrdquo without presenting anyreasoning for such a conclusion

Any group of extant species may justas well be united by a stem-based deni-tion The crown taxon Amniota for examplecould have a stem-based denition such asldquoany Recent tetrapod more closely related toMammalia than to Anura and all extinct de-scendants of their most recent common an-cestorrdquo A stem-based denition for a crowntaxon in fact may be preferable if the basalrelationships among extant taxa within acrown clade are poorly established as seemsto be the case with Amniota (see also Ro-dentia Wyss and Meng 1996) (Fig 2b) Allmembers of the crown taxon will be in-cluded even if basal ingroup relationshipsare poorly resolved because a stem-baseddenition is based on reference to an out-group taxon (or taxa)

A single suboscine bird for examplewould sufce as an outgroup referencetaxon for a stem-based denition of thecrown taxon Oscines A node-based de-nition of the same crown taxon would re-quire listing many oscine subgroups to en-sure inclusion of all living species currentlyregarded as oscines On the other hand ifthe basal dichotomy within a crown taxonis well established and outgroup relation-ships are uncertain (Fig 2a) a node-based

denition will more effectively stabilize tax-onomic content Thus there is no currentjustication for exclusive use of node-baseddenitions to delimit crown taxa Node-based crown taxa in conjunction with stem-based total taxa however yield a local def-initional conguration that can provide jus-tication for preferential use of node-basedcrown taxa (see Node-Stem Triplet below)

Node-based crown taxa are more stable infor-mative and accurate than taxa dened by ex-tinct species and have been (or should be) as-sociated with ldquowidely usedrdquo namesmdashGauthier(198612) stated that restricting ldquoAvesrdquo toldquoliving taxardquo maximizes ldquostability and phy-logenetic informativenessrdquo and de Queirozand Gauthier (1992468) suggested that ldquobi-ologists commonly use the widely knownname Aves when making generalizationsthat apply to extant birds alonerdquo With re-gard to the taxon Mammalia Rowe andGauthier (1992372) observed that ldquono onein several centuries has mistakenly assigneda Recent mammal to some other taxonrdquoMcKenna and Bell (199732) remarked thatldquoa crown group is likely to remain relativelystablerdquo

Despite these and similar statements inthe literature node- or stem-based crowntaxa are not demonstrably more stable interms of meaning or content Any node- orstem-based denition that specically iden-ties a most recent common ancestor clearlyspecies the boundaries of a clade regard-less of the living or extinct status of the ref-erence taxa or the particular phylogeny towhich the denition is applied (Lee 1996)

In terms of taxonomic content reden-ing Mammalia as a crown taxon does notimmediately clarify which extinct taxa willbe included or excluded Mammalia asa crown taxon appears more stable onlyif extinct taxa are ignored and attentionis focused on the great phylogenetic dis-tance to their nearest living sister groupEven so inclusion of some living mam-mals may be uncertain Rowe and Gau-thierrsquos (1992375) preferred node-based def-inition of Mammalia (ldquothe last commonancestor of Monotremata and Theriardquo seealso McKenna and Bell 19973235) maybe synonymous with Theria if monotremes

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 331

TABLE 1 Denitions for terms used in this paper

Term Denition

Taxonomic denition relational statement specifying the taxonomic content of a taxonTaxonomic diagnosis descriptive statement specifying the apomorphies (derived character states)

that serve to identify members of a taxonTaxonomic content existing and potential taxa or individuals that by denition are included

within a taxonReference taxon taxon serving as a reference in a phylogenetic denitionCrown taxon a living species or a clade that can be dened by living species whose

immediate outgroup is extinctTotal taxon clade composed of a crown taxon plus all extinct outgroups more closely

related to it than to another crown cladeStem taxon an extinct species or clade for which immediate outgroup includes at least

one living memberNode-based denition statement specifying a clade composed of the most recent common ancestor

of two or more reference taxa and all descendantsStem-based denition statement specifying a clade composed of all descendants more closely

related to one reference taxon (or taxa) than another (or others)Complementary denitions phylogenetic denitions with the same reference taxa (as in

a node-stem triplet)Reciprocal denitions phylogenetic denitions with the same reference taxa but in reverse order

(as in opposing stem-based taxa of a node-stem triplet)Node-stem triplet trio of taxa with complementary denitions consisting of a node-based

taxon and two subordinate stem-based taxaTaxonomic equivalence statement of equivalence (or equation) in which the taxonomic content of

one taxon equals that of subordinate taxa plus their common ancestor(eg A = B + C as in a node-stem triplet)

prove to be more closely related to marsu-pials as indicated by some recent molecu-lar evidence (Penny and Hasegawa 1997)Stability of taxonomic content is not re-lated to the living or extinct status of ref-erence taxa in a phylogenetic denitionbut rather to the particular choice of ref-erence taxa For many node-based deni-tions of crown taxa stability is also re-lated to the certainty with which a speciedbasal dichotomy has been correctly ascer-tained

That ldquowidely usedrdquo names such as Avesshould be assigned to crown taxa has beenjustied by arguing that (1) neontologistsshould report their observations accuratelyreferring only to crown clades rather thancrown clades plus some of their extinct out-groups and that (2) doing so follows long-standing taxonomic practice (de Queirozand Gauthier 1992) The condition in ex-tinct outgroups for characters that are notpreserved however is ambiguous To sup-pose that Archaeopteryx lacked an apomor-phic molecular sequence found in crown-group birds (Neornithes) is no more justied

than tosuppose that the ancientbird had thissequence (Lee 1996) Accelerated character-state transformation in other words is notmore correct accurate or informative thandelayed transformation

Historical and current usage such as maybe ascertained also does not clearly asso-ciate ldquowidely usedrdquo names with crown taxa(Bryant 1994 Lee 1996) Regarding Mam-malia Rowe and Gauthier (1992375ndash376)stated that ldquoits universal usage by compar-ative biologists except for some paleontol-ogists has long been for the crown cladeMammals viz the last common ancestor ofMonotremata and Theria and all of its de-scendantsrdquo and that ldquocurrent debate overthe name Mammalia highlights the fact thatmany paleontologists continue to operate ina system inuenced in fundamental ways bya pre-Darwinian world viewrdquo Darwin wasfairly clear regarding his conception of thetaxonomic contentof Aves Mammalia Mar-supialia and other groups with living repre-sentatives For Darwin these taxa includedfossil stem groups In The Origin of the Species(1859268) he wrote

332 SYSTEMATIC BIOLOGY VOL 48

FIGURE 2 Crown taxa denitional types and sta-bility (a) A crown clade with resolved basal relation-ships but with unresolved outgroup relationships maybe dened with greatest stability as a node-based taxon(b) A crown clade with unresolved basal relationshipsbut with a stable extinct outgroup may be dened withgreatest stability as a stem-based taxon Crown groupsare encircled dashed lines indicate extinct taxa a dotindicates a node-based denition an arrow indicates astem-based denition

those groups which have in known geological pe-riods undergone much modication should in theolder formations make some slight approach to eachother so that the older members should differ less insome of their characters than do the existing mem-bers of the same groups

About mammals Darwin (1859522) wrote

They [marsupials] appeared in an earlier geologicalperiod and their range was formerly much more ex-tensive than at present Hence the Placentata are gen-erally supposed to have been derived from the Im-

placentata or Marsupials not however from formsclosely resembling the existing Marsupials but fromtheir early progenitors

Most extant biologists also conceive of thetaxonomic content of Mammalia in a simi-lar fashion whether they are evolutionarytaxonomists cladists ecologists or molec-ular biologists (eg Jollie 197376ndash77 Fu-tuyma 1986334 Walker 1975) Higher taxasuch as Mammalia and Aves have been asso-ciated with ldquokeyrdquo characters and their asso-ciated functions Archaeopteryx with feath-ers wings and the capacity for ight willalways be considered a ldquobirdrdquo within Avesby the majority of biologists Altering thisequation for well-established monophyletictaxa is likely to engender confusion

Stem-based denitions are less ldquoconsistentrdquothan node-based denitionsmdashSchander andThollesson (1995264) suggested that stem-based denitions ldquorefer to non-existingclades under some phylogenetic resolu-tionsrdquo They presented an example (Fig 3a)in which relocation of one taxon (taxon A)results in a denition that refers to an ldquoim-possiblerdquo clade Relocation of taxon A cre-ates this internal inconsistency because theoriginal ingroup reference taxon (taxon F)for stem-based taxon Gnow includes by def-inition the original outgroup taxon (taxonD) Schander and Thollesson (1995264) con-cluded that ldquoif a consistent system of taxon-omy is wanted stem-based denitions maybetter be avoidedrdquo

Similar internal ldquoinconsistencyrdquo how-ever can be generated for node-based taxaas well (Fig 3b) In this example relocationof one taxon (taxon A) results in a denitionthat refers to an ldquoimpossiblerdquo or internallyredundant clade Relocation of taxon A cre-ates this internal inconsistency because oneof the original reference taxa (taxon E) fornode-based taxon G now includes the otherreference taxon (taxon F) and the commonancestor Even if taxon E is regarded as a ju-nior synonym of taxon G after relocation oftaxon A the denition of taxon G must bechanged (because taxon E cannot be used asa reference taxon)

Rather than demonstrating any partic-ular structural weakness of stem-baseddenitions these examples (Fig 3) better

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 333

FIGURE 3 Problems associated with maximally in-clusive reference taxa (a) Relocation of taxon A resultsin an ldquoinconsistentrdquo stem-based denition for taxon Gbecause it denes a group that cannot exist under the al-ternative phylogenetic hypothesis (after Schander andThollesson 1995Fig 3) Schander and Thollesson de-ned stem-based taxon G as ldquoF and all taxa sharing amore recent common ancestor with it than with Drdquo andstem-based taxon F as ldquoA and all taxa sharing a morerecent common ancestor with it than with Crdquo TaxonG is ldquoinconsistentrdquo after relocation of taxon A becauseits ingroup reference taxon (taxon F) now includes bydenition the outgroup reference taxon (taxon D) Theldquoinconsistencyrdquo however disappears if a less-inclusiveingroup reference taxon had been chosen for taxon G(ie taxon A or B rather than taxon F) In this casetaxon G exists after relocation of taxon A (either as asynonym of taxon A or as a group including taxa Band C depending on whether taxon A or B was used asthe ingroup reference taxon respectively) (b) Reloca-tion of taxon A results in an ldquoinconsistentrdquo node-baseddenition for taxon G because it denes a group thatcannot existunder thealternativephylogenetichypoth-esis In this example taxon G is node-based and denedas ldquotaxon E taxon F their common ancestor and all

descendantsrdquo taxon E is dened as ldquotaxon A taxon Btheir common ancestor and alldescendantsrdquo and taxonF is node-based and dened as ldquotaxon C taxon D theircommon ancestor and all descendantsrdquo Taxon G is ldquoin-consistentrdquo after relocation of taxon A because one ofits reference taxa (taxon E) now includes by denitionthe other reference taxon (taxon F) and the commonancestor The reference taxa for a node-based taxon arepresumed to be exclusive of one another and their com-mon ancestor after relocation of taxon A these condi-tions are violated for taxon G as originally dened Theldquoinconsistencyrdquo however disappears if a less-inclusivereference taxon had originally been selected for taxon G(ie taxon A or B rather than taxon E) In this case taxonG exists after relocation of taxon A (either unchanged inits taxonomic content or as a less-inclusive taxon thatexcludes taxon B depending on whether taxon B or Awas used as an ingroup reference taxon respectively) Dots indicate node-based denitions and arrows indi-cate stem-based denitions

illustrate problems that arise as a result of se-lecting maximally inclusive reference taxamdashthe most common kind of reference taxacurrently chosen in phylogenetic denitions(see Reference Taxa below) If Schander andThollesson had chosen a more nested in-group reference taxon in their stem-baseddenition of taxon G (ie taxon A or B ratherthan taxon F Fig 3a) relocation of taxonA poses no special problems In this casetaxon G either refers to the same clade mi-nus taxon A (if taxon B is the ingroup ref-erence taxon) or becomes a potential juniorsynonym of taxon A (if taxon A is the in-group reference taxon) Similarly for node-based taxon G (Fig 3b) a more nested refer-ence taxon (taxon A or B rather than taxonE) removes any inconsistency after reloca-tion of taxon A Taxon G refers either to thesame clade (if taxon B is an ingroup refer-ence taxon) or to a less inclusive clade thatexcludes taxon B (if taxon A is an ingroupreference taxon)

Node- or stem-based denitions better re-ect original usage and contentmdashGauthier(198612 13) for example suggested that hisstem-based denition for Haeckelrsquos (1866)taxon Ornithurae (ldquoExtant birds and allother taxa such as Ichthyornis and Hes-perornithes [sic] that are closer to extantbirds than is Archaeopteryxrdquo) was ldquoin keep-ing with its original intentrdquo (Fig 4) Chiappe(1991 1996) on the other hand argued that

334 SYSTEMATIC BIOLOGY VOL 48

Ornithurae is better dened as a less inclu-sive node-based taxon with Hesperornithi-formes and Neornithes (crown-group birds)as reference taxa (Fig 4) Initially Chiappe

FIGURE 4 Cladogram of basal avians showing re-cent use of the higher taxon Ornithurae (Haeckel 1866)The dot indicates a node-based denition and the ar-row indicates a stem-based denition

(1991337) reasoned that a node-based de-nition of Ornithurae was more appropriatebecause it excluded several newly discov-ered basal avians that lacked some of thesynapomorphies in Gauthierrsquos (1986) diag-nosis of Ornithurae Later Chiappe arguedthat a node-based denition of Ornithuraeldquobetter reects the composition of the cladebefore to [sic] the application of phyloge-netic denitionsrdquo (1996205)

Neither Gauthierrsquos nor Chiappersquos deni-tions have historical precedence Haeckel(1866) coined Ornithurae (ldquobird tailrdquo) for thereduced number of vertebrae and uniquecoossication of the distal tail (as a sin-gle bone the pygostyle) that characterizesall extant birds in contrast to the condi-tion in Archaeopteryx Haeckelrsquos taxon owesits existence to the then-recent discovery ofArchaeopteryx a primitive bird with a tailcomprising over 20 vertebrae Contrary toChiappe (1996) Ornithurae predates the de-scription of the extinct short-tailed generaHesperornis (Marsh 1872a) and Ichthyor-nis (Marsh 1872b) or the taxon Hesper-ornithiformes (Furbringer 1888) Ornithu-

rae as noted by Gauthier (1986) has beensupplanted by Neornithes (Gadow 1893)a taxon referring to crown-group aviansBased on original intent an argument couldbe made that Haeckelrsquos Ornithurae shouldsupplant Neornithes on grounds of prior-ity or that it be dened explicitly as anapomorphy-based taxon based on the pres-ence of a pygostyle which is now known tobe lacking in birds other than Archaeopteryx(Forster et al 1998) In either case thedenition would circumscribe clades otherthan those identied by either Gauthier orChiappe

For most taxon names the original de-nition if explicitly stated at all is character-based or taxon-based (a list of included taxa)and lacks a relational phrase about ances-try that would specify potential membershipThe intent of the original author of a taxonto include or exclude unknown or reposi-tioned taxa usually cannot be interpretedunambiguously This problem is not widelyappreciated

APOMORPHY-BASED DEFINITIONS

An apomorphy-based denition speci-es the boundaries of a clade by identi-fying the ldquorst ancestor with a particularsynapomorphyrdquo (de Queiroz and Gauthier1990310) and encompassing all of its de-scendants An apomorphy-based taxon in-cludes all descendants whether or not thesynapomorphy is maintained Apomorphy-based denitions are subject to three prob-lems that are not relevant to node-based andstem-based denitions variation in charac-ter coding optimization ambiguity and ho-moplasy (Bryant 1994 Schander and Thol-lesson 1995 Sereno 1998) These problemshave long been associated with traditionaluse of ldquokeyrdquo characters to dene taxa Thusfar few apomorphy-based denitions havebeen erected and this denitional formshould be avoided

Character ambiguitymdashConsider an apo-morphy-based denition for Aves (includ-ing Archaeopteryx and Ornithurae) based onthe presence of ldquofeathersrdquo the ldquokey char-acterrdquo usually associated with this taxonFirst the character codingmdashldquoFeathers ab-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 335

sent (0) present (1)rdquomdashis only one of sev-eral possibilities A ldquofeatherrdquo is composedof a rachis barbs barbules and other com-ponents that may be better coded as sepa-rate characters The character on which anapomorphy-based denition is based there-fore may be recoded as two or more char-acters with independent character-state dis-tributions Indeed this is an expectation formost qualitative or quantitative charactersthat is once the transformations become bet-ter known intermediate states or multiplecharacters will emerge as recent discoveriesare beginning to reveal with regard to feath-ers (Chen et al 1998 Ji et al 1998) or withregard to the characters historically associ-ated with Mammalia (Rowe 1988 RoweandGauthier 1992 Bryant 1994 de Queiroz1994)

The apomorphy-based denition given asan example by de Queiroz and Gauthier(1990310) is subject on the same groundsto interpretational ambiguity ldquoTetrapodardquois dened as the ldquorst vertebrate to possessdigits (ie hands and feet rather than ns)and all of its descendantsrdquo What exactlyconstitutes a ldquodigitrdquo versus a ldquonrdquo or ldquopad-dlerdquo In fact this is a signicant questiongiven recent fossil discoveries (Daeschlerand Shubin 1998) Can this feature or listof features be observed in extinct formsAnd what happens if we learn that ldquodigitsrdquoevolved rst on the hands and later on thefeet These are common problems in the in-terpretation of character data which shouldnot be extended to taxonomic denitions

Optimization ambiguitymdashApomorphy-based denitions also do not specify a sin-gle ancestor when the chosen apomorphyhas an ambiguous optimization Charac-ters with ambiguous character-state opti-mization are commonplace in systematicsand arise from missing data (lack of preser-vation strong transformation) and homo-plasy (Bryant 1994 Schander and Thol-lesson 1995 Sereno 1998) There may bemany equally parsimonious positions fora particular character state on the shortestcladogram

HomoplasymdashHomoplasy may involve thestable placement of an apomorphy at morethan one node (Bryant 1994) This can create

an impasse for an apomorphy-based deni-tion because the apomorphy might identifymore than one clade Invoking time as an ar-biter (such as ldquothe rst taxon that has rdquo)invites ambiguities associated with tempo-ral origin (such as missing ancestral lin-eages)

PHYLOGENETIC DEFINITIONS TERMINOLOGY

The terminology used in the followingdiscussion of phylogenetic taxonomic def-initions is claried below and tabulated (Ta-bles 1 2) Someof these terms aregiven morespecic meaning than in the current litera-ture others are new or recently introduced(Sereno 1998)

Taxonomic denition and diagnosis are de-ned here much as they have been describedpreviously by phylogenetic systematists (deQueiroz and Gauthier 1990 1992) and theirforebears (Buck and Hull 1966 Farris 1976Ghiselin 1984) (Table 1) De Queiroz andGauthier (1992461) dened phylogeneticdenitions as ldquostatements specifying themeanings of taxon names (words) they arestated in terms of ancestryrdquo The ldquomeaningof a taxon namerdquo in phylogenetic taxonomyconcerns its taxonomic content or mem-bership as delimited by a relational state-ment Taxonomic denitions therefore areregarded here as ldquoextensionalrdquo (ostensive)statements of relationship that specify thetaxonomic content of a taxon (de Queiroz1992)

Taxonomic diagnosis on the other handinvolves the ldquointensionalrdquo characterizationof members of a clade as recognized bythe shared apomorphies present in thecommon ancestor (Buck and Hull 1966Farris 1976 Ghiselin 1984 Rowe 1987de Queiroz and Gauthier 1990 1992 deQueiroz 1992) According to de Queiroz andGauthier (1992461) diagnoses are ldquostate-ments specifying how to determine whethera given species or organism is a representa-tive of the taxon (clade) to which a partic-ular name refers they are most commonlystated in terms of charactersrdquo Strictly speak-ing shared apomorphies (derived-characterstates) rather than characters determine themembership of a particular clade or species

336 SYSTEMATIC BIOLOGY VOL 48

Taxonomic content is used here to refer toall existing and potential taxa included by thephylogenetic denition of a taxon (Table 1)The smallest number of subordinate taxathat can fulll this denition of taxonomiccontent is two a redundant taxon (ie asingle subordinate taxon) would have anidentical phylogenetic denition and thusis uninformative in phylogenetic taxonomy(Farris 1976) The taxonomic content oftaxon C therefore could be ldquotaxon A taxonB their most recent common ancestor andall descendantsrdquo if taxa A and B are denedextensionally to encompass all existing andpotential subordinate taxa currently withintaxon C

A list of included taxa is insufcient to un-ambiguously specify taxonomic content be-cause there will always be potential mem-bers of a taxon that lie outside any list ofincluded taxa whether or not the includedtaxa themselves have phylogenetic deni-tions An exhaustive list of included taxawith stem-based denitions for exampledoes not specify membership for their hy-pothetical immediate common ancestor (anode-based taxon) or more distant knownor hypothetical outgroups (a stem-basedtaxon) Two reference taxa and a relationalstatement about ancestry are necessary andsufcient to specify taxonomic content be-cause they unambiguously specify member-ship for all existing and potential membersof a taxon

A reference taxon is a species or higher-leveltaxon that is used to delimit a phylogeneticdenition (Table 1) For example ldquotaxon Ataxon B their most recent common ancestorand all descendantsrdquo uses A and B as refer-ence taxa for a node-based taxonomic deni-tion Previously cited as ldquoreference pointsrdquo

(Schander and Thollesson 1995) theseimportant components of phylogeneticdenitions have not been distinguishedterminologically

Several kinds of reference taxa can beidentied (Fig 5 Table 2) Ingroup and out-group reference taxa are located within oroutside the taxon they dene respectively(Fig 5a b) An inclusive reference taxon in-cludes several ingroup or outgroup taxaa maximally inclusive reference taxon in-cludes the largest possible ingroup or out-group reference taxon (Fig 5c) Basal andnested reference taxa are less inclusive andas their name suggests are located at thebase or nested within an ingroup or out-group taxon (Fig 5d e)

A crown taxon denotes a living speciesor a clade dened by using living refer-ence taxa whose immediate outgroup isextinct (Figs 6 7 Table 1) The extinct out-group is a necessary component of the def-inition Jefferies (1979449) coined the termldquocrown grouprdquo to replace Hennigrsquos (1969) group which Jefferies dened as ldquothe lat-est common ancestor of all living membersof group 1 plus all descendants of this an-cestor whether these are living or extinctrdquoTheexistence of an immediate outgroup thatis extinct is implied and was clearly presentin the examples given by Hennig and Jef-feries Patterson (1981207) also clearly usedcrown- and stem-groups in this mannerstating that the ldquostem-group contains fos-sil taxa that exhibit some but not all of thecharacters of the crown-grouprdquo Crown- andstem-group terminology has been used ex-clusively for higher taxa bounded by liv-ing species that have extinct outgroups (egde Queiroz and Gauthier 1992 Forey 1992McKenna and Bell 1997) Although it may

TABLE 2 Denitions for the various kinds of reference taxa

Reference taxon type Denition

Ingroup reference taxon included within a taxon by denitionOutgroup reference taxon excluded from a taxon by denitionInclusive most inclusive reference taxon among ingroup or outgroup taxaBasal reference taxon near the basal furcation of a taxonNested reference taxon distant from the basal furcation of a taxonVernacular vernacular name used as a reference taxon (eg ldquobirdsrdquo or ldquoextant birdsrdquo)

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 337

FIGURE 5 Kinds of reference taxa (a) An ingroup reference taxon is included by denition within a taxon anode-based denition is based on two ingroup reference taxa (b) An outgroup reference taxon is excluded bydenition from a taxon a stem-based denition is based on one ingroup and one outgroup reference taxon (c) Aninclusive reference taxon is the most inclusive ingroup or outgroup reference taxon possible inclusive referencetaxa B and C for taxon A allow denitional ambiguity with change in the relationship of a basal taxon (d) Abasal reference taxon is located at or near the base of an ingroup or outgroup taxon basal reference taxa allowdenitional ambiguity when their relationships change (e) A nested reference taxon is remote from the base ofan ingroup or outgroup reference taxon nested reference taxa are unaffected by changes in the relationships ofbasal taxa Reference taxa are encircled a dot indicates a node-based denition an arrow indicates a stem-baseddenition a dashed line indicates a change in position of a basal taxon

FIGURE 6 Crown stem and total taxa (a) Total group composed of a crown group plus a paraphyletic ldquostemgrouprdquo as conceived by Hennig (1969 1983) and Jefferies (1979) (b) Total taxon composed of a crown taxon andmonophyletic stem taxa as used in this article One stem taxon (left) is shown with a node-based denition andthe other (right) with a stem-based denition

338 SYSTEMATIC BIOLOGY VOL 48

FIGURE 7 Crown species or clades (enclosed) musthave an extinct (stem) taxon (dashed) as an immediateoutgroup (a) Crown species and crown clades are de-limited by the presence of an immediate outgroup thatis extinct (b) Often referred to as ldquoextantrdquo taxa crowntaxa may be composed primarily of extinct taxa ldquoLiv-ingrdquo and ldquoextinctrdquo conditions are treated as if they con-stitute plesiomorphic and apomorphic character statesrespectively of an irreversible character ldquostate of be-ingrdquo Crown clades under this conceptualization areplesiomorphy-based taxa bounded by two living taxafor which the immediate outgroup is extinct

be hypothetically true that ldquoevery Recentgroup whatever its position in thehierarchymay be expected to have had a stem-grouprdquo(Patterson 1981207) clades bounded by liv-ing taxa with immediate outgroups that arealso living have not been considered crowntaxa

Some recent denitions of crown taxahowever have set aside Hennigrsquos andJefferiesrsquo formulations as well as generalusage in the literature A less restric-tive meaning has been proposed basedsolely on reference taxa that include liv-

ing representatives De Queiroz and Gau-thier (1992469) dened crown taxa asldquoclades within which both branches ofthe basal dichotomy are represented byextant descendantsrdquo without mention ofthe condition of the immediate outgroupYet all of their examples of crown cladeshave immediate outgroups that are ex-tinct Regarding their craniate taxonomythey stated that each stem-based taxon iscomposed of a ldquocrown taxon plus all ex-tinct taxa more closely related to it thanto any other crown claderdquo (de Queirozand Gauthier 1992475) Lee (19961103)proposed that ldquoa crown clade denitionresults when both taxa implicated in anode-based denition are representedby extant formsrdquo Although the conditionof the immediate outgroup is never men-tioned Lee (19961103) stated that crowntaxon denitions apply to clades thatconsist of ldquothe most recent common an-cestor of all extant forms and all its de-scendantsrdquo (emphasis added) The pres-ence of an immediate extinct outgroup isimplied

The denition of a crown taxon presentedhere in contrast does not specify ingroupstructure (such as the presence or absenceof a basal dichotomy) or denitional type(node- or stem-based) and explicitly statesthat the immediate outgroup(s) is extinct(Fig 7a Table 1) Crown taxa so denedcan be mapped unerringly on a cladogramif (1) ldquolivingrdquo and ldquoextinctrdquo conditions aretreated as primitive and derived character-states respectively of the character ldquocurrentstate of beingrdquo and (2) if this character isregarded as irreversible (ie CaminndashSokalparsimony) The living state in other wordsis always regarded as plesiomorphic Theirreversible status of the character (state ofbeing) is important because it identiescrown groups on cladograms of living andextinct taxa that would have an ambigu-ous status if character-state reversal wereallowed (Fig 7b) This character-state con-ceptualization of the denition of a crowngroup captures the essence of crown taxawhich have always been used to refer tomaximally inclusive taxa dened by extantrepresentatives not simply any ldquoextant-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 339

boundedrdquo taxon Discovery of extinct sistertaxa provides the opportunity to recognizecrown taxa

Crown taxa are often regarded as cladesthat are bounded uniquely by living in-groups The use of living reference taxahowever is most often an arbitrary deci-sion Any extinct ingroup taxon that is mostclosely related to a living reference taxoncould be used to delimit the same clade(Fig 7b) A crown clade is dened uniquelyby living reference taxa only when such taxado not have extinct sister taxa (Fig 7b) afact that is not widely appreciated Crowntaxa in other words do not have any spe-cial biological signicance beyond signal-ing that particular reference taxa have sur-vived to an arbitrary datum (the Recent)Informal reference to crown taxa as ldquolivingrdquoor ldquoextantrdquo taxa therefore can be mislead-ing because a crown taxon may be com-posed predominantly of extinct species asis the case with the crown taxon Crocodylia(Fig 7b Brochu 1997) ldquoExtant crocodil-iansrdquo if taken literally refers to a smallparaphyletic subset of Crocodylia Explicitreference to ldquocrown crocodiliansrdquo or sim-ply ldquoCrocodyliardquo is preferable to use of theterms ldquolivingrdquo or ldquoextantrdquo

A living taxon is also essential to delimita maximally inclusive clade that is entirelyextinctmdasha stem taxon as here dened Astem taxon denotes an extinct species orclade the immediate outgroup of which in-cludes at least one living member (Fig 6b)Many extinct taxa are not stem taxa becausetheir immediate outgroup is itself extinctSuch extinct taxa are members of more in-clusive stem taxa Stem taxa can have eithernode- or stem-based denitions (Fig 6b)Stem taxa and their associated less inclusivecrown taxon make up the taxonomic contentof their respective total taxon

Stem taxa as here dened should notbe confused with ldquostem grouprdquo (Fig 6a)which was rst used in phylogenetic sys-tematics by Hennig (ldquostammgrupperdquo 19691983) Hennigrsquos (198315) vague formulationof the concept specied all extinct speciesldquowhich can be shown probably to belong toa particular monophyletic group of the phy-logenetic system but which are probably no

closer [sic] related to one subgroup amongrecent animals than to anotherrdquo (translatedfrom the German Ax 1987224) Jefferies(1979) claried the term to include all ex-tinct outgroups of a crown taxon (Fig 6a)Other authors have developed more convo-luted denitions that approximate the sameend Wiley (1981217) for example dened aldquostem grouprdquo as ldquoall fossil species of a groupmore primitive than the most primitive Re-cent species of the same grouprdquo Theseldquostem groupsrdquo are paraphyletic and oftencircumscribe the same ldquoancestral groupsrdquoand ldquoevolutionary gradesrdquo that have mud-died phylogenetic thinking since DarwinOnly a few cladists continue to endorse suchconcepts (Smith 1994) Ax (1987224) pro-posed a cumbersome solution to the prob-lem of paraphyletic ldquostem groupsrdquo by divid-ing them into ancestral ldquostem lineagesrdquo andldquorepresentatives of the stem lineagerdquo Thelatter are monophyletic taxa most simply re-ferred to as stem taxa

A total taxon is regarded here as a cladecomposed of a crown taxon plus all ex-tinct outgroups more closely related to itthan to another crown taxon (Fig 3 Ta-ble 1) Hennigrsquos (1969) and Jefferiesrsquo (1979)formulations of the concept are wordy butsimilar in effect because they restricted in-clusion of extinct outgroups to those mostclosely related to a particular crown taxonDe Queiroz and Gauthier (1992470) deneda total clade as a ldquomore inclusive clade con-sisting of thecrown clade plus its extinctout-groupsrdquo This denition is incomplete be-cause it does not specify which extinct out-groups are included A total taxon explicitlyincludes only those extinct outgroups mostclosely related to a particular crown taxon

Node-based and stem-based denitions areconstructed by (1) identifying reference taxaand (2) attaching to them a relational phrasethat either includes their most recent com-mon ancestor and all descendants (node-based) or limits inclusion to all descen-dants closer to one or more of the referencetaxa (stem-based) (de Queiroz and Gauthier1992) (Fig 1 Table 1) In node-based deni-tions the phrase ldquoleast-inclusive claderdquo canreplace ldquocommon ancestor and all descen-dantsrdquo for cladists sensitive to the allusion to

340 SYSTEMATIC BIOLOGY VOL 48

ancestors (Lee 1998) Listing included taxawithout a relational phrase about ancestryor simply designating taxa as node- or stem-based without reference taxa does not con-stitutea phylogenetic denition because theboundaries of the taxon in question are notspecied (de Queiroz and Gauthier 1990Bryant 1996)

In the current literature there exists someconfusion over what is necessary and suf-cient for node- and stem-based denitionsBryant (1996185) suggested that a more in-clusive taxon be inserted within a stem-based denition such as ldquoall members oftaxon A thataremoreclosely related to taxonB than to taxon Crdquo Including taxon A how-ever is not necessary and may lead to inter-nal conicts within the denition (eg whentaxon A is included within taxon B or C onan alternativephylogeny) The goalof a phy-logenetic denition should be to unambigu-ously identify a specic common ancestorgiven any arrangement of taxa rather thanto identify such an ancestor only on a subsetof phylogenetic hypotheses

Meng et al (1994) and Wyss and Meng(1996) introduced a ldquostem-modied node-based denitionrdquo to effectively unite crown-group rodents whose basal relationshipsare uncertain Their preferred denition ofRodentia (ldquothe clade stemming from themost recent common ancestor of Mus andall Recent mammals more closely relatedto Mus than to Lagomorpha or membersof any other eutherian lsquoorderrdquorsquo Wyss andMeng 1996562) is better interpreted as aspecial stem-based rather than node-baseddenition because it uses only one ingroupreference taxon (Mus) and one or more out-group reference taxa (eg Lagomorpha)Their denition merely adds a phrase (ldquotheclade stemming from the most recent com-mon ancestorrdquo) to an otherwise typicalstem-based denition based on living ref-erence taxa That phrase ensures inclusionof all extinct taxa within the crown cladeBryantrsquos (1996185) reformulation of thisdenition on the other hand is clearly anode-based denition because he speci-ed only ingroup reference taxa (ldquothe mostrecent common ancestor of the species re-ferred to Rodentia by Wilson and Reeder

(1993) and all of its descendantsrdquo) These ex-amples underscore the need for claricationof terms (Tables 1 2)

A node-stem triplet (NST) is a denitionalconguration involving three taxa a node-based taxon composed of two stem-basedsubordinate taxa (Sereno 1997 1998 Fig 8)

Taxon A Taxa b and c their most recentcommon ancestor and all its descendants

Taxon B All taxa closer to taxon b than totaxon c

Taxon C All taxa closer to taxon c than totaxon b

FIGURE 8 Denitional triumvirate termed a node-stem triplet composed of a node-based taxon (A) andtwo subordinate stem-based taxa (B and C) which in-corporates added or repositioned taxa (dashed lineswith daggers) without changing the relative taxonomiccontent of taxon A B or C (as expressed by the equiva-lence statementA = B +C) A dot indicates a node-baseddenition an arrow indicates a stem-based denition

Taxa b and c (not shown in Fig 8) constitutereference taxa within taxon B and C respec-tively Although not essential the same ref-erence taxa (b c) may be used for the trio oftaxa in a NST which then specify complemen-tary denitions The pair of stem-based sis-ter taxa in addition may have reciprocal def-initions that utilize the same reference taxain antipodal positions The word ldquotaxardquo inthe stem-based denitions listed above mayoptionally be replaced by reference toa moreinclusive taxon (Bryant 1996) A stem-baseddenition for Saurischia for example mightread ldquoAll dinosaurs more closely related

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 341

to Neornithes than to Triceratopsrdquo (Sereno1998)

Taxonomic equivalence is a statement (orequation) in which one taxon is shownto be equivalent in current and poten-tial taxonomic content to subordinate taxaplus their common ancestor (Table 1)The NST described above is composedof a node-based taxon (A) and twosubordinate stem-based taxa (B C) andcan be written as the equivalence state-ment ldquoA = B + Crdquo because taxon Aby denition is equivalent to taxa Band C plus their most recent commonancestor

PHYLOGENETIC DEFINITIONS RATIONALE

The rst step toward a phylogenetic tax-onomy involved the distinction betweendenition and diagnosis (Buck and Hull1966 Grifths 1973 1974 Farris 1976 Ghis-elin 1984 Rowe 1987) A second step in-volved the formulation of node- and stem-based denitions by use of reference taxaand relational statements about ancestry (deQueiroz and Gauthier 1992) A third stepoutlined here involves the specication ofguidelines for selecting effective referencetaxa and criteria for positioning node- andstem-based denitions

Phylogenetic rearrangements and introduc-tion of new taxa are the principal factorsthat destabilize taxonomic content Stabil-ity of taxonomic content is enhanced bychoosing (1) effective reference taxa and(2) an effective conguration of phyloge-netic denitions

Reference TaxaPositionmdashGiven a particular phyloge-

netic hypothesis the taxonomic content of aphylogenetic denition is altered only if thereference taxa specify an alternative com-mon ancestor Given a basal dichotomy withone reference taxon on each side the iden-tication of the immediate common ances-tor is not affected by relocation of a ref-erence taxon on its side of the basal di-chotomy (Fig 9 move 1) Relocation ofa reference taxon to the opposing side orto a location outside the basal dichotomyidenties a different common ancestor

(Fig 9 moves 2 3) For a taxon with a basaldichotomy an alternative common ances-tor can be identied only if one referencetaxon is relocated to the opposing side oroutside the basal dichotomy Are some ref-erence taxa more likely to be relocated in thismanner than others

FIGURE 9 Three possible outcomes from relocationof a reference taxon 1 = relocation on the same side ofa basal dichotomy has no effect 2 = relocation to theopposite side of the basal dichotomy identies a lessinclusive common ancestor 3 = relocation outside thebasal dichotomy identies a more inclusive commonancestor Reference taxa are encircled and the originalbasal dichotomy is shown with heavy lines

First a taxon positioned near the basaldichotomy of a cladogram is always eas-ier to relocate to the opposing side (ierequires fewer additional steps) than is anested taxon if all other factors are regardedas equal among ingroup taxa (eg complete-ness missing data and homoplasy) Unlikea basal taxon (Fig 5d) a nested taxon sharessynapomorphies that increase its phyloge-netic (patristic) distance from the commonancestor (Fig 5e) Thus stability of taxo-nomic content is enhanced in phylogeneticdenitions if nested reference taxa are cho-sen that are located at some distance (atleast several nodes away) from the basaldichotomy This distance is easy to assessquantitatively for various nested taxa on agiven phylogeny

Second maximally inclusive referencetaxa (Fig 5c) have greater potential to cre-ate unnecessary taxonomic redundancy andinternal inconsistencies after relocation oftaxa (Fig 3) This is true for both node-

342 SYSTEMATIC BIOLOGY VOL 48

and stem-based taxa In the current litera-ture maximally inclusive reference taxa areoften used in phylogenetic denitions Intheir tetrapod classication for example deQueiroz and Gauthier (1992475) used suc-cessive maximally inclusive reference taxafor all denitionsmdasha pattern here termed re-cursive referencing (Fig 10a) Recursive refer-encing can have undesirable consequencesbecause it shifts the burden of a more precisedenition to a slightly less inclusive highertaxon This higher taxon may not be de-ned or it too may be affected by an alter-native conguration of basal relationshipsSalgado et al (1997) for example choseProsauropoda and Sauropoda as maximallyinclusive reference taxa for Sauropodomor-pha If a basal sauropodomorph is reposi-tioned the denitions of Prosauropoda andSauropoda must be determined to assess po-tential taxonomic consequences HoweverSalgado et al (1997) used Prosauropodaas a terminal taxon without denitionSauropoda was dened as a node-basedtaxon on the basis of two maximally in-clusive reference taxa Vulcanodon and Eu-sauropoda (all other sauropods) Thusif the incompletely known basal sauro-pod Vulcanodon is reinterpreted as be-ing more closely related to prosauropodsSauropoda and Sauropodomorpha wouldbecome synonyms

If on the other hand Sauropodomorphais dened on the basis of the nested ref-erence taxamdashthe prosauropod Plateosaurusand sauropod Saltasaurus (Sereno 1998)mdasha change in the phylogenetic positionof any other prosauropod or sauropodhas no effect on the taxonomic con-tents of Sauropodomorpha or SauropodaSauropodomorpha can be dened as ldquoPla-teosaurus Saltasaurus their most recentcommon ancestor and all descendantsrdquoand Sauropoda can be dened as ldquoallsauropodomorphs more closely related toSaltasaurus than to Plateosaurusrdquo The phylo-genetic position of Vulcanodon has no effecton the identication of respective commonancestors

CompletenessmdashPoorly known taxa aremore likely to be phylogenetically unsta-ble than well-known taxa (Wilkinson 1995)

FIGURE 10 Referencing styles (a) Recursive refer-encing selects a succession of the most inclusive ref-erence taxa (b) Nested referencing selects that sameremote reference taxon Reference taxa are encircled

Taxa with numerous missing entries in aphylogenetic analysis usually reduce phylo-genetic resolution by generating numerousequally parsimonious trees Well-known ref-erence taxa even if they are somewhat lessnested are preferable to poorly known taxa(Fig 11) A single poorly known basal taxonis least stable and therefore least desirableas a reference taxon

Use with Recent or extinct ltersmdashldquoRecentrdquo(ldquoextantrdquo or ldquolivingrdquo) or ldquoextinctrdquo may beused in phylogenetic denitions of crownand stem taxa respectively to help to stabi-lize taxonomic content in the face of poorlyresolved basal relationships For a crowntaxon with a stem-based denition a Recentlter can restrict the boundaries of a cladeto living taxa (Meng et al 1994 Wyss andMeng 1996) Rodentia for example could

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 343

be dened as ldquothe least inclusive clade com-posed of Recent mammals more closely re-lated to Mus than to Lepusrdquo For stem taxain contrast a stem-based denition using anldquoextinctrdquo lter can restrict inclusion to ex-tinct taxa when outgroup relationships areuncertain Moas for example could be de-ned as ldquoall extinct species more closelyrelated to Dinornis than to Struthiordquo theostrich

NumbermdashThe number of reference taxathat have been used in phylogenetic deni-tions has varied from two to innity Gau-thier (198614) for example dened Aves asa node-based taxon including ldquoall descen-

FIGURE 11 Nestedness and completeness in the se-lection of reference taxa A nested well-known taxonis more likely to constitute a stable reference taxon (en-circled)

dants of the most recent common ancestorof Ratitae Tinami and Neognathaerdquo Threereference taxa rather than two were chosenpresumably because of unstated uncertaintyin the phylogenetic relationships amongthese three avian subgroups De Queirozand Gauthier (1990310 1992461) suggestedthat phylogenetic denitions should utilizeonly two reference taxamdashbut then erecteddenitions for total taxa that used as refer-ence taxa one crown clade plus ldquoall extinct

taxa more closely related to it than to anyother crown claderdquo (de Queiroz and Gau-thier 1992475)

Listing more than a pair of referencetaxa does not seem to compromise a node-based phylogenetic denition in any dis-cernible way Node-based denitions areconstructed with what may be termed in-ternal inclusion A contradictory relation-ship between multiple ingroup referencetaxa for a node-based denition is im-possible to construct with internal inclu-sion because a clade always exists thatincludes all ingroup reference taxa Multi-ple ingroup reference taxa effectively stabi-lize the taxonomic content of a node-basedtaxon if basal relationships are uncertain(Fig 12a)

For stem-based denitions multiple out-group reference taxa effectively stabilize thetaxonomic content of a taxon if outgroup re-lationships are uncertain (Fig 12b) Multi-ple ingroup reference taxa for stem-baseddenitions can create a contradictory sit-uation if one (or more) ingroup referencetaxa are repositioned closer to the outgroupreference taxon (Fig 12c) or to a positionoutside the clade as originally dened Inthis case the external inclusion fundamen-tal to a stem-based denition creates a con-tradictory relationship among the originalingroup reference taxa which are no longermore closely related to each other than to theoutgroup reference taxon (Fig 12c) In theexample shown here relocation of ingroupreference taxon D creates a contradictory re-lationship with a second ingroup referencetaxon E if taxon A has the stem-based def-inition of ldquoall descendants more closely re-lated to taxon D and E than to taxon Brdquo Aninclusive ingroup reference taxon howevercan manifest the same problems in this cir-cumstance (Fig 12c) If taxon A is denedas ldquoall descendants more closely related totaxon C than to taxon Brdquo and if taxon C isdened on the basis of taxon D a contra-dictory denition results with relocation oftaxon D as shown taxon A now subsumesits original ingroup reference taxon C Suchcontradictory denitions can be avoided ifa nested ingroup reference taxon (or nestedtaxa) are selected such as taxon E its un-

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 2: Sereno, 1999b

330 SYSTEMATIC BIOLOGY VOL 48

FIGURE 1 Node-based and stem-based phyloge-netic denitions which usually have been showngraphically by encircling portions of a cladogram (deQueiroz and Gauthier 1992) are indicated here by a dot(node-based) or arrow (stem-based)

are represented by extant formsrdquo And Wyssand Meng (1996559) stated that the deni-tions of crown taxa ldquoare thus node-based intheir formulationrdquo without presenting anyreasoning for such a conclusion

Any group of extant species may justas well be united by a stem-based deni-tion The crown taxon Amniota for examplecould have a stem-based denition such asldquoany Recent tetrapod more closely related toMammalia than to Anura and all extinct de-scendants of their most recent common an-cestorrdquo A stem-based denition for a crowntaxon in fact may be preferable if the basalrelationships among extant taxa within acrown clade are poorly established as seemsto be the case with Amniota (see also Ro-dentia Wyss and Meng 1996) (Fig 2b) Allmembers of the crown taxon will be in-cluded even if basal ingroup relationshipsare poorly resolved because a stem-baseddenition is based on reference to an out-group taxon (or taxa)

A single suboscine bird for examplewould sufce as an outgroup referencetaxon for a stem-based denition of thecrown taxon Oscines A node-based de-nition of the same crown taxon would re-quire listing many oscine subgroups to en-sure inclusion of all living species currentlyregarded as oscines On the other hand ifthe basal dichotomy within a crown taxonis well established and outgroup relation-ships are uncertain (Fig 2a) a node-based

denition will more effectively stabilize tax-onomic content Thus there is no currentjustication for exclusive use of node-baseddenitions to delimit crown taxa Node-based crown taxa in conjunction with stem-based total taxa however yield a local def-initional conguration that can provide jus-tication for preferential use of node-basedcrown taxa (see Node-Stem Triplet below)

Node-based crown taxa are more stable infor-mative and accurate than taxa dened by ex-tinct species and have been (or should be) as-sociated with ldquowidely usedrdquo namesmdashGauthier(198612) stated that restricting ldquoAvesrdquo toldquoliving taxardquo maximizes ldquostability and phy-logenetic informativenessrdquo and de Queirozand Gauthier (1992468) suggested that ldquobi-ologists commonly use the widely knownname Aves when making generalizationsthat apply to extant birds alonerdquo With re-gard to the taxon Mammalia Rowe andGauthier (1992372) observed that ldquono onein several centuries has mistakenly assigneda Recent mammal to some other taxonrdquoMcKenna and Bell (199732) remarked thatldquoa crown group is likely to remain relativelystablerdquo

Despite these and similar statements inthe literature node- or stem-based crowntaxa are not demonstrably more stable interms of meaning or content Any node- orstem-based denition that specically iden-ties a most recent common ancestor clearlyspecies the boundaries of a clade regard-less of the living or extinct status of the ref-erence taxa or the particular phylogeny towhich the denition is applied (Lee 1996)

In terms of taxonomic content reden-ing Mammalia as a crown taxon does notimmediately clarify which extinct taxa willbe included or excluded Mammalia asa crown taxon appears more stable onlyif extinct taxa are ignored and attentionis focused on the great phylogenetic dis-tance to their nearest living sister groupEven so inclusion of some living mam-mals may be uncertain Rowe and Gau-thierrsquos (1992375) preferred node-based def-inition of Mammalia (ldquothe last commonancestor of Monotremata and Theriardquo seealso McKenna and Bell 19973235) maybe synonymous with Theria if monotremes

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 331

TABLE 1 Denitions for terms used in this paper

Term Denition

Taxonomic denition relational statement specifying the taxonomic content of a taxonTaxonomic diagnosis descriptive statement specifying the apomorphies (derived character states)

that serve to identify members of a taxonTaxonomic content existing and potential taxa or individuals that by denition are included

within a taxonReference taxon taxon serving as a reference in a phylogenetic denitionCrown taxon a living species or a clade that can be dened by living species whose

immediate outgroup is extinctTotal taxon clade composed of a crown taxon plus all extinct outgroups more closely

related to it than to another crown cladeStem taxon an extinct species or clade for which immediate outgroup includes at least

one living memberNode-based denition statement specifying a clade composed of the most recent common ancestor

of two or more reference taxa and all descendantsStem-based denition statement specifying a clade composed of all descendants more closely

related to one reference taxon (or taxa) than another (or others)Complementary denitions phylogenetic denitions with the same reference taxa (as in

a node-stem triplet)Reciprocal denitions phylogenetic denitions with the same reference taxa but in reverse order

(as in opposing stem-based taxa of a node-stem triplet)Node-stem triplet trio of taxa with complementary denitions consisting of a node-based

taxon and two subordinate stem-based taxaTaxonomic equivalence statement of equivalence (or equation) in which the taxonomic content of

one taxon equals that of subordinate taxa plus their common ancestor(eg A = B + C as in a node-stem triplet)

prove to be more closely related to marsu-pials as indicated by some recent molecu-lar evidence (Penny and Hasegawa 1997)Stability of taxonomic content is not re-lated to the living or extinct status of ref-erence taxa in a phylogenetic denitionbut rather to the particular choice of ref-erence taxa For many node-based deni-tions of crown taxa stability is also re-lated to the certainty with which a speciedbasal dichotomy has been correctly ascer-tained

That ldquowidely usedrdquo names such as Avesshould be assigned to crown taxa has beenjustied by arguing that (1) neontologistsshould report their observations accuratelyreferring only to crown clades rather thancrown clades plus some of their extinct out-groups and that (2) doing so follows long-standing taxonomic practice (de Queirozand Gauthier 1992) The condition in ex-tinct outgroups for characters that are notpreserved however is ambiguous To sup-pose that Archaeopteryx lacked an apomor-phic molecular sequence found in crown-group birds (Neornithes) is no more justied

than tosuppose that the ancientbird had thissequence (Lee 1996) Accelerated character-state transformation in other words is notmore correct accurate or informative thandelayed transformation

Historical and current usage such as maybe ascertained also does not clearly asso-ciate ldquowidely usedrdquo names with crown taxa(Bryant 1994 Lee 1996) Regarding Mam-malia Rowe and Gauthier (1992375ndash376)stated that ldquoits universal usage by compar-ative biologists except for some paleontol-ogists has long been for the crown cladeMammals viz the last common ancestor ofMonotremata and Theria and all of its de-scendantsrdquo and that ldquocurrent debate overthe name Mammalia highlights the fact thatmany paleontologists continue to operate ina system inuenced in fundamental ways bya pre-Darwinian world viewrdquo Darwin wasfairly clear regarding his conception of thetaxonomic contentof Aves Mammalia Mar-supialia and other groups with living repre-sentatives For Darwin these taxa includedfossil stem groups In The Origin of the Species(1859268) he wrote

332 SYSTEMATIC BIOLOGY VOL 48

FIGURE 2 Crown taxa denitional types and sta-bility (a) A crown clade with resolved basal relation-ships but with unresolved outgroup relationships maybe dened with greatest stability as a node-based taxon(b) A crown clade with unresolved basal relationshipsbut with a stable extinct outgroup may be dened withgreatest stability as a stem-based taxon Crown groupsare encircled dashed lines indicate extinct taxa a dotindicates a node-based denition an arrow indicates astem-based denition

those groups which have in known geological pe-riods undergone much modication should in theolder formations make some slight approach to eachother so that the older members should differ less insome of their characters than do the existing mem-bers of the same groups

About mammals Darwin (1859522) wrote

They [marsupials] appeared in an earlier geologicalperiod and their range was formerly much more ex-tensive than at present Hence the Placentata are gen-erally supposed to have been derived from the Im-

placentata or Marsupials not however from formsclosely resembling the existing Marsupials but fromtheir early progenitors

Most extant biologists also conceive of thetaxonomic content of Mammalia in a simi-lar fashion whether they are evolutionarytaxonomists cladists ecologists or molec-ular biologists (eg Jollie 197376ndash77 Fu-tuyma 1986334 Walker 1975) Higher taxasuch as Mammalia and Aves have been asso-ciated with ldquokeyrdquo characters and their asso-ciated functions Archaeopteryx with feath-ers wings and the capacity for ight willalways be considered a ldquobirdrdquo within Avesby the majority of biologists Altering thisequation for well-established monophyletictaxa is likely to engender confusion

Stem-based denitions are less ldquoconsistentrdquothan node-based denitionsmdashSchander andThollesson (1995264) suggested that stem-based denitions ldquorefer to non-existingclades under some phylogenetic resolu-tionsrdquo They presented an example (Fig 3a)in which relocation of one taxon (taxon A)results in a denition that refers to an ldquoim-possiblerdquo clade Relocation of taxon A cre-ates this internal inconsistency because theoriginal ingroup reference taxon (taxon F)for stem-based taxon Gnow includes by def-inition the original outgroup taxon (taxonD) Schander and Thollesson (1995264) con-cluded that ldquoif a consistent system of taxon-omy is wanted stem-based denitions maybetter be avoidedrdquo

Similar internal ldquoinconsistencyrdquo how-ever can be generated for node-based taxaas well (Fig 3b) In this example relocationof one taxon (taxon A) results in a denitionthat refers to an ldquoimpossiblerdquo or internallyredundant clade Relocation of taxon A cre-ates this internal inconsistency because oneof the original reference taxa (taxon E) fornode-based taxon G now includes the otherreference taxon (taxon F) and the commonancestor Even if taxon E is regarded as a ju-nior synonym of taxon G after relocation oftaxon A the denition of taxon G must bechanged (because taxon E cannot be used asa reference taxon)

Rather than demonstrating any partic-ular structural weakness of stem-baseddenitions these examples (Fig 3) better

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 333

FIGURE 3 Problems associated with maximally in-clusive reference taxa (a) Relocation of taxon A resultsin an ldquoinconsistentrdquo stem-based denition for taxon Gbecause it denes a group that cannot exist under the al-ternative phylogenetic hypothesis (after Schander andThollesson 1995Fig 3) Schander and Thollesson de-ned stem-based taxon G as ldquoF and all taxa sharing amore recent common ancestor with it than with Drdquo andstem-based taxon F as ldquoA and all taxa sharing a morerecent common ancestor with it than with Crdquo TaxonG is ldquoinconsistentrdquo after relocation of taxon A becauseits ingroup reference taxon (taxon F) now includes bydenition the outgroup reference taxon (taxon D) Theldquoinconsistencyrdquo however disappears if a less-inclusiveingroup reference taxon had been chosen for taxon G(ie taxon A or B rather than taxon F) In this casetaxon G exists after relocation of taxon A (either as asynonym of taxon A or as a group including taxa Band C depending on whether taxon A or B was used asthe ingroup reference taxon respectively) (b) Reloca-tion of taxon A results in an ldquoinconsistentrdquo node-baseddenition for taxon G because it denes a group thatcannot existunder thealternativephylogenetichypoth-esis In this example taxon G is node-based and denedas ldquotaxon E taxon F their common ancestor and all

descendantsrdquo taxon E is dened as ldquotaxon A taxon Btheir common ancestor and alldescendantsrdquo and taxonF is node-based and dened as ldquotaxon C taxon D theircommon ancestor and all descendantsrdquo Taxon G is ldquoin-consistentrdquo after relocation of taxon A because one ofits reference taxa (taxon E) now includes by denitionthe other reference taxon (taxon F) and the commonancestor The reference taxa for a node-based taxon arepresumed to be exclusive of one another and their com-mon ancestor after relocation of taxon A these condi-tions are violated for taxon G as originally dened Theldquoinconsistencyrdquo however disappears if a less-inclusivereference taxon had originally been selected for taxon G(ie taxon A or B rather than taxon E) In this case taxonG exists after relocation of taxon A (either unchanged inits taxonomic content or as a less-inclusive taxon thatexcludes taxon B depending on whether taxon B or Awas used as an ingroup reference taxon respectively) Dots indicate node-based denitions and arrows indi-cate stem-based denitions

illustrate problems that arise as a result of se-lecting maximally inclusive reference taxamdashthe most common kind of reference taxacurrently chosen in phylogenetic denitions(see Reference Taxa below) If Schander andThollesson had chosen a more nested in-group reference taxon in their stem-baseddenition of taxon G (ie taxon A or B ratherthan taxon F Fig 3a) relocation of taxonA poses no special problems In this casetaxon G either refers to the same clade mi-nus taxon A (if taxon B is the ingroup ref-erence taxon) or becomes a potential juniorsynonym of taxon A (if taxon A is the in-group reference taxon) Similarly for node-based taxon G (Fig 3b) a more nested refer-ence taxon (taxon A or B rather than taxonE) removes any inconsistency after reloca-tion of taxon A Taxon G refers either to thesame clade (if taxon B is an ingroup refer-ence taxon) or to a less inclusive clade thatexcludes taxon B (if taxon A is an ingroupreference taxon)

Node- or stem-based denitions better re-ect original usage and contentmdashGauthier(198612 13) for example suggested that hisstem-based denition for Haeckelrsquos (1866)taxon Ornithurae (ldquoExtant birds and allother taxa such as Ichthyornis and Hes-perornithes [sic] that are closer to extantbirds than is Archaeopteryxrdquo) was ldquoin keep-ing with its original intentrdquo (Fig 4) Chiappe(1991 1996) on the other hand argued that

334 SYSTEMATIC BIOLOGY VOL 48

Ornithurae is better dened as a less inclu-sive node-based taxon with Hesperornithi-formes and Neornithes (crown-group birds)as reference taxa (Fig 4) Initially Chiappe

FIGURE 4 Cladogram of basal avians showing re-cent use of the higher taxon Ornithurae (Haeckel 1866)The dot indicates a node-based denition and the ar-row indicates a stem-based denition

(1991337) reasoned that a node-based de-nition of Ornithurae was more appropriatebecause it excluded several newly discov-ered basal avians that lacked some of thesynapomorphies in Gauthierrsquos (1986) diag-nosis of Ornithurae Later Chiappe arguedthat a node-based denition of Ornithuraeldquobetter reects the composition of the cladebefore to [sic] the application of phyloge-netic denitionsrdquo (1996205)

Neither Gauthierrsquos nor Chiappersquos deni-tions have historical precedence Haeckel(1866) coined Ornithurae (ldquobird tailrdquo) for thereduced number of vertebrae and uniquecoossication of the distal tail (as a sin-gle bone the pygostyle) that characterizesall extant birds in contrast to the condi-tion in Archaeopteryx Haeckelrsquos taxon owesits existence to the then-recent discovery ofArchaeopteryx a primitive bird with a tailcomprising over 20 vertebrae Contrary toChiappe (1996) Ornithurae predates the de-scription of the extinct short-tailed generaHesperornis (Marsh 1872a) and Ichthyor-nis (Marsh 1872b) or the taxon Hesper-ornithiformes (Furbringer 1888) Ornithu-

rae as noted by Gauthier (1986) has beensupplanted by Neornithes (Gadow 1893)a taxon referring to crown-group aviansBased on original intent an argument couldbe made that Haeckelrsquos Ornithurae shouldsupplant Neornithes on grounds of prior-ity or that it be dened explicitly as anapomorphy-based taxon based on the pres-ence of a pygostyle which is now known tobe lacking in birds other than Archaeopteryx(Forster et al 1998) In either case thedenition would circumscribe clades otherthan those identied by either Gauthier orChiappe

For most taxon names the original de-nition if explicitly stated at all is character-based or taxon-based (a list of included taxa)and lacks a relational phrase about ances-try that would specify potential membershipThe intent of the original author of a taxonto include or exclude unknown or reposi-tioned taxa usually cannot be interpretedunambiguously This problem is not widelyappreciated

APOMORPHY-BASED DEFINITIONS

An apomorphy-based denition speci-es the boundaries of a clade by identi-fying the ldquorst ancestor with a particularsynapomorphyrdquo (de Queiroz and Gauthier1990310) and encompassing all of its de-scendants An apomorphy-based taxon in-cludes all descendants whether or not thesynapomorphy is maintained Apomorphy-based denitions are subject to three prob-lems that are not relevant to node-based andstem-based denitions variation in charac-ter coding optimization ambiguity and ho-moplasy (Bryant 1994 Schander and Thol-lesson 1995 Sereno 1998) These problemshave long been associated with traditionaluse of ldquokeyrdquo characters to dene taxa Thusfar few apomorphy-based denitions havebeen erected and this denitional formshould be avoided

Character ambiguitymdashConsider an apo-morphy-based denition for Aves (includ-ing Archaeopteryx and Ornithurae) based onthe presence of ldquofeathersrdquo the ldquokey char-acterrdquo usually associated with this taxonFirst the character codingmdashldquoFeathers ab-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 335

sent (0) present (1)rdquomdashis only one of sev-eral possibilities A ldquofeatherrdquo is composedof a rachis barbs barbules and other com-ponents that may be better coded as sepa-rate characters The character on which anapomorphy-based denition is based there-fore may be recoded as two or more char-acters with independent character-state dis-tributions Indeed this is an expectation formost qualitative or quantitative charactersthat is once the transformations become bet-ter known intermediate states or multiplecharacters will emerge as recent discoveriesare beginning to reveal with regard to feath-ers (Chen et al 1998 Ji et al 1998) or withregard to the characters historically associ-ated with Mammalia (Rowe 1988 RoweandGauthier 1992 Bryant 1994 de Queiroz1994)

The apomorphy-based denition given asan example by de Queiroz and Gauthier(1990310) is subject on the same groundsto interpretational ambiguity ldquoTetrapodardquois dened as the ldquorst vertebrate to possessdigits (ie hands and feet rather than ns)and all of its descendantsrdquo What exactlyconstitutes a ldquodigitrdquo versus a ldquonrdquo or ldquopad-dlerdquo In fact this is a signicant questiongiven recent fossil discoveries (Daeschlerand Shubin 1998) Can this feature or listof features be observed in extinct formsAnd what happens if we learn that ldquodigitsrdquoevolved rst on the hands and later on thefeet These are common problems in the in-terpretation of character data which shouldnot be extended to taxonomic denitions

Optimization ambiguitymdashApomorphy-based denitions also do not specify a sin-gle ancestor when the chosen apomorphyhas an ambiguous optimization Charac-ters with ambiguous character-state opti-mization are commonplace in systematicsand arise from missing data (lack of preser-vation strong transformation) and homo-plasy (Bryant 1994 Schander and Thol-lesson 1995 Sereno 1998) There may bemany equally parsimonious positions fora particular character state on the shortestcladogram

HomoplasymdashHomoplasy may involve thestable placement of an apomorphy at morethan one node (Bryant 1994) This can create

an impasse for an apomorphy-based deni-tion because the apomorphy might identifymore than one clade Invoking time as an ar-biter (such as ldquothe rst taxon that has rdquo)invites ambiguities associated with tempo-ral origin (such as missing ancestral lin-eages)

PHYLOGENETIC DEFINITIONS TERMINOLOGY

The terminology used in the followingdiscussion of phylogenetic taxonomic def-initions is claried below and tabulated (Ta-bles 1 2) Someof these terms aregiven morespecic meaning than in the current litera-ture others are new or recently introduced(Sereno 1998)

Taxonomic denition and diagnosis are de-ned here much as they have been describedpreviously by phylogenetic systematists (deQueiroz and Gauthier 1990 1992) and theirforebears (Buck and Hull 1966 Farris 1976Ghiselin 1984) (Table 1) De Queiroz andGauthier (1992461) dened phylogeneticdenitions as ldquostatements specifying themeanings of taxon names (words) they arestated in terms of ancestryrdquo The ldquomeaningof a taxon namerdquo in phylogenetic taxonomyconcerns its taxonomic content or mem-bership as delimited by a relational state-ment Taxonomic denitions therefore areregarded here as ldquoextensionalrdquo (ostensive)statements of relationship that specify thetaxonomic content of a taxon (de Queiroz1992)

Taxonomic diagnosis on the other handinvolves the ldquointensionalrdquo characterizationof members of a clade as recognized bythe shared apomorphies present in thecommon ancestor (Buck and Hull 1966Farris 1976 Ghiselin 1984 Rowe 1987de Queiroz and Gauthier 1990 1992 deQueiroz 1992) According to de Queiroz andGauthier (1992461) diagnoses are ldquostate-ments specifying how to determine whethera given species or organism is a representa-tive of the taxon (clade) to which a partic-ular name refers they are most commonlystated in terms of charactersrdquo Strictly speak-ing shared apomorphies (derived-characterstates) rather than characters determine themembership of a particular clade or species

336 SYSTEMATIC BIOLOGY VOL 48

Taxonomic content is used here to refer toall existing and potential taxa included by thephylogenetic denition of a taxon (Table 1)The smallest number of subordinate taxathat can fulll this denition of taxonomiccontent is two a redundant taxon (ie asingle subordinate taxon) would have anidentical phylogenetic denition and thusis uninformative in phylogenetic taxonomy(Farris 1976) The taxonomic content oftaxon C therefore could be ldquotaxon A taxonB their most recent common ancestor andall descendantsrdquo if taxa A and B are denedextensionally to encompass all existing andpotential subordinate taxa currently withintaxon C

A list of included taxa is insufcient to un-ambiguously specify taxonomic content be-cause there will always be potential mem-bers of a taxon that lie outside any list ofincluded taxa whether or not the includedtaxa themselves have phylogenetic deni-tions An exhaustive list of included taxawith stem-based denitions for exampledoes not specify membership for their hy-pothetical immediate common ancestor (anode-based taxon) or more distant knownor hypothetical outgroups (a stem-basedtaxon) Two reference taxa and a relationalstatement about ancestry are necessary andsufcient to specify taxonomic content be-cause they unambiguously specify member-ship for all existing and potential membersof a taxon

A reference taxon is a species or higher-leveltaxon that is used to delimit a phylogeneticdenition (Table 1) For example ldquotaxon Ataxon B their most recent common ancestorand all descendantsrdquo uses A and B as refer-ence taxa for a node-based taxonomic deni-tion Previously cited as ldquoreference pointsrdquo

(Schander and Thollesson 1995) theseimportant components of phylogeneticdenitions have not been distinguishedterminologically

Several kinds of reference taxa can beidentied (Fig 5 Table 2) Ingroup and out-group reference taxa are located within oroutside the taxon they dene respectively(Fig 5a b) An inclusive reference taxon in-cludes several ingroup or outgroup taxaa maximally inclusive reference taxon in-cludes the largest possible ingroup or out-group reference taxon (Fig 5c) Basal andnested reference taxa are less inclusive andas their name suggests are located at thebase or nested within an ingroup or out-group taxon (Fig 5d e)

A crown taxon denotes a living speciesor a clade dened by using living refer-ence taxa whose immediate outgroup isextinct (Figs 6 7 Table 1) The extinct out-group is a necessary component of the def-inition Jefferies (1979449) coined the termldquocrown grouprdquo to replace Hennigrsquos (1969) group which Jefferies dened as ldquothe lat-est common ancestor of all living membersof group 1 plus all descendants of this an-cestor whether these are living or extinctrdquoTheexistence of an immediate outgroup thatis extinct is implied and was clearly presentin the examples given by Hennig and Jef-feries Patterson (1981207) also clearly usedcrown- and stem-groups in this mannerstating that the ldquostem-group contains fos-sil taxa that exhibit some but not all of thecharacters of the crown-grouprdquo Crown- andstem-group terminology has been used ex-clusively for higher taxa bounded by liv-ing species that have extinct outgroups (egde Queiroz and Gauthier 1992 Forey 1992McKenna and Bell 1997) Although it may

TABLE 2 Denitions for the various kinds of reference taxa

Reference taxon type Denition

Ingroup reference taxon included within a taxon by denitionOutgroup reference taxon excluded from a taxon by denitionInclusive most inclusive reference taxon among ingroup or outgroup taxaBasal reference taxon near the basal furcation of a taxonNested reference taxon distant from the basal furcation of a taxonVernacular vernacular name used as a reference taxon (eg ldquobirdsrdquo or ldquoextant birdsrdquo)

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 337

FIGURE 5 Kinds of reference taxa (a) An ingroup reference taxon is included by denition within a taxon anode-based denition is based on two ingroup reference taxa (b) An outgroup reference taxon is excluded bydenition from a taxon a stem-based denition is based on one ingroup and one outgroup reference taxon (c) Aninclusive reference taxon is the most inclusive ingroup or outgroup reference taxon possible inclusive referencetaxa B and C for taxon A allow denitional ambiguity with change in the relationship of a basal taxon (d) Abasal reference taxon is located at or near the base of an ingroup or outgroup taxon basal reference taxa allowdenitional ambiguity when their relationships change (e) A nested reference taxon is remote from the base ofan ingroup or outgroup reference taxon nested reference taxa are unaffected by changes in the relationships ofbasal taxa Reference taxa are encircled a dot indicates a node-based denition an arrow indicates a stem-baseddenition a dashed line indicates a change in position of a basal taxon

FIGURE 6 Crown stem and total taxa (a) Total group composed of a crown group plus a paraphyletic ldquostemgrouprdquo as conceived by Hennig (1969 1983) and Jefferies (1979) (b) Total taxon composed of a crown taxon andmonophyletic stem taxa as used in this article One stem taxon (left) is shown with a node-based denition andthe other (right) with a stem-based denition

338 SYSTEMATIC BIOLOGY VOL 48

FIGURE 7 Crown species or clades (enclosed) musthave an extinct (stem) taxon (dashed) as an immediateoutgroup (a) Crown species and crown clades are de-limited by the presence of an immediate outgroup thatis extinct (b) Often referred to as ldquoextantrdquo taxa crowntaxa may be composed primarily of extinct taxa ldquoLiv-ingrdquo and ldquoextinctrdquo conditions are treated as if they con-stitute plesiomorphic and apomorphic character statesrespectively of an irreversible character ldquostate of be-ingrdquo Crown clades under this conceptualization areplesiomorphy-based taxa bounded by two living taxafor which the immediate outgroup is extinct

be hypothetically true that ldquoevery Recentgroup whatever its position in thehierarchymay be expected to have had a stem-grouprdquo(Patterson 1981207) clades bounded by liv-ing taxa with immediate outgroups that arealso living have not been considered crowntaxa

Some recent denitions of crown taxahowever have set aside Hennigrsquos andJefferiesrsquo formulations as well as generalusage in the literature A less restric-tive meaning has been proposed basedsolely on reference taxa that include liv-

ing representatives De Queiroz and Gau-thier (1992469) dened crown taxa asldquoclades within which both branches ofthe basal dichotomy are represented byextant descendantsrdquo without mention ofthe condition of the immediate outgroupYet all of their examples of crown cladeshave immediate outgroups that are ex-tinct Regarding their craniate taxonomythey stated that each stem-based taxon iscomposed of a ldquocrown taxon plus all ex-tinct taxa more closely related to it thanto any other crown claderdquo (de Queirozand Gauthier 1992475) Lee (19961103)proposed that ldquoa crown clade denitionresults when both taxa implicated in anode-based denition are representedby extant formsrdquo Although the conditionof the immediate outgroup is never men-tioned Lee (19961103) stated that crowntaxon denitions apply to clades thatconsist of ldquothe most recent common an-cestor of all extant forms and all its de-scendantsrdquo (emphasis added) The pres-ence of an immediate extinct outgroup isimplied

The denition of a crown taxon presentedhere in contrast does not specify ingroupstructure (such as the presence or absenceof a basal dichotomy) or denitional type(node- or stem-based) and explicitly statesthat the immediate outgroup(s) is extinct(Fig 7a Table 1) Crown taxa so denedcan be mapped unerringly on a cladogramif (1) ldquolivingrdquo and ldquoextinctrdquo conditions aretreated as primitive and derived character-states respectively of the character ldquocurrentstate of beingrdquo and (2) if this character isregarded as irreversible (ie CaminndashSokalparsimony) The living state in other wordsis always regarded as plesiomorphic Theirreversible status of the character (state ofbeing) is important because it identiescrown groups on cladograms of living andextinct taxa that would have an ambigu-ous status if character-state reversal wereallowed (Fig 7b) This character-state con-ceptualization of the denition of a crowngroup captures the essence of crown taxawhich have always been used to refer tomaximally inclusive taxa dened by extantrepresentatives not simply any ldquoextant-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 339

boundedrdquo taxon Discovery of extinct sistertaxa provides the opportunity to recognizecrown taxa

Crown taxa are often regarded as cladesthat are bounded uniquely by living in-groups The use of living reference taxahowever is most often an arbitrary deci-sion Any extinct ingroup taxon that is mostclosely related to a living reference taxoncould be used to delimit the same clade(Fig 7b) A crown clade is dened uniquelyby living reference taxa only when such taxado not have extinct sister taxa (Fig 7b) afact that is not widely appreciated Crowntaxa in other words do not have any spe-cial biological signicance beyond signal-ing that particular reference taxa have sur-vived to an arbitrary datum (the Recent)Informal reference to crown taxa as ldquolivingrdquoor ldquoextantrdquo taxa therefore can be mislead-ing because a crown taxon may be com-posed predominantly of extinct species asis the case with the crown taxon Crocodylia(Fig 7b Brochu 1997) ldquoExtant crocodil-iansrdquo if taken literally refers to a smallparaphyletic subset of Crocodylia Explicitreference to ldquocrown crocodiliansrdquo or sim-ply ldquoCrocodyliardquo is preferable to use of theterms ldquolivingrdquo or ldquoextantrdquo

A living taxon is also essential to delimita maximally inclusive clade that is entirelyextinctmdasha stem taxon as here dened Astem taxon denotes an extinct species orclade the immediate outgroup of which in-cludes at least one living member (Fig 6b)Many extinct taxa are not stem taxa becausetheir immediate outgroup is itself extinctSuch extinct taxa are members of more in-clusive stem taxa Stem taxa can have eithernode- or stem-based denitions (Fig 6b)Stem taxa and their associated less inclusivecrown taxon make up the taxonomic contentof their respective total taxon

Stem taxa as here dened should notbe confused with ldquostem grouprdquo (Fig 6a)which was rst used in phylogenetic sys-tematics by Hennig (ldquostammgrupperdquo 19691983) Hennigrsquos (198315) vague formulationof the concept specied all extinct speciesldquowhich can be shown probably to belong toa particular monophyletic group of the phy-logenetic system but which are probably no

closer [sic] related to one subgroup amongrecent animals than to anotherrdquo (translatedfrom the German Ax 1987224) Jefferies(1979) claried the term to include all ex-tinct outgroups of a crown taxon (Fig 6a)Other authors have developed more convo-luted denitions that approximate the sameend Wiley (1981217) for example dened aldquostem grouprdquo as ldquoall fossil species of a groupmore primitive than the most primitive Re-cent species of the same grouprdquo Theseldquostem groupsrdquo are paraphyletic and oftencircumscribe the same ldquoancestral groupsrdquoand ldquoevolutionary gradesrdquo that have mud-died phylogenetic thinking since DarwinOnly a few cladists continue to endorse suchconcepts (Smith 1994) Ax (1987224) pro-posed a cumbersome solution to the prob-lem of paraphyletic ldquostem groupsrdquo by divid-ing them into ancestral ldquostem lineagesrdquo andldquorepresentatives of the stem lineagerdquo Thelatter are monophyletic taxa most simply re-ferred to as stem taxa

A total taxon is regarded here as a cladecomposed of a crown taxon plus all ex-tinct outgroups more closely related to itthan to another crown taxon (Fig 3 Ta-ble 1) Hennigrsquos (1969) and Jefferiesrsquo (1979)formulations of the concept are wordy butsimilar in effect because they restricted in-clusion of extinct outgroups to those mostclosely related to a particular crown taxonDe Queiroz and Gauthier (1992470) deneda total clade as a ldquomore inclusive clade con-sisting of thecrown clade plus its extinctout-groupsrdquo This denition is incomplete be-cause it does not specify which extinct out-groups are included A total taxon explicitlyincludes only those extinct outgroups mostclosely related to a particular crown taxon

Node-based and stem-based denitions areconstructed by (1) identifying reference taxaand (2) attaching to them a relational phrasethat either includes their most recent com-mon ancestor and all descendants (node-based) or limits inclusion to all descen-dants closer to one or more of the referencetaxa (stem-based) (de Queiroz and Gauthier1992) (Fig 1 Table 1) In node-based deni-tions the phrase ldquoleast-inclusive claderdquo canreplace ldquocommon ancestor and all descen-dantsrdquo for cladists sensitive to the allusion to

340 SYSTEMATIC BIOLOGY VOL 48

ancestors (Lee 1998) Listing included taxawithout a relational phrase about ancestryor simply designating taxa as node- or stem-based without reference taxa does not con-stitutea phylogenetic denition because theboundaries of the taxon in question are notspecied (de Queiroz and Gauthier 1990Bryant 1996)

In the current literature there exists someconfusion over what is necessary and suf-cient for node- and stem-based denitionsBryant (1996185) suggested that a more in-clusive taxon be inserted within a stem-based denition such as ldquoall members oftaxon A thataremoreclosely related to taxonB than to taxon Crdquo Including taxon A how-ever is not necessary and may lead to inter-nal conicts within the denition (eg whentaxon A is included within taxon B or C onan alternativephylogeny) The goalof a phy-logenetic denition should be to unambigu-ously identify a specic common ancestorgiven any arrangement of taxa rather thanto identify such an ancestor only on a subsetof phylogenetic hypotheses

Meng et al (1994) and Wyss and Meng(1996) introduced a ldquostem-modied node-based denitionrdquo to effectively unite crown-group rodents whose basal relationshipsare uncertain Their preferred denition ofRodentia (ldquothe clade stemming from themost recent common ancestor of Mus andall Recent mammals more closely relatedto Mus than to Lagomorpha or membersof any other eutherian lsquoorderrdquorsquo Wyss andMeng 1996562) is better interpreted as aspecial stem-based rather than node-baseddenition because it uses only one ingroupreference taxon (Mus) and one or more out-group reference taxa (eg Lagomorpha)Their denition merely adds a phrase (ldquotheclade stemming from the most recent com-mon ancestorrdquo) to an otherwise typicalstem-based denition based on living ref-erence taxa That phrase ensures inclusionof all extinct taxa within the crown cladeBryantrsquos (1996185) reformulation of thisdenition on the other hand is clearly anode-based denition because he speci-ed only ingroup reference taxa (ldquothe mostrecent common ancestor of the species re-ferred to Rodentia by Wilson and Reeder

(1993) and all of its descendantsrdquo) These ex-amples underscore the need for claricationof terms (Tables 1 2)

A node-stem triplet (NST) is a denitionalconguration involving three taxa a node-based taxon composed of two stem-basedsubordinate taxa (Sereno 1997 1998 Fig 8)

Taxon A Taxa b and c their most recentcommon ancestor and all its descendants

Taxon B All taxa closer to taxon b than totaxon c

Taxon C All taxa closer to taxon c than totaxon b

FIGURE 8 Denitional triumvirate termed a node-stem triplet composed of a node-based taxon (A) andtwo subordinate stem-based taxa (B and C) which in-corporates added or repositioned taxa (dashed lineswith daggers) without changing the relative taxonomiccontent of taxon A B or C (as expressed by the equiva-lence statementA = B +C) A dot indicates a node-baseddenition an arrow indicates a stem-based denition

Taxa b and c (not shown in Fig 8) constitutereference taxa within taxon B and C respec-tively Although not essential the same ref-erence taxa (b c) may be used for the trio oftaxa in a NST which then specify complemen-tary denitions The pair of stem-based sis-ter taxa in addition may have reciprocal def-initions that utilize the same reference taxain antipodal positions The word ldquotaxardquo inthe stem-based denitions listed above mayoptionally be replaced by reference toa moreinclusive taxon (Bryant 1996) A stem-baseddenition for Saurischia for example mightread ldquoAll dinosaurs more closely related

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 341

to Neornithes than to Triceratopsrdquo (Sereno1998)

Taxonomic equivalence is a statement (orequation) in which one taxon is shownto be equivalent in current and poten-tial taxonomic content to subordinate taxaplus their common ancestor (Table 1)The NST described above is composedof a node-based taxon (A) and twosubordinate stem-based taxa (B C) andcan be written as the equivalence state-ment ldquoA = B + Crdquo because taxon Aby denition is equivalent to taxa Band C plus their most recent commonancestor

PHYLOGENETIC DEFINITIONS RATIONALE

The rst step toward a phylogenetic tax-onomy involved the distinction betweendenition and diagnosis (Buck and Hull1966 Grifths 1973 1974 Farris 1976 Ghis-elin 1984 Rowe 1987) A second step in-volved the formulation of node- and stem-based denitions by use of reference taxaand relational statements about ancestry (deQueiroz and Gauthier 1992) A third stepoutlined here involves the specication ofguidelines for selecting effective referencetaxa and criteria for positioning node- andstem-based denitions

Phylogenetic rearrangements and introduc-tion of new taxa are the principal factorsthat destabilize taxonomic content Stabil-ity of taxonomic content is enhanced bychoosing (1) effective reference taxa and(2) an effective conguration of phyloge-netic denitions

Reference TaxaPositionmdashGiven a particular phyloge-

netic hypothesis the taxonomic content of aphylogenetic denition is altered only if thereference taxa specify an alternative com-mon ancestor Given a basal dichotomy withone reference taxon on each side the iden-tication of the immediate common ances-tor is not affected by relocation of a ref-erence taxon on its side of the basal di-chotomy (Fig 9 move 1) Relocation ofa reference taxon to the opposing side orto a location outside the basal dichotomyidenties a different common ancestor

(Fig 9 moves 2 3) For a taxon with a basaldichotomy an alternative common ances-tor can be identied only if one referencetaxon is relocated to the opposing side oroutside the basal dichotomy Are some ref-erence taxa more likely to be relocated in thismanner than others

FIGURE 9 Three possible outcomes from relocationof a reference taxon 1 = relocation on the same side ofa basal dichotomy has no effect 2 = relocation to theopposite side of the basal dichotomy identies a lessinclusive common ancestor 3 = relocation outside thebasal dichotomy identies a more inclusive commonancestor Reference taxa are encircled and the originalbasal dichotomy is shown with heavy lines

First a taxon positioned near the basaldichotomy of a cladogram is always eas-ier to relocate to the opposing side (ierequires fewer additional steps) than is anested taxon if all other factors are regardedas equal among ingroup taxa (eg complete-ness missing data and homoplasy) Unlikea basal taxon (Fig 5d) a nested taxon sharessynapomorphies that increase its phyloge-netic (patristic) distance from the commonancestor (Fig 5e) Thus stability of taxo-nomic content is enhanced in phylogeneticdenitions if nested reference taxa are cho-sen that are located at some distance (atleast several nodes away) from the basaldichotomy This distance is easy to assessquantitatively for various nested taxa on agiven phylogeny

Second maximally inclusive referencetaxa (Fig 5c) have greater potential to cre-ate unnecessary taxonomic redundancy andinternal inconsistencies after relocation oftaxa (Fig 3) This is true for both node-

342 SYSTEMATIC BIOLOGY VOL 48

and stem-based taxa In the current litera-ture maximally inclusive reference taxa areoften used in phylogenetic denitions Intheir tetrapod classication for example deQueiroz and Gauthier (1992475) used suc-cessive maximally inclusive reference taxafor all denitionsmdasha pattern here termed re-cursive referencing (Fig 10a) Recursive refer-encing can have undesirable consequencesbecause it shifts the burden of a more precisedenition to a slightly less inclusive highertaxon This higher taxon may not be de-ned or it too may be affected by an alter-native conguration of basal relationshipsSalgado et al (1997) for example choseProsauropoda and Sauropoda as maximallyinclusive reference taxa for Sauropodomor-pha If a basal sauropodomorph is reposi-tioned the denitions of Prosauropoda andSauropoda must be determined to assess po-tential taxonomic consequences HoweverSalgado et al (1997) used Prosauropodaas a terminal taxon without denitionSauropoda was dened as a node-basedtaxon on the basis of two maximally in-clusive reference taxa Vulcanodon and Eu-sauropoda (all other sauropods) Thusif the incompletely known basal sauro-pod Vulcanodon is reinterpreted as be-ing more closely related to prosauropodsSauropoda and Sauropodomorpha wouldbecome synonyms

If on the other hand Sauropodomorphais dened on the basis of the nested ref-erence taxamdashthe prosauropod Plateosaurusand sauropod Saltasaurus (Sereno 1998)mdasha change in the phylogenetic positionof any other prosauropod or sauropodhas no effect on the taxonomic con-tents of Sauropodomorpha or SauropodaSauropodomorpha can be dened as ldquoPla-teosaurus Saltasaurus their most recentcommon ancestor and all descendantsrdquoand Sauropoda can be dened as ldquoallsauropodomorphs more closely related toSaltasaurus than to Plateosaurusrdquo The phylo-genetic position of Vulcanodon has no effecton the identication of respective commonancestors

CompletenessmdashPoorly known taxa aremore likely to be phylogenetically unsta-ble than well-known taxa (Wilkinson 1995)

FIGURE 10 Referencing styles (a) Recursive refer-encing selects a succession of the most inclusive ref-erence taxa (b) Nested referencing selects that sameremote reference taxon Reference taxa are encircled

Taxa with numerous missing entries in aphylogenetic analysis usually reduce phylo-genetic resolution by generating numerousequally parsimonious trees Well-known ref-erence taxa even if they are somewhat lessnested are preferable to poorly known taxa(Fig 11) A single poorly known basal taxonis least stable and therefore least desirableas a reference taxon

Use with Recent or extinct ltersmdashldquoRecentrdquo(ldquoextantrdquo or ldquolivingrdquo) or ldquoextinctrdquo may beused in phylogenetic denitions of crownand stem taxa respectively to help to stabi-lize taxonomic content in the face of poorlyresolved basal relationships For a crowntaxon with a stem-based denition a Recentlter can restrict the boundaries of a cladeto living taxa (Meng et al 1994 Wyss andMeng 1996) Rodentia for example could

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 343

be dened as ldquothe least inclusive clade com-posed of Recent mammals more closely re-lated to Mus than to Lepusrdquo For stem taxain contrast a stem-based denition using anldquoextinctrdquo lter can restrict inclusion to ex-tinct taxa when outgroup relationships areuncertain Moas for example could be de-ned as ldquoall extinct species more closelyrelated to Dinornis than to Struthiordquo theostrich

NumbermdashThe number of reference taxathat have been used in phylogenetic deni-tions has varied from two to innity Gau-thier (198614) for example dened Aves asa node-based taxon including ldquoall descen-

FIGURE 11 Nestedness and completeness in the se-lection of reference taxa A nested well-known taxonis more likely to constitute a stable reference taxon (en-circled)

dants of the most recent common ancestorof Ratitae Tinami and Neognathaerdquo Threereference taxa rather than two were chosenpresumably because of unstated uncertaintyin the phylogenetic relationships amongthese three avian subgroups De Queirozand Gauthier (1990310 1992461) suggestedthat phylogenetic denitions should utilizeonly two reference taxamdashbut then erecteddenitions for total taxa that used as refer-ence taxa one crown clade plus ldquoall extinct

taxa more closely related to it than to anyother crown claderdquo (de Queiroz and Gau-thier 1992475)

Listing more than a pair of referencetaxa does not seem to compromise a node-based phylogenetic denition in any dis-cernible way Node-based denitions areconstructed with what may be termed in-ternal inclusion A contradictory relation-ship between multiple ingroup referencetaxa for a node-based denition is im-possible to construct with internal inclu-sion because a clade always exists thatincludes all ingroup reference taxa Multi-ple ingroup reference taxa effectively stabi-lize the taxonomic content of a node-basedtaxon if basal relationships are uncertain(Fig 12a)

For stem-based denitions multiple out-group reference taxa effectively stabilize thetaxonomic content of a taxon if outgroup re-lationships are uncertain (Fig 12b) Multi-ple ingroup reference taxa for stem-baseddenitions can create a contradictory sit-uation if one (or more) ingroup referencetaxa are repositioned closer to the outgroupreference taxon (Fig 12c) or to a positionoutside the clade as originally dened Inthis case the external inclusion fundamen-tal to a stem-based denition creates a con-tradictory relationship among the originalingroup reference taxa which are no longermore closely related to each other than to theoutgroup reference taxon (Fig 12c) In theexample shown here relocation of ingroupreference taxon D creates a contradictory re-lationship with a second ingroup referencetaxon E if taxon A has the stem-based def-inition of ldquoall descendants more closely re-lated to taxon D and E than to taxon Brdquo Aninclusive ingroup reference taxon howevercan manifest the same problems in this cir-cumstance (Fig 12c) If taxon A is denedas ldquoall descendants more closely related totaxon C than to taxon Brdquo and if taxon C isdened on the basis of taxon D a contra-dictory denition results with relocation oftaxon D as shown taxon A now subsumesits original ingroup reference taxon C Suchcontradictory denitions can be avoided ifa nested ingroup reference taxon (or nestedtaxa) are selected such as taxon E its un-

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 3: Sereno, 1999b

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 331

TABLE 1 Denitions for terms used in this paper

Term Denition

Taxonomic denition relational statement specifying the taxonomic content of a taxonTaxonomic diagnosis descriptive statement specifying the apomorphies (derived character states)

that serve to identify members of a taxonTaxonomic content existing and potential taxa or individuals that by denition are included

within a taxonReference taxon taxon serving as a reference in a phylogenetic denitionCrown taxon a living species or a clade that can be dened by living species whose

immediate outgroup is extinctTotal taxon clade composed of a crown taxon plus all extinct outgroups more closely

related to it than to another crown cladeStem taxon an extinct species or clade for which immediate outgroup includes at least

one living memberNode-based denition statement specifying a clade composed of the most recent common ancestor

of two or more reference taxa and all descendantsStem-based denition statement specifying a clade composed of all descendants more closely

related to one reference taxon (or taxa) than another (or others)Complementary denitions phylogenetic denitions with the same reference taxa (as in

a node-stem triplet)Reciprocal denitions phylogenetic denitions with the same reference taxa but in reverse order

(as in opposing stem-based taxa of a node-stem triplet)Node-stem triplet trio of taxa with complementary denitions consisting of a node-based

taxon and two subordinate stem-based taxaTaxonomic equivalence statement of equivalence (or equation) in which the taxonomic content of

one taxon equals that of subordinate taxa plus their common ancestor(eg A = B + C as in a node-stem triplet)

prove to be more closely related to marsu-pials as indicated by some recent molecu-lar evidence (Penny and Hasegawa 1997)Stability of taxonomic content is not re-lated to the living or extinct status of ref-erence taxa in a phylogenetic denitionbut rather to the particular choice of ref-erence taxa For many node-based deni-tions of crown taxa stability is also re-lated to the certainty with which a speciedbasal dichotomy has been correctly ascer-tained

That ldquowidely usedrdquo names such as Avesshould be assigned to crown taxa has beenjustied by arguing that (1) neontologistsshould report their observations accuratelyreferring only to crown clades rather thancrown clades plus some of their extinct out-groups and that (2) doing so follows long-standing taxonomic practice (de Queirozand Gauthier 1992) The condition in ex-tinct outgroups for characters that are notpreserved however is ambiguous To sup-pose that Archaeopteryx lacked an apomor-phic molecular sequence found in crown-group birds (Neornithes) is no more justied

than tosuppose that the ancientbird had thissequence (Lee 1996) Accelerated character-state transformation in other words is notmore correct accurate or informative thandelayed transformation

Historical and current usage such as maybe ascertained also does not clearly asso-ciate ldquowidely usedrdquo names with crown taxa(Bryant 1994 Lee 1996) Regarding Mam-malia Rowe and Gauthier (1992375ndash376)stated that ldquoits universal usage by compar-ative biologists except for some paleontol-ogists has long been for the crown cladeMammals viz the last common ancestor ofMonotremata and Theria and all of its de-scendantsrdquo and that ldquocurrent debate overthe name Mammalia highlights the fact thatmany paleontologists continue to operate ina system inuenced in fundamental ways bya pre-Darwinian world viewrdquo Darwin wasfairly clear regarding his conception of thetaxonomic contentof Aves Mammalia Mar-supialia and other groups with living repre-sentatives For Darwin these taxa includedfossil stem groups In The Origin of the Species(1859268) he wrote

332 SYSTEMATIC BIOLOGY VOL 48

FIGURE 2 Crown taxa denitional types and sta-bility (a) A crown clade with resolved basal relation-ships but with unresolved outgroup relationships maybe dened with greatest stability as a node-based taxon(b) A crown clade with unresolved basal relationshipsbut with a stable extinct outgroup may be dened withgreatest stability as a stem-based taxon Crown groupsare encircled dashed lines indicate extinct taxa a dotindicates a node-based denition an arrow indicates astem-based denition

those groups which have in known geological pe-riods undergone much modication should in theolder formations make some slight approach to eachother so that the older members should differ less insome of their characters than do the existing mem-bers of the same groups

About mammals Darwin (1859522) wrote

They [marsupials] appeared in an earlier geologicalperiod and their range was formerly much more ex-tensive than at present Hence the Placentata are gen-erally supposed to have been derived from the Im-

placentata or Marsupials not however from formsclosely resembling the existing Marsupials but fromtheir early progenitors

Most extant biologists also conceive of thetaxonomic content of Mammalia in a simi-lar fashion whether they are evolutionarytaxonomists cladists ecologists or molec-ular biologists (eg Jollie 197376ndash77 Fu-tuyma 1986334 Walker 1975) Higher taxasuch as Mammalia and Aves have been asso-ciated with ldquokeyrdquo characters and their asso-ciated functions Archaeopteryx with feath-ers wings and the capacity for ight willalways be considered a ldquobirdrdquo within Avesby the majority of biologists Altering thisequation for well-established monophyletictaxa is likely to engender confusion

Stem-based denitions are less ldquoconsistentrdquothan node-based denitionsmdashSchander andThollesson (1995264) suggested that stem-based denitions ldquorefer to non-existingclades under some phylogenetic resolu-tionsrdquo They presented an example (Fig 3a)in which relocation of one taxon (taxon A)results in a denition that refers to an ldquoim-possiblerdquo clade Relocation of taxon A cre-ates this internal inconsistency because theoriginal ingroup reference taxon (taxon F)for stem-based taxon Gnow includes by def-inition the original outgroup taxon (taxonD) Schander and Thollesson (1995264) con-cluded that ldquoif a consistent system of taxon-omy is wanted stem-based denitions maybetter be avoidedrdquo

Similar internal ldquoinconsistencyrdquo how-ever can be generated for node-based taxaas well (Fig 3b) In this example relocationof one taxon (taxon A) results in a denitionthat refers to an ldquoimpossiblerdquo or internallyredundant clade Relocation of taxon A cre-ates this internal inconsistency because oneof the original reference taxa (taxon E) fornode-based taxon G now includes the otherreference taxon (taxon F) and the commonancestor Even if taxon E is regarded as a ju-nior synonym of taxon G after relocation oftaxon A the denition of taxon G must bechanged (because taxon E cannot be used asa reference taxon)

Rather than demonstrating any partic-ular structural weakness of stem-baseddenitions these examples (Fig 3) better

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 333

FIGURE 3 Problems associated with maximally in-clusive reference taxa (a) Relocation of taxon A resultsin an ldquoinconsistentrdquo stem-based denition for taxon Gbecause it denes a group that cannot exist under the al-ternative phylogenetic hypothesis (after Schander andThollesson 1995Fig 3) Schander and Thollesson de-ned stem-based taxon G as ldquoF and all taxa sharing amore recent common ancestor with it than with Drdquo andstem-based taxon F as ldquoA and all taxa sharing a morerecent common ancestor with it than with Crdquo TaxonG is ldquoinconsistentrdquo after relocation of taxon A becauseits ingroup reference taxon (taxon F) now includes bydenition the outgroup reference taxon (taxon D) Theldquoinconsistencyrdquo however disappears if a less-inclusiveingroup reference taxon had been chosen for taxon G(ie taxon A or B rather than taxon F) In this casetaxon G exists after relocation of taxon A (either as asynonym of taxon A or as a group including taxa Band C depending on whether taxon A or B was used asthe ingroup reference taxon respectively) (b) Reloca-tion of taxon A results in an ldquoinconsistentrdquo node-baseddenition for taxon G because it denes a group thatcannot existunder thealternativephylogenetichypoth-esis In this example taxon G is node-based and denedas ldquotaxon E taxon F their common ancestor and all

descendantsrdquo taxon E is dened as ldquotaxon A taxon Btheir common ancestor and alldescendantsrdquo and taxonF is node-based and dened as ldquotaxon C taxon D theircommon ancestor and all descendantsrdquo Taxon G is ldquoin-consistentrdquo after relocation of taxon A because one ofits reference taxa (taxon E) now includes by denitionthe other reference taxon (taxon F) and the commonancestor The reference taxa for a node-based taxon arepresumed to be exclusive of one another and their com-mon ancestor after relocation of taxon A these condi-tions are violated for taxon G as originally dened Theldquoinconsistencyrdquo however disappears if a less-inclusivereference taxon had originally been selected for taxon G(ie taxon A or B rather than taxon E) In this case taxonG exists after relocation of taxon A (either unchanged inits taxonomic content or as a less-inclusive taxon thatexcludes taxon B depending on whether taxon B or Awas used as an ingroup reference taxon respectively) Dots indicate node-based denitions and arrows indi-cate stem-based denitions

illustrate problems that arise as a result of se-lecting maximally inclusive reference taxamdashthe most common kind of reference taxacurrently chosen in phylogenetic denitions(see Reference Taxa below) If Schander andThollesson had chosen a more nested in-group reference taxon in their stem-baseddenition of taxon G (ie taxon A or B ratherthan taxon F Fig 3a) relocation of taxonA poses no special problems In this casetaxon G either refers to the same clade mi-nus taxon A (if taxon B is the ingroup ref-erence taxon) or becomes a potential juniorsynonym of taxon A (if taxon A is the in-group reference taxon) Similarly for node-based taxon G (Fig 3b) a more nested refer-ence taxon (taxon A or B rather than taxonE) removes any inconsistency after reloca-tion of taxon A Taxon G refers either to thesame clade (if taxon B is an ingroup refer-ence taxon) or to a less inclusive clade thatexcludes taxon B (if taxon A is an ingroupreference taxon)

Node- or stem-based denitions better re-ect original usage and contentmdashGauthier(198612 13) for example suggested that hisstem-based denition for Haeckelrsquos (1866)taxon Ornithurae (ldquoExtant birds and allother taxa such as Ichthyornis and Hes-perornithes [sic] that are closer to extantbirds than is Archaeopteryxrdquo) was ldquoin keep-ing with its original intentrdquo (Fig 4) Chiappe(1991 1996) on the other hand argued that

334 SYSTEMATIC BIOLOGY VOL 48

Ornithurae is better dened as a less inclu-sive node-based taxon with Hesperornithi-formes and Neornithes (crown-group birds)as reference taxa (Fig 4) Initially Chiappe

FIGURE 4 Cladogram of basal avians showing re-cent use of the higher taxon Ornithurae (Haeckel 1866)The dot indicates a node-based denition and the ar-row indicates a stem-based denition

(1991337) reasoned that a node-based de-nition of Ornithurae was more appropriatebecause it excluded several newly discov-ered basal avians that lacked some of thesynapomorphies in Gauthierrsquos (1986) diag-nosis of Ornithurae Later Chiappe arguedthat a node-based denition of Ornithuraeldquobetter reects the composition of the cladebefore to [sic] the application of phyloge-netic denitionsrdquo (1996205)

Neither Gauthierrsquos nor Chiappersquos deni-tions have historical precedence Haeckel(1866) coined Ornithurae (ldquobird tailrdquo) for thereduced number of vertebrae and uniquecoossication of the distal tail (as a sin-gle bone the pygostyle) that characterizesall extant birds in contrast to the condi-tion in Archaeopteryx Haeckelrsquos taxon owesits existence to the then-recent discovery ofArchaeopteryx a primitive bird with a tailcomprising over 20 vertebrae Contrary toChiappe (1996) Ornithurae predates the de-scription of the extinct short-tailed generaHesperornis (Marsh 1872a) and Ichthyor-nis (Marsh 1872b) or the taxon Hesper-ornithiformes (Furbringer 1888) Ornithu-

rae as noted by Gauthier (1986) has beensupplanted by Neornithes (Gadow 1893)a taxon referring to crown-group aviansBased on original intent an argument couldbe made that Haeckelrsquos Ornithurae shouldsupplant Neornithes on grounds of prior-ity or that it be dened explicitly as anapomorphy-based taxon based on the pres-ence of a pygostyle which is now known tobe lacking in birds other than Archaeopteryx(Forster et al 1998) In either case thedenition would circumscribe clades otherthan those identied by either Gauthier orChiappe

For most taxon names the original de-nition if explicitly stated at all is character-based or taxon-based (a list of included taxa)and lacks a relational phrase about ances-try that would specify potential membershipThe intent of the original author of a taxonto include or exclude unknown or reposi-tioned taxa usually cannot be interpretedunambiguously This problem is not widelyappreciated

APOMORPHY-BASED DEFINITIONS

An apomorphy-based denition speci-es the boundaries of a clade by identi-fying the ldquorst ancestor with a particularsynapomorphyrdquo (de Queiroz and Gauthier1990310) and encompassing all of its de-scendants An apomorphy-based taxon in-cludes all descendants whether or not thesynapomorphy is maintained Apomorphy-based denitions are subject to three prob-lems that are not relevant to node-based andstem-based denitions variation in charac-ter coding optimization ambiguity and ho-moplasy (Bryant 1994 Schander and Thol-lesson 1995 Sereno 1998) These problemshave long been associated with traditionaluse of ldquokeyrdquo characters to dene taxa Thusfar few apomorphy-based denitions havebeen erected and this denitional formshould be avoided

Character ambiguitymdashConsider an apo-morphy-based denition for Aves (includ-ing Archaeopteryx and Ornithurae) based onthe presence of ldquofeathersrdquo the ldquokey char-acterrdquo usually associated with this taxonFirst the character codingmdashldquoFeathers ab-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 335

sent (0) present (1)rdquomdashis only one of sev-eral possibilities A ldquofeatherrdquo is composedof a rachis barbs barbules and other com-ponents that may be better coded as sepa-rate characters The character on which anapomorphy-based denition is based there-fore may be recoded as two or more char-acters with independent character-state dis-tributions Indeed this is an expectation formost qualitative or quantitative charactersthat is once the transformations become bet-ter known intermediate states or multiplecharacters will emerge as recent discoveriesare beginning to reveal with regard to feath-ers (Chen et al 1998 Ji et al 1998) or withregard to the characters historically associ-ated with Mammalia (Rowe 1988 RoweandGauthier 1992 Bryant 1994 de Queiroz1994)

The apomorphy-based denition given asan example by de Queiroz and Gauthier(1990310) is subject on the same groundsto interpretational ambiguity ldquoTetrapodardquois dened as the ldquorst vertebrate to possessdigits (ie hands and feet rather than ns)and all of its descendantsrdquo What exactlyconstitutes a ldquodigitrdquo versus a ldquonrdquo or ldquopad-dlerdquo In fact this is a signicant questiongiven recent fossil discoveries (Daeschlerand Shubin 1998) Can this feature or listof features be observed in extinct formsAnd what happens if we learn that ldquodigitsrdquoevolved rst on the hands and later on thefeet These are common problems in the in-terpretation of character data which shouldnot be extended to taxonomic denitions

Optimization ambiguitymdashApomorphy-based denitions also do not specify a sin-gle ancestor when the chosen apomorphyhas an ambiguous optimization Charac-ters with ambiguous character-state opti-mization are commonplace in systematicsand arise from missing data (lack of preser-vation strong transformation) and homo-plasy (Bryant 1994 Schander and Thol-lesson 1995 Sereno 1998) There may bemany equally parsimonious positions fora particular character state on the shortestcladogram

HomoplasymdashHomoplasy may involve thestable placement of an apomorphy at morethan one node (Bryant 1994) This can create

an impasse for an apomorphy-based deni-tion because the apomorphy might identifymore than one clade Invoking time as an ar-biter (such as ldquothe rst taxon that has rdquo)invites ambiguities associated with tempo-ral origin (such as missing ancestral lin-eages)

PHYLOGENETIC DEFINITIONS TERMINOLOGY

The terminology used in the followingdiscussion of phylogenetic taxonomic def-initions is claried below and tabulated (Ta-bles 1 2) Someof these terms aregiven morespecic meaning than in the current litera-ture others are new or recently introduced(Sereno 1998)

Taxonomic denition and diagnosis are de-ned here much as they have been describedpreviously by phylogenetic systematists (deQueiroz and Gauthier 1990 1992) and theirforebears (Buck and Hull 1966 Farris 1976Ghiselin 1984) (Table 1) De Queiroz andGauthier (1992461) dened phylogeneticdenitions as ldquostatements specifying themeanings of taxon names (words) they arestated in terms of ancestryrdquo The ldquomeaningof a taxon namerdquo in phylogenetic taxonomyconcerns its taxonomic content or mem-bership as delimited by a relational state-ment Taxonomic denitions therefore areregarded here as ldquoextensionalrdquo (ostensive)statements of relationship that specify thetaxonomic content of a taxon (de Queiroz1992)

Taxonomic diagnosis on the other handinvolves the ldquointensionalrdquo characterizationof members of a clade as recognized bythe shared apomorphies present in thecommon ancestor (Buck and Hull 1966Farris 1976 Ghiselin 1984 Rowe 1987de Queiroz and Gauthier 1990 1992 deQueiroz 1992) According to de Queiroz andGauthier (1992461) diagnoses are ldquostate-ments specifying how to determine whethera given species or organism is a representa-tive of the taxon (clade) to which a partic-ular name refers they are most commonlystated in terms of charactersrdquo Strictly speak-ing shared apomorphies (derived-characterstates) rather than characters determine themembership of a particular clade or species

336 SYSTEMATIC BIOLOGY VOL 48

Taxonomic content is used here to refer toall existing and potential taxa included by thephylogenetic denition of a taxon (Table 1)The smallest number of subordinate taxathat can fulll this denition of taxonomiccontent is two a redundant taxon (ie asingle subordinate taxon) would have anidentical phylogenetic denition and thusis uninformative in phylogenetic taxonomy(Farris 1976) The taxonomic content oftaxon C therefore could be ldquotaxon A taxonB their most recent common ancestor andall descendantsrdquo if taxa A and B are denedextensionally to encompass all existing andpotential subordinate taxa currently withintaxon C

A list of included taxa is insufcient to un-ambiguously specify taxonomic content be-cause there will always be potential mem-bers of a taxon that lie outside any list ofincluded taxa whether or not the includedtaxa themselves have phylogenetic deni-tions An exhaustive list of included taxawith stem-based denitions for exampledoes not specify membership for their hy-pothetical immediate common ancestor (anode-based taxon) or more distant knownor hypothetical outgroups (a stem-basedtaxon) Two reference taxa and a relationalstatement about ancestry are necessary andsufcient to specify taxonomic content be-cause they unambiguously specify member-ship for all existing and potential membersof a taxon

A reference taxon is a species or higher-leveltaxon that is used to delimit a phylogeneticdenition (Table 1) For example ldquotaxon Ataxon B their most recent common ancestorand all descendantsrdquo uses A and B as refer-ence taxa for a node-based taxonomic deni-tion Previously cited as ldquoreference pointsrdquo

(Schander and Thollesson 1995) theseimportant components of phylogeneticdenitions have not been distinguishedterminologically

Several kinds of reference taxa can beidentied (Fig 5 Table 2) Ingroup and out-group reference taxa are located within oroutside the taxon they dene respectively(Fig 5a b) An inclusive reference taxon in-cludes several ingroup or outgroup taxaa maximally inclusive reference taxon in-cludes the largest possible ingroup or out-group reference taxon (Fig 5c) Basal andnested reference taxa are less inclusive andas their name suggests are located at thebase or nested within an ingroup or out-group taxon (Fig 5d e)

A crown taxon denotes a living speciesor a clade dened by using living refer-ence taxa whose immediate outgroup isextinct (Figs 6 7 Table 1) The extinct out-group is a necessary component of the def-inition Jefferies (1979449) coined the termldquocrown grouprdquo to replace Hennigrsquos (1969) group which Jefferies dened as ldquothe lat-est common ancestor of all living membersof group 1 plus all descendants of this an-cestor whether these are living or extinctrdquoTheexistence of an immediate outgroup thatis extinct is implied and was clearly presentin the examples given by Hennig and Jef-feries Patterson (1981207) also clearly usedcrown- and stem-groups in this mannerstating that the ldquostem-group contains fos-sil taxa that exhibit some but not all of thecharacters of the crown-grouprdquo Crown- andstem-group terminology has been used ex-clusively for higher taxa bounded by liv-ing species that have extinct outgroups (egde Queiroz and Gauthier 1992 Forey 1992McKenna and Bell 1997) Although it may

TABLE 2 Denitions for the various kinds of reference taxa

Reference taxon type Denition

Ingroup reference taxon included within a taxon by denitionOutgroup reference taxon excluded from a taxon by denitionInclusive most inclusive reference taxon among ingroup or outgroup taxaBasal reference taxon near the basal furcation of a taxonNested reference taxon distant from the basal furcation of a taxonVernacular vernacular name used as a reference taxon (eg ldquobirdsrdquo or ldquoextant birdsrdquo)

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 337

FIGURE 5 Kinds of reference taxa (a) An ingroup reference taxon is included by denition within a taxon anode-based denition is based on two ingroup reference taxa (b) An outgroup reference taxon is excluded bydenition from a taxon a stem-based denition is based on one ingroup and one outgroup reference taxon (c) Aninclusive reference taxon is the most inclusive ingroup or outgroup reference taxon possible inclusive referencetaxa B and C for taxon A allow denitional ambiguity with change in the relationship of a basal taxon (d) Abasal reference taxon is located at or near the base of an ingroup or outgroup taxon basal reference taxa allowdenitional ambiguity when their relationships change (e) A nested reference taxon is remote from the base ofan ingroup or outgroup reference taxon nested reference taxa are unaffected by changes in the relationships ofbasal taxa Reference taxa are encircled a dot indicates a node-based denition an arrow indicates a stem-baseddenition a dashed line indicates a change in position of a basal taxon

FIGURE 6 Crown stem and total taxa (a) Total group composed of a crown group plus a paraphyletic ldquostemgrouprdquo as conceived by Hennig (1969 1983) and Jefferies (1979) (b) Total taxon composed of a crown taxon andmonophyletic stem taxa as used in this article One stem taxon (left) is shown with a node-based denition andthe other (right) with a stem-based denition

338 SYSTEMATIC BIOLOGY VOL 48

FIGURE 7 Crown species or clades (enclosed) musthave an extinct (stem) taxon (dashed) as an immediateoutgroup (a) Crown species and crown clades are de-limited by the presence of an immediate outgroup thatis extinct (b) Often referred to as ldquoextantrdquo taxa crowntaxa may be composed primarily of extinct taxa ldquoLiv-ingrdquo and ldquoextinctrdquo conditions are treated as if they con-stitute plesiomorphic and apomorphic character statesrespectively of an irreversible character ldquostate of be-ingrdquo Crown clades under this conceptualization areplesiomorphy-based taxa bounded by two living taxafor which the immediate outgroup is extinct

be hypothetically true that ldquoevery Recentgroup whatever its position in thehierarchymay be expected to have had a stem-grouprdquo(Patterson 1981207) clades bounded by liv-ing taxa with immediate outgroups that arealso living have not been considered crowntaxa

Some recent denitions of crown taxahowever have set aside Hennigrsquos andJefferiesrsquo formulations as well as generalusage in the literature A less restric-tive meaning has been proposed basedsolely on reference taxa that include liv-

ing representatives De Queiroz and Gau-thier (1992469) dened crown taxa asldquoclades within which both branches ofthe basal dichotomy are represented byextant descendantsrdquo without mention ofthe condition of the immediate outgroupYet all of their examples of crown cladeshave immediate outgroups that are ex-tinct Regarding their craniate taxonomythey stated that each stem-based taxon iscomposed of a ldquocrown taxon plus all ex-tinct taxa more closely related to it thanto any other crown claderdquo (de Queirozand Gauthier 1992475) Lee (19961103)proposed that ldquoa crown clade denitionresults when both taxa implicated in anode-based denition are representedby extant formsrdquo Although the conditionof the immediate outgroup is never men-tioned Lee (19961103) stated that crowntaxon denitions apply to clades thatconsist of ldquothe most recent common an-cestor of all extant forms and all its de-scendantsrdquo (emphasis added) The pres-ence of an immediate extinct outgroup isimplied

The denition of a crown taxon presentedhere in contrast does not specify ingroupstructure (such as the presence or absenceof a basal dichotomy) or denitional type(node- or stem-based) and explicitly statesthat the immediate outgroup(s) is extinct(Fig 7a Table 1) Crown taxa so denedcan be mapped unerringly on a cladogramif (1) ldquolivingrdquo and ldquoextinctrdquo conditions aretreated as primitive and derived character-states respectively of the character ldquocurrentstate of beingrdquo and (2) if this character isregarded as irreversible (ie CaminndashSokalparsimony) The living state in other wordsis always regarded as plesiomorphic Theirreversible status of the character (state ofbeing) is important because it identiescrown groups on cladograms of living andextinct taxa that would have an ambigu-ous status if character-state reversal wereallowed (Fig 7b) This character-state con-ceptualization of the denition of a crowngroup captures the essence of crown taxawhich have always been used to refer tomaximally inclusive taxa dened by extantrepresentatives not simply any ldquoextant-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 339

boundedrdquo taxon Discovery of extinct sistertaxa provides the opportunity to recognizecrown taxa

Crown taxa are often regarded as cladesthat are bounded uniquely by living in-groups The use of living reference taxahowever is most often an arbitrary deci-sion Any extinct ingroup taxon that is mostclosely related to a living reference taxoncould be used to delimit the same clade(Fig 7b) A crown clade is dened uniquelyby living reference taxa only when such taxado not have extinct sister taxa (Fig 7b) afact that is not widely appreciated Crowntaxa in other words do not have any spe-cial biological signicance beyond signal-ing that particular reference taxa have sur-vived to an arbitrary datum (the Recent)Informal reference to crown taxa as ldquolivingrdquoor ldquoextantrdquo taxa therefore can be mislead-ing because a crown taxon may be com-posed predominantly of extinct species asis the case with the crown taxon Crocodylia(Fig 7b Brochu 1997) ldquoExtant crocodil-iansrdquo if taken literally refers to a smallparaphyletic subset of Crocodylia Explicitreference to ldquocrown crocodiliansrdquo or sim-ply ldquoCrocodyliardquo is preferable to use of theterms ldquolivingrdquo or ldquoextantrdquo

A living taxon is also essential to delimita maximally inclusive clade that is entirelyextinctmdasha stem taxon as here dened Astem taxon denotes an extinct species orclade the immediate outgroup of which in-cludes at least one living member (Fig 6b)Many extinct taxa are not stem taxa becausetheir immediate outgroup is itself extinctSuch extinct taxa are members of more in-clusive stem taxa Stem taxa can have eithernode- or stem-based denitions (Fig 6b)Stem taxa and their associated less inclusivecrown taxon make up the taxonomic contentof their respective total taxon

Stem taxa as here dened should notbe confused with ldquostem grouprdquo (Fig 6a)which was rst used in phylogenetic sys-tematics by Hennig (ldquostammgrupperdquo 19691983) Hennigrsquos (198315) vague formulationof the concept specied all extinct speciesldquowhich can be shown probably to belong toa particular monophyletic group of the phy-logenetic system but which are probably no

closer [sic] related to one subgroup amongrecent animals than to anotherrdquo (translatedfrom the German Ax 1987224) Jefferies(1979) claried the term to include all ex-tinct outgroups of a crown taxon (Fig 6a)Other authors have developed more convo-luted denitions that approximate the sameend Wiley (1981217) for example dened aldquostem grouprdquo as ldquoall fossil species of a groupmore primitive than the most primitive Re-cent species of the same grouprdquo Theseldquostem groupsrdquo are paraphyletic and oftencircumscribe the same ldquoancestral groupsrdquoand ldquoevolutionary gradesrdquo that have mud-died phylogenetic thinking since DarwinOnly a few cladists continue to endorse suchconcepts (Smith 1994) Ax (1987224) pro-posed a cumbersome solution to the prob-lem of paraphyletic ldquostem groupsrdquo by divid-ing them into ancestral ldquostem lineagesrdquo andldquorepresentatives of the stem lineagerdquo Thelatter are monophyletic taxa most simply re-ferred to as stem taxa

A total taxon is regarded here as a cladecomposed of a crown taxon plus all ex-tinct outgroups more closely related to itthan to another crown taxon (Fig 3 Ta-ble 1) Hennigrsquos (1969) and Jefferiesrsquo (1979)formulations of the concept are wordy butsimilar in effect because they restricted in-clusion of extinct outgroups to those mostclosely related to a particular crown taxonDe Queiroz and Gauthier (1992470) deneda total clade as a ldquomore inclusive clade con-sisting of thecrown clade plus its extinctout-groupsrdquo This denition is incomplete be-cause it does not specify which extinct out-groups are included A total taxon explicitlyincludes only those extinct outgroups mostclosely related to a particular crown taxon

Node-based and stem-based denitions areconstructed by (1) identifying reference taxaand (2) attaching to them a relational phrasethat either includes their most recent com-mon ancestor and all descendants (node-based) or limits inclusion to all descen-dants closer to one or more of the referencetaxa (stem-based) (de Queiroz and Gauthier1992) (Fig 1 Table 1) In node-based deni-tions the phrase ldquoleast-inclusive claderdquo canreplace ldquocommon ancestor and all descen-dantsrdquo for cladists sensitive to the allusion to

340 SYSTEMATIC BIOLOGY VOL 48

ancestors (Lee 1998) Listing included taxawithout a relational phrase about ancestryor simply designating taxa as node- or stem-based without reference taxa does not con-stitutea phylogenetic denition because theboundaries of the taxon in question are notspecied (de Queiroz and Gauthier 1990Bryant 1996)

In the current literature there exists someconfusion over what is necessary and suf-cient for node- and stem-based denitionsBryant (1996185) suggested that a more in-clusive taxon be inserted within a stem-based denition such as ldquoall members oftaxon A thataremoreclosely related to taxonB than to taxon Crdquo Including taxon A how-ever is not necessary and may lead to inter-nal conicts within the denition (eg whentaxon A is included within taxon B or C onan alternativephylogeny) The goalof a phy-logenetic denition should be to unambigu-ously identify a specic common ancestorgiven any arrangement of taxa rather thanto identify such an ancestor only on a subsetof phylogenetic hypotheses

Meng et al (1994) and Wyss and Meng(1996) introduced a ldquostem-modied node-based denitionrdquo to effectively unite crown-group rodents whose basal relationshipsare uncertain Their preferred denition ofRodentia (ldquothe clade stemming from themost recent common ancestor of Mus andall Recent mammals more closely relatedto Mus than to Lagomorpha or membersof any other eutherian lsquoorderrdquorsquo Wyss andMeng 1996562) is better interpreted as aspecial stem-based rather than node-baseddenition because it uses only one ingroupreference taxon (Mus) and one or more out-group reference taxa (eg Lagomorpha)Their denition merely adds a phrase (ldquotheclade stemming from the most recent com-mon ancestorrdquo) to an otherwise typicalstem-based denition based on living ref-erence taxa That phrase ensures inclusionof all extinct taxa within the crown cladeBryantrsquos (1996185) reformulation of thisdenition on the other hand is clearly anode-based denition because he speci-ed only ingroup reference taxa (ldquothe mostrecent common ancestor of the species re-ferred to Rodentia by Wilson and Reeder

(1993) and all of its descendantsrdquo) These ex-amples underscore the need for claricationof terms (Tables 1 2)

A node-stem triplet (NST) is a denitionalconguration involving three taxa a node-based taxon composed of two stem-basedsubordinate taxa (Sereno 1997 1998 Fig 8)

Taxon A Taxa b and c their most recentcommon ancestor and all its descendants

Taxon B All taxa closer to taxon b than totaxon c

Taxon C All taxa closer to taxon c than totaxon b

FIGURE 8 Denitional triumvirate termed a node-stem triplet composed of a node-based taxon (A) andtwo subordinate stem-based taxa (B and C) which in-corporates added or repositioned taxa (dashed lineswith daggers) without changing the relative taxonomiccontent of taxon A B or C (as expressed by the equiva-lence statementA = B +C) A dot indicates a node-baseddenition an arrow indicates a stem-based denition

Taxa b and c (not shown in Fig 8) constitutereference taxa within taxon B and C respec-tively Although not essential the same ref-erence taxa (b c) may be used for the trio oftaxa in a NST which then specify complemen-tary denitions The pair of stem-based sis-ter taxa in addition may have reciprocal def-initions that utilize the same reference taxain antipodal positions The word ldquotaxardquo inthe stem-based denitions listed above mayoptionally be replaced by reference toa moreinclusive taxon (Bryant 1996) A stem-baseddenition for Saurischia for example mightread ldquoAll dinosaurs more closely related

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 341

to Neornithes than to Triceratopsrdquo (Sereno1998)

Taxonomic equivalence is a statement (orequation) in which one taxon is shownto be equivalent in current and poten-tial taxonomic content to subordinate taxaplus their common ancestor (Table 1)The NST described above is composedof a node-based taxon (A) and twosubordinate stem-based taxa (B C) andcan be written as the equivalence state-ment ldquoA = B + Crdquo because taxon Aby denition is equivalent to taxa Band C plus their most recent commonancestor

PHYLOGENETIC DEFINITIONS RATIONALE

The rst step toward a phylogenetic tax-onomy involved the distinction betweendenition and diagnosis (Buck and Hull1966 Grifths 1973 1974 Farris 1976 Ghis-elin 1984 Rowe 1987) A second step in-volved the formulation of node- and stem-based denitions by use of reference taxaand relational statements about ancestry (deQueiroz and Gauthier 1992) A third stepoutlined here involves the specication ofguidelines for selecting effective referencetaxa and criteria for positioning node- andstem-based denitions

Phylogenetic rearrangements and introduc-tion of new taxa are the principal factorsthat destabilize taxonomic content Stabil-ity of taxonomic content is enhanced bychoosing (1) effective reference taxa and(2) an effective conguration of phyloge-netic denitions

Reference TaxaPositionmdashGiven a particular phyloge-

netic hypothesis the taxonomic content of aphylogenetic denition is altered only if thereference taxa specify an alternative com-mon ancestor Given a basal dichotomy withone reference taxon on each side the iden-tication of the immediate common ances-tor is not affected by relocation of a ref-erence taxon on its side of the basal di-chotomy (Fig 9 move 1) Relocation ofa reference taxon to the opposing side orto a location outside the basal dichotomyidenties a different common ancestor

(Fig 9 moves 2 3) For a taxon with a basaldichotomy an alternative common ances-tor can be identied only if one referencetaxon is relocated to the opposing side oroutside the basal dichotomy Are some ref-erence taxa more likely to be relocated in thismanner than others

FIGURE 9 Three possible outcomes from relocationof a reference taxon 1 = relocation on the same side ofa basal dichotomy has no effect 2 = relocation to theopposite side of the basal dichotomy identies a lessinclusive common ancestor 3 = relocation outside thebasal dichotomy identies a more inclusive commonancestor Reference taxa are encircled and the originalbasal dichotomy is shown with heavy lines

First a taxon positioned near the basaldichotomy of a cladogram is always eas-ier to relocate to the opposing side (ierequires fewer additional steps) than is anested taxon if all other factors are regardedas equal among ingroup taxa (eg complete-ness missing data and homoplasy) Unlikea basal taxon (Fig 5d) a nested taxon sharessynapomorphies that increase its phyloge-netic (patristic) distance from the commonancestor (Fig 5e) Thus stability of taxo-nomic content is enhanced in phylogeneticdenitions if nested reference taxa are cho-sen that are located at some distance (atleast several nodes away) from the basaldichotomy This distance is easy to assessquantitatively for various nested taxa on agiven phylogeny

Second maximally inclusive referencetaxa (Fig 5c) have greater potential to cre-ate unnecessary taxonomic redundancy andinternal inconsistencies after relocation oftaxa (Fig 3) This is true for both node-

342 SYSTEMATIC BIOLOGY VOL 48

and stem-based taxa In the current litera-ture maximally inclusive reference taxa areoften used in phylogenetic denitions Intheir tetrapod classication for example deQueiroz and Gauthier (1992475) used suc-cessive maximally inclusive reference taxafor all denitionsmdasha pattern here termed re-cursive referencing (Fig 10a) Recursive refer-encing can have undesirable consequencesbecause it shifts the burden of a more precisedenition to a slightly less inclusive highertaxon This higher taxon may not be de-ned or it too may be affected by an alter-native conguration of basal relationshipsSalgado et al (1997) for example choseProsauropoda and Sauropoda as maximallyinclusive reference taxa for Sauropodomor-pha If a basal sauropodomorph is reposi-tioned the denitions of Prosauropoda andSauropoda must be determined to assess po-tential taxonomic consequences HoweverSalgado et al (1997) used Prosauropodaas a terminal taxon without denitionSauropoda was dened as a node-basedtaxon on the basis of two maximally in-clusive reference taxa Vulcanodon and Eu-sauropoda (all other sauropods) Thusif the incompletely known basal sauro-pod Vulcanodon is reinterpreted as be-ing more closely related to prosauropodsSauropoda and Sauropodomorpha wouldbecome synonyms

If on the other hand Sauropodomorphais dened on the basis of the nested ref-erence taxamdashthe prosauropod Plateosaurusand sauropod Saltasaurus (Sereno 1998)mdasha change in the phylogenetic positionof any other prosauropod or sauropodhas no effect on the taxonomic con-tents of Sauropodomorpha or SauropodaSauropodomorpha can be dened as ldquoPla-teosaurus Saltasaurus their most recentcommon ancestor and all descendantsrdquoand Sauropoda can be dened as ldquoallsauropodomorphs more closely related toSaltasaurus than to Plateosaurusrdquo The phylo-genetic position of Vulcanodon has no effecton the identication of respective commonancestors

CompletenessmdashPoorly known taxa aremore likely to be phylogenetically unsta-ble than well-known taxa (Wilkinson 1995)

FIGURE 10 Referencing styles (a) Recursive refer-encing selects a succession of the most inclusive ref-erence taxa (b) Nested referencing selects that sameremote reference taxon Reference taxa are encircled

Taxa with numerous missing entries in aphylogenetic analysis usually reduce phylo-genetic resolution by generating numerousequally parsimonious trees Well-known ref-erence taxa even if they are somewhat lessnested are preferable to poorly known taxa(Fig 11) A single poorly known basal taxonis least stable and therefore least desirableas a reference taxon

Use with Recent or extinct ltersmdashldquoRecentrdquo(ldquoextantrdquo or ldquolivingrdquo) or ldquoextinctrdquo may beused in phylogenetic denitions of crownand stem taxa respectively to help to stabi-lize taxonomic content in the face of poorlyresolved basal relationships For a crowntaxon with a stem-based denition a Recentlter can restrict the boundaries of a cladeto living taxa (Meng et al 1994 Wyss andMeng 1996) Rodentia for example could

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 343

be dened as ldquothe least inclusive clade com-posed of Recent mammals more closely re-lated to Mus than to Lepusrdquo For stem taxain contrast a stem-based denition using anldquoextinctrdquo lter can restrict inclusion to ex-tinct taxa when outgroup relationships areuncertain Moas for example could be de-ned as ldquoall extinct species more closelyrelated to Dinornis than to Struthiordquo theostrich

NumbermdashThe number of reference taxathat have been used in phylogenetic deni-tions has varied from two to innity Gau-thier (198614) for example dened Aves asa node-based taxon including ldquoall descen-

FIGURE 11 Nestedness and completeness in the se-lection of reference taxa A nested well-known taxonis more likely to constitute a stable reference taxon (en-circled)

dants of the most recent common ancestorof Ratitae Tinami and Neognathaerdquo Threereference taxa rather than two were chosenpresumably because of unstated uncertaintyin the phylogenetic relationships amongthese three avian subgroups De Queirozand Gauthier (1990310 1992461) suggestedthat phylogenetic denitions should utilizeonly two reference taxamdashbut then erecteddenitions for total taxa that used as refer-ence taxa one crown clade plus ldquoall extinct

taxa more closely related to it than to anyother crown claderdquo (de Queiroz and Gau-thier 1992475)

Listing more than a pair of referencetaxa does not seem to compromise a node-based phylogenetic denition in any dis-cernible way Node-based denitions areconstructed with what may be termed in-ternal inclusion A contradictory relation-ship between multiple ingroup referencetaxa for a node-based denition is im-possible to construct with internal inclu-sion because a clade always exists thatincludes all ingroup reference taxa Multi-ple ingroup reference taxa effectively stabi-lize the taxonomic content of a node-basedtaxon if basal relationships are uncertain(Fig 12a)

For stem-based denitions multiple out-group reference taxa effectively stabilize thetaxonomic content of a taxon if outgroup re-lationships are uncertain (Fig 12b) Multi-ple ingroup reference taxa for stem-baseddenitions can create a contradictory sit-uation if one (or more) ingroup referencetaxa are repositioned closer to the outgroupreference taxon (Fig 12c) or to a positionoutside the clade as originally dened Inthis case the external inclusion fundamen-tal to a stem-based denition creates a con-tradictory relationship among the originalingroup reference taxa which are no longermore closely related to each other than to theoutgroup reference taxon (Fig 12c) In theexample shown here relocation of ingroupreference taxon D creates a contradictory re-lationship with a second ingroup referencetaxon E if taxon A has the stem-based def-inition of ldquoall descendants more closely re-lated to taxon D and E than to taxon Brdquo Aninclusive ingroup reference taxon howevercan manifest the same problems in this cir-cumstance (Fig 12c) If taxon A is denedas ldquoall descendants more closely related totaxon C than to taxon Brdquo and if taxon C isdened on the basis of taxon D a contra-dictory denition results with relocation oftaxon D as shown taxon A now subsumesits original ingroup reference taxon C Suchcontradictory denitions can be avoided ifa nested ingroup reference taxon (or nestedtaxa) are selected such as taxon E its un-

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 4: Sereno, 1999b

332 SYSTEMATIC BIOLOGY VOL 48

FIGURE 2 Crown taxa denitional types and sta-bility (a) A crown clade with resolved basal relation-ships but with unresolved outgroup relationships maybe dened with greatest stability as a node-based taxon(b) A crown clade with unresolved basal relationshipsbut with a stable extinct outgroup may be dened withgreatest stability as a stem-based taxon Crown groupsare encircled dashed lines indicate extinct taxa a dotindicates a node-based denition an arrow indicates astem-based denition

those groups which have in known geological pe-riods undergone much modication should in theolder formations make some slight approach to eachother so that the older members should differ less insome of their characters than do the existing mem-bers of the same groups

About mammals Darwin (1859522) wrote

They [marsupials] appeared in an earlier geologicalperiod and their range was formerly much more ex-tensive than at present Hence the Placentata are gen-erally supposed to have been derived from the Im-

placentata or Marsupials not however from formsclosely resembling the existing Marsupials but fromtheir early progenitors

Most extant biologists also conceive of thetaxonomic content of Mammalia in a simi-lar fashion whether they are evolutionarytaxonomists cladists ecologists or molec-ular biologists (eg Jollie 197376ndash77 Fu-tuyma 1986334 Walker 1975) Higher taxasuch as Mammalia and Aves have been asso-ciated with ldquokeyrdquo characters and their asso-ciated functions Archaeopteryx with feath-ers wings and the capacity for ight willalways be considered a ldquobirdrdquo within Avesby the majority of biologists Altering thisequation for well-established monophyletictaxa is likely to engender confusion

Stem-based denitions are less ldquoconsistentrdquothan node-based denitionsmdashSchander andThollesson (1995264) suggested that stem-based denitions ldquorefer to non-existingclades under some phylogenetic resolu-tionsrdquo They presented an example (Fig 3a)in which relocation of one taxon (taxon A)results in a denition that refers to an ldquoim-possiblerdquo clade Relocation of taxon A cre-ates this internal inconsistency because theoriginal ingroup reference taxon (taxon F)for stem-based taxon Gnow includes by def-inition the original outgroup taxon (taxonD) Schander and Thollesson (1995264) con-cluded that ldquoif a consistent system of taxon-omy is wanted stem-based denitions maybetter be avoidedrdquo

Similar internal ldquoinconsistencyrdquo how-ever can be generated for node-based taxaas well (Fig 3b) In this example relocationof one taxon (taxon A) results in a denitionthat refers to an ldquoimpossiblerdquo or internallyredundant clade Relocation of taxon A cre-ates this internal inconsistency because oneof the original reference taxa (taxon E) fornode-based taxon G now includes the otherreference taxon (taxon F) and the commonancestor Even if taxon E is regarded as a ju-nior synonym of taxon G after relocation oftaxon A the denition of taxon G must bechanged (because taxon E cannot be used asa reference taxon)

Rather than demonstrating any partic-ular structural weakness of stem-baseddenitions these examples (Fig 3) better

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 333

FIGURE 3 Problems associated with maximally in-clusive reference taxa (a) Relocation of taxon A resultsin an ldquoinconsistentrdquo stem-based denition for taxon Gbecause it denes a group that cannot exist under the al-ternative phylogenetic hypothesis (after Schander andThollesson 1995Fig 3) Schander and Thollesson de-ned stem-based taxon G as ldquoF and all taxa sharing amore recent common ancestor with it than with Drdquo andstem-based taxon F as ldquoA and all taxa sharing a morerecent common ancestor with it than with Crdquo TaxonG is ldquoinconsistentrdquo after relocation of taxon A becauseits ingroup reference taxon (taxon F) now includes bydenition the outgroup reference taxon (taxon D) Theldquoinconsistencyrdquo however disappears if a less-inclusiveingroup reference taxon had been chosen for taxon G(ie taxon A or B rather than taxon F) In this casetaxon G exists after relocation of taxon A (either as asynonym of taxon A or as a group including taxa Band C depending on whether taxon A or B was used asthe ingroup reference taxon respectively) (b) Reloca-tion of taxon A results in an ldquoinconsistentrdquo node-baseddenition for taxon G because it denes a group thatcannot existunder thealternativephylogenetichypoth-esis In this example taxon G is node-based and denedas ldquotaxon E taxon F their common ancestor and all

descendantsrdquo taxon E is dened as ldquotaxon A taxon Btheir common ancestor and alldescendantsrdquo and taxonF is node-based and dened as ldquotaxon C taxon D theircommon ancestor and all descendantsrdquo Taxon G is ldquoin-consistentrdquo after relocation of taxon A because one ofits reference taxa (taxon E) now includes by denitionthe other reference taxon (taxon F) and the commonancestor The reference taxa for a node-based taxon arepresumed to be exclusive of one another and their com-mon ancestor after relocation of taxon A these condi-tions are violated for taxon G as originally dened Theldquoinconsistencyrdquo however disappears if a less-inclusivereference taxon had originally been selected for taxon G(ie taxon A or B rather than taxon E) In this case taxonG exists after relocation of taxon A (either unchanged inits taxonomic content or as a less-inclusive taxon thatexcludes taxon B depending on whether taxon B or Awas used as an ingroup reference taxon respectively) Dots indicate node-based denitions and arrows indi-cate stem-based denitions

illustrate problems that arise as a result of se-lecting maximally inclusive reference taxamdashthe most common kind of reference taxacurrently chosen in phylogenetic denitions(see Reference Taxa below) If Schander andThollesson had chosen a more nested in-group reference taxon in their stem-baseddenition of taxon G (ie taxon A or B ratherthan taxon F Fig 3a) relocation of taxonA poses no special problems In this casetaxon G either refers to the same clade mi-nus taxon A (if taxon B is the ingroup ref-erence taxon) or becomes a potential juniorsynonym of taxon A (if taxon A is the in-group reference taxon) Similarly for node-based taxon G (Fig 3b) a more nested refer-ence taxon (taxon A or B rather than taxonE) removes any inconsistency after reloca-tion of taxon A Taxon G refers either to thesame clade (if taxon B is an ingroup refer-ence taxon) or to a less inclusive clade thatexcludes taxon B (if taxon A is an ingroupreference taxon)

Node- or stem-based denitions better re-ect original usage and contentmdashGauthier(198612 13) for example suggested that hisstem-based denition for Haeckelrsquos (1866)taxon Ornithurae (ldquoExtant birds and allother taxa such as Ichthyornis and Hes-perornithes [sic] that are closer to extantbirds than is Archaeopteryxrdquo) was ldquoin keep-ing with its original intentrdquo (Fig 4) Chiappe(1991 1996) on the other hand argued that

334 SYSTEMATIC BIOLOGY VOL 48

Ornithurae is better dened as a less inclu-sive node-based taxon with Hesperornithi-formes and Neornithes (crown-group birds)as reference taxa (Fig 4) Initially Chiappe

FIGURE 4 Cladogram of basal avians showing re-cent use of the higher taxon Ornithurae (Haeckel 1866)The dot indicates a node-based denition and the ar-row indicates a stem-based denition

(1991337) reasoned that a node-based de-nition of Ornithurae was more appropriatebecause it excluded several newly discov-ered basal avians that lacked some of thesynapomorphies in Gauthierrsquos (1986) diag-nosis of Ornithurae Later Chiappe arguedthat a node-based denition of Ornithuraeldquobetter reects the composition of the cladebefore to [sic] the application of phyloge-netic denitionsrdquo (1996205)

Neither Gauthierrsquos nor Chiappersquos deni-tions have historical precedence Haeckel(1866) coined Ornithurae (ldquobird tailrdquo) for thereduced number of vertebrae and uniquecoossication of the distal tail (as a sin-gle bone the pygostyle) that characterizesall extant birds in contrast to the condi-tion in Archaeopteryx Haeckelrsquos taxon owesits existence to the then-recent discovery ofArchaeopteryx a primitive bird with a tailcomprising over 20 vertebrae Contrary toChiappe (1996) Ornithurae predates the de-scription of the extinct short-tailed generaHesperornis (Marsh 1872a) and Ichthyor-nis (Marsh 1872b) or the taxon Hesper-ornithiformes (Furbringer 1888) Ornithu-

rae as noted by Gauthier (1986) has beensupplanted by Neornithes (Gadow 1893)a taxon referring to crown-group aviansBased on original intent an argument couldbe made that Haeckelrsquos Ornithurae shouldsupplant Neornithes on grounds of prior-ity or that it be dened explicitly as anapomorphy-based taxon based on the pres-ence of a pygostyle which is now known tobe lacking in birds other than Archaeopteryx(Forster et al 1998) In either case thedenition would circumscribe clades otherthan those identied by either Gauthier orChiappe

For most taxon names the original de-nition if explicitly stated at all is character-based or taxon-based (a list of included taxa)and lacks a relational phrase about ances-try that would specify potential membershipThe intent of the original author of a taxonto include or exclude unknown or reposi-tioned taxa usually cannot be interpretedunambiguously This problem is not widelyappreciated

APOMORPHY-BASED DEFINITIONS

An apomorphy-based denition speci-es the boundaries of a clade by identi-fying the ldquorst ancestor with a particularsynapomorphyrdquo (de Queiroz and Gauthier1990310) and encompassing all of its de-scendants An apomorphy-based taxon in-cludes all descendants whether or not thesynapomorphy is maintained Apomorphy-based denitions are subject to three prob-lems that are not relevant to node-based andstem-based denitions variation in charac-ter coding optimization ambiguity and ho-moplasy (Bryant 1994 Schander and Thol-lesson 1995 Sereno 1998) These problemshave long been associated with traditionaluse of ldquokeyrdquo characters to dene taxa Thusfar few apomorphy-based denitions havebeen erected and this denitional formshould be avoided

Character ambiguitymdashConsider an apo-morphy-based denition for Aves (includ-ing Archaeopteryx and Ornithurae) based onthe presence of ldquofeathersrdquo the ldquokey char-acterrdquo usually associated with this taxonFirst the character codingmdashldquoFeathers ab-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 335

sent (0) present (1)rdquomdashis only one of sev-eral possibilities A ldquofeatherrdquo is composedof a rachis barbs barbules and other com-ponents that may be better coded as sepa-rate characters The character on which anapomorphy-based denition is based there-fore may be recoded as two or more char-acters with independent character-state dis-tributions Indeed this is an expectation formost qualitative or quantitative charactersthat is once the transformations become bet-ter known intermediate states or multiplecharacters will emerge as recent discoveriesare beginning to reveal with regard to feath-ers (Chen et al 1998 Ji et al 1998) or withregard to the characters historically associ-ated with Mammalia (Rowe 1988 RoweandGauthier 1992 Bryant 1994 de Queiroz1994)

The apomorphy-based denition given asan example by de Queiroz and Gauthier(1990310) is subject on the same groundsto interpretational ambiguity ldquoTetrapodardquois dened as the ldquorst vertebrate to possessdigits (ie hands and feet rather than ns)and all of its descendantsrdquo What exactlyconstitutes a ldquodigitrdquo versus a ldquonrdquo or ldquopad-dlerdquo In fact this is a signicant questiongiven recent fossil discoveries (Daeschlerand Shubin 1998) Can this feature or listof features be observed in extinct formsAnd what happens if we learn that ldquodigitsrdquoevolved rst on the hands and later on thefeet These are common problems in the in-terpretation of character data which shouldnot be extended to taxonomic denitions

Optimization ambiguitymdashApomorphy-based denitions also do not specify a sin-gle ancestor when the chosen apomorphyhas an ambiguous optimization Charac-ters with ambiguous character-state opti-mization are commonplace in systematicsand arise from missing data (lack of preser-vation strong transformation) and homo-plasy (Bryant 1994 Schander and Thol-lesson 1995 Sereno 1998) There may bemany equally parsimonious positions fora particular character state on the shortestcladogram

HomoplasymdashHomoplasy may involve thestable placement of an apomorphy at morethan one node (Bryant 1994) This can create

an impasse for an apomorphy-based deni-tion because the apomorphy might identifymore than one clade Invoking time as an ar-biter (such as ldquothe rst taxon that has rdquo)invites ambiguities associated with tempo-ral origin (such as missing ancestral lin-eages)

PHYLOGENETIC DEFINITIONS TERMINOLOGY

The terminology used in the followingdiscussion of phylogenetic taxonomic def-initions is claried below and tabulated (Ta-bles 1 2) Someof these terms aregiven morespecic meaning than in the current litera-ture others are new or recently introduced(Sereno 1998)

Taxonomic denition and diagnosis are de-ned here much as they have been describedpreviously by phylogenetic systematists (deQueiroz and Gauthier 1990 1992) and theirforebears (Buck and Hull 1966 Farris 1976Ghiselin 1984) (Table 1) De Queiroz andGauthier (1992461) dened phylogeneticdenitions as ldquostatements specifying themeanings of taxon names (words) they arestated in terms of ancestryrdquo The ldquomeaningof a taxon namerdquo in phylogenetic taxonomyconcerns its taxonomic content or mem-bership as delimited by a relational state-ment Taxonomic denitions therefore areregarded here as ldquoextensionalrdquo (ostensive)statements of relationship that specify thetaxonomic content of a taxon (de Queiroz1992)

Taxonomic diagnosis on the other handinvolves the ldquointensionalrdquo characterizationof members of a clade as recognized bythe shared apomorphies present in thecommon ancestor (Buck and Hull 1966Farris 1976 Ghiselin 1984 Rowe 1987de Queiroz and Gauthier 1990 1992 deQueiroz 1992) According to de Queiroz andGauthier (1992461) diagnoses are ldquostate-ments specifying how to determine whethera given species or organism is a representa-tive of the taxon (clade) to which a partic-ular name refers they are most commonlystated in terms of charactersrdquo Strictly speak-ing shared apomorphies (derived-characterstates) rather than characters determine themembership of a particular clade or species

336 SYSTEMATIC BIOLOGY VOL 48

Taxonomic content is used here to refer toall existing and potential taxa included by thephylogenetic denition of a taxon (Table 1)The smallest number of subordinate taxathat can fulll this denition of taxonomiccontent is two a redundant taxon (ie asingle subordinate taxon) would have anidentical phylogenetic denition and thusis uninformative in phylogenetic taxonomy(Farris 1976) The taxonomic content oftaxon C therefore could be ldquotaxon A taxonB their most recent common ancestor andall descendantsrdquo if taxa A and B are denedextensionally to encompass all existing andpotential subordinate taxa currently withintaxon C

A list of included taxa is insufcient to un-ambiguously specify taxonomic content be-cause there will always be potential mem-bers of a taxon that lie outside any list ofincluded taxa whether or not the includedtaxa themselves have phylogenetic deni-tions An exhaustive list of included taxawith stem-based denitions for exampledoes not specify membership for their hy-pothetical immediate common ancestor (anode-based taxon) or more distant knownor hypothetical outgroups (a stem-basedtaxon) Two reference taxa and a relationalstatement about ancestry are necessary andsufcient to specify taxonomic content be-cause they unambiguously specify member-ship for all existing and potential membersof a taxon

A reference taxon is a species or higher-leveltaxon that is used to delimit a phylogeneticdenition (Table 1) For example ldquotaxon Ataxon B their most recent common ancestorand all descendantsrdquo uses A and B as refer-ence taxa for a node-based taxonomic deni-tion Previously cited as ldquoreference pointsrdquo

(Schander and Thollesson 1995) theseimportant components of phylogeneticdenitions have not been distinguishedterminologically

Several kinds of reference taxa can beidentied (Fig 5 Table 2) Ingroup and out-group reference taxa are located within oroutside the taxon they dene respectively(Fig 5a b) An inclusive reference taxon in-cludes several ingroup or outgroup taxaa maximally inclusive reference taxon in-cludes the largest possible ingroup or out-group reference taxon (Fig 5c) Basal andnested reference taxa are less inclusive andas their name suggests are located at thebase or nested within an ingroup or out-group taxon (Fig 5d e)

A crown taxon denotes a living speciesor a clade dened by using living refer-ence taxa whose immediate outgroup isextinct (Figs 6 7 Table 1) The extinct out-group is a necessary component of the def-inition Jefferies (1979449) coined the termldquocrown grouprdquo to replace Hennigrsquos (1969) group which Jefferies dened as ldquothe lat-est common ancestor of all living membersof group 1 plus all descendants of this an-cestor whether these are living or extinctrdquoTheexistence of an immediate outgroup thatis extinct is implied and was clearly presentin the examples given by Hennig and Jef-feries Patterson (1981207) also clearly usedcrown- and stem-groups in this mannerstating that the ldquostem-group contains fos-sil taxa that exhibit some but not all of thecharacters of the crown-grouprdquo Crown- andstem-group terminology has been used ex-clusively for higher taxa bounded by liv-ing species that have extinct outgroups (egde Queiroz and Gauthier 1992 Forey 1992McKenna and Bell 1997) Although it may

TABLE 2 Denitions for the various kinds of reference taxa

Reference taxon type Denition

Ingroup reference taxon included within a taxon by denitionOutgroup reference taxon excluded from a taxon by denitionInclusive most inclusive reference taxon among ingroup or outgroup taxaBasal reference taxon near the basal furcation of a taxonNested reference taxon distant from the basal furcation of a taxonVernacular vernacular name used as a reference taxon (eg ldquobirdsrdquo or ldquoextant birdsrdquo)

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 337

FIGURE 5 Kinds of reference taxa (a) An ingroup reference taxon is included by denition within a taxon anode-based denition is based on two ingroup reference taxa (b) An outgroup reference taxon is excluded bydenition from a taxon a stem-based denition is based on one ingroup and one outgroup reference taxon (c) Aninclusive reference taxon is the most inclusive ingroup or outgroup reference taxon possible inclusive referencetaxa B and C for taxon A allow denitional ambiguity with change in the relationship of a basal taxon (d) Abasal reference taxon is located at or near the base of an ingroup or outgroup taxon basal reference taxa allowdenitional ambiguity when their relationships change (e) A nested reference taxon is remote from the base ofan ingroup or outgroup reference taxon nested reference taxa are unaffected by changes in the relationships ofbasal taxa Reference taxa are encircled a dot indicates a node-based denition an arrow indicates a stem-baseddenition a dashed line indicates a change in position of a basal taxon

FIGURE 6 Crown stem and total taxa (a) Total group composed of a crown group plus a paraphyletic ldquostemgrouprdquo as conceived by Hennig (1969 1983) and Jefferies (1979) (b) Total taxon composed of a crown taxon andmonophyletic stem taxa as used in this article One stem taxon (left) is shown with a node-based denition andthe other (right) with a stem-based denition

338 SYSTEMATIC BIOLOGY VOL 48

FIGURE 7 Crown species or clades (enclosed) musthave an extinct (stem) taxon (dashed) as an immediateoutgroup (a) Crown species and crown clades are de-limited by the presence of an immediate outgroup thatis extinct (b) Often referred to as ldquoextantrdquo taxa crowntaxa may be composed primarily of extinct taxa ldquoLiv-ingrdquo and ldquoextinctrdquo conditions are treated as if they con-stitute plesiomorphic and apomorphic character statesrespectively of an irreversible character ldquostate of be-ingrdquo Crown clades under this conceptualization areplesiomorphy-based taxa bounded by two living taxafor which the immediate outgroup is extinct

be hypothetically true that ldquoevery Recentgroup whatever its position in thehierarchymay be expected to have had a stem-grouprdquo(Patterson 1981207) clades bounded by liv-ing taxa with immediate outgroups that arealso living have not been considered crowntaxa

Some recent denitions of crown taxahowever have set aside Hennigrsquos andJefferiesrsquo formulations as well as generalusage in the literature A less restric-tive meaning has been proposed basedsolely on reference taxa that include liv-

ing representatives De Queiroz and Gau-thier (1992469) dened crown taxa asldquoclades within which both branches ofthe basal dichotomy are represented byextant descendantsrdquo without mention ofthe condition of the immediate outgroupYet all of their examples of crown cladeshave immediate outgroups that are ex-tinct Regarding their craniate taxonomythey stated that each stem-based taxon iscomposed of a ldquocrown taxon plus all ex-tinct taxa more closely related to it thanto any other crown claderdquo (de Queirozand Gauthier 1992475) Lee (19961103)proposed that ldquoa crown clade denitionresults when both taxa implicated in anode-based denition are representedby extant formsrdquo Although the conditionof the immediate outgroup is never men-tioned Lee (19961103) stated that crowntaxon denitions apply to clades thatconsist of ldquothe most recent common an-cestor of all extant forms and all its de-scendantsrdquo (emphasis added) The pres-ence of an immediate extinct outgroup isimplied

The denition of a crown taxon presentedhere in contrast does not specify ingroupstructure (such as the presence or absenceof a basal dichotomy) or denitional type(node- or stem-based) and explicitly statesthat the immediate outgroup(s) is extinct(Fig 7a Table 1) Crown taxa so denedcan be mapped unerringly on a cladogramif (1) ldquolivingrdquo and ldquoextinctrdquo conditions aretreated as primitive and derived character-states respectively of the character ldquocurrentstate of beingrdquo and (2) if this character isregarded as irreversible (ie CaminndashSokalparsimony) The living state in other wordsis always regarded as plesiomorphic Theirreversible status of the character (state ofbeing) is important because it identiescrown groups on cladograms of living andextinct taxa that would have an ambigu-ous status if character-state reversal wereallowed (Fig 7b) This character-state con-ceptualization of the denition of a crowngroup captures the essence of crown taxawhich have always been used to refer tomaximally inclusive taxa dened by extantrepresentatives not simply any ldquoextant-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 339

boundedrdquo taxon Discovery of extinct sistertaxa provides the opportunity to recognizecrown taxa

Crown taxa are often regarded as cladesthat are bounded uniquely by living in-groups The use of living reference taxahowever is most often an arbitrary deci-sion Any extinct ingroup taxon that is mostclosely related to a living reference taxoncould be used to delimit the same clade(Fig 7b) A crown clade is dened uniquelyby living reference taxa only when such taxado not have extinct sister taxa (Fig 7b) afact that is not widely appreciated Crowntaxa in other words do not have any spe-cial biological signicance beyond signal-ing that particular reference taxa have sur-vived to an arbitrary datum (the Recent)Informal reference to crown taxa as ldquolivingrdquoor ldquoextantrdquo taxa therefore can be mislead-ing because a crown taxon may be com-posed predominantly of extinct species asis the case with the crown taxon Crocodylia(Fig 7b Brochu 1997) ldquoExtant crocodil-iansrdquo if taken literally refers to a smallparaphyletic subset of Crocodylia Explicitreference to ldquocrown crocodiliansrdquo or sim-ply ldquoCrocodyliardquo is preferable to use of theterms ldquolivingrdquo or ldquoextantrdquo

A living taxon is also essential to delimita maximally inclusive clade that is entirelyextinctmdasha stem taxon as here dened Astem taxon denotes an extinct species orclade the immediate outgroup of which in-cludes at least one living member (Fig 6b)Many extinct taxa are not stem taxa becausetheir immediate outgroup is itself extinctSuch extinct taxa are members of more in-clusive stem taxa Stem taxa can have eithernode- or stem-based denitions (Fig 6b)Stem taxa and their associated less inclusivecrown taxon make up the taxonomic contentof their respective total taxon

Stem taxa as here dened should notbe confused with ldquostem grouprdquo (Fig 6a)which was rst used in phylogenetic sys-tematics by Hennig (ldquostammgrupperdquo 19691983) Hennigrsquos (198315) vague formulationof the concept specied all extinct speciesldquowhich can be shown probably to belong toa particular monophyletic group of the phy-logenetic system but which are probably no

closer [sic] related to one subgroup amongrecent animals than to anotherrdquo (translatedfrom the German Ax 1987224) Jefferies(1979) claried the term to include all ex-tinct outgroups of a crown taxon (Fig 6a)Other authors have developed more convo-luted denitions that approximate the sameend Wiley (1981217) for example dened aldquostem grouprdquo as ldquoall fossil species of a groupmore primitive than the most primitive Re-cent species of the same grouprdquo Theseldquostem groupsrdquo are paraphyletic and oftencircumscribe the same ldquoancestral groupsrdquoand ldquoevolutionary gradesrdquo that have mud-died phylogenetic thinking since DarwinOnly a few cladists continue to endorse suchconcepts (Smith 1994) Ax (1987224) pro-posed a cumbersome solution to the prob-lem of paraphyletic ldquostem groupsrdquo by divid-ing them into ancestral ldquostem lineagesrdquo andldquorepresentatives of the stem lineagerdquo Thelatter are monophyletic taxa most simply re-ferred to as stem taxa

A total taxon is regarded here as a cladecomposed of a crown taxon plus all ex-tinct outgroups more closely related to itthan to another crown taxon (Fig 3 Ta-ble 1) Hennigrsquos (1969) and Jefferiesrsquo (1979)formulations of the concept are wordy butsimilar in effect because they restricted in-clusion of extinct outgroups to those mostclosely related to a particular crown taxonDe Queiroz and Gauthier (1992470) deneda total clade as a ldquomore inclusive clade con-sisting of thecrown clade plus its extinctout-groupsrdquo This denition is incomplete be-cause it does not specify which extinct out-groups are included A total taxon explicitlyincludes only those extinct outgroups mostclosely related to a particular crown taxon

Node-based and stem-based denitions areconstructed by (1) identifying reference taxaand (2) attaching to them a relational phrasethat either includes their most recent com-mon ancestor and all descendants (node-based) or limits inclusion to all descen-dants closer to one or more of the referencetaxa (stem-based) (de Queiroz and Gauthier1992) (Fig 1 Table 1) In node-based deni-tions the phrase ldquoleast-inclusive claderdquo canreplace ldquocommon ancestor and all descen-dantsrdquo for cladists sensitive to the allusion to

340 SYSTEMATIC BIOLOGY VOL 48

ancestors (Lee 1998) Listing included taxawithout a relational phrase about ancestryor simply designating taxa as node- or stem-based without reference taxa does not con-stitutea phylogenetic denition because theboundaries of the taxon in question are notspecied (de Queiroz and Gauthier 1990Bryant 1996)

In the current literature there exists someconfusion over what is necessary and suf-cient for node- and stem-based denitionsBryant (1996185) suggested that a more in-clusive taxon be inserted within a stem-based denition such as ldquoall members oftaxon A thataremoreclosely related to taxonB than to taxon Crdquo Including taxon A how-ever is not necessary and may lead to inter-nal conicts within the denition (eg whentaxon A is included within taxon B or C onan alternativephylogeny) The goalof a phy-logenetic denition should be to unambigu-ously identify a specic common ancestorgiven any arrangement of taxa rather thanto identify such an ancestor only on a subsetof phylogenetic hypotheses

Meng et al (1994) and Wyss and Meng(1996) introduced a ldquostem-modied node-based denitionrdquo to effectively unite crown-group rodents whose basal relationshipsare uncertain Their preferred denition ofRodentia (ldquothe clade stemming from themost recent common ancestor of Mus andall Recent mammals more closely relatedto Mus than to Lagomorpha or membersof any other eutherian lsquoorderrdquorsquo Wyss andMeng 1996562) is better interpreted as aspecial stem-based rather than node-baseddenition because it uses only one ingroupreference taxon (Mus) and one or more out-group reference taxa (eg Lagomorpha)Their denition merely adds a phrase (ldquotheclade stemming from the most recent com-mon ancestorrdquo) to an otherwise typicalstem-based denition based on living ref-erence taxa That phrase ensures inclusionof all extinct taxa within the crown cladeBryantrsquos (1996185) reformulation of thisdenition on the other hand is clearly anode-based denition because he speci-ed only ingroup reference taxa (ldquothe mostrecent common ancestor of the species re-ferred to Rodentia by Wilson and Reeder

(1993) and all of its descendantsrdquo) These ex-amples underscore the need for claricationof terms (Tables 1 2)

A node-stem triplet (NST) is a denitionalconguration involving three taxa a node-based taxon composed of two stem-basedsubordinate taxa (Sereno 1997 1998 Fig 8)

Taxon A Taxa b and c their most recentcommon ancestor and all its descendants

Taxon B All taxa closer to taxon b than totaxon c

Taxon C All taxa closer to taxon c than totaxon b

FIGURE 8 Denitional triumvirate termed a node-stem triplet composed of a node-based taxon (A) andtwo subordinate stem-based taxa (B and C) which in-corporates added or repositioned taxa (dashed lineswith daggers) without changing the relative taxonomiccontent of taxon A B or C (as expressed by the equiva-lence statementA = B +C) A dot indicates a node-baseddenition an arrow indicates a stem-based denition

Taxa b and c (not shown in Fig 8) constitutereference taxa within taxon B and C respec-tively Although not essential the same ref-erence taxa (b c) may be used for the trio oftaxa in a NST which then specify complemen-tary denitions The pair of stem-based sis-ter taxa in addition may have reciprocal def-initions that utilize the same reference taxain antipodal positions The word ldquotaxardquo inthe stem-based denitions listed above mayoptionally be replaced by reference toa moreinclusive taxon (Bryant 1996) A stem-baseddenition for Saurischia for example mightread ldquoAll dinosaurs more closely related

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 341

to Neornithes than to Triceratopsrdquo (Sereno1998)

Taxonomic equivalence is a statement (orequation) in which one taxon is shownto be equivalent in current and poten-tial taxonomic content to subordinate taxaplus their common ancestor (Table 1)The NST described above is composedof a node-based taxon (A) and twosubordinate stem-based taxa (B C) andcan be written as the equivalence state-ment ldquoA = B + Crdquo because taxon Aby denition is equivalent to taxa Band C plus their most recent commonancestor

PHYLOGENETIC DEFINITIONS RATIONALE

The rst step toward a phylogenetic tax-onomy involved the distinction betweendenition and diagnosis (Buck and Hull1966 Grifths 1973 1974 Farris 1976 Ghis-elin 1984 Rowe 1987) A second step in-volved the formulation of node- and stem-based denitions by use of reference taxaand relational statements about ancestry (deQueiroz and Gauthier 1992) A third stepoutlined here involves the specication ofguidelines for selecting effective referencetaxa and criteria for positioning node- andstem-based denitions

Phylogenetic rearrangements and introduc-tion of new taxa are the principal factorsthat destabilize taxonomic content Stabil-ity of taxonomic content is enhanced bychoosing (1) effective reference taxa and(2) an effective conguration of phyloge-netic denitions

Reference TaxaPositionmdashGiven a particular phyloge-

netic hypothesis the taxonomic content of aphylogenetic denition is altered only if thereference taxa specify an alternative com-mon ancestor Given a basal dichotomy withone reference taxon on each side the iden-tication of the immediate common ances-tor is not affected by relocation of a ref-erence taxon on its side of the basal di-chotomy (Fig 9 move 1) Relocation ofa reference taxon to the opposing side orto a location outside the basal dichotomyidenties a different common ancestor

(Fig 9 moves 2 3) For a taxon with a basaldichotomy an alternative common ances-tor can be identied only if one referencetaxon is relocated to the opposing side oroutside the basal dichotomy Are some ref-erence taxa more likely to be relocated in thismanner than others

FIGURE 9 Three possible outcomes from relocationof a reference taxon 1 = relocation on the same side ofa basal dichotomy has no effect 2 = relocation to theopposite side of the basal dichotomy identies a lessinclusive common ancestor 3 = relocation outside thebasal dichotomy identies a more inclusive commonancestor Reference taxa are encircled and the originalbasal dichotomy is shown with heavy lines

First a taxon positioned near the basaldichotomy of a cladogram is always eas-ier to relocate to the opposing side (ierequires fewer additional steps) than is anested taxon if all other factors are regardedas equal among ingroup taxa (eg complete-ness missing data and homoplasy) Unlikea basal taxon (Fig 5d) a nested taxon sharessynapomorphies that increase its phyloge-netic (patristic) distance from the commonancestor (Fig 5e) Thus stability of taxo-nomic content is enhanced in phylogeneticdenitions if nested reference taxa are cho-sen that are located at some distance (atleast several nodes away) from the basaldichotomy This distance is easy to assessquantitatively for various nested taxa on agiven phylogeny

Second maximally inclusive referencetaxa (Fig 5c) have greater potential to cre-ate unnecessary taxonomic redundancy andinternal inconsistencies after relocation oftaxa (Fig 3) This is true for both node-

342 SYSTEMATIC BIOLOGY VOL 48

and stem-based taxa In the current litera-ture maximally inclusive reference taxa areoften used in phylogenetic denitions Intheir tetrapod classication for example deQueiroz and Gauthier (1992475) used suc-cessive maximally inclusive reference taxafor all denitionsmdasha pattern here termed re-cursive referencing (Fig 10a) Recursive refer-encing can have undesirable consequencesbecause it shifts the burden of a more precisedenition to a slightly less inclusive highertaxon This higher taxon may not be de-ned or it too may be affected by an alter-native conguration of basal relationshipsSalgado et al (1997) for example choseProsauropoda and Sauropoda as maximallyinclusive reference taxa for Sauropodomor-pha If a basal sauropodomorph is reposi-tioned the denitions of Prosauropoda andSauropoda must be determined to assess po-tential taxonomic consequences HoweverSalgado et al (1997) used Prosauropodaas a terminal taxon without denitionSauropoda was dened as a node-basedtaxon on the basis of two maximally in-clusive reference taxa Vulcanodon and Eu-sauropoda (all other sauropods) Thusif the incompletely known basal sauro-pod Vulcanodon is reinterpreted as be-ing more closely related to prosauropodsSauropoda and Sauropodomorpha wouldbecome synonyms

If on the other hand Sauropodomorphais dened on the basis of the nested ref-erence taxamdashthe prosauropod Plateosaurusand sauropod Saltasaurus (Sereno 1998)mdasha change in the phylogenetic positionof any other prosauropod or sauropodhas no effect on the taxonomic con-tents of Sauropodomorpha or SauropodaSauropodomorpha can be dened as ldquoPla-teosaurus Saltasaurus their most recentcommon ancestor and all descendantsrdquoand Sauropoda can be dened as ldquoallsauropodomorphs more closely related toSaltasaurus than to Plateosaurusrdquo The phylo-genetic position of Vulcanodon has no effecton the identication of respective commonancestors

CompletenessmdashPoorly known taxa aremore likely to be phylogenetically unsta-ble than well-known taxa (Wilkinson 1995)

FIGURE 10 Referencing styles (a) Recursive refer-encing selects a succession of the most inclusive ref-erence taxa (b) Nested referencing selects that sameremote reference taxon Reference taxa are encircled

Taxa with numerous missing entries in aphylogenetic analysis usually reduce phylo-genetic resolution by generating numerousequally parsimonious trees Well-known ref-erence taxa even if they are somewhat lessnested are preferable to poorly known taxa(Fig 11) A single poorly known basal taxonis least stable and therefore least desirableas a reference taxon

Use with Recent or extinct ltersmdashldquoRecentrdquo(ldquoextantrdquo or ldquolivingrdquo) or ldquoextinctrdquo may beused in phylogenetic denitions of crownand stem taxa respectively to help to stabi-lize taxonomic content in the face of poorlyresolved basal relationships For a crowntaxon with a stem-based denition a Recentlter can restrict the boundaries of a cladeto living taxa (Meng et al 1994 Wyss andMeng 1996) Rodentia for example could

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 343

be dened as ldquothe least inclusive clade com-posed of Recent mammals more closely re-lated to Mus than to Lepusrdquo For stem taxain contrast a stem-based denition using anldquoextinctrdquo lter can restrict inclusion to ex-tinct taxa when outgroup relationships areuncertain Moas for example could be de-ned as ldquoall extinct species more closelyrelated to Dinornis than to Struthiordquo theostrich

NumbermdashThe number of reference taxathat have been used in phylogenetic deni-tions has varied from two to innity Gau-thier (198614) for example dened Aves asa node-based taxon including ldquoall descen-

FIGURE 11 Nestedness and completeness in the se-lection of reference taxa A nested well-known taxonis more likely to constitute a stable reference taxon (en-circled)

dants of the most recent common ancestorof Ratitae Tinami and Neognathaerdquo Threereference taxa rather than two were chosenpresumably because of unstated uncertaintyin the phylogenetic relationships amongthese three avian subgroups De Queirozand Gauthier (1990310 1992461) suggestedthat phylogenetic denitions should utilizeonly two reference taxamdashbut then erecteddenitions for total taxa that used as refer-ence taxa one crown clade plus ldquoall extinct

taxa more closely related to it than to anyother crown claderdquo (de Queiroz and Gau-thier 1992475)

Listing more than a pair of referencetaxa does not seem to compromise a node-based phylogenetic denition in any dis-cernible way Node-based denitions areconstructed with what may be termed in-ternal inclusion A contradictory relation-ship between multiple ingroup referencetaxa for a node-based denition is im-possible to construct with internal inclu-sion because a clade always exists thatincludes all ingroup reference taxa Multi-ple ingroup reference taxa effectively stabi-lize the taxonomic content of a node-basedtaxon if basal relationships are uncertain(Fig 12a)

For stem-based denitions multiple out-group reference taxa effectively stabilize thetaxonomic content of a taxon if outgroup re-lationships are uncertain (Fig 12b) Multi-ple ingroup reference taxa for stem-baseddenitions can create a contradictory sit-uation if one (or more) ingroup referencetaxa are repositioned closer to the outgroupreference taxon (Fig 12c) or to a positionoutside the clade as originally dened Inthis case the external inclusion fundamen-tal to a stem-based denition creates a con-tradictory relationship among the originalingroup reference taxa which are no longermore closely related to each other than to theoutgroup reference taxon (Fig 12c) In theexample shown here relocation of ingroupreference taxon D creates a contradictory re-lationship with a second ingroup referencetaxon E if taxon A has the stem-based def-inition of ldquoall descendants more closely re-lated to taxon D and E than to taxon Brdquo Aninclusive ingroup reference taxon howevercan manifest the same problems in this cir-cumstance (Fig 12c) If taxon A is denedas ldquoall descendants more closely related totaxon C than to taxon Brdquo and if taxon C isdened on the basis of taxon D a contra-dictory denition results with relocation oftaxon D as shown taxon A now subsumesits original ingroup reference taxon C Suchcontradictory denitions can be avoided ifa nested ingroup reference taxon (or nestedtaxa) are selected such as taxon E its un-

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 5: Sereno, 1999b

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 333

FIGURE 3 Problems associated with maximally in-clusive reference taxa (a) Relocation of taxon A resultsin an ldquoinconsistentrdquo stem-based denition for taxon Gbecause it denes a group that cannot exist under the al-ternative phylogenetic hypothesis (after Schander andThollesson 1995Fig 3) Schander and Thollesson de-ned stem-based taxon G as ldquoF and all taxa sharing amore recent common ancestor with it than with Drdquo andstem-based taxon F as ldquoA and all taxa sharing a morerecent common ancestor with it than with Crdquo TaxonG is ldquoinconsistentrdquo after relocation of taxon A becauseits ingroup reference taxon (taxon F) now includes bydenition the outgroup reference taxon (taxon D) Theldquoinconsistencyrdquo however disappears if a less-inclusiveingroup reference taxon had been chosen for taxon G(ie taxon A or B rather than taxon F) In this casetaxon G exists after relocation of taxon A (either as asynonym of taxon A or as a group including taxa Band C depending on whether taxon A or B was used asthe ingroup reference taxon respectively) (b) Reloca-tion of taxon A results in an ldquoinconsistentrdquo node-baseddenition for taxon G because it denes a group thatcannot existunder thealternativephylogenetichypoth-esis In this example taxon G is node-based and denedas ldquotaxon E taxon F their common ancestor and all

descendantsrdquo taxon E is dened as ldquotaxon A taxon Btheir common ancestor and alldescendantsrdquo and taxonF is node-based and dened as ldquotaxon C taxon D theircommon ancestor and all descendantsrdquo Taxon G is ldquoin-consistentrdquo after relocation of taxon A because one ofits reference taxa (taxon E) now includes by denitionthe other reference taxon (taxon F) and the commonancestor The reference taxa for a node-based taxon arepresumed to be exclusive of one another and their com-mon ancestor after relocation of taxon A these condi-tions are violated for taxon G as originally dened Theldquoinconsistencyrdquo however disappears if a less-inclusivereference taxon had originally been selected for taxon G(ie taxon A or B rather than taxon E) In this case taxonG exists after relocation of taxon A (either unchanged inits taxonomic content or as a less-inclusive taxon thatexcludes taxon B depending on whether taxon B or Awas used as an ingroup reference taxon respectively) Dots indicate node-based denitions and arrows indi-cate stem-based denitions

illustrate problems that arise as a result of se-lecting maximally inclusive reference taxamdashthe most common kind of reference taxacurrently chosen in phylogenetic denitions(see Reference Taxa below) If Schander andThollesson had chosen a more nested in-group reference taxon in their stem-baseddenition of taxon G (ie taxon A or B ratherthan taxon F Fig 3a) relocation of taxonA poses no special problems In this casetaxon G either refers to the same clade mi-nus taxon A (if taxon B is the ingroup ref-erence taxon) or becomes a potential juniorsynonym of taxon A (if taxon A is the in-group reference taxon) Similarly for node-based taxon G (Fig 3b) a more nested refer-ence taxon (taxon A or B rather than taxonE) removes any inconsistency after reloca-tion of taxon A Taxon G refers either to thesame clade (if taxon B is an ingroup refer-ence taxon) or to a less inclusive clade thatexcludes taxon B (if taxon A is an ingroupreference taxon)

Node- or stem-based denitions better re-ect original usage and contentmdashGauthier(198612 13) for example suggested that hisstem-based denition for Haeckelrsquos (1866)taxon Ornithurae (ldquoExtant birds and allother taxa such as Ichthyornis and Hes-perornithes [sic] that are closer to extantbirds than is Archaeopteryxrdquo) was ldquoin keep-ing with its original intentrdquo (Fig 4) Chiappe(1991 1996) on the other hand argued that

334 SYSTEMATIC BIOLOGY VOL 48

Ornithurae is better dened as a less inclu-sive node-based taxon with Hesperornithi-formes and Neornithes (crown-group birds)as reference taxa (Fig 4) Initially Chiappe

FIGURE 4 Cladogram of basal avians showing re-cent use of the higher taxon Ornithurae (Haeckel 1866)The dot indicates a node-based denition and the ar-row indicates a stem-based denition

(1991337) reasoned that a node-based de-nition of Ornithurae was more appropriatebecause it excluded several newly discov-ered basal avians that lacked some of thesynapomorphies in Gauthierrsquos (1986) diag-nosis of Ornithurae Later Chiappe arguedthat a node-based denition of Ornithuraeldquobetter reects the composition of the cladebefore to [sic] the application of phyloge-netic denitionsrdquo (1996205)

Neither Gauthierrsquos nor Chiappersquos deni-tions have historical precedence Haeckel(1866) coined Ornithurae (ldquobird tailrdquo) for thereduced number of vertebrae and uniquecoossication of the distal tail (as a sin-gle bone the pygostyle) that characterizesall extant birds in contrast to the condi-tion in Archaeopteryx Haeckelrsquos taxon owesits existence to the then-recent discovery ofArchaeopteryx a primitive bird with a tailcomprising over 20 vertebrae Contrary toChiappe (1996) Ornithurae predates the de-scription of the extinct short-tailed generaHesperornis (Marsh 1872a) and Ichthyor-nis (Marsh 1872b) or the taxon Hesper-ornithiformes (Furbringer 1888) Ornithu-

rae as noted by Gauthier (1986) has beensupplanted by Neornithes (Gadow 1893)a taxon referring to crown-group aviansBased on original intent an argument couldbe made that Haeckelrsquos Ornithurae shouldsupplant Neornithes on grounds of prior-ity or that it be dened explicitly as anapomorphy-based taxon based on the pres-ence of a pygostyle which is now known tobe lacking in birds other than Archaeopteryx(Forster et al 1998) In either case thedenition would circumscribe clades otherthan those identied by either Gauthier orChiappe

For most taxon names the original de-nition if explicitly stated at all is character-based or taxon-based (a list of included taxa)and lacks a relational phrase about ances-try that would specify potential membershipThe intent of the original author of a taxonto include or exclude unknown or reposi-tioned taxa usually cannot be interpretedunambiguously This problem is not widelyappreciated

APOMORPHY-BASED DEFINITIONS

An apomorphy-based denition speci-es the boundaries of a clade by identi-fying the ldquorst ancestor with a particularsynapomorphyrdquo (de Queiroz and Gauthier1990310) and encompassing all of its de-scendants An apomorphy-based taxon in-cludes all descendants whether or not thesynapomorphy is maintained Apomorphy-based denitions are subject to three prob-lems that are not relevant to node-based andstem-based denitions variation in charac-ter coding optimization ambiguity and ho-moplasy (Bryant 1994 Schander and Thol-lesson 1995 Sereno 1998) These problemshave long been associated with traditionaluse of ldquokeyrdquo characters to dene taxa Thusfar few apomorphy-based denitions havebeen erected and this denitional formshould be avoided

Character ambiguitymdashConsider an apo-morphy-based denition for Aves (includ-ing Archaeopteryx and Ornithurae) based onthe presence of ldquofeathersrdquo the ldquokey char-acterrdquo usually associated with this taxonFirst the character codingmdashldquoFeathers ab-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 335

sent (0) present (1)rdquomdashis only one of sev-eral possibilities A ldquofeatherrdquo is composedof a rachis barbs barbules and other com-ponents that may be better coded as sepa-rate characters The character on which anapomorphy-based denition is based there-fore may be recoded as two or more char-acters with independent character-state dis-tributions Indeed this is an expectation formost qualitative or quantitative charactersthat is once the transformations become bet-ter known intermediate states or multiplecharacters will emerge as recent discoveriesare beginning to reveal with regard to feath-ers (Chen et al 1998 Ji et al 1998) or withregard to the characters historically associ-ated with Mammalia (Rowe 1988 RoweandGauthier 1992 Bryant 1994 de Queiroz1994)

The apomorphy-based denition given asan example by de Queiroz and Gauthier(1990310) is subject on the same groundsto interpretational ambiguity ldquoTetrapodardquois dened as the ldquorst vertebrate to possessdigits (ie hands and feet rather than ns)and all of its descendantsrdquo What exactlyconstitutes a ldquodigitrdquo versus a ldquonrdquo or ldquopad-dlerdquo In fact this is a signicant questiongiven recent fossil discoveries (Daeschlerand Shubin 1998) Can this feature or listof features be observed in extinct formsAnd what happens if we learn that ldquodigitsrdquoevolved rst on the hands and later on thefeet These are common problems in the in-terpretation of character data which shouldnot be extended to taxonomic denitions

Optimization ambiguitymdashApomorphy-based denitions also do not specify a sin-gle ancestor when the chosen apomorphyhas an ambiguous optimization Charac-ters with ambiguous character-state opti-mization are commonplace in systematicsand arise from missing data (lack of preser-vation strong transformation) and homo-plasy (Bryant 1994 Schander and Thol-lesson 1995 Sereno 1998) There may bemany equally parsimonious positions fora particular character state on the shortestcladogram

HomoplasymdashHomoplasy may involve thestable placement of an apomorphy at morethan one node (Bryant 1994) This can create

an impasse for an apomorphy-based deni-tion because the apomorphy might identifymore than one clade Invoking time as an ar-biter (such as ldquothe rst taxon that has rdquo)invites ambiguities associated with tempo-ral origin (such as missing ancestral lin-eages)

PHYLOGENETIC DEFINITIONS TERMINOLOGY

The terminology used in the followingdiscussion of phylogenetic taxonomic def-initions is claried below and tabulated (Ta-bles 1 2) Someof these terms aregiven morespecic meaning than in the current litera-ture others are new or recently introduced(Sereno 1998)

Taxonomic denition and diagnosis are de-ned here much as they have been describedpreviously by phylogenetic systematists (deQueiroz and Gauthier 1990 1992) and theirforebears (Buck and Hull 1966 Farris 1976Ghiselin 1984) (Table 1) De Queiroz andGauthier (1992461) dened phylogeneticdenitions as ldquostatements specifying themeanings of taxon names (words) they arestated in terms of ancestryrdquo The ldquomeaningof a taxon namerdquo in phylogenetic taxonomyconcerns its taxonomic content or mem-bership as delimited by a relational state-ment Taxonomic denitions therefore areregarded here as ldquoextensionalrdquo (ostensive)statements of relationship that specify thetaxonomic content of a taxon (de Queiroz1992)

Taxonomic diagnosis on the other handinvolves the ldquointensionalrdquo characterizationof members of a clade as recognized bythe shared apomorphies present in thecommon ancestor (Buck and Hull 1966Farris 1976 Ghiselin 1984 Rowe 1987de Queiroz and Gauthier 1990 1992 deQueiroz 1992) According to de Queiroz andGauthier (1992461) diagnoses are ldquostate-ments specifying how to determine whethera given species or organism is a representa-tive of the taxon (clade) to which a partic-ular name refers they are most commonlystated in terms of charactersrdquo Strictly speak-ing shared apomorphies (derived-characterstates) rather than characters determine themembership of a particular clade or species

336 SYSTEMATIC BIOLOGY VOL 48

Taxonomic content is used here to refer toall existing and potential taxa included by thephylogenetic denition of a taxon (Table 1)The smallest number of subordinate taxathat can fulll this denition of taxonomiccontent is two a redundant taxon (ie asingle subordinate taxon) would have anidentical phylogenetic denition and thusis uninformative in phylogenetic taxonomy(Farris 1976) The taxonomic content oftaxon C therefore could be ldquotaxon A taxonB their most recent common ancestor andall descendantsrdquo if taxa A and B are denedextensionally to encompass all existing andpotential subordinate taxa currently withintaxon C

A list of included taxa is insufcient to un-ambiguously specify taxonomic content be-cause there will always be potential mem-bers of a taxon that lie outside any list ofincluded taxa whether or not the includedtaxa themselves have phylogenetic deni-tions An exhaustive list of included taxawith stem-based denitions for exampledoes not specify membership for their hy-pothetical immediate common ancestor (anode-based taxon) or more distant knownor hypothetical outgroups (a stem-basedtaxon) Two reference taxa and a relationalstatement about ancestry are necessary andsufcient to specify taxonomic content be-cause they unambiguously specify member-ship for all existing and potential membersof a taxon

A reference taxon is a species or higher-leveltaxon that is used to delimit a phylogeneticdenition (Table 1) For example ldquotaxon Ataxon B their most recent common ancestorand all descendantsrdquo uses A and B as refer-ence taxa for a node-based taxonomic deni-tion Previously cited as ldquoreference pointsrdquo

(Schander and Thollesson 1995) theseimportant components of phylogeneticdenitions have not been distinguishedterminologically

Several kinds of reference taxa can beidentied (Fig 5 Table 2) Ingroup and out-group reference taxa are located within oroutside the taxon they dene respectively(Fig 5a b) An inclusive reference taxon in-cludes several ingroup or outgroup taxaa maximally inclusive reference taxon in-cludes the largest possible ingroup or out-group reference taxon (Fig 5c) Basal andnested reference taxa are less inclusive andas their name suggests are located at thebase or nested within an ingroup or out-group taxon (Fig 5d e)

A crown taxon denotes a living speciesor a clade dened by using living refer-ence taxa whose immediate outgroup isextinct (Figs 6 7 Table 1) The extinct out-group is a necessary component of the def-inition Jefferies (1979449) coined the termldquocrown grouprdquo to replace Hennigrsquos (1969) group which Jefferies dened as ldquothe lat-est common ancestor of all living membersof group 1 plus all descendants of this an-cestor whether these are living or extinctrdquoTheexistence of an immediate outgroup thatis extinct is implied and was clearly presentin the examples given by Hennig and Jef-feries Patterson (1981207) also clearly usedcrown- and stem-groups in this mannerstating that the ldquostem-group contains fos-sil taxa that exhibit some but not all of thecharacters of the crown-grouprdquo Crown- andstem-group terminology has been used ex-clusively for higher taxa bounded by liv-ing species that have extinct outgroups (egde Queiroz and Gauthier 1992 Forey 1992McKenna and Bell 1997) Although it may

TABLE 2 Denitions for the various kinds of reference taxa

Reference taxon type Denition

Ingroup reference taxon included within a taxon by denitionOutgroup reference taxon excluded from a taxon by denitionInclusive most inclusive reference taxon among ingroup or outgroup taxaBasal reference taxon near the basal furcation of a taxonNested reference taxon distant from the basal furcation of a taxonVernacular vernacular name used as a reference taxon (eg ldquobirdsrdquo or ldquoextant birdsrdquo)

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 337

FIGURE 5 Kinds of reference taxa (a) An ingroup reference taxon is included by denition within a taxon anode-based denition is based on two ingroup reference taxa (b) An outgroup reference taxon is excluded bydenition from a taxon a stem-based denition is based on one ingroup and one outgroup reference taxon (c) Aninclusive reference taxon is the most inclusive ingroup or outgroup reference taxon possible inclusive referencetaxa B and C for taxon A allow denitional ambiguity with change in the relationship of a basal taxon (d) Abasal reference taxon is located at or near the base of an ingroup or outgroup taxon basal reference taxa allowdenitional ambiguity when their relationships change (e) A nested reference taxon is remote from the base ofan ingroup or outgroup reference taxon nested reference taxa are unaffected by changes in the relationships ofbasal taxa Reference taxa are encircled a dot indicates a node-based denition an arrow indicates a stem-baseddenition a dashed line indicates a change in position of a basal taxon

FIGURE 6 Crown stem and total taxa (a) Total group composed of a crown group plus a paraphyletic ldquostemgrouprdquo as conceived by Hennig (1969 1983) and Jefferies (1979) (b) Total taxon composed of a crown taxon andmonophyletic stem taxa as used in this article One stem taxon (left) is shown with a node-based denition andthe other (right) with a stem-based denition

338 SYSTEMATIC BIOLOGY VOL 48

FIGURE 7 Crown species or clades (enclosed) musthave an extinct (stem) taxon (dashed) as an immediateoutgroup (a) Crown species and crown clades are de-limited by the presence of an immediate outgroup thatis extinct (b) Often referred to as ldquoextantrdquo taxa crowntaxa may be composed primarily of extinct taxa ldquoLiv-ingrdquo and ldquoextinctrdquo conditions are treated as if they con-stitute plesiomorphic and apomorphic character statesrespectively of an irreversible character ldquostate of be-ingrdquo Crown clades under this conceptualization areplesiomorphy-based taxa bounded by two living taxafor which the immediate outgroup is extinct

be hypothetically true that ldquoevery Recentgroup whatever its position in thehierarchymay be expected to have had a stem-grouprdquo(Patterson 1981207) clades bounded by liv-ing taxa with immediate outgroups that arealso living have not been considered crowntaxa

Some recent denitions of crown taxahowever have set aside Hennigrsquos andJefferiesrsquo formulations as well as generalusage in the literature A less restric-tive meaning has been proposed basedsolely on reference taxa that include liv-

ing representatives De Queiroz and Gau-thier (1992469) dened crown taxa asldquoclades within which both branches ofthe basal dichotomy are represented byextant descendantsrdquo without mention ofthe condition of the immediate outgroupYet all of their examples of crown cladeshave immediate outgroups that are ex-tinct Regarding their craniate taxonomythey stated that each stem-based taxon iscomposed of a ldquocrown taxon plus all ex-tinct taxa more closely related to it thanto any other crown claderdquo (de Queirozand Gauthier 1992475) Lee (19961103)proposed that ldquoa crown clade denitionresults when both taxa implicated in anode-based denition are representedby extant formsrdquo Although the conditionof the immediate outgroup is never men-tioned Lee (19961103) stated that crowntaxon denitions apply to clades thatconsist of ldquothe most recent common an-cestor of all extant forms and all its de-scendantsrdquo (emphasis added) The pres-ence of an immediate extinct outgroup isimplied

The denition of a crown taxon presentedhere in contrast does not specify ingroupstructure (such as the presence or absenceof a basal dichotomy) or denitional type(node- or stem-based) and explicitly statesthat the immediate outgroup(s) is extinct(Fig 7a Table 1) Crown taxa so denedcan be mapped unerringly on a cladogramif (1) ldquolivingrdquo and ldquoextinctrdquo conditions aretreated as primitive and derived character-states respectively of the character ldquocurrentstate of beingrdquo and (2) if this character isregarded as irreversible (ie CaminndashSokalparsimony) The living state in other wordsis always regarded as plesiomorphic Theirreversible status of the character (state ofbeing) is important because it identiescrown groups on cladograms of living andextinct taxa that would have an ambigu-ous status if character-state reversal wereallowed (Fig 7b) This character-state con-ceptualization of the denition of a crowngroup captures the essence of crown taxawhich have always been used to refer tomaximally inclusive taxa dened by extantrepresentatives not simply any ldquoextant-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 339

boundedrdquo taxon Discovery of extinct sistertaxa provides the opportunity to recognizecrown taxa

Crown taxa are often regarded as cladesthat are bounded uniquely by living in-groups The use of living reference taxahowever is most often an arbitrary deci-sion Any extinct ingroup taxon that is mostclosely related to a living reference taxoncould be used to delimit the same clade(Fig 7b) A crown clade is dened uniquelyby living reference taxa only when such taxado not have extinct sister taxa (Fig 7b) afact that is not widely appreciated Crowntaxa in other words do not have any spe-cial biological signicance beyond signal-ing that particular reference taxa have sur-vived to an arbitrary datum (the Recent)Informal reference to crown taxa as ldquolivingrdquoor ldquoextantrdquo taxa therefore can be mislead-ing because a crown taxon may be com-posed predominantly of extinct species asis the case with the crown taxon Crocodylia(Fig 7b Brochu 1997) ldquoExtant crocodil-iansrdquo if taken literally refers to a smallparaphyletic subset of Crocodylia Explicitreference to ldquocrown crocodiliansrdquo or sim-ply ldquoCrocodyliardquo is preferable to use of theterms ldquolivingrdquo or ldquoextantrdquo

A living taxon is also essential to delimita maximally inclusive clade that is entirelyextinctmdasha stem taxon as here dened Astem taxon denotes an extinct species orclade the immediate outgroup of which in-cludes at least one living member (Fig 6b)Many extinct taxa are not stem taxa becausetheir immediate outgroup is itself extinctSuch extinct taxa are members of more in-clusive stem taxa Stem taxa can have eithernode- or stem-based denitions (Fig 6b)Stem taxa and their associated less inclusivecrown taxon make up the taxonomic contentof their respective total taxon

Stem taxa as here dened should notbe confused with ldquostem grouprdquo (Fig 6a)which was rst used in phylogenetic sys-tematics by Hennig (ldquostammgrupperdquo 19691983) Hennigrsquos (198315) vague formulationof the concept specied all extinct speciesldquowhich can be shown probably to belong toa particular monophyletic group of the phy-logenetic system but which are probably no

closer [sic] related to one subgroup amongrecent animals than to anotherrdquo (translatedfrom the German Ax 1987224) Jefferies(1979) claried the term to include all ex-tinct outgroups of a crown taxon (Fig 6a)Other authors have developed more convo-luted denitions that approximate the sameend Wiley (1981217) for example dened aldquostem grouprdquo as ldquoall fossil species of a groupmore primitive than the most primitive Re-cent species of the same grouprdquo Theseldquostem groupsrdquo are paraphyletic and oftencircumscribe the same ldquoancestral groupsrdquoand ldquoevolutionary gradesrdquo that have mud-died phylogenetic thinking since DarwinOnly a few cladists continue to endorse suchconcepts (Smith 1994) Ax (1987224) pro-posed a cumbersome solution to the prob-lem of paraphyletic ldquostem groupsrdquo by divid-ing them into ancestral ldquostem lineagesrdquo andldquorepresentatives of the stem lineagerdquo Thelatter are monophyletic taxa most simply re-ferred to as stem taxa

A total taxon is regarded here as a cladecomposed of a crown taxon plus all ex-tinct outgroups more closely related to itthan to another crown taxon (Fig 3 Ta-ble 1) Hennigrsquos (1969) and Jefferiesrsquo (1979)formulations of the concept are wordy butsimilar in effect because they restricted in-clusion of extinct outgroups to those mostclosely related to a particular crown taxonDe Queiroz and Gauthier (1992470) deneda total clade as a ldquomore inclusive clade con-sisting of thecrown clade plus its extinctout-groupsrdquo This denition is incomplete be-cause it does not specify which extinct out-groups are included A total taxon explicitlyincludes only those extinct outgroups mostclosely related to a particular crown taxon

Node-based and stem-based denitions areconstructed by (1) identifying reference taxaand (2) attaching to them a relational phrasethat either includes their most recent com-mon ancestor and all descendants (node-based) or limits inclusion to all descen-dants closer to one or more of the referencetaxa (stem-based) (de Queiroz and Gauthier1992) (Fig 1 Table 1) In node-based deni-tions the phrase ldquoleast-inclusive claderdquo canreplace ldquocommon ancestor and all descen-dantsrdquo for cladists sensitive to the allusion to

340 SYSTEMATIC BIOLOGY VOL 48

ancestors (Lee 1998) Listing included taxawithout a relational phrase about ancestryor simply designating taxa as node- or stem-based without reference taxa does not con-stitutea phylogenetic denition because theboundaries of the taxon in question are notspecied (de Queiroz and Gauthier 1990Bryant 1996)

In the current literature there exists someconfusion over what is necessary and suf-cient for node- and stem-based denitionsBryant (1996185) suggested that a more in-clusive taxon be inserted within a stem-based denition such as ldquoall members oftaxon A thataremoreclosely related to taxonB than to taxon Crdquo Including taxon A how-ever is not necessary and may lead to inter-nal conicts within the denition (eg whentaxon A is included within taxon B or C onan alternativephylogeny) The goalof a phy-logenetic denition should be to unambigu-ously identify a specic common ancestorgiven any arrangement of taxa rather thanto identify such an ancestor only on a subsetof phylogenetic hypotheses

Meng et al (1994) and Wyss and Meng(1996) introduced a ldquostem-modied node-based denitionrdquo to effectively unite crown-group rodents whose basal relationshipsare uncertain Their preferred denition ofRodentia (ldquothe clade stemming from themost recent common ancestor of Mus andall Recent mammals more closely relatedto Mus than to Lagomorpha or membersof any other eutherian lsquoorderrdquorsquo Wyss andMeng 1996562) is better interpreted as aspecial stem-based rather than node-baseddenition because it uses only one ingroupreference taxon (Mus) and one or more out-group reference taxa (eg Lagomorpha)Their denition merely adds a phrase (ldquotheclade stemming from the most recent com-mon ancestorrdquo) to an otherwise typicalstem-based denition based on living ref-erence taxa That phrase ensures inclusionof all extinct taxa within the crown cladeBryantrsquos (1996185) reformulation of thisdenition on the other hand is clearly anode-based denition because he speci-ed only ingroup reference taxa (ldquothe mostrecent common ancestor of the species re-ferred to Rodentia by Wilson and Reeder

(1993) and all of its descendantsrdquo) These ex-amples underscore the need for claricationof terms (Tables 1 2)

A node-stem triplet (NST) is a denitionalconguration involving three taxa a node-based taxon composed of two stem-basedsubordinate taxa (Sereno 1997 1998 Fig 8)

Taxon A Taxa b and c their most recentcommon ancestor and all its descendants

Taxon B All taxa closer to taxon b than totaxon c

Taxon C All taxa closer to taxon c than totaxon b

FIGURE 8 Denitional triumvirate termed a node-stem triplet composed of a node-based taxon (A) andtwo subordinate stem-based taxa (B and C) which in-corporates added or repositioned taxa (dashed lineswith daggers) without changing the relative taxonomiccontent of taxon A B or C (as expressed by the equiva-lence statementA = B +C) A dot indicates a node-baseddenition an arrow indicates a stem-based denition

Taxa b and c (not shown in Fig 8) constitutereference taxa within taxon B and C respec-tively Although not essential the same ref-erence taxa (b c) may be used for the trio oftaxa in a NST which then specify complemen-tary denitions The pair of stem-based sis-ter taxa in addition may have reciprocal def-initions that utilize the same reference taxain antipodal positions The word ldquotaxardquo inthe stem-based denitions listed above mayoptionally be replaced by reference toa moreinclusive taxon (Bryant 1996) A stem-baseddenition for Saurischia for example mightread ldquoAll dinosaurs more closely related

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 341

to Neornithes than to Triceratopsrdquo (Sereno1998)

Taxonomic equivalence is a statement (orequation) in which one taxon is shownto be equivalent in current and poten-tial taxonomic content to subordinate taxaplus their common ancestor (Table 1)The NST described above is composedof a node-based taxon (A) and twosubordinate stem-based taxa (B C) andcan be written as the equivalence state-ment ldquoA = B + Crdquo because taxon Aby denition is equivalent to taxa Band C plus their most recent commonancestor

PHYLOGENETIC DEFINITIONS RATIONALE

The rst step toward a phylogenetic tax-onomy involved the distinction betweendenition and diagnosis (Buck and Hull1966 Grifths 1973 1974 Farris 1976 Ghis-elin 1984 Rowe 1987) A second step in-volved the formulation of node- and stem-based denitions by use of reference taxaand relational statements about ancestry (deQueiroz and Gauthier 1992) A third stepoutlined here involves the specication ofguidelines for selecting effective referencetaxa and criteria for positioning node- andstem-based denitions

Phylogenetic rearrangements and introduc-tion of new taxa are the principal factorsthat destabilize taxonomic content Stabil-ity of taxonomic content is enhanced bychoosing (1) effective reference taxa and(2) an effective conguration of phyloge-netic denitions

Reference TaxaPositionmdashGiven a particular phyloge-

netic hypothesis the taxonomic content of aphylogenetic denition is altered only if thereference taxa specify an alternative com-mon ancestor Given a basal dichotomy withone reference taxon on each side the iden-tication of the immediate common ances-tor is not affected by relocation of a ref-erence taxon on its side of the basal di-chotomy (Fig 9 move 1) Relocation ofa reference taxon to the opposing side orto a location outside the basal dichotomyidenties a different common ancestor

(Fig 9 moves 2 3) For a taxon with a basaldichotomy an alternative common ances-tor can be identied only if one referencetaxon is relocated to the opposing side oroutside the basal dichotomy Are some ref-erence taxa more likely to be relocated in thismanner than others

FIGURE 9 Three possible outcomes from relocationof a reference taxon 1 = relocation on the same side ofa basal dichotomy has no effect 2 = relocation to theopposite side of the basal dichotomy identies a lessinclusive common ancestor 3 = relocation outside thebasal dichotomy identies a more inclusive commonancestor Reference taxa are encircled and the originalbasal dichotomy is shown with heavy lines

First a taxon positioned near the basaldichotomy of a cladogram is always eas-ier to relocate to the opposing side (ierequires fewer additional steps) than is anested taxon if all other factors are regardedas equal among ingroup taxa (eg complete-ness missing data and homoplasy) Unlikea basal taxon (Fig 5d) a nested taxon sharessynapomorphies that increase its phyloge-netic (patristic) distance from the commonancestor (Fig 5e) Thus stability of taxo-nomic content is enhanced in phylogeneticdenitions if nested reference taxa are cho-sen that are located at some distance (atleast several nodes away) from the basaldichotomy This distance is easy to assessquantitatively for various nested taxa on agiven phylogeny

Second maximally inclusive referencetaxa (Fig 5c) have greater potential to cre-ate unnecessary taxonomic redundancy andinternal inconsistencies after relocation oftaxa (Fig 3) This is true for both node-

342 SYSTEMATIC BIOLOGY VOL 48

and stem-based taxa In the current litera-ture maximally inclusive reference taxa areoften used in phylogenetic denitions Intheir tetrapod classication for example deQueiroz and Gauthier (1992475) used suc-cessive maximally inclusive reference taxafor all denitionsmdasha pattern here termed re-cursive referencing (Fig 10a) Recursive refer-encing can have undesirable consequencesbecause it shifts the burden of a more precisedenition to a slightly less inclusive highertaxon This higher taxon may not be de-ned or it too may be affected by an alter-native conguration of basal relationshipsSalgado et al (1997) for example choseProsauropoda and Sauropoda as maximallyinclusive reference taxa for Sauropodomor-pha If a basal sauropodomorph is reposi-tioned the denitions of Prosauropoda andSauropoda must be determined to assess po-tential taxonomic consequences HoweverSalgado et al (1997) used Prosauropodaas a terminal taxon without denitionSauropoda was dened as a node-basedtaxon on the basis of two maximally in-clusive reference taxa Vulcanodon and Eu-sauropoda (all other sauropods) Thusif the incompletely known basal sauro-pod Vulcanodon is reinterpreted as be-ing more closely related to prosauropodsSauropoda and Sauropodomorpha wouldbecome synonyms

If on the other hand Sauropodomorphais dened on the basis of the nested ref-erence taxamdashthe prosauropod Plateosaurusand sauropod Saltasaurus (Sereno 1998)mdasha change in the phylogenetic positionof any other prosauropod or sauropodhas no effect on the taxonomic con-tents of Sauropodomorpha or SauropodaSauropodomorpha can be dened as ldquoPla-teosaurus Saltasaurus their most recentcommon ancestor and all descendantsrdquoand Sauropoda can be dened as ldquoallsauropodomorphs more closely related toSaltasaurus than to Plateosaurusrdquo The phylo-genetic position of Vulcanodon has no effecton the identication of respective commonancestors

CompletenessmdashPoorly known taxa aremore likely to be phylogenetically unsta-ble than well-known taxa (Wilkinson 1995)

FIGURE 10 Referencing styles (a) Recursive refer-encing selects a succession of the most inclusive ref-erence taxa (b) Nested referencing selects that sameremote reference taxon Reference taxa are encircled

Taxa with numerous missing entries in aphylogenetic analysis usually reduce phylo-genetic resolution by generating numerousequally parsimonious trees Well-known ref-erence taxa even if they are somewhat lessnested are preferable to poorly known taxa(Fig 11) A single poorly known basal taxonis least stable and therefore least desirableas a reference taxon

Use with Recent or extinct ltersmdashldquoRecentrdquo(ldquoextantrdquo or ldquolivingrdquo) or ldquoextinctrdquo may beused in phylogenetic denitions of crownand stem taxa respectively to help to stabi-lize taxonomic content in the face of poorlyresolved basal relationships For a crowntaxon with a stem-based denition a Recentlter can restrict the boundaries of a cladeto living taxa (Meng et al 1994 Wyss andMeng 1996) Rodentia for example could

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 343

be dened as ldquothe least inclusive clade com-posed of Recent mammals more closely re-lated to Mus than to Lepusrdquo For stem taxain contrast a stem-based denition using anldquoextinctrdquo lter can restrict inclusion to ex-tinct taxa when outgroup relationships areuncertain Moas for example could be de-ned as ldquoall extinct species more closelyrelated to Dinornis than to Struthiordquo theostrich

NumbermdashThe number of reference taxathat have been used in phylogenetic deni-tions has varied from two to innity Gau-thier (198614) for example dened Aves asa node-based taxon including ldquoall descen-

FIGURE 11 Nestedness and completeness in the se-lection of reference taxa A nested well-known taxonis more likely to constitute a stable reference taxon (en-circled)

dants of the most recent common ancestorof Ratitae Tinami and Neognathaerdquo Threereference taxa rather than two were chosenpresumably because of unstated uncertaintyin the phylogenetic relationships amongthese three avian subgroups De Queirozand Gauthier (1990310 1992461) suggestedthat phylogenetic denitions should utilizeonly two reference taxamdashbut then erecteddenitions for total taxa that used as refer-ence taxa one crown clade plus ldquoall extinct

taxa more closely related to it than to anyother crown claderdquo (de Queiroz and Gau-thier 1992475)

Listing more than a pair of referencetaxa does not seem to compromise a node-based phylogenetic denition in any dis-cernible way Node-based denitions areconstructed with what may be termed in-ternal inclusion A contradictory relation-ship between multiple ingroup referencetaxa for a node-based denition is im-possible to construct with internal inclu-sion because a clade always exists thatincludes all ingroup reference taxa Multi-ple ingroup reference taxa effectively stabi-lize the taxonomic content of a node-basedtaxon if basal relationships are uncertain(Fig 12a)

For stem-based denitions multiple out-group reference taxa effectively stabilize thetaxonomic content of a taxon if outgroup re-lationships are uncertain (Fig 12b) Multi-ple ingroup reference taxa for stem-baseddenitions can create a contradictory sit-uation if one (or more) ingroup referencetaxa are repositioned closer to the outgroupreference taxon (Fig 12c) or to a positionoutside the clade as originally dened Inthis case the external inclusion fundamen-tal to a stem-based denition creates a con-tradictory relationship among the originalingroup reference taxa which are no longermore closely related to each other than to theoutgroup reference taxon (Fig 12c) In theexample shown here relocation of ingroupreference taxon D creates a contradictory re-lationship with a second ingroup referencetaxon E if taxon A has the stem-based def-inition of ldquoall descendants more closely re-lated to taxon D and E than to taxon Brdquo Aninclusive ingroup reference taxon howevercan manifest the same problems in this cir-cumstance (Fig 12c) If taxon A is denedas ldquoall descendants more closely related totaxon C than to taxon Brdquo and if taxon C isdened on the basis of taxon D a contra-dictory denition results with relocation oftaxon D as shown taxon A now subsumesits original ingroup reference taxon C Suchcontradictory denitions can be avoided ifa nested ingroup reference taxon (or nestedtaxa) are selected such as taxon E its un-

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 6: Sereno, 1999b

334 SYSTEMATIC BIOLOGY VOL 48

Ornithurae is better dened as a less inclu-sive node-based taxon with Hesperornithi-formes and Neornithes (crown-group birds)as reference taxa (Fig 4) Initially Chiappe

FIGURE 4 Cladogram of basal avians showing re-cent use of the higher taxon Ornithurae (Haeckel 1866)The dot indicates a node-based denition and the ar-row indicates a stem-based denition

(1991337) reasoned that a node-based de-nition of Ornithurae was more appropriatebecause it excluded several newly discov-ered basal avians that lacked some of thesynapomorphies in Gauthierrsquos (1986) diag-nosis of Ornithurae Later Chiappe arguedthat a node-based denition of Ornithuraeldquobetter reects the composition of the cladebefore to [sic] the application of phyloge-netic denitionsrdquo (1996205)

Neither Gauthierrsquos nor Chiappersquos deni-tions have historical precedence Haeckel(1866) coined Ornithurae (ldquobird tailrdquo) for thereduced number of vertebrae and uniquecoossication of the distal tail (as a sin-gle bone the pygostyle) that characterizesall extant birds in contrast to the condi-tion in Archaeopteryx Haeckelrsquos taxon owesits existence to the then-recent discovery ofArchaeopteryx a primitive bird with a tailcomprising over 20 vertebrae Contrary toChiappe (1996) Ornithurae predates the de-scription of the extinct short-tailed generaHesperornis (Marsh 1872a) and Ichthyor-nis (Marsh 1872b) or the taxon Hesper-ornithiformes (Furbringer 1888) Ornithu-

rae as noted by Gauthier (1986) has beensupplanted by Neornithes (Gadow 1893)a taxon referring to crown-group aviansBased on original intent an argument couldbe made that Haeckelrsquos Ornithurae shouldsupplant Neornithes on grounds of prior-ity or that it be dened explicitly as anapomorphy-based taxon based on the pres-ence of a pygostyle which is now known tobe lacking in birds other than Archaeopteryx(Forster et al 1998) In either case thedenition would circumscribe clades otherthan those identied by either Gauthier orChiappe

For most taxon names the original de-nition if explicitly stated at all is character-based or taxon-based (a list of included taxa)and lacks a relational phrase about ances-try that would specify potential membershipThe intent of the original author of a taxonto include or exclude unknown or reposi-tioned taxa usually cannot be interpretedunambiguously This problem is not widelyappreciated

APOMORPHY-BASED DEFINITIONS

An apomorphy-based denition speci-es the boundaries of a clade by identi-fying the ldquorst ancestor with a particularsynapomorphyrdquo (de Queiroz and Gauthier1990310) and encompassing all of its de-scendants An apomorphy-based taxon in-cludes all descendants whether or not thesynapomorphy is maintained Apomorphy-based denitions are subject to three prob-lems that are not relevant to node-based andstem-based denitions variation in charac-ter coding optimization ambiguity and ho-moplasy (Bryant 1994 Schander and Thol-lesson 1995 Sereno 1998) These problemshave long been associated with traditionaluse of ldquokeyrdquo characters to dene taxa Thusfar few apomorphy-based denitions havebeen erected and this denitional formshould be avoided

Character ambiguitymdashConsider an apo-morphy-based denition for Aves (includ-ing Archaeopteryx and Ornithurae) based onthe presence of ldquofeathersrdquo the ldquokey char-acterrdquo usually associated with this taxonFirst the character codingmdashldquoFeathers ab-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 335

sent (0) present (1)rdquomdashis only one of sev-eral possibilities A ldquofeatherrdquo is composedof a rachis barbs barbules and other com-ponents that may be better coded as sepa-rate characters The character on which anapomorphy-based denition is based there-fore may be recoded as two or more char-acters with independent character-state dis-tributions Indeed this is an expectation formost qualitative or quantitative charactersthat is once the transformations become bet-ter known intermediate states or multiplecharacters will emerge as recent discoveriesare beginning to reveal with regard to feath-ers (Chen et al 1998 Ji et al 1998) or withregard to the characters historically associ-ated with Mammalia (Rowe 1988 RoweandGauthier 1992 Bryant 1994 de Queiroz1994)

The apomorphy-based denition given asan example by de Queiroz and Gauthier(1990310) is subject on the same groundsto interpretational ambiguity ldquoTetrapodardquois dened as the ldquorst vertebrate to possessdigits (ie hands and feet rather than ns)and all of its descendantsrdquo What exactlyconstitutes a ldquodigitrdquo versus a ldquonrdquo or ldquopad-dlerdquo In fact this is a signicant questiongiven recent fossil discoveries (Daeschlerand Shubin 1998) Can this feature or listof features be observed in extinct formsAnd what happens if we learn that ldquodigitsrdquoevolved rst on the hands and later on thefeet These are common problems in the in-terpretation of character data which shouldnot be extended to taxonomic denitions

Optimization ambiguitymdashApomorphy-based denitions also do not specify a sin-gle ancestor when the chosen apomorphyhas an ambiguous optimization Charac-ters with ambiguous character-state opti-mization are commonplace in systematicsand arise from missing data (lack of preser-vation strong transformation) and homo-plasy (Bryant 1994 Schander and Thol-lesson 1995 Sereno 1998) There may bemany equally parsimonious positions fora particular character state on the shortestcladogram

HomoplasymdashHomoplasy may involve thestable placement of an apomorphy at morethan one node (Bryant 1994) This can create

an impasse for an apomorphy-based deni-tion because the apomorphy might identifymore than one clade Invoking time as an ar-biter (such as ldquothe rst taxon that has rdquo)invites ambiguities associated with tempo-ral origin (such as missing ancestral lin-eages)

PHYLOGENETIC DEFINITIONS TERMINOLOGY

The terminology used in the followingdiscussion of phylogenetic taxonomic def-initions is claried below and tabulated (Ta-bles 1 2) Someof these terms aregiven morespecic meaning than in the current litera-ture others are new or recently introduced(Sereno 1998)

Taxonomic denition and diagnosis are de-ned here much as they have been describedpreviously by phylogenetic systematists (deQueiroz and Gauthier 1990 1992) and theirforebears (Buck and Hull 1966 Farris 1976Ghiselin 1984) (Table 1) De Queiroz andGauthier (1992461) dened phylogeneticdenitions as ldquostatements specifying themeanings of taxon names (words) they arestated in terms of ancestryrdquo The ldquomeaningof a taxon namerdquo in phylogenetic taxonomyconcerns its taxonomic content or mem-bership as delimited by a relational state-ment Taxonomic denitions therefore areregarded here as ldquoextensionalrdquo (ostensive)statements of relationship that specify thetaxonomic content of a taxon (de Queiroz1992)

Taxonomic diagnosis on the other handinvolves the ldquointensionalrdquo characterizationof members of a clade as recognized bythe shared apomorphies present in thecommon ancestor (Buck and Hull 1966Farris 1976 Ghiselin 1984 Rowe 1987de Queiroz and Gauthier 1990 1992 deQueiroz 1992) According to de Queiroz andGauthier (1992461) diagnoses are ldquostate-ments specifying how to determine whethera given species or organism is a representa-tive of the taxon (clade) to which a partic-ular name refers they are most commonlystated in terms of charactersrdquo Strictly speak-ing shared apomorphies (derived-characterstates) rather than characters determine themembership of a particular clade or species

336 SYSTEMATIC BIOLOGY VOL 48

Taxonomic content is used here to refer toall existing and potential taxa included by thephylogenetic denition of a taxon (Table 1)The smallest number of subordinate taxathat can fulll this denition of taxonomiccontent is two a redundant taxon (ie asingle subordinate taxon) would have anidentical phylogenetic denition and thusis uninformative in phylogenetic taxonomy(Farris 1976) The taxonomic content oftaxon C therefore could be ldquotaxon A taxonB their most recent common ancestor andall descendantsrdquo if taxa A and B are denedextensionally to encompass all existing andpotential subordinate taxa currently withintaxon C

A list of included taxa is insufcient to un-ambiguously specify taxonomic content be-cause there will always be potential mem-bers of a taxon that lie outside any list ofincluded taxa whether or not the includedtaxa themselves have phylogenetic deni-tions An exhaustive list of included taxawith stem-based denitions for exampledoes not specify membership for their hy-pothetical immediate common ancestor (anode-based taxon) or more distant knownor hypothetical outgroups (a stem-basedtaxon) Two reference taxa and a relationalstatement about ancestry are necessary andsufcient to specify taxonomic content be-cause they unambiguously specify member-ship for all existing and potential membersof a taxon

A reference taxon is a species or higher-leveltaxon that is used to delimit a phylogeneticdenition (Table 1) For example ldquotaxon Ataxon B their most recent common ancestorand all descendantsrdquo uses A and B as refer-ence taxa for a node-based taxonomic deni-tion Previously cited as ldquoreference pointsrdquo

(Schander and Thollesson 1995) theseimportant components of phylogeneticdenitions have not been distinguishedterminologically

Several kinds of reference taxa can beidentied (Fig 5 Table 2) Ingroup and out-group reference taxa are located within oroutside the taxon they dene respectively(Fig 5a b) An inclusive reference taxon in-cludes several ingroup or outgroup taxaa maximally inclusive reference taxon in-cludes the largest possible ingroup or out-group reference taxon (Fig 5c) Basal andnested reference taxa are less inclusive andas their name suggests are located at thebase or nested within an ingroup or out-group taxon (Fig 5d e)

A crown taxon denotes a living speciesor a clade dened by using living refer-ence taxa whose immediate outgroup isextinct (Figs 6 7 Table 1) The extinct out-group is a necessary component of the def-inition Jefferies (1979449) coined the termldquocrown grouprdquo to replace Hennigrsquos (1969) group which Jefferies dened as ldquothe lat-est common ancestor of all living membersof group 1 plus all descendants of this an-cestor whether these are living or extinctrdquoTheexistence of an immediate outgroup thatis extinct is implied and was clearly presentin the examples given by Hennig and Jef-feries Patterson (1981207) also clearly usedcrown- and stem-groups in this mannerstating that the ldquostem-group contains fos-sil taxa that exhibit some but not all of thecharacters of the crown-grouprdquo Crown- andstem-group terminology has been used ex-clusively for higher taxa bounded by liv-ing species that have extinct outgroups (egde Queiroz and Gauthier 1992 Forey 1992McKenna and Bell 1997) Although it may

TABLE 2 Denitions for the various kinds of reference taxa

Reference taxon type Denition

Ingroup reference taxon included within a taxon by denitionOutgroup reference taxon excluded from a taxon by denitionInclusive most inclusive reference taxon among ingroup or outgroup taxaBasal reference taxon near the basal furcation of a taxonNested reference taxon distant from the basal furcation of a taxonVernacular vernacular name used as a reference taxon (eg ldquobirdsrdquo or ldquoextant birdsrdquo)

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 337

FIGURE 5 Kinds of reference taxa (a) An ingroup reference taxon is included by denition within a taxon anode-based denition is based on two ingroup reference taxa (b) An outgroup reference taxon is excluded bydenition from a taxon a stem-based denition is based on one ingroup and one outgroup reference taxon (c) Aninclusive reference taxon is the most inclusive ingroup or outgroup reference taxon possible inclusive referencetaxa B and C for taxon A allow denitional ambiguity with change in the relationship of a basal taxon (d) Abasal reference taxon is located at or near the base of an ingroup or outgroup taxon basal reference taxa allowdenitional ambiguity when their relationships change (e) A nested reference taxon is remote from the base ofan ingroup or outgroup reference taxon nested reference taxa are unaffected by changes in the relationships ofbasal taxa Reference taxa are encircled a dot indicates a node-based denition an arrow indicates a stem-baseddenition a dashed line indicates a change in position of a basal taxon

FIGURE 6 Crown stem and total taxa (a) Total group composed of a crown group plus a paraphyletic ldquostemgrouprdquo as conceived by Hennig (1969 1983) and Jefferies (1979) (b) Total taxon composed of a crown taxon andmonophyletic stem taxa as used in this article One stem taxon (left) is shown with a node-based denition andthe other (right) with a stem-based denition

338 SYSTEMATIC BIOLOGY VOL 48

FIGURE 7 Crown species or clades (enclosed) musthave an extinct (stem) taxon (dashed) as an immediateoutgroup (a) Crown species and crown clades are de-limited by the presence of an immediate outgroup thatis extinct (b) Often referred to as ldquoextantrdquo taxa crowntaxa may be composed primarily of extinct taxa ldquoLiv-ingrdquo and ldquoextinctrdquo conditions are treated as if they con-stitute plesiomorphic and apomorphic character statesrespectively of an irreversible character ldquostate of be-ingrdquo Crown clades under this conceptualization areplesiomorphy-based taxa bounded by two living taxafor which the immediate outgroup is extinct

be hypothetically true that ldquoevery Recentgroup whatever its position in thehierarchymay be expected to have had a stem-grouprdquo(Patterson 1981207) clades bounded by liv-ing taxa with immediate outgroups that arealso living have not been considered crowntaxa

Some recent denitions of crown taxahowever have set aside Hennigrsquos andJefferiesrsquo formulations as well as generalusage in the literature A less restric-tive meaning has been proposed basedsolely on reference taxa that include liv-

ing representatives De Queiroz and Gau-thier (1992469) dened crown taxa asldquoclades within which both branches ofthe basal dichotomy are represented byextant descendantsrdquo without mention ofthe condition of the immediate outgroupYet all of their examples of crown cladeshave immediate outgroups that are ex-tinct Regarding their craniate taxonomythey stated that each stem-based taxon iscomposed of a ldquocrown taxon plus all ex-tinct taxa more closely related to it thanto any other crown claderdquo (de Queirozand Gauthier 1992475) Lee (19961103)proposed that ldquoa crown clade denitionresults when both taxa implicated in anode-based denition are representedby extant formsrdquo Although the conditionof the immediate outgroup is never men-tioned Lee (19961103) stated that crowntaxon denitions apply to clades thatconsist of ldquothe most recent common an-cestor of all extant forms and all its de-scendantsrdquo (emphasis added) The pres-ence of an immediate extinct outgroup isimplied

The denition of a crown taxon presentedhere in contrast does not specify ingroupstructure (such as the presence or absenceof a basal dichotomy) or denitional type(node- or stem-based) and explicitly statesthat the immediate outgroup(s) is extinct(Fig 7a Table 1) Crown taxa so denedcan be mapped unerringly on a cladogramif (1) ldquolivingrdquo and ldquoextinctrdquo conditions aretreated as primitive and derived character-states respectively of the character ldquocurrentstate of beingrdquo and (2) if this character isregarded as irreversible (ie CaminndashSokalparsimony) The living state in other wordsis always regarded as plesiomorphic Theirreversible status of the character (state ofbeing) is important because it identiescrown groups on cladograms of living andextinct taxa that would have an ambigu-ous status if character-state reversal wereallowed (Fig 7b) This character-state con-ceptualization of the denition of a crowngroup captures the essence of crown taxawhich have always been used to refer tomaximally inclusive taxa dened by extantrepresentatives not simply any ldquoextant-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 339

boundedrdquo taxon Discovery of extinct sistertaxa provides the opportunity to recognizecrown taxa

Crown taxa are often regarded as cladesthat are bounded uniquely by living in-groups The use of living reference taxahowever is most often an arbitrary deci-sion Any extinct ingroup taxon that is mostclosely related to a living reference taxoncould be used to delimit the same clade(Fig 7b) A crown clade is dened uniquelyby living reference taxa only when such taxado not have extinct sister taxa (Fig 7b) afact that is not widely appreciated Crowntaxa in other words do not have any spe-cial biological signicance beyond signal-ing that particular reference taxa have sur-vived to an arbitrary datum (the Recent)Informal reference to crown taxa as ldquolivingrdquoor ldquoextantrdquo taxa therefore can be mislead-ing because a crown taxon may be com-posed predominantly of extinct species asis the case with the crown taxon Crocodylia(Fig 7b Brochu 1997) ldquoExtant crocodil-iansrdquo if taken literally refers to a smallparaphyletic subset of Crocodylia Explicitreference to ldquocrown crocodiliansrdquo or sim-ply ldquoCrocodyliardquo is preferable to use of theterms ldquolivingrdquo or ldquoextantrdquo

A living taxon is also essential to delimita maximally inclusive clade that is entirelyextinctmdasha stem taxon as here dened Astem taxon denotes an extinct species orclade the immediate outgroup of which in-cludes at least one living member (Fig 6b)Many extinct taxa are not stem taxa becausetheir immediate outgroup is itself extinctSuch extinct taxa are members of more in-clusive stem taxa Stem taxa can have eithernode- or stem-based denitions (Fig 6b)Stem taxa and their associated less inclusivecrown taxon make up the taxonomic contentof their respective total taxon

Stem taxa as here dened should notbe confused with ldquostem grouprdquo (Fig 6a)which was rst used in phylogenetic sys-tematics by Hennig (ldquostammgrupperdquo 19691983) Hennigrsquos (198315) vague formulationof the concept specied all extinct speciesldquowhich can be shown probably to belong toa particular monophyletic group of the phy-logenetic system but which are probably no

closer [sic] related to one subgroup amongrecent animals than to anotherrdquo (translatedfrom the German Ax 1987224) Jefferies(1979) claried the term to include all ex-tinct outgroups of a crown taxon (Fig 6a)Other authors have developed more convo-luted denitions that approximate the sameend Wiley (1981217) for example dened aldquostem grouprdquo as ldquoall fossil species of a groupmore primitive than the most primitive Re-cent species of the same grouprdquo Theseldquostem groupsrdquo are paraphyletic and oftencircumscribe the same ldquoancestral groupsrdquoand ldquoevolutionary gradesrdquo that have mud-died phylogenetic thinking since DarwinOnly a few cladists continue to endorse suchconcepts (Smith 1994) Ax (1987224) pro-posed a cumbersome solution to the prob-lem of paraphyletic ldquostem groupsrdquo by divid-ing them into ancestral ldquostem lineagesrdquo andldquorepresentatives of the stem lineagerdquo Thelatter are monophyletic taxa most simply re-ferred to as stem taxa

A total taxon is regarded here as a cladecomposed of a crown taxon plus all ex-tinct outgroups more closely related to itthan to another crown taxon (Fig 3 Ta-ble 1) Hennigrsquos (1969) and Jefferiesrsquo (1979)formulations of the concept are wordy butsimilar in effect because they restricted in-clusion of extinct outgroups to those mostclosely related to a particular crown taxonDe Queiroz and Gauthier (1992470) deneda total clade as a ldquomore inclusive clade con-sisting of thecrown clade plus its extinctout-groupsrdquo This denition is incomplete be-cause it does not specify which extinct out-groups are included A total taxon explicitlyincludes only those extinct outgroups mostclosely related to a particular crown taxon

Node-based and stem-based denitions areconstructed by (1) identifying reference taxaand (2) attaching to them a relational phrasethat either includes their most recent com-mon ancestor and all descendants (node-based) or limits inclusion to all descen-dants closer to one or more of the referencetaxa (stem-based) (de Queiroz and Gauthier1992) (Fig 1 Table 1) In node-based deni-tions the phrase ldquoleast-inclusive claderdquo canreplace ldquocommon ancestor and all descen-dantsrdquo for cladists sensitive to the allusion to

340 SYSTEMATIC BIOLOGY VOL 48

ancestors (Lee 1998) Listing included taxawithout a relational phrase about ancestryor simply designating taxa as node- or stem-based without reference taxa does not con-stitutea phylogenetic denition because theboundaries of the taxon in question are notspecied (de Queiroz and Gauthier 1990Bryant 1996)

In the current literature there exists someconfusion over what is necessary and suf-cient for node- and stem-based denitionsBryant (1996185) suggested that a more in-clusive taxon be inserted within a stem-based denition such as ldquoall members oftaxon A thataremoreclosely related to taxonB than to taxon Crdquo Including taxon A how-ever is not necessary and may lead to inter-nal conicts within the denition (eg whentaxon A is included within taxon B or C onan alternativephylogeny) The goalof a phy-logenetic denition should be to unambigu-ously identify a specic common ancestorgiven any arrangement of taxa rather thanto identify such an ancestor only on a subsetof phylogenetic hypotheses

Meng et al (1994) and Wyss and Meng(1996) introduced a ldquostem-modied node-based denitionrdquo to effectively unite crown-group rodents whose basal relationshipsare uncertain Their preferred denition ofRodentia (ldquothe clade stemming from themost recent common ancestor of Mus andall Recent mammals more closely relatedto Mus than to Lagomorpha or membersof any other eutherian lsquoorderrdquorsquo Wyss andMeng 1996562) is better interpreted as aspecial stem-based rather than node-baseddenition because it uses only one ingroupreference taxon (Mus) and one or more out-group reference taxa (eg Lagomorpha)Their denition merely adds a phrase (ldquotheclade stemming from the most recent com-mon ancestorrdquo) to an otherwise typicalstem-based denition based on living ref-erence taxa That phrase ensures inclusionof all extinct taxa within the crown cladeBryantrsquos (1996185) reformulation of thisdenition on the other hand is clearly anode-based denition because he speci-ed only ingroup reference taxa (ldquothe mostrecent common ancestor of the species re-ferred to Rodentia by Wilson and Reeder

(1993) and all of its descendantsrdquo) These ex-amples underscore the need for claricationof terms (Tables 1 2)

A node-stem triplet (NST) is a denitionalconguration involving three taxa a node-based taxon composed of two stem-basedsubordinate taxa (Sereno 1997 1998 Fig 8)

Taxon A Taxa b and c their most recentcommon ancestor and all its descendants

Taxon B All taxa closer to taxon b than totaxon c

Taxon C All taxa closer to taxon c than totaxon b

FIGURE 8 Denitional triumvirate termed a node-stem triplet composed of a node-based taxon (A) andtwo subordinate stem-based taxa (B and C) which in-corporates added or repositioned taxa (dashed lineswith daggers) without changing the relative taxonomiccontent of taxon A B or C (as expressed by the equiva-lence statementA = B +C) A dot indicates a node-baseddenition an arrow indicates a stem-based denition

Taxa b and c (not shown in Fig 8) constitutereference taxa within taxon B and C respec-tively Although not essential the same ref-erence taxa (b c) may be used for the trio oftaxa in a NST which then specify complemen-tary denitions The pair of stem-based sis-ter taxa in addition may have reciprocal def-initions that utilize the same reference taxain antipodal positions The word ldquotaxardquo inthe stem-based denitions listed above mayoptionally be replaced by reference toa moreinclusive taxon (Bryant 1996) A stem-baseddenition for Saurischia for example mightread ldquoAll dinosaurs more closely related

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 341

to Neornithes than to Triceratopsrdquo (Sereno1998)

Taxonomic equivalence is a statement (orequation) in which one taxon is shownto be equivalent in current and poten-tial taxonomic content to subordinate taxaplus their common ancestor (Table 1)The NST described above is composedof a node-based taxon (A) and twosubordinate stem-based taxa (B C) andcan be written as the equivalence state-ment ldquoA = B + Crdquo because taxon Aby denition is equivalent to taxa Band C plus their most recent commonancestor

PHYLOGENETIC DEFINITIONS RATIONALE

The rst step toward a phylogenetic tax-onomy involved the distinction betweendenition and diagnosis (Buck and Hull1966 Grifths 1973 1974 Farris 1976 Ghis-elin 1984 Rowe 1987) A second step in-volved the formulation of node- and stem-based denitions by use of reference taxaand relational statements about ancestry (deQueiroz and Gauthier 1992) A third stepoutlined here involves the specication ofguidelines for selecting effective referencetaxa and criteria for positioning node- andstem-based denitions

Phylogenetic rearrangements and introduc-tion of new taxa are the principal factorsthat destabilize taxonomic content Stabil-ity of taxonomic content is enhanced bychoosing (1) effective reference taxa and(2) an effective conguration of phyloge-netic denitions

Reference TaxaPositionmdashGiven a particular phyloge-

netic hypothesis the taxonomic content of aphylogenetic denition is altered only if thereference taxa specify an alternative com-mon ancestor Given a basal dichotomy withone reference taxon on each side the iden-tication of the immediate common ances-tor is not affected by relocation of a ref-erence taxon on its side of the basal di-chotomy (Fig 9 move 1) Relocation ofa reference taxon to the opposing side orto a location outside the basal dichotomyidenties a different common ancestor

(Fig 9 moves 2 3) For a taxon with a basaldichotomy an alternative common ances-tor can be identied only if one referencetaxon is relocated to the opposing side oroutside the basal dichotomy Are some ref-erence taxa more likely to be relocated in thismanner than others

FIGURE 9 Three possible outcomes from relocationof a reference taxon 1 = relocation on the same side ofa basal dichotomy has no effect 2 = relocation to theopposite side of the basal dichotomy identies a lessinclusive common ancestor 3 = relocation outside thebasal dichotomy identies a more inclusive commonancestor Reference taxa are encircled and the originalbasal dichotomy is shown with heavy lines

First a taxon positioned near the basaldichotomy of a cladogram is always eas-ier to relocate to the opposing side (ierequires fewer additional steps) than is anested taxon if all other factors are regardedas equal among ingroup taxa (eg complete-ness missing data and homoplasy) Unlikea basal taxon (Fig 5d) a nested taxon sharessynapomorphies that increase its phyloge-netic (patristic) distance from the commonancestor (Fig 5e) Thus stability of taxo-nomic content is enhanced in phylogeneticdenitions if nested reference taxa are cho-sen that are located at some distance (atleast several nodes away) from the basaldichotomy This distance is easy to assessquantitatively for various nested taxa on agiven phylogeny

Second maximally inclusive referencetaxa (Fig 5c) have greater potential to cre-ate unnecessary taxonomic redundancy andinternal inconsistencies after relocation oftaxa (Fig 3) This is true for both node-

342 SYSTEMATIC BIOLOGY VOL 48

and stem-based taxa In the current litera-ture maximally inclusive reference taxa areoften used in phylogenetic denitions Intheir tetrapod classication for example deQueiroz and Gauthier (1992475) used suc-cessive maximally inclusive reference taxafor all denitionsmdasha pattern here termed re-cursive referencing (Fig 10a) Recursive refer-encing can have undesirable consequencesbecause it shifts the burden of a more precisedenition to a slightly less inclusive highertaxon This higher taxon may not be de-ned or it too may be affected by an alter-native conguration of basal relationshipsSalgado et al (1997) for example choseProsauropoda and Sauropoda as maximallyinclusive reference taxa for Sauropodomor-pha If a basal sauropodomorph is reposi-tioned the denitions of Prosauropoda andSauropoda must be determined to assess po-tential taxonomic consequences HoweverSalgado et al (1997) used Prosauropodaas a terminal taxon without denitionSauropoda was dened as a node-basedtaxon on the basis of two maximally in-clusive reference taxa Vulcanodon and Eu-sauropoda (all other sauropods) Thusif the incompletely known basal sauro-pod Vulcanodon is reinterpreted as be-ing more closely related to prosauropodsSauropoda and Sauropodomorpha wouldbecome synonyms

If on the other hand Sauropodomorphais dened on the basis of the nested ref-erence taxamdashthe prosauropod Plateosaurusand sauropod Saltasaurus (Sereno 1998)mdasha change in the phylogenetic positionof any other prosauropod or sauropodhas no effect on the taxonomic con-tents of Sauropodomorpha or SauropodaSauropodomorpha can be dened as ldquoPla-teosaurus Saltasaurus their most recentcommon ancestor and all descendantsrdquoand Sauropoda can be dened as ldquoallsauropodomorphs more closely related toSaltasaurus than to Plateosaurusrdquo The phylo-genetic position of Vulcanodon has no effecton the identication of respective commonancestors

CompletenessmdashPoorly known taxa aremore likely to be phylogenetically unsta-ble than well-known taxa (Wilkinson 1995)

FIGURE 10 Referencing styles (a) Recursive refer-encing selects a succession of the most inclusive ref-erence taxa (b) Nested referencing selects that sameremote reference taxon Reference taxa are encircled

Taxa with numerous missing entries in aphylogenetic analysis usually reduce phylo-genetic resolution by generating numerousequally parsimonious trees Well-known ref-erence taxa even if they are somewhat lessnested are preferable to poorly known taxa(Fig 11) A single poorly known basal taxonis least stable and therefore least desirableas a reference taxon

Use with Recent or extinct ltersmdashldquoRecentrdquo(ldquoextantrdquo or ldquolivingrdquo) or ldquoextinctrdquo may beused in phylogenetic denitions of crownand stem taxa respectively to help to stabi-lize taxonomic content in the face of poorlyresolved basal relationships For a crowntaxon with a stem-based denition a Recentlter can restrict the boundaries of a cladeto living taxa (Meng et al 1994 Wyss andMeng 1996) Rodentia for example could

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 343

be dened as ldquothe least inclusive clade com-posed of Recent mammals more closely re-lated to Mus than to Lepusrdquo For stem taxain contrast a stem-based denition using anldquoextinctrdquo lter can restrict inclusion to ex-tinct taxa when outgroup relationships areuncertain Moas for example could be de-ned as ldquoall extinct species more closelyrelated to Dinornis than to Struthiordquo theostrich

NumbermdashThe number of reference taxathat have been used in phylogenetic deni-tions has varied from two to innity Gau-thier (198614) for example dened Aves asa node-based taxon including ldquoall descen-

FIGURE 11 Nestedness and completeness in the se-lection of reference taxa A nested well-known taxonis more likely to constitute a stable reference taxon (en-circled)

dants of the most recent common ancestorof Ratitae Tinami and Neognathaerdquo Threereference taxa rather than two were chosenpresumably because of unstated uncertaintyin the phylogenetic relationships amongthese three avian subgroups De Queirozand Gauthier (1990310 1992461) suggestedthat phylogenetic denitions should utilizeonly two reference taxamdashbut then erecteddenitions for total taxa that used as refer-ence taxa one crown clade plus ldquoall extinct

taxa more closely related to it than to anyother crown claderdquo (de Queiroz and Gau-thier 1992475)

Listing more than a pair of referencetaxa does not seem to compromise a node-based phylogenetic denition in any dis-cernible way Node-based denitions areconstructed with what may be termed in-ternal inclusion A contradictory relation-ship between multiple ingroup referencetaxa for a node-based denition is im-possible to construct with internal inclu-sion because a clade always exists thatincludes all ingroup reference taxa Multi-ple ingroup reference taxa effectively stabi-lize the taxonomic content of a node-basedtaxon if basal relationships are uncertain(Fig 12a)

For stem-based denitions multiple out-group reference taxa effectively stabilize thetaxonomic content of a taxon if outgroup re-lationships are uncertain (Fig 12b) Multi-ple ingroup reference taxa for stem-baseddenitions can create a contradictory sit-uation if one (or more) ingroup referencetaxa are repositioned closer to the outgroupreference taxon (Fig 12c) or to a positionoutside the clade as originally dened Inthis case the external inclusion fundamen-tal to a stem-based denition creates a con-tradictory relationship among the originalingroup reference taxa which are no longermore closely related to each other than to theoutgroup reference taxon (Fig 12c) In theexample shown here relocation of ingroupreference taxon D creates a contradictory re-lationship with a second ingroup referencetaxon E if taxon A has the stem-based def-inition of ldquoall descendants more closely re-lated to taxon D and E than to taxon Brdquo Aninclusive ingroup reference taxon howevercan manifest the same problems in this cir-cumstance (Fig 12c) If taxon A is denedas ldquoall descendants more closely related totaxon C than to taxon Brdquo and if taxon C isdened on the basis of taxon D a contra-dictory denition results with relocation oftaxon D as shown taxon A now subsumesits original ingroup reference taxon C Suchcontradictory denitions can be avoided ifa nested ingroup reference taxon (or nestedtaxa) are selected such as taxon E its un-

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 7: Sereno, 1999b

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 335

sent (0) present (1)rdquomdashis only one of sev-eral possibilities A ldquofeatherrdquo is composedof a rachis barbs barbules and other com-ponents that may be better coded as sepa-rate characters The character on which anapomorphy-based denition is based there-fore may be recoded as two or more char-acters with independent character-state dis-tributions Indeed this is an expectation formost qualitative or quantitative charactersthat is once the transformations become bet-ter known intermediate states or multiplecharacters will emerge as recent discoveriesare beginning to reveal with regard to feath-ers (Chen et al 1998 Ji et al 1998) or withregard to the characters historically associ-ated with Mammalia (Rowe 1988 RoweandGauthier 1992 Bryant 1994 de Queiroz1994)

The apomorphy-based denition given asan example by de Queiroz and Gauthier(1990310) is subject on the same groundsto interpretational ambiguity ldquoTetrapodardquois dened as the ldquorst vertebrate to possessdigits (ie hands and feet rather than ns)and all of its descendantsrdquo What exactlyconstitutes a ldquodigitrdquo versus a ldquonrdquo or ldquopad-dlerdquo In fact this is a signicant questiongiven recent fossil discoveries (Daeschlerand Shubin 1998) Can this feature or listof features be observed in extinct formsAnd what happens if we learn that ldquodigitsrdquoevolved rst on the hands and later on thefeet These are common problems in the in-terpretation of character data which shouldnot be extended to taxonomic denitions

Optimization ambiguitymdashApomorphy-based denitions also do not specify a sin-gle ancestor when the chosen apomorphyhas an ambiguous optimization Charac-ters with ambiguous character-state opti-mization are commonplace in systematicsand arise from missing data (lack of preser-vation strong transformation) and homo-plasy (Bryant 1994 Schander and Thol-lesson 1995 Sereno 1998) There may bemany equally parsimonious positions fora particular character state on the shortestcladogram

HomoplasymdashHomoplasy may involve thestable placement of an apomorphy at morethan one node (Bryant 1994) This can create

an impasse for an apomorphy-based deni-tion because the apomorphy might identifymore than one clade Invoking time as an ar-biter (such as ldquothe rst taxon that has rdquo)invites ambiguities associated with tempo-ral origin (such as missing ancestral lin-eages)

PHYLOGENETIC DEFINITIONS TERMINOLOGY

The terminology used in the followingdiscussion of phylogenetic taxonomic def-initions is claried below and tabulated (Ta-bles 1 2) Someof these terms aregiven morespecic meaning than in the current litera-ture others are new or recently introduced(Sereno 1998)

Taxonomic denition and diagnosis are de-ned here much as they have been describedpreviously by phylogenetic systematists (deQueiroz and Gauthier 1990 1992) and theirforebears (Buck and Hull 1966 Farris 1976Ghiselin 1984) (Table 1) De Queiroz andGauthier (1992461) dened phylogeneticdenitions as ldquostatements specifying themeanings of taxon names (words) they arestated in terms of ancestryrdquo The ldquomeaningof a taxon namerdquo in phylogenetic taxonomyconcerns its taxonomic content or mem-bership as delimited by a relational state-ment Taxonomic denitions therefore areregarded here as ldquoextensionalrdquo (ostensive)statements of relationship that specify thetaxonomic content of a taxon (de Queiroz1992)

Taxonomic diagnosis on the other handinvolves the ldquointensionalrdquo characterizationof members of a clade as recognized bythe shared apomorphies present in thecommon ancestor (Buck and Hull 1966Farris 1976 Ghiselin 1984 Rowe 1987de Queiroz and Gauthier 1990 1992 deQueiroz 1992) According to de Queiroz andGauthier (1992461) diagnoses are ldquostate-ments specifying how to determine whethera given species or organism is a representa-tive of the taxon (clade) to which a partic-ular name refers they are most commonlystated in terms of charactersrdquo Strictly speak-ing shared apomorphies (derived-characterstates) rather than characters determine themembership of a particular clade or species

336 SYSTEMATIC BIOLOGY VOL 48

Taxonomic content is used here to refer toall existing and potential taxa included by thephylogenetic denition of a taxon (Table 1)The smallest number of subordinate taxathat can fulll this denition of taxonomiccontent is two a redundant taxon (ie asingle subordinate taxon) would have anidentical phylogenetic denition and thusis uninformative in phylogenetic taxonomy(Farris 1976) The taxonomic content oftaxon C therefore could be ldquotaxon A taxonB their most recent common ancestor andall descendantsrdquo if taxa A and B are denedextensionally to encompass all existing andpotential subordinate taxa currently withintaxon C

A list of included taxa is insufcient to un-ambiguously specify taxonomic content be-cause there will always be potential mem-bers of a taxon that lie outside any list ofincluded taxa whether or not the includedtaxa themselves have phylogenetic deni-tions An exhaustive list of included taxawith stem-based denitions for exampledoes not specify membership for their hy-pothetical immediate common ancestor (anode-based taxon) or more distant knownor hypothetical outgroups (a stem-basedtaxon) Two reference taxa and a relationalstatement about ancestry are necessary andsufcient to specify taxonomic content be-cause they unambiguously specify member-ship for all existing and potential membersof a taxon

A reference taxon is a species or higher-leveltaxon that is used to delimit a phylogeneticdenition (Table 1) For example ldquotaxon Ataxon B their most recent common ancestorand all descendantsrdquo uses A and B as refer-ence taxa for a node-based taxonomic deni-tion Previously cited as ldquoreference pointsrdquo

(Schander and Thollesson 1995) theseimportant components of phylogeneticdenitions have not been distinguishedterminologically

Several kinds of reference taxa can beidentied (Fig 5 Table 2) Ingroup and out-group reference taxa are located within oroutside the taxon they dene respectively(Fig 5a b) An inclusive reference taxon in-cludes several ingroup or outgroup taxaa maximally inclusive reference taxon in-cludes the largest possible ingroup or out-group reference taxon (Fig 5c) Basal andnested reference taxa are less inclusive andas their name suggests are located at thebase or nested within an ingroup or out-group taxon (Fig 5d e)

A crown taxon denotes a living speciesor a clade dened by using living refer-ence taxa whose immediate outgroup isextinct (Figs 6 7 Table 1) The extinct out-group is a necessary component of the def-inition Jefferies (1979449) coined the termldquocrown grouprdquo to replace Hennigrsquos (1969) group which Jefferies dened as ldquothe lat-est common ancestor of all living membersof group 1 plus all descendants of this an-cestor whether these are living or extinctrdquoTheexistence of an immediate outgroup thatis extinct is implied and was clearly presentin the examples given by Hennig and Jef-feries Patterson (1981207) also clearly usedcrown- and stem-groups in this mannerstating that the ldquostem-group contains fos-sil taxa that exhibit some but not all of thecharacters of the crown-grouprdquo Crown- andstem-group terminology has been used ex-clusively for higher taxa bounded by liv-ing species that have extinct outgroups (egde Queiroz and Gauthier 1992 Forey 1992McKenna and Bell 1997) Although it may

TABLE 2 Denitions for the various kinds of reference taxa

Reference taxon type Denition

Ingroup reference taxon included within a taxon by denitionOutgroup reference taxon excluded from a taxon by denitionInclusive most inclusive reference taxon among ingroup or outgroup taxaBasal reference taxon near the basal furcation of a taxonNested reference taxon distant from the basal furcation of a taxonVernacular vernacular name used as a reference taxon (eg ldquobirdsrdquo or ldquoextant birdsrdquo)

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 337

FIGURE 5 Kinds of reference taxa (a) An ingroup reference taxon is included by denition within a taxon anode-based denition is based on two ingroup reference taxa (b) An outgroup reference taxon is excluded bydenition from a taxon a stem-based denition is based on one ingroup and one outgroup reference taxon (c) Aninclusive reference taxon is the most inclusive ingroup or outgroup reference taxon possible inclusive referencetaxa B and C for taxon A allow denitional ambiguity with change in the relationship of a basal taxon (d) Abasal reference taxon is located at or near the base of an ingroup or outgroup taxon basal reference taxa allowdenitional ambiguity when their relationships change (e) A nested reference taxon is remote from the base ofan ingroup or outgroup reference taxon nested reference taxa are unaffected by changes in the relationships ofbasal taxa Reference taxa are encircled a dot indicates a node-based denition an arrow indicates a stem-baseddenition a dashed line indicates a change in position of a basal taxon

FIGURE 6 Crown stem and total taxa (a) Total group composed of a crown group plus a paraphyletic ldquostemgrouprdquo as conceived by Hennig (1969 1983) and Jefferies (1979) (b) Total taxon composed of a crown taxon andmonophyletic stem taxa as used in this article One stem taxon (left) is shown with a node-based denition andthe other (right) with a stem-based denition

338 SYSTEMATIC BIOLOGY VOL 48

FIGURE 7 Crown species or clades (enclosed) musthave an extinct (stem) taxon (dashed) as an immediateoutgroup (a) Crown species and crown clades are de-limited by the presence of an immediate outgroup thatis extinct (b) Often referred to as ldquoextantrdquo taxa crowntaxa may be composed primarily of extinct taxa ldquoLiv-ingrdquo and ldquoextinctrdquo conditions are treated as if they con-stitute plesiomorphic and apomorphic character statesrespectively of an irreversible character ldquostate of be-ingrdquo Crown clades under this conceptualization areplesiomorphy-based taxa bounded by two living taxafor which the immediate outgroup is extinct

be hypothetically true that ldquoevery Recentgroup whatever its position in thehierarchymay be expected to have had a stem-grouprdquo(Patterson 1981207) clades bounded by liv-ing taxa with immediate outgroups that arealso living have not been considered crowntaxa

Some recent denitions of crown taxahowever have set aside Hennigrsquos andJefferiesrsquo formulations as well as generalusage in the literature A less restric-tive meaning has been proposed basedsolely on reference taxa that include liv-

ing representatives De Queiroz and Gau-thier (1992469) dened crown taxa asldquoclades within which both branches ofthe basal dichotomy are represented byextant descendantsrdquo without mention ofthe condition of the immediate outgroupYet all of their examples of crown cladeshave immediate outgroups that are ex-tinct Regarding their craniate taxonomythey stated that each stem-based taxon iscomposed of a ldquocrown taxon plus all ex-tinct taxa more closely related to it thanto any other crown claderdquo (de Queirozand Gauthier 1992475) Lee (19961103)proposed that ldquoa crown clade denitionresults when both taxa implicated in anode-based denition are representedby extant formsrdquo Although the conditionof the immediate outgroup is never men-tioned Lee (19961103) stated that crowntaxon denitions apply to clades thatconsist of ldquothe most recent common an-cestor of all extant forms and all its de-scendantsrdquo (emphasis added) The pres-ence of an immediate extinct outgroup isimplied

The denition of a crown taxon presentedhere in contrast does not specify ingroupstructure (such as the presence or absenceof a basal dichotomy) or denitional type(node- or stem-based) and explicitly statesthat the immediate outgroup(s) is extinct(Fig 7a Table 1) Crown taxa so denedcan be mapped unerringly on a cladogramif (1) ldquolivingrdquo and ldquoextinctrdquo conditions aretreated as primitive and derived character-states respectively of the character ldquocurrentstate of beingrdquo and (2) if this character isregarded as irreversible (ie CaminndashSokalparsimony) The living state in other wordsis always regarded as plesiomorphic Theirreversible status of the character (state ofbeing) is important because it identiescrown groups on cladograms of living andextinct taxa that would have an ambigu-ous status if character-state reversal wereallowed (Fig 7b) This character-state con-ceptualization of the denition of a crowngroup captures the essence of crown taxawhich have always been used to refer tomaximally inclusive taxa dened by extantrepresentatives not simply any ldquoextant-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 339

boundedrdquo taxon Discovery of extinct sistertaxa provides the opportunity to recognizecrown taxa

Crown taxa are often regarded as cladesthat are bounded uniquely by living in-groups The use of living reference taxahowever is most often an arbitrary deci-sion Any extinct ingroup taxon that is mostclosely related to a living reference taxoncould be used to delimit the same clade(Fig 7b) A crown clade is dened uniquelyby living reference taxa only when such taxado not have extinct sister taxa (Fig 7b) afact that is not widely appreciated Crowntaxa in other words do not have any spe-cial biological signicance beyond signal-ing that particular reference taxa have sur-vived to an arbitrary datum (the Recent)Informal reference to crown taxa as ldquolivingrdquoor ldquoextantrdquo taxa therefore can be mislead-ing because a crown taxon may be com-posed predominantly of extinct species asis the case with the crown taxon Crocodylia(Fig 7b Brochu 1997) ldquoExtant crocodil-iansrdquo if taken literally refers to a smallparaphyletic subset of Crocodylia Explicitreference to ldquocrown crocodiliansrdquo or sim-ply ldquoCrocodyliardquo is preferable to use of theterms ldquolivingrdquo or ldquoextantrdquo

A living taxon is also essential to delimita maximally inclusive clade that is entirelyextinctmdasha stem taxon as here dened Astem taxon denotes an extinct species orclade the immediate outgroup of which in-cludes at least one living member (Fig 6b)Many extinct taxa are not stem taxa becausetheir immediate outgroup is itself extinctSuch extinct taxa are members of more in-clusive stem taxa Stem taxa can have eithernode- or stem-based denitions (Fig 6b)Stem taxa and their associated less inclusivecrown taxon make up the taxonomic contentof their respective total taxon

Stem taxa as here dened should notbe confused with ldquostem grouprdquo (Fig 6a)which was rst used in phylogenetic sys-tematics by Hennig (ldquostammgrupperdquo 19691983) Hennigrsquos (198315) vague formulationof the concept specied all extinct speciesldquowhich can be shown probably to belong toa particular monophyletic group of the phy-logenetic system but which are probably no

closer [sic] related to one subgroup amongrecent animals than to anotherrdquo (translatedfrom the German Ax 1987224) Jefferies(1979) claried the term to include all ex-tinct outgroups of a crown taxon (Fig 6a)Other authors have developed more convo-luted denitions that approximate the sameend Wiley (1981217) for example dened aldquostem grouprdquo as ldquoall fossil species of a groupmore primitive than the most primitive Re-cent species of the same grouprdquo Theseldquostem groupsrdquo are paraphyletic and oftencircumscribe the same ldquoancestral groupsrdquoand ldquoevolutionary gradesrdquo that have mud-died phylogenetic thinking since DarwinOnly a few cladists continue to endorse suchconcepts (Smith 1994) Ax (1987224) pro-posed a cumbersome solution to the prob-lem of paraphyletic ldquostem groupsrdquo by divid-ing them into ancestral ldquostem lineagesrdquo andldquorepresentatives of the stem lineagerdquo Thelatter are monophyletic taxa most simply re-ferred to as stem taxa

A total taxon is regarded here as a cladecomposed of a crown taxon plus all ex-tinct outgroups more closely related to itthan to another crown taxon (Fig 3 Ta-ble 1) Hennigrsquos (1969) and Jefferiesrsquo (1979)formulations of the concept are wordy butsimilar in effect because they restricted in-clusion of extinct outgroups to those mostclosely related to a particular crown taxonDe Queiroz and Gauthier (1992470) deneda total clade as a ldquomore inclusive clade con-sisting of thecrown clade plus its extinctout-groupsrdquo This denition is incomplete be-cause it does not specify which extinct out-groups are included A total taxon explicitlyincludes only those extinct outgroups mostclosely related to a particular crown taxon

Node-based and stem-based denitions areconstructed by (1) identifying reference taxaand (2) attaching to them a relational phrasethat either includes their most recent com-mon ancestor and all descendants (node-based) or limits inclusion to all descen-dants closer to one or more of the referencetaxa (stem-based) (de Queiroz and Gauthier1992) (Fig 1 Table 1) In node-based deni-tions the phrase ldquoleast-inclusive claderdquo canreplace ldquocommon ancestor and all descen-dantsrdquo for cladists sensitive to the allusion to

340 SYSTEMATIC BIOLOGY VOL 48

ancestors (Lee 1998) Listing included taxawithout a relational phrase about ancestryor simply designating taxa as node- or stem-based without reference taxa does not con-stitutea phylogenetic denition because theboundaries of the taxon in question are notspecied (de Queiroz and Gauthier 1990Bryant 1996)

In the current literature there exists someconfusion over what is necessary and suf-cient for node- and stem-based denitionsBryant (1996185) suggested that a more in-clusive taxon be inserted within a stem-based denition such as ldquoall members oftaxon A thataremoreclosely related to taxonB than to taxon Crdquo Including taxon A how-ever is not necessary and may lead to inter-nal conicts within the denition (eg whentaxon A is included within taxon B or C onan alternativephylogeny) The goalof a phy-logenetic denition should be to unambigu-ously identify a specic common ancestorgiven any arrangement of taxa rather thanto identify such an ancestor only on a subsetof phylogenetic hypotheses

Meng et al (1994) and Wyss and Meng(1996) introduced a ldquostem-modied node-based denitionrdquo to effectively unite crown-group rodents whose basal relationshipsare uncertain Their preferred denition ofRodentia (ldquothe clade stemming from themost recent common ancestor of Mus andall Recent mammals more closely relatedto Mus than to Lagomorpha or membersof any other eutherian lsquoorderrdquorsquo Wyss andMeng 1996562) is better interpreted as aspecial stem-based rather than node-baseddenition because it uses only one ingroupreference taxon (Mus) and one or more out-group reference taxa (eg Lagomorpha)Their denition merely adds a phrase (ldquotheclade stemming from the most recent com-mon ancestorrdquo) to an otherwise typicalstem-based denition based on living ref-erence taxa That phrase ensures inclusionof all extinct taxa within the crown cladeBryantrsquos (1996185) reformulation of thisdenition on the other hand is clearly anode-based denition because he speci-ed only ingroup reference taxa (ldquothe mostrecent common ancestor of the species re-ferred to Rodentia by Wilson and Reeder

(1993) and all of its descendantsrdquo) These ex-amples underscore the need for claricationof terms (Tables 1 2)

A node-stem triplet (NST) is a denitionalconguration involving three taxa a node-based taxon composed of two stem-basedsubordinate taxa (Sereno 1997 1998 Fig 8)

Taxon A Taxa b and c their most recentcommon ancestor and all its descendants

Taxon B All taxa closer to taxon b than totaxon c

Taxon C All taxa closer to taxon c than totaxon b

FIGURE 8 Denitional triumvirate termed a node-stem triplet composed of a node-based taxon (A) andtwo subordinate stem-based taxa (B and C) which in-corporates added or repositioned taxa (dashed lineswith daggers) without changing the relative taxonomiccontent of taxon A B or C (as expressed by the equiva-lence statementA = B +C) A dot indicates a node-baseddenition an arrow indicates a stem-based denition

Taxa b and c (not shown in Fig 8) constitutereference taxa within taxon B and C respec-tively Although not essential the same ref-erence taxa (b c) may be used for the trio oftaxa in a NST which then specify complemen-tary denitions The pair of stem-based sis-ter taxa in addition may have reciprocal def-initions that utilize the same reference taxain antipodal positions The word ldquotaxardquo inthe stem-based denitions listed above mayoptionally be replaced by reference toa moreinclusive taxon (Bryant 1996) A stem-baseddenition for Saurischia for example mightread ldquoAll dinosaurs more closely related

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 341

to Neornithes than to Triceratopsrdquo (Sereno1998)

Taxonomic equivalence is a statement (orequation) in which one taxon is shownto be equivalent in current and poten-tial taxonomic content to subordinate taxaplus their common ancestor (Table 1)The NST described above is composedof a node-based taxon (A) and twosubordinate stem-based taxa (B C) andcan be written as the equivalence state-ment ldquoA = B + Crdquo because taxon Aby denition is equivalent to taxa Band C plus their most recent commonancestor

PHYLOGENETIC DEFINITIONS RATIONALE

The rst step toward a phylogenetic tax-onomy involved the distinction betweendenition and diagnosis (Buck and Hull1966 Grifths 1973 1974 Farris 1976 Ghis-elin 1984 Rowe 1987) A second step in-volved the formulation of node- and stem-based denitions by use of reference taxaand relational statements about ancestry (deQueiroz and Gauthier 1992) A third stepoutlined here involves the specication ofguidelines for selecting effective referencetaxa and criteria for positioning node- andstem-based denitions

Phylogenetic rearrangements and introduc-tion of new taxa are the principal factorsthat destabilize taxonomic content Stabil-ity of taxonomic content is enhanced bychoosing (1) effective reference taxa and(2) an effective conguration of phyloge-netic denitions

Reference TaxaPositionmdashGiven a particular phyloge-

netic hypothesis the taxonomic content of aphylogenetic denition is altered only if thereference taxa specify an alternative com-mon ancestor Given a basal dichotomy withone reference taxon on each side the iden-tication of the immediate common ances-tor is not affected by relocation of a ref-erence taxon on its side of the basal di-chotomy (Fig 9 move 1) Relocation ofa reference taxon to the opposing side orto a location outside the basal dichotomyidenties a different common ancestor

(Fig 9 moves 2 3) For a taxon with a basaldichotomy an alternative common ances-tor can be identied only if one referencetaxon is relocated to the opposing side oroutside the basal dichotomy Are some ref-erence taxa more likely to be relocated in thismanner than others

FIGURE 9 Three possible outcomes from relocationof a reference taxon 1 = relocation on the same side ofa basal dichotomy has no effect 2 = relocation to theopposite side of the basal dichotomy identies a lessinclusive common ancestor 3 = relocation outside thebasal dichotomy identies a more inclusive commonancestor Reference taxa are encircled and the originalbasal dichotomy is shown with heavy lines

First a taxon positioned near the basaldichotomy of a cladogram is always eas-ier to relocate to the opposing side (ierequires fewer additional steps) than is anested taxon if all other factors are regardedas equal among ingroup taxa (eg complete-ness missing data and homoplasy) Unlikea basal taxon (Fig 5d) a nested taxon sharessynapomorphies that increase its phyloge-netic (patristic) distance from the commonancestor (Fig 5e) Thus stability of taxo-nomic content is enhanced in phylogeneticdenitions if nested reference taxa are cho-sen that are located at some distance (atleast several nodes away) from the basaldichotomy This distance is easy to assessquantitatively for various nested taxa on agiven phylogeny

Second maximally inclusive referencetaxa (Fig 5c) have greater potential to cre-ate unnecessary taxonomic redundancy andinternal inconsistencies after relocation oftaxa (Fig 3) This is true for both node-

342 SYSTEMATIC BIOLOGY VOL 48

and stem-based taxa In the current litera-ture maximally inclusive reference taxa areoften used in phylogenetic denitions Intheir tetrapod classication for example deQueiroz and Gauthier (1992475) used suc-cessive maximally inclusive reference taxafor all denitionsmdasha pattern here termed re-cursive referencing (Fig 10a) Recursive refer-encing can have undesirable consequencesbecause it shifts the burden of a more precisedenition to a slightly less inclusive highertaxon This higher taxon may not be de-ned or it too may be affected by an alter-native conguration of basal relationshipsSalgado et al (1997) for example choseProsauropoda and Sauropoda as maximallyinclusive reference taxa for Sauropodomor-pha If a basal sauropodomorph is reposi-tioned the denitions of Prosauropoda andSauropoda must be determined to assess po-tential taxonomic consequences HoweverSalgado et al (1997) used Prosauropodaas a terminal taxon without denitionSauropoda was dened as a node-basedtaxon on the basis of two maximally in-clusive reference taxa Vulcanodon and Eu-sauropoda (all other sauropods) Thusif the incompletely known basal sauro-pod Vulcanodon is reinterpreted as be-ing more closely related to prosauropodsSauropoda and Sauropodomorpha wouldbecome synonyms

If on the other hand Sauropodomorphais dened on the basis of the nested ref-erence taxamdashthe prosauropod Plateosaurusand sauropod Saltasaurus (Sereno 1998)mdasha change in the phylogenetic positionof any other prosauropod or sauropodhas no effect on the taxonomic con-tents of Sauropodomorpha or SauropodaSauropodomorpha can be dened as ldquoPla-teosaurus Saltasaurus their most recentcommon ancestor and all descendantsrdquoand Sauropoda can be dened as ldquoallsauropodomorphs more closely related toSaltasaurus than to Plateosaurusrdquo The phylo-genetic position of Vulcanodon has no effecton the identication of respective commonancestors

CompletenessmdashPoorly known taxa aremore likely to be phylogenetically unsta-ble than well-known taxa (Wilkinson 1995)

FIGURE 10 Referencing styles (a) Recursive refer-encing selects a succession of the most inclusive ref-erence taxa (b) Nested referencing selects that sameremote reference taxon Reference taxa are encircled

Taxa with numerous missing entries in aphylogenetic analysis usually reduce phylo-genetic resolution by generating numerousequally parsimonious trees Well-known ref-erence taxa even if they are somewhat lessnested are preferable to poorly known taxa(Fig 11) A single poorly known basal taxonis least stable and therefore least desirableas a reference taxon

Use with Recent or extinct ltersmdashldquoRecentrdquo(ldquoextantrdquo or ldquolivingrdquo) or ldquoextinctrdquo may beused in phylogenetic denitions of crownand stem taxa respectively to help to stabi-lize taxonomic content in the face of poorlyresolved basal relationships For a crowntaxon with a stem-based denition a Recentlter can restrict the boundaries of a cladeto living taxa (Meng et al 1994 Wyss andMeng 1996) Rodentia for example could

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 343

be dened as ldquothe least inclusive clade com-posed of Recent mammals more closely re-lated to Mus than to Lepusrdquo For stem taxain contrast a stem-based denition using anldquoextinctrdquo lter can restrict inclusion to ex-tinct taxa when outgroup relationships areuncertain Moas for example could be de-ned as ldquoall extinct species more closelyrelated to Dinornis than to Struthiordquo theostrich

NumbermdashThe number of reference taxathat have been used in phylogenetic deni-tions has varied from two to innity Gau-thier (198614) for example dened Aves asa node-based taxon including ldquoall descen-

FIGURE 11 Nestedness and completeness in the se-lection of reference taxa A nested well-known taxonis more likely to constitute a stable reference taxon (en-circled)

dants of the most recent common ancestorof Ratitae Tinami and Neognathaerdquo Threereference taxa rather than two were chosenpresumably because of unstated uncertaintyin the phylogenetic relationships amongthese three avian subgroups De Queirozand Gauthier (1990310 1992461) suggestedthat phylogenetic denitions should utilizeonly two reference taxamdashbut then erecteddenitions for total taxa that used as refer-ence taxa one crown clade plus ldquoall extinct

taxa more closely related to it than to anyother crown claderdquo (de Queiroz and Gau-thier 1992475)

Listing more than a pair of referencetaxa does not seem to compromise a node-based phylogenetic denition in any dis-cernible way Node-based denitions areconstructed with what may be termed in-ternal inclusion A contradictory relation-ship between multiple ingroup referencetaxa for a node-based denition is im-possible to construct with internal inclu-sion because a clade always exists thatincludes all ingroup reference taxa Multi-ple ingroup reference taxa effectively stabi-lize the taxonomic content of a node-basedtaxon if basal relationships are uncertain(Fig 12a)

For stem-based denitions multiple out-group reference taxa effectively stabilize thetaxonomic content of a taxon if outgroup re-lationships are uncertain (Fig 12b) Multi-ple ingroup reference taxa for stem-baseddenitions can create a contradictory sit-uation if one (or more) ingroup referencetaxa are repositioned closer to the outgroupreference taxon (Fig 12c) or to a positionoutside the clade as originally dened Inthis case the external inclusion fundamen-tal to a stem-based denition creates a con-tradictory relationship among the originalingroup reference taxa which are no longermore closely related to each other than to theoutgroup reference taxon (Fig 12c) In theexample shown here relocation of ingroupreference taxon D creates a contradictory re-lationship with a second ingroup referencetaxon E if taxon A has the stem-based def-inition of ldquoall descendants more closely re-lated to taxon D and E than to taxon Brdquo Aninclusive ingroup reference taxon howevercan manifest the same problems in this cir-cumstance (Fig 12c) If taxon A is denedas ldquoall descendants more closely related totaxon C than to taxon Brdquo and if taxon C isdened on the basis of taxon D a contra-dictory denition results with relocation oftaxon D as shown taxon A now subsumesits original ingroup reference taxon C Suchcontradictory denitions can be avoided ifa nested ingroup reference taxon (or nestedtaxa) are selected such as taxon E its un-

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 8: Sereno, 1999b

336 SYSTEMATIC BIOLOGY VOL 48

Taxonomic content is used here to refer toall existing and potential taxa included by thephylogenetic denition of a taxon (Table 1)The smallest number of subordinate taxathat can fulll this denition of taxonomiccontent is two a redundant taxon (ie asingle subordinate taxon) would have anidentical phylogenetic denition and thusis uninformative in phylogenetic taxonomy(Farris 1976) The taxonomic content oftaxon C therefore could be ldquotaxon A taxonB their most recent common ancestor andall descendantsrdquo if taxa A and B are denedextensionally to encompass all existing andpotential subordinate taxa currently withintaxon C

A list of included taxa is insufcient to un-ambiguously specify taxonomic content be-cause there will always be potential mem-bers of a taxon that lie outside any list ofincluded taxa whether or not the includedtaxa themselves have phylogenetic deni-tions An exhaustive list of included taxawith stem-based denitions for exampledoes not specify membership for their hy-pothetical immediate common ancestor (anode-based taxon) or more distant knownor hypothetical outgroups (a stem-basedtaxon) Two reference taxa and a relationalstatement about ancestry are necessary andsufcient to specify taxonomic content be-cause they unambiguously specify member-ship for all existing and potential membersof a taxon

A reference taxon is a species or higher-leveltaxon that is used to delimit a phylogeneticdenition (Table 1) For example ldquotaxon Ataxon B their most recent common ancestorand all descendantsrdquo uses A and B as refer-ence taxa for a node-based taxonomic deni-tion Previously cited as ldquoreference pointsrdquo

(Schander and Thollesson 1995) theseimportant components of phylogeneticdenitions have not been distinguishedterminologically

Several kinds of reference taxa can beidentied (Fig 5 Table 2) Ingroup and out-group reference taxa are located within oroutside the taxon they dene respectively(Fig 5a b) An inclusive reference taxon in-cludes several ingroup or outgroup taxaa maximally inclusive reference taxon in-cludes the largest possible ingroup or out-group reference taxon (Fig 5c) Basal andnested reference taxa are less inclusive andas their name suggests are located at thebase or nested within an ingroup or out-group taxon (Fig 5d e)

A crown taxon denotes a living speciesor a clade dened by using living refer-ence taxa whose immediate outgroup isextinct (Figs 6 7 Table 1) The extinct out-group is a necessary component of the def-inition Jefferies (1979449) coined the termldquocrown grouprdquo to replace Hennigrsquos (1969) group which Jefferies dened as ldquothe lat-est common ancestor of all living membersof group 1 plus all descendants of this an-cestor whether these are living or extinctrdquoTheexistence of an immediate outgroup thatis extinct is implied and was clearly presentin the examples given by Hennig and Jef-feries Patterson (1981207) also clearly usedcrown- and stem-groups in this mannerstating that the ldquostem-group contains fos-sil taxa that exhibit some but not all of thecharacters of the crown-grouprdquo Crown- andstem-group terminology has been used ex-clusively for higher taxa bounded by liv-ing species that have extinct outgroups (egde Queiroz and Gauthier 1992 Forey 1992McKenna and Bell 1997) Although it may

TABLE 2 Denitions for the various kinds of reference taxa

Reference taxon type Denition

Ingroup reference taxon included within a taxon by denitionOutgroup reference taxon excluded from a taxon by denitionInclusive most inclusive reference taxon among ingroup or outgroup taxaBasal reference taxon near the basal furcation of a taxonNested reference taxon distant from the basal furcation of a taxonVernacular vernacular name used as a reference taxon (eg ldquobirdsrdquo or ldquoextant birdsrdquo)

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 337

FIGURE 5 Kinds of reference taxa (a) An ingroup reference taxon is included by denition within a taxon anode-based denition is based on two ingroup reference taxa (b) An outgroup reference taxon is excluded bydenition from a taxon a stem-based denition is based on one ingroup and one outgroup reference taxon (c) Aninclusive reference taxon is the most inclusive ingroup or outgroup reference taxon possible inclusive referencetaxa B and C for taxon A allow denitional ambiguity with change in the relationship of a basal taxon (d) Abasal reference taxon is located at or near the base of an ingroup or outgroup taxon basal reference taxa allowdenitional ambiguity when their relationships change (e) A nested reference taxon is remote from the base ofan ingroup or outgroup reference taxon nested reference taxa are unaffected by changes in the relationships ofbasal taxa Reference taxa are encircled a dot indicates a node-based denition an arrow indicates a stem-baseddenition a dashed line indicates a change in position of a basal taxon

FIGURE 6 Crown stem and total taxa (a) Total group composed of a crown group plus a paraphyletic ldquostemgrouprdquo as conceived by Hennig (1969 1983) and Jefferies (1979) (b) Total taxon composed of a crown taxon andmonophyletic stem taxa as used in this article One stem taxon (left) is shown with a node-based denition andthe other (right) with a stem-based denition

338 SYSTEMATIC BIOLOGY VOL 48

FIGURE 7 Crown species or clades (enclosed) musthave an extinct (stem) taxon (dashed) as an immediateoutgroup (a) Crown species and crown clades are de-limited by the presence of an immediate outgroup thatis extinct (b) Often referred to as ldquoextantrdquo taxa crowntaxa may be composed primarily of extinct taxa ldquoLiv-ingrdquo and ldquoextinctrdquo conditions are treated as if they con-stitute plesiomorphic and apomorphic character statesrespectively of an irreversible character ldquostate of be-ingrdquo Crown clades under this conceptualization areplesiomorphy-based taxa bounded by two living taxafor which the immediate outgroup is extinct

be hypothetically true that ldquoevery Recentgroup whatever its position in thehierarchymay be expected to have had a stem-grouprdquo(Patterson 1981207) clades bounded by liv-ing taxa with immediate outgroups that arealso living have not been considered crowntaxa

Some recent denitions of crown taxahowever have set aside Hennigrsquos andJefferiesrsquo formulations as well as generalusage in the literature A less restric-tive meaning has been proposed basedsolely on reference taxa that include liv-

ing representatives De Queiroz and Gau-thier (1992469) dened crown taxa asldquoclades within which both branches ofthe basal dichotomy are represented byextant descendantsrdquo without mention ofthe condition of the immediate outgroupYet all of their examples of crown cladeshave immediate outgroups that are ex-tinct Regarding their craniate taxonomythey stated that each stem-based taxon iscomposed of a ldquocrown taxon plus all ex-tinct taxa more closely related to it thanto any other crown claderdquo (de Queirozand Gauthier 1992475) Lee (19961103)proposed that ldquoa crown clade denitionresults when both taxa implicated in anode-based denition are representedby extant formsrdquo Although the conditionof the immediate outgroup is never men-tioned Lee (19961103) stated that crowntaxon denitions apply to clades thatconsist of ldquothe most recent common an-cestor of all extant forms and all its de-scendantsrdquo (emphasis added) The pres-ence of an immediate extinct outgroup isimplied

The denition of a crown taxon presentedhere in contrast does not specify ingroupstructure (such as the presence or absenceof a basal dichotomy) or denitional type(node- or stem-based) and explicitly statesthat the immediate outgroup(s) is extinct(Fig 7a Table 1) Crown taxa so denedcan be mapped unerringly on a cladogramif (1) ldquolivingrdquo and ldquoextinctrdquo conditions aretreated as primitive and derived character-states respectively of the character ldquocurrentstate of beingrdquo and (2) if this character isregarded as irreversible (ie CaminndashSokalparsimony) The living state in other wordsis always regarded as plesiomorphic Theirreversible status of the character (state ofbeing) is important because it identiescrown groups on cladograms of living andextinct taxa that would have an ambigu-ous status if character-state reversal wereallowed (Fig 7b) This character-state con-ceptualization of the denition of a crowngroup captures the essence of crown taxawhich have always been used to refer tomaximally inclusive taxa dened by extantrepresentatives not simply any ldquoextant-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 339

boundedrdquo taxon Discovery of extinct sistertaxa provides the opportunity to recognizecrown taxa

Crown taxa are often regarded as cladesthat are bounded uniquely by living in-groups The use of living reference taxahowever is most often an arbitrary deci-sion Any extinct ingroup taxon that is mostclosely related to a living reference taxoncould be used to delimit the same clade(Fig 7b) A crown clade is dened uniquelyby living reference taxa only when such taxado not have extinct sister taxa (Fig 7b) afact that is not widely appreciated Crowntaxa in other words do not have any spe-cial biological signicance beyond signal-ing that particular reference taxa have sur-vived to an arbitrary datum (the Recent)Informal reference to crown taxa as ldquolivingrdquoor ldquoextantrdquo taxa therefore can be mislead-ing because a crown taxon may be com-posed predominantly of extinct species asis the case with the crown taxon Crocodylia(Fig 7b Brochu 1997) ldquoExtant crocodil-iansrdquo if taken literally refers to a smallparaphyletic subset of Crocodylia Explicitreference to ldquocrown crocodiliansrdquo or sim-ply ldquoCrocodyliardquo is preferable to use of theterms ldquolivingrdquo or ldquoextantrdquo

A living taxon is also essential to delimita maximally inclusive clade that is entirelyextinctmdasha stem taxon as here dened Astem taxon denotes an extinct species orclade the immediate outgroup of which in-cludes at least one living member (Fig 6b)Many extinct taxa are not stem taxa becausetheir immediate outgroup is itself extinctSuch extinct taxa are members of more in-clusive stem taxa Stem taxa can have eithernode- or stem-based denitions (Fig 6b)Stem taxa and their associated less inclusivecrown taxon make up the taxonomic contentof their respective total taxon

Stem taxa as here dened should notbe confused with ldquostem grouprdquo (Fig 6a)which was rst used in phylogenetic sys-tematics by Hennig (ldquostammgrupperdquo 19691983) Hennigrsquos (198315) vague formulationof the concept specied all extinct speciesldquowhich can be shown probably to belong toa particular monophyletic group of the phy-logenetic system but which are probably no

closer [sic] related to one subgroup amongrecent animals than to anotherrdquo (translatedfrom the German Ax 1987224) Jefferies(1979) claried the term to include all ex-tinct outgroups of a crown taxon (Fig 6a)Other authors have developed more convo-luted denitions that approximate the sameend Wiley (1981217) for example dened aldquostem grouprdquo as ldquoall fossil species of a groupmore primitive than the most primitive Re-cent species of the same grouprdquo Theseldquostem groupsrdquo are paraphyletic and oftencircumscribe the same ldquoancestral groupsrdquoand ldquoevolutionary gradesrdquo that have mud-died phylogenetic thinking since DarwinOnly a few cladists continue to endorse suchconcepts (Smith 1994) Ax (1987224) pro-posed a cumbersome solution to the prob-lem of paraphyletic ldquostem groupsrdquo by divid-ing them into ancestral ldquostem lineagesrdquo andldquorepresentatives of the stem lineagerdquo Thelatter are monophyletic taxa most simply re-ferred to as stem taxa

A total taxon is regarded here as a cladecomposed of a crown taxon plus all ex-tinct outgroups more closely related to itthan to another crown taxon (Fig 3 Ta-ble 1) Hennigrsquos (1969) and Jefferiesrsquo (1979)formulations of the concept are wordy butsimilar in effect because they restricted in-clusion of extinct outgroups to those mostclosely related to a particular crown taxonDe Queiroz and Gauthier (1992470) deneda total clade as a ldquomore inclusive clade con-sisting of thecrown clade plus its extinctout-groupsrdquo This denition is incomplete be-cause it does not specify which extinct out-groups are included A total taxon explicitlyincludes only those extinct outgroups mostclosely related to a particular crown taxon

Node-based and stem-based denitions areconstructed by (1) identifying reference taxaand (2) attaching to them a relational phrasethat either includes their most recent com-mon ancestor and all descendants (node-based) or limits inclusion to all descen-dants closer to one or more of the referencetaxa (stem-based) (de Queiroz and Gauthier1992) (Fig 1 Table 1) In node-based deni-tions the phrase ldquoleast-inclusive claderdquo canreplace ldquocommon ancestor and all descen-dantsrdquo for cladists sensitive to the allusion to

340 SYSTEMATIC BIOLOGY VOL 48

ancestors (Lee 1998) Listing included taxawithout a relational phrase about ancestryor simply designating taxa as node- or stem-based without reference taxa does not con-stitutea phylogenetic denition because theboundaries of the taxon in question are notspecied (de Queiroz and Gauthier 1990Bryant 1996)

In the current literature there exists someconfusion over what is necessary and suf-cient for node- and stem-based denitionsBryant (1996185) suggested that a more in-clusive taxon be inserted within a stem-based denition such as ldquoall members oftaxon A thataremoreclosely related to taxonB than to taxon Crdquo Including taxon A how-ever is not necessary and may lead to inter-nal conicts within the denition (eg whentaxon A is included within taxon B or C onan alternativephylogeny) The goalof a phy-logenetic denition should be to unambigu-ously identify a specic common ancestorgiven any arrangement of taxa rather thanto identify such an ancestor only on a subsetof phylogenetic hypotheses

Meng et al (1994) and Wyss and Meng(1996) introduced a ldquostem-modied node-based denitionrdquo to effectively unite crown-group rodents whose basal relationshipsare uncertain Their preferred denition ofRodentia (ldquothe clade stemming from themost recent common ancestor of Mus andall Recent mammals more closely relatedto Mus than to Lagomorpha or membersof any other eutherian lsquoorderrdquorsquo Wyss andMeng 1996562) is better interpreted as aspecial stem-based rather than node-baseddenition because it uses only one ingroupreference taxon (Mus) and one or more out-group reference taxa (eg Lagomorpha)Their denition merely adds a phrase (ldquotheclade stemming from the most recent com-mon ancestorrdquo) to an otherwise typicalstem-based denition based on living ref-erence taxa That phrase ensures inclusionof all extinct taxa within the crown cladeBryantrsquos (1996185) reformulation of thisdenition on the other hand is clearly anode-based denition because he speci-ed only ingroup reference taxa (ldquothe mostrecent common ancestor of the species re-ferred to Rodentia by Wilson and Reeder

(1993) and all of its descendantsrdquo) These ex-amples underscore the need for claricationof terms (Tables 1 2)

A node-stem triplet (NST) is a denitionalconguration involving three taxa a node-based taxon composed of two stem-basedsubordinate taxa (Sereno 1997 1998 Fig 8)

Taxon A Taxa b and c their most recentcommon ancestor and all its descendants

Taxon B All taxa closer to taxon b than totaxon c

Taxon C All taxa closer to taxon c than totaxon b

FIGURE 8 Denitional triumvirate termed a node-stem triplet composed of a node-based taxon (A) andtwo subordinate stem-based taxa (B and C) which in-corporates added or repositioned taxa (dashed lineswith daggers) without changing the relative taxonomiccontent of taxon A B or C (as expressed by the equiva-lence statementA = B +C) A dot indicates a node-baseddenition an arrow indicates a stem-based denition

Taxa b and c (not shown in Fig 8) constitutereference taxa within taxon B and C respec-tively Although not essential the same ref-erence taxa (b c) may be used for the trio oftaxa in a NST which then specify complemen-tary denitions The pair of stem-based sis-ter taxa in addition may have reciprocal def-initions that utilize the same reference taxain antipodal positions The word ldquotaxardquo inthe stem-based denitions listed above mayoptionally be replaced by reference toa moreinclusive taxon (Bryant 1996) A stem-baseddenition for Saurischia for example mightread ldquoAll dinosaurs more closely related

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 341

to Neornithes than to Triceratopsrdquo (Sereno1998)

Taxonomic equivalence is a statement (orequation) in which one taxon is shownto be equivalent in current and poten-tial taxonomic content to subordinate taxaplus their common ancestor (Table 1)The NST described above is composedof a node-based taxon (A) and twosubordinate stem-based taxa (B C) andcan be written as the equivalence state-ment ldquoA = B + Crdquo because taxon Aby denition is equivalent to taxa Band C plus their most recent commonancestor

PHYLOGENETIC DEFINITIONS RATIONALE

The rst step toward a phylogenetic tax-onomy involved the distinction betweendenition and diagnosis (Buck and Hull1966 Grifths 1973 1974 Farris 1976 Ghis-elin 1984 Rowe 1987) A second step in-volved the formulation of node- and stem-based denitions by use of reference taxaand relational statements about ancestry (deQueiroz and Gauthier 1992) A third stepoutlined here involves the specication ofguidelines for selecting effective referencetaxa and criteria for positioning node- andstem-based denitions

Phylogenetic rearrangements and introduc-tion of new taxa are the principal factorsthat destabilize taxonomic content Stabil-ity of taxonomic content is enhanced bychoosing (1) effective reference taxa and(2) an effective conguration of phyloge-netic denitions

Reference TaxaPositionmdashGiven a particular phyloge-

netic hypothesis the taxonomic content of aphylogenetic denition is altered only if thereference taxa specify an alternative com-mon ancestor Given a basal dichotomy withone reference taxon on each side the iden-tication of the immediate common ances-tor is not affected by relocation of a ref-erence taxon on its side of the basal di-chotomy (Fig 9 move 1) Relocation ofa reference taxon to the opposing side orto a location outside the basal dichotomyidenties a different common ancestor

(Fig 9 moves 2 3) For a taxon with a basaldichotomy an alternative common ances-tor can be identied only if one referencetaxon is relocated to the opposing side oroutside the basal dichotomy Are some ref-erence taxa more likely to be relocated in thismanner than others

FIGURE 9 Three possible outcomes from relocationof a reference taxon 1 = relocation on the same side ofa basal dichotomy has no effect 2 = relocation to theopposite side of the basal dichotomy identies a lessinclusive common ancestor 3 = relocation outside thebasal dichotomy identies a more inclusive commonancestor Reference taxa are encircled and the originalbasal dichotomy is shown with heavy lines

First a taxon positioned near the basaldichotomy of a cladogram is always eas-ier to relocate to the opposing side (ierequires fewer additional steps) than is anested taxon if all other factors are regardedas equal among ingroup taxa (eg complete-ness missing data and homoplasy) Unlikea basal taxon (Fig 5d) a nested taxon sharessynapomorphies that increase its phyloge-netic (patristic) distance from the commonancestor (Fig 5e) Thus stability of taxo-nomic content is enhanced in phylogeneticdenitions if nested reference taxa are cho-sen that are located at some distance (atleast several nodes away) from the basaldichotomy This distance is easy to assessquantitatively for various nested taxa on agiven phylogeny

Second maximally inclusive referencetaxa (Fig 5c) have greater potential to cre-ate unnecessary taxonomic redundancy andinternal inconsistencies after relocation oftaxa (Fig 3) This is true for both node-

342 SYSTEMATIC BIOLOGY VOL 48

and stem-based taxa In the current litera-ture maximally inclusive reference taxa areoften used in phylogenetic denitions Intheir tetrapod classication for example deQueiroz and Gauthier (1992475) used suc-cessive maximally inclusive reference taxafor all denitionsmdasha pattern here termed re-cursive referencing (Fig 10a) Recursive refer-encing can have undesirable consequencesbecause it shifts the burden of a more precisedenition to a slightly less inclusive highertaxon This higher taxon may not be de-ned or it too may be affected by an alter-native conguration of basal relationshipsSalgado et al (1997) for example choseProsauropoda and Sauropoda as maximallyinclusive reference taxa for Sauropodomor-pha If a basal sauropodomorph is reposi-tioned the denitions of Prosauropoda andSauropoda must be determined to assess po-tential taxonomic consequences HoweverSalgado et al (1997) used Prosauropodaas a terminal taxon without denitionSauropoda was dened as a node-basedtaxon on the basis of two maximally in-clusive reference taxa Vulcanodon and Eu-sauropoda (all other sauropods) Thusif the incompletely known basal sauro-pod Vulcanodon is reinterpreted as be-ing more closely related to prosauropodsSauropoda and Sauropodomorpha wouldbecome synonyms

If on the other hand Sauropodomorphais dened on the basis of the nested ref-erence taxamdashthe prosauropod Plateosaurusand sauropod Saltasaurus (Sereno 1998)mdasha change in the phylogenetic positionof any other prosauropod or sauropodhas no effect on the taxonomic con-tents of Sauropodomorpha or SauropodaSauropodomorpha can be dened as ldquoPla-teosaurus Saltasaurus their most recentcommon ancestor and all descendantsrdquoand Sauropoda can be dened as ldquoallsauropodomorphs more closely related toSaltasaurus than to Plateosaurusrdquo The phylo-genetic position of Vulcanodon has no effecton the identication of respective commonancestors

CompletenessmdashPoorly known taxa aremore likely to be phylogenetically unsta-ble than well-known taxa (Wilkinson 1995)

FIGURE 10 Referencing styles (a) Recursive refer-encing selects a succession of the most inclusive ref-erence taxa (b) Nested referencing selects that sameremote reference taxon Reference taxa are encircled

Taxa with numerous missing entries in aphylogenetic analysis usually reduce phylo-genetic resolution by generating numerousequally parsimonious trees Well-known ref-erence taxa even if they are somewhat lessnested are preferable to poorly known taxa(Fig 11) A single poorly known basal taxonis least stable and therefore least desirableas a reference taxon

Use with Recent or extinct ltersmdashldquoRecentrdquo(ldquoextantrdquo or ldquolivingrdquo) or ldquoextinctrdquo may beused in phylogenetic denitions of crownand stem taxa respectively to help to stabi-lize taxonomic content in the face of poorlyresolved basal relationships For a crowntaxon with a stem-based denition a Recentlter can restrict the boundaries of a cladeto living taxa (Meng et al 1994 Wyss andMeng 1996) Rodentia for example could

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 343

be dened as ldquothe least inclusive clade com-posed of Recent mammals more closely re-lated to Mus than to Lepusrdquo For stem taxain contrast a stem-based denition using anldquoextinctrdquo lter can restrict inclusion to ex-tinct taxa when outgroup relationships areuncertain Moas for example could be de-ned as ldquoall extinct species more closelyrelated to Dinornis than to Struthiordquo theostrich

NumbermdashThe number of reference taxathat have been used in phylogenetic deni-tions has varied from two to innity Gau-thier (198614) for example dened Aves asa node-based taxon including ldquoall descen-

FIGURE 11 Nestedness and completeness in the se-lection of reference taxa A nested well-known taxonis more likely to constitute a stable reference taxon (en-circled)

dants of the most recent common ancestorof Ratitae Tinami and Neognathaerdquo Threereference taxa rather than two were chosenpresumably because of unstated uncertaintyin the phylogenetic relationships amongthese three avian subgroups De Queirozand Gauthier (1990310 1992461) suggestedthat phylogenetic denitions should utilizeonly two reference taxamdashbut then erecteddenitions for total taxa that used as refer-ence taxa one crown clade plus ldquoall extinct

taxa more closely related to it than to anyother crown claderdquo (de Queiroz and Gau-thier 1992475)

Listing more than a pair of referencetaxa does not seem to compromise a node-based phylogenetic denition in any dis-cernible way Node-based denitions areconstructed with what may be termed in-ternal inclusion A contradictory relation-ship between multiple ingroup referencetaxa for a node-based denition is im-possible to construct with internal inclu-sion because a clade always exists thatincludes all ingroup reference taxa Multi-ple ingroup reference taxa effectively stabi-lize the taxonomic content of a node-basedtaxon if basal relationships are uncertain(Fig 12a)

For stem-based denitions multiple out-group reference taxa effectively stabilize thetaxonomic content of a taxon if outgroup re-lationships are uncertain (Fig 12b) Multi-ple ingroup reference taxa for stem-baseddenitions can create a contradictory sit-uation if one (or more) ingroup referencetaxa are repositioned closer to the outgroupreference taxon (Fig 12c) or to a positionoutside the clade as originally dened Inthis case the external inclusion fundamen-tal to a stem-based denition creates a con-tradictory relationship among the originalingroup reference taxa which are no longermore closely related to each other than to theoutgroup reference taxon (Fig 12c) In theexample shown here relocation of ingroupreference taxon D creates a contradictory re-lationship with a second ingroup referencetaxon E if taxon A has the stem-based def-inition of ldquoall descendants more closely re-lated to taxon D and E than to taxon Brdquo Aninclusive ingroup reference taxon howevercan manifest the same problems in this cir-cumstance (Fig 12c) If taxon A is denedas ldquoall descendants more closely related totaxon C than to taxon Brdquo and if taxon C isdened on the basis of taxon D a contra-dictory denition results with relocation oftaxon D as shown taxon A now subsumesits original ingroup reference taxon C Suchcontradictory denitions can be avoided ifa nested ingroup reference taxon (or nestedtaxa) are selected such as taxon E its un-

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 9: Sereno, 1999b

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 337

FIGURE 5 Kinds of reference taxa (a) An ingroup reference taxon is included by denition within a taxon anode-based denition is based on two ingroup reference taxa (b) An outgroup reference taxon is excluded bydenition from a taxon a stem-based denition is based on one ingroup and one outgroup reference taxon (c) Aninclusive reference taxon is the most inclusive ingroup or outgroup reference taxon possible inclusive referencetaxa B and C for taxon A allow denitional ambiguity with change in the relationship of a basal taxon (d) Abasal reference taxon is located at or near the base of an ingroup or outgroup taxon basal reference taxa allowdenitional ambiguity when their relationships change (e) A nested reference taxon is remote from the base ofan ingroup or outgroup reference taxon nested reference taxa are unaffected by changes in the relationships ofbasal taxa Reference taxa are encircled a dot indicates a node-based denition an arrow indicates a stem-baseddenition a dashed line indicates a change in position of a basal taxon

FIGURE 6 Crown stem and total taxa (a) Total group composed of a crown group plus a paraphyletic ldquostemgrouprdquo as conceived by Hennig (1969 1983) and Jefferies (1979) (b) Total taxon composed of a crown taxon andmonophyletic stem taxa as used in this article One stem taxon (left) is shown with a node-based denition andthe other (right) with a stem-based denition

338 SYSTEMATIC BIOLOGY VOL 48

FIGURE 7 Crown species or clades (enclosed) musthave an extinct (stem) taxon (dashed) as an immediateoutgroup (a) Crown species and crown clades are de-limited by the presence of an immediate outgroup thatis extinct (b) Often referred to as ldquoextantrdquo taxa crowntaxa may be composed primarily of extinct taxa ldquoLiv-ingrdquo and ldquoextinctrdquo conditions are treated as if they con-stitute plesiomorphic and apomorphic character statesrespectively of an irreversible character ldquostate of be-ingrdquo Crown clades under this conceptualization areplesiomorphy-based taxa bounded by two living taxafor which the immediate outgroup is extinct

be hypothetically true that ldquoevery Recentgroup whatever its position in thehierarchymay be expected to have had a stem-grouprdquo(Patterson 1981207) clades bounded by liv-ing taxa with immediate outgroups that arealso living have not been considered crowntaxa

Some recent denitions of crown taxahowever have set aside Hennigrsquos andJefferiesrsquo formulations as well as generalusage in the literature A less restric-tive meaning has been proposed basedsolely on reference taxa that include liv-

ing representatives De Queiroz and Gau-thier (1992469) dened crown taxa asldquoclades within which both branches ofthe basal dichotomy are represented byextant descendantsrdquo without mention ofthe condition of the immediate outgroupYet all of their examples of crown cladeshave immediate outgroups that are ex-tinct Regarding their craniate taxonomythey stated that each stem-based taxon iscomposed of a ldquocrown taxon plus all ex-tinct taxa more closely related to it thanto any other crown claderdquo (de Queirozand Gauthier 1992475) Lee (19961103)proposed that ldquoa crown clade denitionresults when both taxa implicated in anode-based denition are representedby extant formsrdquo Although the conditionof the immediate outgroup is never men-tioned Lee (19961103) stated that crowntaxon denitions apply to clades thatconsist of ldquothe most recent common an-cestor of all extant forms and all its de-scendantsrdquo (emphasis added) The pres-ence of an immediate extinct outgroup isimplied

The denition of a crown taxon presentedhere in contrast does not specify ingroupstructure (such as the presence or absenceof a basal dichotomy) or denitional type(node- or stem-based) and explicitly statesthat the immediate outgroup(s) is extinct(Fig 7a Table 1) Crown taxa so denedcan be mapped unerringly on a cladogramif (1) ldquolivingrdquo and ldquoextinctrdquo conditions aretreated as primitive and derived character-states respectively of the character ldquocurrentstate of beingrdquo and (2) if this character isregarded as irreversible (ie CaminndashSokalparsimony) The living state in other wordsis always regarded as plesiomorphic Theirreversible status of the character (state ofbeing) is important because it identiescrown groups on cladograms of living andextinct taxa that would have an ambigu-ous status if character-state reversal wereallowed (Fig 7b) This character-state con-ceptualization of the denition of a crowngroup captures the essence of crown taxawhich have always been used to refer tomaximally inclusive taxa dened by extantrepresentatives not simply any ldquoextant-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 339

boundedrdquo taxon Discovery of extinct sistertaxa provides the opportunity to recognizecrown taxa

Crown taxa are often regarded as cladesthat are bounded uniquely by living in-groups The use of living reference taxahowever is most often an arbitrary deci-sion Any extinct ingroup taxon that is mostclosely related to a living reference taxoncould be used to delimit the same clade(Fig 7b) A crown clade is dened uniquelyby living reference taxa only when such taxado not have extinct sister taxa (Fig 7b) afact that is not widely appreciated Crowntaxa in other words do not have any spe-cial biological signicance beyond signal-ing that particular reference taxa have sur-vived to an arbitrary datum (the Recent)Informal reference to crown taxa as ldquolivingrdquoor ldquoextantrdquo taxa therefore can be mislead-ing because a crown taxon may be com-posed predominantly of extinct species asis the case with the crown taxon Crocodylia(Fig 7b Brochu 1997) ldquoExtant crocodil-iansrdquo if taken literally refers to a smallparaphyletic subset of Crocodylia Explicitreference to ldquocrown crocodiliansrdquo or sim-ply ldquoCrocodyliardquo is preferable to use of theterms ldquolivingrdquo or ldquoextantrdquo

A living taxon is also essential to delimita maximally inclusive clade that is entirelyextinctmdasha stem taxon as here dened Astem taxon denotes an extinct species orclade the immediate outgroup of which in-cludes at least one living member (Fig 6b)Many extinct taxa are not stem taxa becausetheir immediate outgroup is itself extinctSuch extinct taxa are members of more in-clusive stem taxa Stem taxa can have eithernode- or stem-based denitions (Fig 6b)Stem taxa and their associated less inclusivecrown taxon make up the taxonomic contentof their respective total taxon

Stem taxa as here dened should notbe confused with ldquostem grouprdquo (Fig 6a)which was rst used in phylogenetic sys-tematics by Hennig (ldquostammgrupperdquo 19691983) Hennigrsquos (198315) vague formulationof the concept specied all extinct speciesldquowhich can be shown probably to belong toa particular monophyletic group of the phy-logenetic system but which are probably no

closer [sic] related to one subgroup amongrecent animals than to anotherrdquo (translatedfrom the German Ax 1987224) Jefferies(1979) claried the term to include all ex-tinct outgroups of a crown taxon (Fig 6a)Other authors have developed more convo-luted denitions that approximate the sameend Wiley (1981217) for example dened aldquostem grouprdquo as ldquoall fossil species of a groupmore primitive than the most primitive Re-cent species of the same grouprdquo Theseldquostem groupsrdquo are paraphyletic and oftencircumscribe the same ldquoancestral groupsrdquoand ldquoevolutionary gradesrdquo that have mud-died phylogenetic thinking since DarwinOnly a few cladists continue to endorse suchconcepts (Smith 1994) Ax (1987224) pro-posed a cumbersome solution to the prob-lem of paraphyletic ldquostem groupsrdquo by divid-ing them into ancestral ldquostem lineagesrdquo andldquorepresentatives of the stem lineagerdquo Thelatter are monophyletic taxa most simply re-ferred to as stem taxa

A total taxon is regarded here as a cladecomposed of a crown taxon plus all ex-tinct outgroups more closely related to itthan to another crown taxon (Fig 3 Ta-ble 1) Hennigrsquos (1969) and Jefferiesrsquo (1979)formulations of the concept are wordy butsimilar in effect because they restricted in-clusion of extinct outgroups to those mostclosely related to a particular crown taxonDe Queiroz and Gauthier (1992470) deneda total clade as a ldquomore inclusive clade con-sisting of thecrown clade plus its extinctout-groupsrdquo This denition is incomplete be-cause it does not specify which extinct out-groups are included A total taxon explicitlyincludes only those extinct outgroups mostclosely related to a particular crown taxon

Node-based and stem-based denitions areconstructed by (1) identifying reference taxaand (2) attaching to them a relational phrasethat either includes their most recent com-mon ancestor and all descendants (node-based) or limits inclusion to all descen-dants closer to one or more of the referencetaxa (stem-based) (de Queiroz and Gauthier1992) (Fig 1 Table 1) In node-based deni-tions the phrase ldquoleast-inclusive claderdquo canreplace ldquocommon ancestor and all descen-dantsrdquo for cladists sensitive to the allusion to

340 SYSTEMATIC BIOLOGY VOL 48

ancestors (Lee 1998) Listing included taxawithout a relational phrase about ancestryor simply designating taxa as node- or stem-based without reference taxa does not con-stitutea phylogenetic denition because theboundaries of the taxon in question are notspecied (de Queiroz and Gauthier 1990Bryant 1996)

In the current literature there exists someconfusion over what is necessary and suf-cient for node- and stem-based denitionsBryant (1996185) suggested that a more in-clusive taxon be inserted within a stem-based denition such as ldquoall members oftaxon A thataremoreclosely related to taxonB than to taxon Crdquo Including taxon A how-ever is not necessary and may lead to inter-nal conicts within the denition (eg whentaxon A is included within taxon B or C onan alternativephylogeny) The goalof a phy-logenetic denition should be to unambigu-ously identify a specic common ancestorgiven any arrangement of taxa rather thanto identify such an ancestor only on a subsetof phylogenetic hypotheses

Meng et al (1994) and Wyss and Meng(1996) introduced a ldquostem-modied node-based denitionrdquo to effectively unite crown-group rodents whose basal relationshipsare uncertain Their preferred denition ofRodentia (ldquothe clade stemming from themost recent common ancestor of Mus andall Recent mammals more closely relatedto Mus than to Lagomorpha or membersof any other eutherian lsquoorderrdquorsquo Wyss andMeng 1996562) is better interpreted as aspecial stem-based rather than node-baseddenition because it uses only one ingroupreference taxon (Mus) and one or more out-group reference taxa (eg Lagomorpha)Their denition merely adds a phrase (ldquotheclade stemming from the most recent com-mon ancestorrdquo) to an otherwise typicalstem-based denition based on living ref-erence taxa That phrase ensures inclusionof all extinct taxa within the crown cladeBryantrsquos (1996185) reformulation of thisdenition on the other hand is clearly anode-based denition because he speci-ed only ingroup reference taxa (ldquothe mostrecent common ancestor of the species re-ferred to Rodentia by Wilson and Reeder

(1993) and all of its descendantsrdquo) These ex-amples underscore the need for claricationof terms (Tables 1 2)

A node-stem triplet (NST) is a denitionalconguration involving three taxa a node-based taxon composed of two stem-basedsubordinate taxa (Sereno 1997 1998 Fig 8)

Taxon A Taxa b and c their most recentcommon ancestor and all its descendants

Taxon B All taxa closer to taxon b than totaxon c

Taxon C All taxa closer to taxon c than totaxon b

FIGURE 8 Denitional triumvirate termed a node-stem triplet composed of a node-based taxon (A) andtwo subordinate stem-based taxa (B and C) which in-corporates added or repositioned taxa (dashed lineswith daggers) without changing the relative taxonomiccontent of taxon A B or C (as expressed by the equiva-lence statementA = B +C) A dot indicates a node-baseddenition an arrow indicates a stem-based denition

Taxa b and c (not shown in Fig 8) constitutereference taxa within taxon B and C respec-tively Although not essential the same ref-erence taxa (b c) may be used for the trio oftaxa in a NST which then specify complemen-tary denitions The pair of stem-based sis-ter taxa in addition may have reciprocal def-initions that utilize the same reference taxain antipodal positions The word ldquotaxardquo inthe stem-based denitions listed above mayoptionally be replaced by reference toa moreinclusive taxon (Bryant 1996) A stem-baseddenition for Saurischia for example mightread ldquoAll dinosaurs more closely related

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 341

to Neornithes than to Triceratopsrdquo (Sereno1998)

Taxonomic equivalence is a statement (orequation) in which one taxon is shownto be equivalent in current and poten-tial taxonomic content to subordinate taxaplus their common ancestor (Table 1)The NST described above is composedof a node-based taxon (A) and twosubordinate stem-based taxa (B C) andcan be written as the equivalence state-ment ldquoA = B + Crdquo because taxon Aby denition is equivalent to taxa Band C plus their most recent commonancestor

PHYLOGENETIC DEFINITIONS RATIONALE

The rst step toward a phylogenetic tax-onomy involved the distinction betweendenition and diagnosis (Buck and Hull1966 Grifths 1973 1974 Farris 1976 Ghis-elin 1984 Rowe 1987) A second step in-volved the formulation of node- and stem-based denitions by use of reference taxaand relational statements about ancestry (deQueiroz and Gauthier 1992) A third stepoutlined here involves the specication ofguidelines for selecting effective referencetaxa and criteria for positioning node- andstem-based denitions

Phylogenetic rearrangements and introduc-tion of new taxa are the principal factorsthat destabilize taxonomic content Stabil-ity of taxonomic content is enhanced bychoosing (1) effective reference taxa and(2) an effective conguration of phyloge-netic denitions

Reference TaxaPositionmdashGiven a particular phyloge-

netic hypothesis the taxonomic content of aphylogenetic denition is altered only if thereference taxa specify an alternative com-mon ancestor Given a basal dichotomy withone reference taxon on each side the iden-tication of the immediate common ances-tor is not affected by relocation of a ref-erence taxon on its side of the basal di-chotomy (Fig 9 move 1) Relocation ofa reference taxon to the opposing side orto a location outside the basal dichotomyidenties a different common ancestor

(Fig 9 moves 2 3) For a taxon with a basaldichotomy an alternative common ances-tor can be identied only if one referencetaxon is relocated to the opposing side oroutside the basal dichotomy Are some ref-erence taxa more likely to be relocated in thismanner than others

FIGURE 9 Three possible outcomes from relocationof a reference taxon 1 = relocation on the same side ofa basal dichotomy has no effect 2 = relocation to theopposite side of the basal dichotomy identies a lessinclusive common ancestor 3 = relocation outside thebasal dichotomy identies a more inclusive commonancestor Reference taxa are encircled and the originalbasal dichotomy is shown with heavy lines

First a taxon positioned near the basaldichotomy of a cladogram is always eas-ier to relocate to the opposing side (ierequires fewer additional steps) than is anested taxon if all other factors are regardedas equal among ingroup taxa (eg complete-ness missing data and homoplasy) Unlikea basal taxon (Fig 5d) a nested taxon sharessynapomorphies that increase its phyloge-netic (patristic) distance from the commonancestor (Fig 5e) Thus stability of taxo-nomic content is enhanced in phylogeneticdenitions if nested reference taxa are cho-sen that are located at some distance (atleast several nodes away) from the basaldichotomy This distance is easy to assessquantitatively for various nested taxa on agiven phylogeny

Second maximally inclusive referencetaxa (Fig 5c) have greater potential to cre-ate unnecessary taxonomic redundancy andinternal inconsistencies after relocation oftaxa (Fig 3) This is true for both node-

342 SYSTEMATIC BIOLOGY VOL 48

and stem-based taxa In the current litera-ture maximally inclusive reference taxa areoften used in phylogenetic denitions Intheir tetrapod classication for example deQueiroz and Gauthier (1992475) used suc-cessive maximally inclusive reference taxafor all denitionsmdasha pattern here termed re-cursive referencing (Fig 10a) Recursive refer-encing can have undesirable consequencesbecause it shifts the burden of a more precisedenition to a slightly less inclusive highertaxon This higher taxon may not be de-ned or it too may be affected by an alter-native conguration of basal relationshipsSalgado et al (1997) for example choseProsauropoda and Sauropoda as maximallyinclusive reference taxa for Sauropodomor-pha If a basal sauropodomorph is reposi-tioned the denitions of Prosauropoda andSauropoda must be determined to assess po-tential taxonomic consequences HoweverSalgado et al (1997) used Prosauropodaas a terminal taxon without denitionSauropoda was dened as a node-basedtaxon on the basis of two maximally in-clusive reference taxa Vulcanodon and Eu-sauropoda (all other sauropods) Thusif the incompletely known basal sauro-pod Vulcanodon is reinterpreted as be-ing more closely related to prosauropodsSauropoda and Sauropodomorpha wouldbecome synonyms

If on the other hand Sauropodomorphais dened on the basis of the nested ref-erence taxamdashthe prosauropod Plateosaurusand sauropod Saltasaurus (Sereno 1998)mdasha change in the phylogenetic positionof any other prosauropod or sauropodhas no effect on the taxonomic con-tents of Sauropodomorpha or SauropodaSauropodomorpha can be dened as ldquoPla-teosaurus Saltasaurus their most recentcommon ancestor and all descendantsrdquoand Sauropoda can be dened as ldquoallsauropodomorphs more closely related toSaltasaurus than to Plateosaurusrdquo The phylo-genetic position of Vulcanodon has no effecton the identication of respective commonancestors

CompletenessmdashPoorly known taxa aremore likely to be phylogenetically unsta-ble than well-known taxa (Wilkinson 1995)

FIGURE 10 Referencing styles (a) Recursive refer-encing selects a succession of the most inclusive ref-erence taxa (b) Nested referencing selects that sameremote reference taxon Reference taxa are encircled

Taxa with numerous missing entries in aphylogenetic analysis usually reduce phylo-genetic resolution by generating numerousequally parsimonious trees Well-known ref-erence taxa even if they are somewhat lessnested are preferable to poorly known taxa(Fig 11) A single poorly known basal taxonis least stable and therefore least desirableas a reference taxon

Use with Recent or extinct ltersmdashldquoRecentrdquo(ldquoextantrdquo or ldquolivingrdquo) or ldquoextinctrdquo may beused in phylogenetic denitions of crownand stem taxa respectively to help to stabi-lize taxonomic content in the face of poorlyresolved basal relationships For a crowntaxon with a stem-based denition a Recentlter can restrict the boundaries of a cladeto living taxa (Meng et al 1994 Wyss andMeng 1996) Rodentia for example could

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 343

be dened as ldquothe least inclusive clade com-posed of Recent mammals more closely re-lated to Mus than to Lepusrdquo For stem taxain contrast a stem-based denition using anldquoextinctrdquo lter can restrict inclusion to ex-tinct taxa when outgroup relationships areuncertain Moas for example could be de-ned as ldquoall extinct species more closelyrelated to Dinornis than to Struthiordquo theostrich

NumbermdashThe number of reference taxathat have been used in phylogenetic deni-tions has varied from two to innity Gau-thier (198614) for example dened Aves asa node-based taxon including ldquoall descen-

FIGURE 11 Nestedness and completeness in the se-lection of reference taxa A nested well-known taxonis more likely to constitute a stable reference taxon (en-circled)

dants of the most recent common ancestorof Ratitae Tinami and Neognathaerdquo Threereference taxa rather than two were chosenpresumably because of unstated uncertaintyin the phylogenetic relationships amongthese three avian subgroups De Queirozand Gauthier (1990310 1992461) suggestedthat phylogenetic denitions should utilizeonly two reference taxamdashbut then erecteddenitions for total taxa that used as refer-ence taxa one crown clade plus ldquoall extinct

taxa more closely related to it than to anyother crown claderdquo (de Queiroz and Gau-thier 1992475)

Listing more than a pair of referencetaxa does not seem to compromise a node-based phylogenetic denition in any dis-cernible way Node-based denitions areconstructed with what may be termed in-ternal inclusion A contradictory relation-ship between multiple ingroup referencetaxa for a node-based denition is im-possible to construct with internal inclu-sion because a clade always exists thatincludes all ingroup reference taxa Multi-ple ingroup reference taxa effectively stabi-lize the taxonomic content of a node-basedtaxon if basal relationships are uncertain(Fig 12a)

For stem-based denitions multiple out-group reference taxa effectively stabilize thetaxonomic content of a taxon if outgroup re-lationships are uncertain (Fig 12b) Multi-ple ingroup reference taxa for stem-baseddenitions can create a contradictory sit-uation if one (or more) ingroup referencetaxa are repositioned closer to the outgroupreference taxon (Fig 12c) or to a positionoutside the clade as originally dened Inthis case the external inclusion fundamen-tal to a stem-based denition creates a con-tradictory relationship among the originalingroup reference taxa which are no longermore closely related to each other than to theoutgroup reference taxon (Fig 12c) In theexample shown here relocation of ingroupreference taxon D creates a contradictory re-lationship with a second ingroup referencetaxon E if taxon A has the stem-based def-inition of ldquoall descendants more closely re-lated to taxon D and E than to taxon Brdquo Aninclusive ingroup reference taxon howevercan manifest the same problems in this cir-cumstance (Fig 12c) If taxon A is denedas ldquoall descendants more closely related totaxon C than to taxon Brdquo and if taxon C isdened on the basis of taxon D a contra-dictory denition results with relocation oftaxon D as shown taxon A now subsumesits original ingroup reference taxon C Suchcontradictory denitions can be avoided ifa nested ingroup reference taxon (or nestedtaxa) are selected such as taxon E its un-

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 10: Sereno, 1999b

338 SYSTEMATIC BIOLOGY VOL 48

FIGURE 7 Crown species or clades (enclosed) musthave an extinct (stem) taxon (dashed) as an immediateoutgroup (a) Crown species and crown clades are de-limited by the presence of an immediate outgroup thatis extinct (b) Often referred to as ldquoextantrdquo taxa crowntaxa may be composed primarily of extinct taxa ldquoLiv-ingrdquo and ldquoextinctrdquo conditions are treated as if they con-stitute plesiomorphic and apomorphic character statesrespectively of an irreversible character ldquostate of be-ingrdquo Crown clades under this conceptualization areplesiomorphy-based taxa bounded by two living taxafor which the immediate outgroup is extinct

be hypothetically true that ldquoevery Recentgroup whatever its position in thehierarchymay be expected to have had a stem-grouprdquo(Patterson 1981207) clades bounded by liv-ing taxa with immediate outgroups that arealso living have not been considered crowntaxa

Some recent denitions of crown taxahowever have set aside Hennigrsquos andJefferiesrsquo formulations as well as generalusage in the literature A less restric-tive meaning has been proposed basedsolely on reference taxa that include liv-

ing representatives De Queiroz and Gau-thier (1992469) dened crown taxa asldquoclades within which both branches ofthe basal dichotomy are represented byextant descendantsrdquo without mention ofthe condition of the immediate outgroupYet all of their examples of crown cladeshave immediate outgroups that are ex-tinct Regarding their craniate taxonomythey stated that each stem-based taxon iscomposed of a ldquocrown taxon plus all ex-tinct taxa more closely related to it thanto any other crown claderdquo (de Queirozand Gauthier 1992475) Lee (19961103)proposed that ldquoa crown clade denitionresults when both taxa implicated in anode-based denition are representedby extant formsrdquo Although the conditionof the immediate outgroup is never men-tioned Lee (19961103) stated that crowntaxon denitions apply to clades thatconsist of ldquothe most recent common an-cestor of all extant forms and all its de-scendantsrdquo (emphasis added) The pres-ence of an immediate extinct outgroup isimplied

The denition of a crown taxon presentedhere in contrast does not specify ingroupstructure (such as the presence or absenceof a basal dichotomy) or denitional type(node- or stem-based) and explicitly statesthat the immediate outgroup(s) is extinct(Fig 7a Table 1) Crown taxa so denedcan be mapped unerringly on a cladogramif (1) ldquolivingrdquo and ldquoextinctrdquo conditions aretreated as primitive and derived character-states respectively of the character ldquocurrentstate of beingrdquo and (2) if this character isregarded as irreversible (ie CaminndashSokalparsimony) The living state in other wordsis always regarded as plesiomorphic Theirreversible status of the character (state ofbeing) is important because it identiescrown groups on cladograms of living andextinct taxa that would have an ambigu-ous status if character-state reversal wereallowed (Fig 7b) This character-state con-ceptualization of the denition of a crowngroup captures the essence of crown taxawhich have always been used to refer tomaximally inclusive taxa dened by extantrepresentatives not simply any ldquoextant-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 339

boundedrdquo taxon Discovery of extinct sistertaxa provides the opportunity to recognizecrown taxa

Crown taxa are often regarded as cladesthat are bounded uniquely by living in-groups The use of living reference taxahowever is most often an arbitrary deci-sion Any extinct ingroup taxon that is mostclosely related to a living reference taxoncould be used to delimit the same clade(Fig 7b) A crown clade is dened uniquelyby living reference taxa only when such taxado not have extinct sister taxa (Fig 7b) afact that is not widely appreciated Crowntaxa in other words do not have any spe-cial biological signicance beyond signal-ing that particular reference taxa have sur-vived to an arbitrary datum (the Recent)Informal reference to crown taxa as ldquolivingrdquoor ldquoextantrdquo taxa therefore can be mislead-ing because a crown taxon may be com-posed predominantly of extinct species asis the case with the crown taxon Crocodylia(Fig 7b Brochu 1997) ldquoExtant crocodil-iansrdquo if taken literally refers to a smallparaphyletic subset of Crocodylia Explicitreference to ldquocrown crocodiliansrdquo or sim-ply ldquoCrocodyliardquo is preferable to use of theterms ldquolivingrdquo or ldquoextantrdquo

A living taxon is also essential to delimita maximally inclusive clade that is entirelyextinctmdasha stem taxon as here dened Astem taxon denotes an extinct species orclade the immediate outgroup of which in-cludes at least one living member (Fig 6b)Many extinct taxa are not stem taxa becausetheir immediate outgroup is itself extinctSuch extinct taxa are members of more in-clusive stem taxa Stem taxa can have eithernode- or stem-based denitions (Fig 6b)Stem taxa and their associated less inclusivecrown taxon make up the taxonomic contentof their respective total taxon

Stem taxa as here dened should notbe confused with ldquostem grouprdquo (Fig 6a)which was rst used in phylogenetic sys-tematics by Hennig (ldquostammgrupperdquo 19691983) Hennigrsquos (198315) vague formulationof the concept specied all extinct speciesldquowhich can be shown probably to belong toa particular monophyletic group of the phy-logenetic system but which are probably no

closer [sic] related to one subgroup amongrecent animals than to anotherrdquo (translatedfrom the German Ax 1987224) Jefferies(1979) claried the term to include all ex-tinct outgroups of a crown taxon (Fig 6a)Other authors have developed more convo-luted denitions that approximate the sameend Wiley (1981217) for example dened aldquostem grouprdquo as ldquoall fossil species of a groupmore primitive than the most primitive Re-cent species of the same grouprdquo Theseldquostem groupsrdquo are paraphyletic and oftencircumscribe the same ldquoancestral groupsrdquoand ldquoevolutionary gradesrdquo that have mud-died phylogenetic thinking since DarwinOnly a few cladists continue to endorse suchconcepts (Smith 1994) Ax (1987224) pro-posed a cumbersome solution to the prob-lem of paraphyletic ldquostem groupsrdquo by divid-ing them into ancestral ldquostem lineagesrdquo andldquorepresentatives of the stem lineagerdquo Thelatter are monophyletic taxa most simply re-ferred to as stem taxa

A total taxon is regarded here as a cladecomposed of a crown taxon plus all ex-tinct outgroups more closely related to itthan to another crown taxon (Fig 3 Ta-ble 1) Hennigrsquos (1969) and Jefferiesrsquo (1979)formulations of the concept are wordy butsimilar in effect because they restricted in-clusion of extinct outgroups to those mostclosely related to a particular crown taxonDe Queiroz and Gauthier (1992470) deneda total clade as a ldquomore inclusive clade con-sisting of thecrown clade plus its extinctout-groupsrdquo This denition is incomplete be-cause it does not specify which extinct out-groups are included A total taxon explicitlyincludes only those extinct outgroups mostclosely related to a particular crown taxon

Node-based and stem-based denitions areconstructed by (1) identifying reference taxaand (2) attaching to them a relational phrasethat either includes their most recent com-mon ancestor and all descendants (node-based) or limits inclusion to all descen-dants closer to one or more of the referencetaxa (stem-based) (de Queiroz and Gauthier1992) (Fig 1 Table 1) In node-based deni-tions the phrase ldquoleast-inclusive claderdquo canreplace ldquocommon ancestor and all descen-dantsrdquo for cladists sensitive to the allusion to

340 SYSTEMATIC BIOLOGY VOL 48

ancestors (Lee 1998) Listing included taxawithout a relational phrase about ancestryor simply designating taxa as node- or stem-based without reference taxa does not con-stitutea phylogenetic denition because theboundaries of the taxon in question are notspecied (de Queiroz and Gauthier 1990Bryant 1996)

In the current literature there exists someconfusion over what is necessary and suf-cient for node- and stem-based denitionsBryant (1996185) suggested that a more in-clusive taxon be inserted within a stem-based denition such as ldquoall members oftaxon A thataremoreclosely related to taxonB than to taxon Crdquo Including taxon A how-ever is not necessary and may lead to inter-nal conicts within the denition (eg whentaxon A is included within taxon B or C onan alternativephylogeny) The goalof a phy-logenetic denition should be to unambigu-ously identify a specic common ancestorgiven any arrangement of taxa rather thanto identify such an ancestor only on a subsetof phylogenetic hypotheses

Meng et al (1994) and Wyss and Meng(1996) introduced a ldquostem-modied node-based denitionrdquo to effectively unite crown-group rodents whose basal relationshipsare uncertain Their preferred denition ofRodentia (ldquothe clade stemming from themost recent common ancestor of Mus andall Recent mammals more closely relatedto Mus than to Lagomorpha or membersof any other eutherian lsquoorderrdquorsquo Wyss andMeng 1996562) is better interpreted as aspecial stem-based rather than node-baseddenition because it uses only one ingroupreference taxon (Mus) and one or more out-group reference taxa (eg Lagomorpha)Their denition merely adds a phrase (ldquotheclade stemming from the most recent com-mon ancestorrdquo) to an otherwise typicalstem-based denition based on living ref-erence taxa That phrase ensures inclusionof all extinct taxa within the crown cladeBryantrsquos (1996185) reformulation of thisdenition on the other hand is clearly anode-based denition because he speci-ed only ingroup reference taxa (ldquothe mostrecent common ancestor of the species re-ferred to Rodentia by Wilson and Reeder

(1993) and all of its descendantsrdquo) These ex-amples underscore the need for claricationof terms (Tables 1 2)

A node-stem triplet (NST) is a denitionalconguration involving three taxa a node-based taxon composed of two stem-basedsubordinate taxa (Sereno 1997 1998 Fig 8)

Taxon A Taxa b and c their most recentcommon ancestor and all its descendants

Taxon B All taxa closer to taxon b than totaxon c

Taxon C All taxa closer to taxon c than totaxon b

FIGURE 8 Denitional triumvirate termed a node-stem triplet composed of a node-based taxon (A) andtwo subordinate stem-based taxa (B and C) which in-corporates added or repositioned taxa (dashed lineswith daggers) without changing the relative taxonomiccontent of taxon A B or C (as expressed by the equiva-lence statementA = B +C) A dot indicates a node-baseddenition an arrow indicates a stem-based denition

Taxa b and c (not shown in Fig 8) constitutereference taxa within taxon B and C respec-tively Although not essential the same ref-erence taxa (b c) may be used for the trio oftaxa in a NST which then specify complemen-tary denitions The pair of stem-based sis-ter taxa in addition may have reciprocal def-initions that utilize the same reference taxain antipodal positions The word ldquotaxardquo inthe stem-based denitions listed above mayoptionally be replaced by reference toa moreinclusive taxon (Bryant 1996) A stem-baseddenition for Saurischia for example mightread ldquoAll dinosaurs more closely related

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 341

to Neornithes than to Triceratopsrdquo (Sereno1998)

Taxonomic equivalence is a statement (orequation) in which one taxon is shownto be equivalent in current and poten-tial taxonomic content to subordinate taxaplus their common ancestor (Table 1)The NST described above is composedof a node-based taxon (A) and twosubordinate stem-based taxa (B C) andcan be written as the equivalence state-ment ldquoA = B + Crdquo because taxon Aby denition is equivalent to taxa Band C plus their most recent commonancestor

PHYLOGENETIC DEFINITIONS RATIONALE

The rst step toward a phylogenetic tax-onomy involved the distinction betweendenition and diagnosis (Buck and Hull1966 Grifths 1973 1974 Farris 1976 Ghis-elin 1984 Rowe 1987) A second step in-volved the formulation of node- and stem-based denitions by use of reference taxaand relational statements about ancestry (deQueiroz and Gauthier 1992) A third stepoutlined here involves the specication ofguidelines for selecting effective referencetaxa and criteria for positioning node- andstem-based denitions

Phylogenetic rearrangements and introduc-tion of new taxa are the principal factorsthat destabilize taxonomic content Stabil-ity of taxonomic content is enhanced bychoosing (1) effective reference taxa and(2) an effective conguration of phyloge-netic denitions

Reference TaxaPositionmdashGiven a particular phyloge-

netic hypothesis the taxonomic content of aphylogenetic denition is altered only if thereference taxa specify an alternative com-mon ancestor Given a basal dichotomy withone reference taxon on each side the iden-tication of the immediate common ances-tor is not affected by relocation of a ref-erence taxon on its side of the basal di-chotomy (Fig 9 move 1) Relocation ofa reference taxon to the opposing side orto a location outside the basal dichotomyidenties a different common ancestor

(Fig 9 moves 2 3) For a taxon with a basaldichotomy an alternative common ances-tor can be identied only if one referencetaxon is relocated to the opposing side oroutside the basal dichotomy Are some ref-erence taxa more likely to be relocated in thismanner than others

FIGURE 9 Three possible outcomes from relocationof a reference taxon 1 = relocation on the same side ofa basal dichotomy has no effect 2 = relocation to theopposite side of the basal dichotomy identies a lessinclusive common ancestor 3 = relocation outside thebasal dichotomy identies a more inclusive commonancestor Reference taxa are encircled and the originalbasal dichotomy is shown with heavy lines

First a taxon positioned near the basaldichotomy of a cladogram is always eas-ier to relocate to the opposing side (ierequires fewer additional steps) than is anested taxon if all other factors are regardedas equal among ingroup taxa (eg complete-ness missing data and homoplasy) Unlikea basal taxon (Fig 5d) a nested taxon sharessynapomorphies that increase its phyloge-netic (patristic) distance from the commonancestor (Fig 5e) Thus stability of taxo-nomic content is enhanced in phylogeneticdenitions if nested reference taxa are cho-sen that are located at some distance (atleast several nodes away) from the basaldichotomy This distance is easy to assessquantitatively for various nested taxa on agiven phylogeny

Second maximally inclusive referencetaxa (Fig 5c) have greater potential to cre-ate unnecessary taxonomic redundancy andinternal inconsistencies after relocation oftaxa (Fig 3) This is true for both node-

342 SYSTEMATIC BIOLOGY VOL 48

and stem-based taxa In the current litera-ture maximally inclusive reference taxa areoften used in phylogenetic denitions Intheir tetrapod classication for example deQueiroz and Gauthier (1992475) used suc-cessive maximally inclusive reference taxafor all denitionsmdasha pattern here termed re-cursive referencing (Fig 10a) Recursive refer-encing can have undesirable consequencesbecause it shifts the burden of a more precisedenition to a slightly less inclusive highertaxon This higher taxon may not be de-ned or it too may be affected by an alter-native conguration of basal relationshipsSalgado et al (1997) for example choseProsauropoda and Sauropoda as maximallyinclusive reference taxa for Sauropodomor-pha If a basal sauropodomorph is reposi-tioned the denitions of Prosauropoda andSauropoda must be determined to assess po-tential taxonomic consequences HoweverSalgado et al (1997) used Prosauropodaas a terminal taxon without denitionSauropoda was dened as a node-basedtaxon on the basis of two maximally in-clusive reference taxa Vulcanodon and Eu-sauropoda (all other sauropods) Thusif the incompletely known basal sauro-pod Vulcanodon is reinterpreted as be-ing more closely related to prosauropodsSauropoda and Sauropodomorpha wouldbecome synonyms

If on the other hand Sauropodomorphais dened on the basis of the nested ref-erence taxamdashthe prosauropod Plateosaurusand sauropod Saltasaurus (Sereno 1998)mdasha change in the phylogenetic positionof any other prosauropod or sauropodhas no effect on the taxonomic con-tents of Sauropodomorpha or SauropodaSauropodomorpha can be dened as ldquoPla-teosaurus Saltasaurus their most recentcommon ancestor and all descendantsrdquoand Sauropoda can be dened as ldquoallsauropodomorphs more closely related toSaltasaurus than to Plateosaurusrdquo The phylo-genetic position of Vulcanodon has no effecton the identication of respective commonancestors

CompletenessmdashPoorly known taxa aremore likely to be phylogenetically unsta-ble than well-known taxa (Wilkinson 1995)

FIGURE 10 Referencing styles (a) Recursive refer-encing selects a succession of the most inclusive ref-erence taxa (b) Nested referencing selects that sameremote reference taxon Reference taxa are encircled

Taxa with numerous missing entries in aphylogenetic analysis usually reduce phylo-genetic resolution by generating numerousequally parsimonious trees Well-known ref-erence taxa even if they are somewhat lessnested are preferable to poorly known taxa(Fig 11) A single poorly known basal taxonis least stable and therefore least desirableas a reference taxon

Use with Recent or extinct ltersmdashldquoRecentrdquo(ldquoextantrdquo or ldquolivingrdquo) or ldquoextinctrdquo may beused in phylogenetic denitions of crownand stem taxa respectively to help to stabi-lize taxonomic content in the face of poorlyresolved basal relationships For a crowntaxon with a stem-based denition a Recentlter can restrict the boundaries of a cladeto living taxa (Meng et al 1994 Wyss andMeng 1996) Rodentia for example could

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 343

be dened as ldquothe least inclusive clade com-posed of Recent mammals more closely re-lated to Mus than to Lepusrdquo For stem taxain contrast a stem-based denition using anldquoextinctrdquo lter can restrict inclusion to ex-tinct taxa when outgroup relationships areuncertain Moas for example could be de-ned as ldquoall extinct species more closelyrelated to Dinornis than to Struthiordquo theostrich

NumbermdashThe number of reference taxathat have been used in phylogenetic deni-tions has varied from two to innity Gau-thier (198614) for example dened Aves asa node-based taxon including ldquoall descen-

FIGURE 11 Nestedness and completeness in the se-lection of reference taxa A nested well-known taxonis more likely to constitute a stable reference taxon (en-circled)

dants of the most recent common ancestorof Ratitae Tinami and Neognathaerdquo Threereference taxa rather than two were chosenpresumably because of unstated uncertaintyin the phylogenetic relationships amongthese three avian subgroups De Queirozand Gauthier (1990310 1992461) suggestedthat phylogenetic denitions should utilizeonly two reference taxamdashbut then erecteddenitions for total taxa that used as refer-ence taxa one crown clade plus ldquoall extinct

taxa more closely related to it than to anyother crown claderdquo (de Queiroz and Gau-thier 1992475)

Listing more than a pair of referencetaxa does not seem to compromise a node-based phylogenetic denition in any dis-cernible way Node-based denitions areconstructed with what may be termed in-ternal inclusion A contradictory relation-ship between multiple ingroup referencetaxa for a node-based denition is im-possible to construct with internal inclu-sion because a clade always exists thatincludes all ingroup reference taxa Multi-ple ingroup reference taxa effectively stabi-lize the taxonomic content of a node-basedtaxon if basal relationships are uncertain(Fig 12a)

For stem-based denitions multiple out-group reference taxa effectively stabilize thetaxonomic content of a taxon if outgroup re-lationships are uncertain (Fig 12b) Multi-ple ingroup reference taxa for stem-baseddenitions can create a contradictory sit-uation if one (or more) ingroup referencetaxa are repositioned closer to the outgroupreference taxon (Fig 12c) or to a positionoutside the clade as originally dened Inthis case the external inclusion fundamen-tal to a stem-based denition creates a con-tradictory relationship among the originalingroup reference taxa which are no longermore closely related to each other than to theoutgroup reference taxon (Fig 12c) In theexample shown here relocation of ingroupreference taxon D creates a contradictory re-lationship with a second ingroup referencetaxon E if taxon A has the stem-based def-inition of ldquoall descendants more closely re-lated to taxon D and E than to taxon Brdquo Aninclusive ingroup reference taxon howevercan manifest the same problems in this cir-cumstance (Fig 12c) If taxon A is denedas ldquoall descendants more closely related totaxon C than to taxon Brdquo and if taxon C isdened on the basis of taxon D a contra-dictory denition results with relocation oftaxon D as shown taxon A now subsumesits original ingroup reference taxon C Suchcontradictory denitions can be avoided ifa nested ingroup reference taxon (or nestedtaxa) are selected such as taxon E its un-

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 11: Sereno, 1999b

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 339

boundedrdquo taxon Discovery of extinct sistertaxa provides the opportunity to recognizecrown taxa

Crown taxa are often regarded as cladesthat are bounded uniquely by living in-groups The use of living reference taxahowever is most often an arbitrary deci-sion Any extinct ingroup taxon that is mostclosely related to a living reference taxoncould be used to delimit the same clade(Fig 7b) A crown clade is dened uniquelyby living reference taxa only when such taxado not have extinct sister taxa (Fig 7b) afact that is not widely appreciated Crowntaxa in other words do not have any spe-cial biological signicance beyond signal-ing that particular reference taxa have sur-vived to an arbitrary datum (the Recent)Informal reference to crown taxa as ldquolivingrdquoor ldquoextantrdquo taxa therefore can be mislead-ing because a crown taxon may be com-posed predominantly of extinct species asis the case with the crown taxon Crocodylia(Fig 7b Brochu 1997) ldquoExtant crocodil-iansrdquo if taken literally refers to a smallparaphyletic subset of Crocodylia Explicitreference to ldquocrown crocodiliansrdquo or sim-ply ldquoCrocodyliardquo is preferable to use of theterms ldquolivingrdquo or ldquoextantrdquo

A living taxon is also essential to delimita maximally inclusive clade that is entirelyextinctmdasha stem taxon as here dened Astem taxon denotes an extinct species orclade the immediate outgroup of which in-cludes at least one living member (Fig 6b)Many extinct taxa are not stem taxa becausetheir immediate outgroup is itself extinctSuch extinct taxa are members of more in-clusive stem taxa Stem taxa can have eithernode- or stem-based denitions (Fig 6b)Stem taxa and their associated less inclusivecrown taxon make up the taxonomic contentof their respective total taxon

Stem taxa as here dened should notbe confused with ldquostem grouprdquo (Fig 6a)which was rst used in phylogenetic sys-tematics by Hennig (ldquostammgrupperdquo 19691983) Hennigrsquos (198315) vague formulationof the concept specied all extinct speciesldquowhich can be shown probably to belong toa particular monophyletic group of the phy-logenetic system but which are probably no

closer [sic] related to one subgroup amongrecent animals than to anotherrdquo (translatedfrom the German Ax 1987224) Jefferies(1979) claried the term to include all ex-tinct outgroups of a crown taxon (Fig 6a)Other authors have developed more convo-luted denitions that approximate the sameend Wiley (1981217) for example dened aldquostem grouprdquo as ldquoall fossil species of a groupmore primitive than the most primitive Re-cent species of the same grouprdquo Theseldquostem groupsrdquo are paraphyletic and oftencircumscribe the same ldquoancestral groupsrdquoand ldquoevolutionary gradesrdquo that have mud-died phylogenetic thinking since DarwinOnly a few cladists continue to endorse suchconcepts (Smith 1994) Ax (1987224) pro-posed a cumbersome solution to the prob-lem of paraphyletic ldquostem groupsrdquo by divid-ing them into ancestral ldquostem lineagesrdquo andldquorepresentatives of the stem lineagerdquo Thelatter are monophyletic taxa most simply re-ferred to as stem taxa

A total taxon is regarded here as a cladecomposed of a crown taxon plus all ex-tinct outgroups more closely related to itthan to another crown taxon (Fig 3 Ta-ble 1) Hennigrsquos (1969) and Jefferiesrsquo (1979)formulations of the concept are wordy butsimilar in effect because they restricted in-clusion of extinct outgroups to those mostclosely related to a particular crown taxonDe Queiroz and Gauthier (1992470) deneda total clade as a ldquomore inclusive clade con-sisting of thecrown clade plus its extinctout-groupsrdquo This denition is incomplete be-cause it does not specify which extinct out-groups are included A total taxon explicitlyincludes only those extinct outgroups mostclosely related to a particular crown taxon

Node-based and stem-based denitions areconstructed by (1) identifying reference taxaand (2) attaching to them a relational phrasethat either includes their most recent com-mon ancestor and all descendants (node-based) or limits inclusion to all descen-dants closer to one or more of the referencetaxa (stem-based) (de Queiroz and Gauthier1992) (Fig 1 Table 1) In node-based deni-tions the phrase ldquoleast-inclusive claderdquo canreplace ldquocommon ancestor and all descen-dantsrdquo for cladists sensitive to the allusion to

340 SYSTEMATIC BIOLOGY VOL 48

ancestors (Lee 1998) Listing included taxawithout a relational phrase about ancestryor simply designating taxa as node- or stem-based without reference taxa does not con-stitutea phylogenetic denition because theboundaries of the taxon in question are notspecied (de Queiroz and Gauthier 1990Bryant 1996)

In the current literature there exists someconfusion over what is necessary and suf-cient for node- and stem-based denitionsBryant (1996185) suggested that a more in-clusive taxon be inserted within a stem-based denition such as ldquoall members oftaxon A thataremoreclosely related to taxonB than to taxon Crdquo Including taxon A how-ever is not necessary and may lead to inter-nal conicts within the denition (eg whentaxon A is included within taxon B or C onan alternativephylogeny) The goalof a phy-logenetic denition should be to unambigu-ously identify a specic common ancestorgiven any arrangement of taxa rather thanto identify such an ancestor only on a subsetof phylogenetic hypotheses

Meng et al (1994) and Wyss and Meng(1996) introduced a ldquostem-modied node-based denitionrdquo to effectively unite crown-group rodents whose basal relationshipsare uncertain Their preferred denition ofRodentia (ldquothe clade stemming from themost recent common ancestor of Mus andall Recent mammals more closely relatedto Mus than to Lagomorpha or membersof any other eutherian lsquoorderrdquorsquo Wyss andMeng 1996562) is better interpreted as aspecial stem-based rather than node-baseddenition because it uses only one ingroupreference taxon (Mus) and one or more out-group reference taxa (eg Lagomorpha)Their denition merely adds a phrase (ldquotheclade stemming from the most recent com-mon ancestorrdquo) to an otherwise typicalstem-based denition based on living ref-erence taxa That phrase ensures inclusionof all extinct taxa within the crown cladeBryantrsquos (1996185) reformulation of thisdenition on the other hand is clearly anode-based denition because he speci-ed only ingroup reference taxa (ldquothe mostrecent common ancestor of the species re-ferred to Rodentia by Wilson and Reeder

(1993) and all of its descendantsrdquo) These ex-amples underscore the need for claricationof terms (Tables 1 2)

A node-stem triplet (NST) is a denitionalconguration involving three taxa a node-based taxon composed of two stem-basedsubordinate taxa (Sereno 1997 1998 Fig 8)

Taxon A Taxa b and c their most recentcommon ancestor and all its descendants

Taxon B All taxa closer to taxon b than totaxon c

Taxon C All taxa closer to taxon c than totaxon b

FIGURE 8 Denitional triumvirate termed a node-stem triplet composed of a node-based taxon (A) andtwo subordinate stem-based taxa (B and C) which in-corporates added or repositioned taxa (dashed lineswith daggers) without changing the relative taxonomiccontent of taxon A B or C (as expressed by the equiva-lence statementA = B +C) A dot indicates a node-baseddenition an arrow indicates a stem-based denition

Taxa b and c (not shown in Fig 8) constitutereference taxa within taxon B and C respec-tively Although not essential the same ref-erence taxa (b c) may be used for the trio oftaxa in a NST which then specify complemen-tary denitions The pair of stem-based sis-ter taxa in addition may have reciprocal def-initions that utilize the same reference taxain antipodal positions The word ldquotaxardquo inthe stem-based denitions listed above mayoptionally be replaced by reference toa moreinclusive taxon (Bryant 1996) A stem-baseddenition for Saurischia for example mightread ldquoAll dinosaurs more closely related

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 341

to Neornithes than to Triceratopsrdquo (Sereno1998)

Taxonomic equivalence is a statement (orequation) in which one taxon is shownto be equivalent in current and poten-tial taxonomic content to subordinate taxaplus their common ancestor (Table 1)The NST described above is composedof a node-based taxon (A) and twosubordinate stem-based taxa (B C) andcan be written as the equivalence state-ment ldquoA = B + Crdquo because taxon Aby denition is equivalent to taxa Band C plus their most recent commonancestor

PHYLOGENETIC DEFINITIONS RATIONALE

The rst step toward a phylogenetic tax-onomy involved the distinction betweendenition and diagnosis (Buck and Hull1966 Grifths 1973 1974 Farris 1976 Ghis-elin 1984 Rowe 1987) A second step in-volved the formulation of node- and stem-based denitions by use of reference taxaand relational statements about ancestry (deQueiroz and Gauthier 1992) A third stepoutlined here involves the specication ofguidelines for selecting effective referencetaxa and criteria for positioning node- andstem-based denitions

Phylogenetic rearrangements and introduc-tion of new taxa are the principal factorsthat destabilize taxonomic content Stabil-ity of taxonomic content is enhanced bychoosing (1) effective reference taxa and(2) an effective conguration of phyloge-netic denitions

Reference TaxaPositionmdashGiven a particular phyloge-

netic hypothesis the taxonomic content of aphylogenetic denition is altered only if thereference taxa specify an alternative com-mon ancestor Given a basal dichotomy withone reference taxon on each side the iden-tication of the immediate common ances-tor is not affected by relocation of a ref-erence taxon on its side of the basal di-chotomy (Fig 9 move 1) Relocation ofa reference taxon to the opposing side orto a location outside the basal dichotomyidenties a different common ancestor

(Fig 9 moves 2 3) For a taxon with a basaldichotomy an alternative common ances-tor can be identied only if one referencetaxon is relocated to the opposing side oroutside the basal dichotomy Are some ref-erence taxa more likely to be relocated in thismanner than others

FIGURE 9 Three possible outcomes from relocationof a reference taxon 1 = relocation on the same side ofa basal dichotomy has no effect 2 = relocation to theopposite side of the basal dichotomy identies a lessinclusive common ancestor 3 = relocation outside thebasal dichotomy identies a more inclusive commonancestor Reference taxa are encircled and the originalbasal dichotomy is shown with heavy lines

First a taxon positioned near the basaldichotomy of a cladogram is always eas-ier to relocate to the opposing side (ierequires fewer additional steps) than is anested taxon if all other factors are regardedas equal among ingroup taxa (eg complete-ness missing data and homoplasy) Unlikea basal taxon (Fig 5d) a nested taxon sharessynapomorphies that increase its phyloge-netic (patristic) distance from the commonancestor (Fig 5e) Thus stability of taxo-nomic content is enhanced in phylogeneticdenitions if nested reference taxa are cho-sen that are located at some distance (atleast several nodes away) from the basaldichotomy This distance is easy to assessquantitatively for various nested taxa on agiven phylogeny

Second maximally inclusive referencetaxa (Fig 5c) have greater potential to cre-ate unnecessary taxonomic redundancy andinternal inconsistencies after relocation oftaxa (Fig 3) This is true for both node-

342 SYSTEMATIC BIOLOGY VOL 48

and stem-based taxa In the current litera-ture maximally inclusive reference taxa areoften used in phylogenetic denitions Intheir tetrapod classication for example deQueiroz and Gauthier (1992475) used suc-cessive maximally inclusive reference taxafor all denitionsmdasha pattern here termed re-cursive referencing (Fig 10a) Recursive refer-encing can have undesirable consequencesbecause it shifts the burden of a more precisedenition to a slightly less inclusive highertaxon This higher taxon may not be de-ned or it too may be affected by an alter-native conguration of basal relationshipsSalgado et al (1997) for example choseProsauropoda and Sauropoda as maximallyinclusive reference taxa for Sauropodomor-pha If a basal sauropodomorph is reposi-tioned the denitions of Prosauropoda andSauropoda must be determined to assess po-tential taxonomic consequences HoweverSalgado et al (1997) used Prosauropodaas a terminal taxon without denitionSauropoda was dened as a node-basedtaxon on the basis of two maximally in-clusive reference taxa Vulcanodon and Eu-sauropoda (all other sauropods) Thusif the incompletely known basal sauro-pod Vulcanodon is reinterpreted as be-ing more closely related to prosauropodsSauropoda and Sauropodomorpha wouldbecome synonyms

If on the other hand Sauropodomorphais dened on the basis of the nested ref-erence taxamdashthe prosauropod Plateosaurusand sauropod Saltasaurus (Sereno 1998)mdasha change in the phylogenetic positionof any other prosauropod or sauropodhas no effect on the taxonomic con-tents of Sauropodomorpha or SauropodaSauropodomorpha can be dened as ldquoPla-teosaurus Saltasaurus their most recentcommon ancestor and all descendantsrdquoand Sauropoda can be dened as ldquoallsauropodomorphs more closely related toSaltasaurus than to Plateosaurusrdquo The phylo-genetic position of Vulcanodon has no effecton the identication of respective commonancestors

CompletenessmdashPoorly known taxa aremore likely to be phylogenetically unsta-ble than well-known taxa (Wilkinson 1995)

FIGURE 10 Referencing styles (a) Recursive refer-encing selects a succession of the most inclusive ref-erence taxa (b) Nested referencing selects that sameremote reference taxon Reference taxa are encircled

Taxa with numerous missing entries in aphylogenetic analysis usually reduce phylo-genetic resolution by generating numerousequally parsimonious trees Well-known ref-erence taxa even if they are somewhat lessnested are preferable to poorly known taxa(Fig 11) A single poorly known basal taxonis least stable and therefore least desirableas a reference taxon

Use with Recent or extinct ltersmdashldquoRecentrdquo(ldquoextantrdquo or ldquolivingrdquo) or ldquoextinctrdquo may beused in phylogenetic denitions of crownand stem taxa respectively to help to stabi-lize taxonomic content in the face of poorlyresolved basal relationships For a crowntaxon with a stem-based denition a Recentlter can restrict the boundaries of a cladeto living taxa (Meng et al 1994 Wyss andMeng 1996) Rodentia for example could

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 343

be dened as ldquothe least inclusive clade com-posed of Recent mammals more closely re-lated to Mus than to Lepusrdquo For stem taxain contrast a stem-based denition using anldquoextinctrdquo lter can restrict inclusion to ex-tinct taxa when outgroup relationships areuncertain Moas for example could be de-ned as ldquoall extinct species more closelyrelated to Dinornis than to Struthiordquo theostrich

NumbermdashThe number of reference taxathat have been used in phylogenetic deni-tions has varied from two to innity Gau-thier (198614) for example dened Aves asa node-based taxon including ldquoall descen-

FIGURE 11 Nestedness and completeness in the se-lection of reference taxa A nested well-known taxonis more likely to constitute a stable reference taxon (en-circled)

dants of the most recent common ancestorof Ratitae Tinami and Neognathaerdquo Threereference taxa rather than two were chosenpresumably because of unstated uncertaintyin the phylogenetic relationships amongthese three avian subgroups De Queirozand Gauthier (1990310 1992461) suggestedthat phylogenetic denitions should utilizeonly two reference taxamdashbut then erecteddenitions for total taxa that used as refer-ence taxa one crown clade plus ldquoall extinct

taxa more closely related to it than to anyother crown claderdquo (de Queiroz and Gau-thier 1992475)

Listing more than a pair of referencetaxa does not seem to compromise a node-based phylogenetic denition in any dis-cernible way Node-based denitions areconstructed with what may be termed in-ternal inclusion A contradictory relation-ship between multiple ingroup referencetaxa for a node-based denition is im-possible to construct with internal inclu-sion because a clade always exists thatincludes all ingroup reference taxa Multi-ple ingroup reference taxa effectively stabi-lize the taxonomic content of a node-basedtaxon if basal relationships are uncertain(Fig 12a)

For stem-based denitions multiple out-group reference taxa effectively stabilize thetaxonomic content of a taxon if outgroup re-lationships are uncertain (Fig 12b) Multi-ple ingroup reference taxa for stem-baseddenitions can create a contradictory sit-uation if one (or more) ingroup referencetaxa are repositioned closer to the outgroupreference taxon (Fig 12c) or to a positionoutside the clade as originally dened Inthis case the external inclusion fundamen-tal to a stem-based denition creates a con-tradictory relationship among the originalingroup reference taxa which are no longermore closely related to each other than to theoutgroup reference taxon (Fig 12c) In theexample shown here relocation of ingroupreference taxon D creates a contradictory re-lationship with a second ingroup referencetaxon E if taxon A has the stem-based def-inition of ldquoall descendants more closely re-lated to taxon D and E than to taxon Brdquo Aninclusive ingroup reference taxon howevercan manifest the same problems in this cir-cumstance (Fig 12c) If taxon A is denedas ldquoall descendants more closely related totaxon C than to taxon Brdquo and if taxon C isdened on the basis of taxon D a contra-dictory denition results with relocation oftaxon D as shown taxon A now subsumesits original ingroup reference taxon C Suchcontradictory denitions can be avoided ifa nested ingroup reference taxon (or nestedtaxa) are selected such as taxon E its un-

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 12: Sereno, 1999b

340 SYSTEMATIC BIOLOGY VOL 48

ancestors (Lee 1998) Listing included taxawithout a relational phrase about ancestryor simply designating taxa as node- or stem-based without reference taxa does not con-stitutea phylogenetic denition because theboundaries of the taxon in question are notspecied (de Queiroz and Gauthier 1990Bryant 1996)

In the current literature there exists someconfusion over what is necessary and suf-cient for node- and stem-based denitionsBryant (1996185) suggested that a more in-clusive taxon be inserted within a stem-based denition such as ldquoall members oftaxon A thataremoreclosely related to taxonB than to taxon Crdquo Including taxon A how-ever is not necessary and may lead to inter-nal conicts within the denition (eg whentaxon A is included within taxon B or C onan alternativephylogeny) The goalof a phy-logenetic denition should be to unambigu-ously identify a specic common ancestorgiven any arrangement of taxa rather thanto identify such an ancestor only on a subsetof phylogenetic hypotheses

Meng et al (1994) and Wyss and Meng(1996) introduced a ldquostem-modied node-based denitionrdquo to effectively unite crown-group rodents whose basal relationshipsare uncertain Their preferred denition ofRodentia (ldquothe clade stemming from themost recent common ancestor of Mus andall Recent mammals more closely relatedto Mus than to Lagomorpha or membersof any other eutherian lsquoorderrdquorsquo Wyss andMeng 1996562) is better interpreted as aspecial stem-based rather than node-baseddenition because it uses only one ingroupreference taxon (Mus) and one or more out-group reference taxa (eg Lagomorpha)Their denition merely adds a phrase (ldquotheclade stemming from the most recent com-mon ancestorrdquo) to an otherwise typicalstem-based denition based on living ref-erence taxa That phrase ensures inclusionof all extinct taxa within the crown cladeBryantrsquos (1996185) reformulation of thisdenition on the other hand is clearly anode-based denition because he speci-ed only ingroup reference taxa (ldquothe mostrecent common ancestor of the species re-ferred to Rodentia by Wilson and Reeder

(1993) and all of its descendantsrdquo) These ex-amples underscore the need for claricationof terms (Tables 1 2)

A node-stem triplet (NST) is a denitionalconguration involving three taxa a node-based taxon composed of two stem-basedsubordinate taxa (Sereno 1997 1998 Fig 8)

Taxon A Taxa b and c their most recentcommon ancestor and all its descendants

Taxon B All taxa closer to taxon b than totaxon c

Taxon C All taxa closer to taxon c than totaxon b

FIGURE 8 Denitional triumvirate termed a node-stem triplet composed of a node-based taxon (A) andtwo subordinate stem-based taxa (B and C) which in-corporates added or repositioned taxa (dashed lineswith daggers) without changing the relative taxonomiccontent of taxon A B or C (as expressed by the equiva-lence statementA = B +C) A dot indicates a node-baseddenition an arrow indicates a stem-based denition

Taxa b and c (not shown in Fig 8) constitutereference taxa within taxon B and C respec-tively Although not essential the same ref-erence taxa (b c) may be used for the trio oftaxa in a NST which then specify complemen-tary denitions The pair of stem-based sis-ter taxa in addition may have reciprocal def-initions that utilize the same reference taxain antipodal positions The word ldquotaxardquo inthe stem-based denitions listed above mayoptionally be replaced by reference toa moreinclusive taxon (Bryant 1996) A stem-baseddenition for Saurischia for example mightread ldquoAll dinosaurs more closely related

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 341

to Neornithes than to Triceratopsrdquo (Sereno1998)

Taxonomic equivalence is a statement (orequation) in which one taxon is shownto be equivalent in current and poten-tial taxonomic content to subordinate taxaplus their common ancestor (Table 1)The NST described above is composedof a node-based taxon (A) and twosubordinate stem-based taxa (B C) andcan be written as the equivalence state-ment ldquoA = B + Crdquo because taxon Aby denition is equivalent to taxa Band C plus their most recent commonancestor

PHYLOGENETIC DEFINITIONS RATIONALE

The rst step toward a phylogenetic tax-onomy involved the distinction betweendenition and diagnosis (Buck and Hull1966 Grifths 1973 1974 Farris 1976 Ghis-elin 1984 Rowe 1987) A second step in-volved the formulation of node- and stem-based denitions by use of reference taxaand relational statements about ancestry (deQueiroz and Gauthier 1992) A third stepoutlined here involves the specication ofguidelines for selecting effective referencetaxa and criteria for positioning node- andstem-based denitions

Phylogenetic rearrangements and introduc-tion of new taxa are the principal factorsthat destabilize taxonomic content Stabil-ity of taxonomic content is enhanced bychoosing (1) effective reference taxa and(2) an effective conguration of phyloge-netic denitions

Reference TaxaPositionmdashGiven a particular phyloge-

netic hypothesis the taxonomic content of aphylogenetic denition is altered only if thereference taxa specify an alternative com-mon ancestor Given a basal dichotomy withone reference taxon on each side the iden-tication of the immediate common ances-tor is not affected by relocation of a ref-erence taxon on its side of the basal di-chotomy (Fig 9 move 1) Relocation ofa reference taxon to the opposing side orto a location outside the basal dichotomyidenties a different common ancestor

(Fig 9 moves 2 3) For a taxon with a basaldichotomy an alternative common ances-tor can be identied only if one referencetaxon is relocated to the opposing side oroutside the basal dichotomy Are some ref-erence taxa more likely to be relocated in thismanner than others

FIGURE 9 Three possible outcomes from relocationof a reference taxon 1 = relocation on the same side ofa basal dichotomy has no effect 2 = relocation to theopposite side of the basal dichotomy identies a lessinclusive common ancestor 3 = relocation outside thebasal dichotomy identies a more inclusive commonancestor Reference taxa are encircled and the originalbasal dichotomy is shown with heavy lines

First a taxon positioned near the basaldichotomy of a cladogram is always eas-ier to relocate to the opposing side (ierequires fewer additional steps) than is anested taxon if all other factors are regardedas equal among ingroup taxa (eg complete-ness missing data and homoplasy) Unlikea basal taxon (Fig 5d) a nested taxon sharessynapomorphies that increase its phyloge-netic (patristic) distance from the commonancestor (Fig 5e) Thus stability of taxo-nomic content is enhanced in phylogeneticdenitions if nested reference taxa are cho-sen that are located at some distance (atleast several nodes away) from the basaldichotomy This distance is easy to assessquantitatively for various nested taxa on agiven phylogeny

Second maximally inclusive referencetaxa (Fig 5c) have greater potential to cre-ate unnecessary taxonomic redundancy andinternal inconsistencies after relocation oftaxa (Fig 3) This is true for both node-

342 SYSTEMATIC BIOLOGY VOL 48

and stem-based taxa In the current litera-ture maximally inclusive reference taxa areoften used in phylogenetic denitions Intheir tetrapod classication for example deQueiroz and Gauthier (1992475) used suc-cessive maximally inclusive reference taxafor all denitionsmdasha pattern here termed re-cursive referencing (Fig 10a) Recursive refer-encing can have undesirable consequencesbecause it shifts the burden of a more precisedenition to a slightly less inclusive highertaxon This higher taxon may not be de-ned or it too may be affected by an alter-native conguration of basal relationshipsSalgado et al (1997) for example choseProsauropoda and Sauropoda as maximallyinclusive reference taxa for Sauropodomor-pha If a basal sauropodomorph is reposi-tioned the denitions of Prosauropoda andSauropoda must be determined to assess po-tential taxonomic consequences HoweverSalgado et al (1997) used Prosauropodaas a terminal taxon without denitionSauropoda was dened as a node-basedtaxon on the basis of two maximally in-clusive reference taxa Vulcanodon and Eu-sauropoda (all other sauropods) Thusif the incompletely known basal sauro-pod Vulcanodon is reinterpreted as be-ing more closely related to prosauropodsSauropoda and Sauropodomorpha wouldbecome synonyms

If on the other hand Sauropodomorphais dened on the basis of the nested ref-erence taxamdashthe prosauropod Plateosaurusand sauropod Saltasaurus (Sereno 1998)mdasha change in the phylogenetic positionof any other prosauropod or sauropodhas no effect on the taxonomic con-tents of Sauropodomorpha or SauropodaSauropodomorpha can be dened as ldquoPla-teosaurus Saltasaurus their most recentcommon ancestor and all descendantsrdquoand Sauropoda can be dened as ldquoallsauropodomorphs more closely related toSaltasaurus than to Plateosaurusrdquo The phylo-genetic position of Vulcanodon has no effecton the identication of respective commonancestors

CompletenessmdashPoorly known taxa aremore likely to be phylogenetically unsta-ble than well-known taxa (Wilkinson 1995)

FIGURE 10 Referencing styles (a) Recursive refer-encing selects a succession of the most inclusive ref-erence taxa (b) Nested referencing selects that sameremote reference taxon Reference taxa are encircled

Taxa with numerous missing entries in aphylogenetic analysis usually reduce phylo-genetic resolution by generating numerousequally parsimonious trees Well-known ref-erence taxa even if they are somewhat lessnested are preferable to poorly known taxa(Fig 11) A single poorly known basal taxonis least stable and therefore least desirableas a reference taxon

Use with Recent or extinct ltersmdashldquoRecentrdquo(ldquoextantrdquo or ldquolivingrdquo) or ldquoextinctrdquo may beused in phylogenetic denitions of crownand stem taxa respectively to help to stabi-lize taxonomic content in the face of poorlyresolved basal relationships For a crowntaxon with a stem-based denition a Recentlter can restrict the boundaries of a cladeto living taxa (Meng et al 1994 Wyss andMeng 1996) Rodentia for example could

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 343

be dened as ldquothe least inclusive clade com-posed of Recent mammals more closely re-lated to Mus than to Lepusrdquo For stem taxain contrast a stem-based denition using anldquoextinctrdquo lter can restrict inclusion to ex-tinct taxa when outgroup relationships areuncertain Moas for example could be de-ned as ldquoall extinct species more closelyrelated to Dinornis than to Struthiordquo theostrich

NumbermdashThe number of reference taxathat have been used in phylogenetic deni-tions has varied from two to innity Gau-thier (198614) for example dened Aves asa node-based taxon including ldquoall descen-

FIGURE 11 Nestedness and completeness in the se-lection of reference taxa A nested well-known taxonis more likely to constitute a stable reference taxon (en-circled)

dants of the most recent common ancestorof Ratitae Tinami and Neognathaerdquo Threereference taxa rather than two were chosenpresumably because of unstated uncertaintyin the phylogenetic relationships amongthese three avian subgroups De Queirozand Gauthier (1990310 1992461) suggestedthat phylogenetic denitions should utilizeonly two reference taxamdashbut then erecteddenitions for total taxa that used as refer-ence taxa one crown clade plus ldquoall extinct

taxa more closely related to it than to anyother crown claderdquo (de Queiroz and Gau-thier 1992475)

Listing more than a pair of referencetaxa does not seem to compromise a node-based phylogenetic denition in any dis-cernible way Node-based denitions areconstructed with what may be termed in-ternal inclusion A contradictory relation-ship between multiple ingroup referencetaxa for a node-based denition is im-possible to construct with internal inclu-sion because a clade always exists thatincludes all ingroup reference taxa Multi-ple ingroup reference taxa effectively stabi-lize the taxonomic content of a node-basedtaxon if basal relationships are uncertain(Fig 12a)

For stem-based denitions multiple out-group reference taxa effectively stabilize thetaxonomic content of a taxon if outgroup re-lationships are uncertain (Fig 12b) Multi-ple ingroup reference taxa for stem-baseddenitions can create a contradictory sit-uation if one (or more) ingroup referencetaxa are repositioned closer to the outgroupreference taxon (Fig 12c) or to a positionoutside the clade as originally dened Inthis case the external inclusion fundamen-tal to a stem-based denition creates a con-tradictory relationship among the originalingroup reference taxa which are no longermore closely related to each other than to theoutgroup reference taxon (Fig 12c) In theexample shown here relocation of ingroupreference taxon D creates a contradictory re-lationship with a second ingroup referencetaxon E if taxon A has the stem-based def-inition of ldquoall descendants more closely re-lated to taxon D and E than to taxon Brdquo Aninclusive ingroup reference taxon howevercan manifest the same problems in this cir-cumstance (Fig 12c) If taxon A is denedas ldquoall descendants more closely related totaxon C than to taxon Brdquo and if taxon C isdened on the basis of taxon D a contra-dictory denition results with relocation oftaxon D as shown taxon A now subsumesits original ingroup reference taxon C Suchcontradictory denitions can be avoided ifa nested ingroup reference taxon (or nestedtaxa) are selected such as taxon E its un-

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 13: Sereno, 1999b

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 341

to Neornithes than to Triceratopsrdquo (Sereno1998)

Taxonomic equivalence is a statement (orequation) in which one taxon is shownto be equivalent in current and poten-tial taxonomic content to subordinate taxaplus their common ancestor (Table 1)The NST described above is composedof a node-based taxon (A) and twosubordinate stem-based taxa (B C) andcan be written as the equivalence state-ment ldquoA = B + Crdquo because taxon Aby denition is equivalent to taxa Band C plus their most recent commonancestor

PHYLOGENETIC DEFINITIONS RATIONALE

The rst step toward a phylogenetic tax-onomy involved the distinction betweendenition and diagnosis (Buck and Hull1966 Grifths 1973 1974 Farris 1976 Ghis-elin 1984 Rowe 1987) A second step in-volved the formulation of node- and stem-based denitions by use of reference taxaand relational statements about ancestry (deQueiroz and Gauthier 1992) A third stepoutlined here involves the specication ofguidelines for selecting effective referencetaxa and criteria for positioning node- andstem-based denitions

Phylogenetic rearrangements and introduc-tion of new taxa are the principal factorsthat destabilize taxonomic content Stabil-ity of taxonomic content is enhanced bychoosing (1) effective reference taxa and(2) an effective conguration of phyloge-netic denitions

Reference TaxaPositionmdashGiven a particular phyloge-

netic hypothesis the taxonomic content of aphylogenetic denition is altered only if thereference taxa specify an alternative com-mon ancestor Given a basal dichotomy withone reference taxon on each side the iden-tication of the immediate common ances-tor is not affected by relocation of a ref-erence taxon on its side of the basal di-chotomy (Fig 9 move 1) Relocation ofa reference taxon to the opposing side orto a location outside the basal dichotomyidenties a different common ancestor

(Fig 9 moves 2 3) For a taxon with a basaldichotomy an alternative common ances-tor can be identied only if one referencetaxon is relocated to the opposing side oroutside the basal dichotomy Are some ref-erence taxa more likely to be relocated in thismanner than others

FIGURE 9 Three possible outcomes from relocationof a reference taxon 1 = relocation on the same side ofa basal dichotomy has no effect 2 = relocation to theopposite side of the basal dichotomy identies a lessinclusive common ancestor 3 = relocation outside thebasal dichotomy identies a more inclusive commonancestor Reference taxa are encircled and the originalbasal dichotomy is shown with heavy lines

First a taxon positioned near the basaldichotomy of a cladogram is always eas-ier to relocate to the opposing side (ierequires fewer additional steps) than is anested taxon if all other factors are regardedas equal among ingroup taxa (eg complete-ness missing data and homoplasy) Unlikea basal taxon (Fig 5d) a nested taxon sharessynapomorphies that increase its phyloge-netic (patristic) distance from the commonancestor (Fig 5e) Thus stability of taxo-nomic content is enhanced in phylogeneticdenitions if nested reference taxa are cho-sen that are located at some distance (atleast several nodes away) from the basaldichotomy This distance is easy to assessquantitatively for various nested taxa on agiven phylogeny

Second maximally inclusive referencetaxa (Fig 5c) have greater potential to cre-ate unnecessary taxonomic redundancy andinternal inconsistencies after relocation oftaxa (Fig 3) This is true for both node-

342 SYSTEMATIC BIOLOGY VOL 48

and stem-based taxa In the current litera-ture maximally inclusive reference taxa areoften used in phylogenetic denitions Intheir tetrapod classication for example deQueiroz and Gauthier (1992475) used suc-cessive maximally inclusive reference taxafor all denitionsmdasha pattern here termed re-cursive referencing (Fig 10a) Recursive refer-encing can have undesirable consequencesbecause it shifts the burden of a more precisedenition to a slightly less inclusive highertaxon This higher taxon may not be de-ned or it too may be affected by an alter-native conguration of basal relationshipsSalgado et al (1997) for example choseProsauropoda and Sauropoda as maximallyinclusive reference taxa for Sauropodomor-pha If a basal sauropodomorph is reposi-tioned the denitions of Prosauropoda andSauropoda must be determined to assess po-tential taxonomic consequences HoweverSalgado et al (1997) used Prosauropodaas a terminal taxon without denitionSauropoda was dened as a node-basedtaxon on the basis of two maximally in-clusive reference taxa Vulcanodon and Eu-sauropoda (all other sauropods) Thusif the incompletely known basal sauro-pod Vulcanodon is reinterpreted as be-ing more closely related to prosauropodsSauropoda and Sauropodomorpha wouldbecome synonyms

If on the other hand Sauropodomorphais dened on the basis of the nested ref-erence taxamdashthe prosauropod Plateosaurusand sauropod Saltasaurus (Sereno 1998)mdasha change in the phylogenetic positionof any other prosauropod or sauropodhas no effect on the taxonomic con-tents of Sauropodomorpha or SauropodaSauropodomorpha can be dened as ldquoPla-teosaurus Saltasaurus their most recentcommon ancestor and all descendantsrdquoand Sauropoda can be dened as ldquoallsauropodomorphs more closely related toSaltasaurus than to Plateosaurusrdquo The phylo-genetic position of Vulcanodon has no effecton the identication of respective commonancestors

CompletenessmdashPoorly known taxa aremore likely to be phylogenetically unsta-ble than well-known taxa (Wilkinson 1995)

FIGURE 10 Referencing styles (a) Recursive refer-encing selects a succession of the most inclusive ref-erence taxa (b) Nested referencing selects that sameremote reference taxon Reference taxa are encircled

Taxa with numerous missing entries in aphylogenetic analysis usually reduce phylo-genetic resolution by generating numerousequally parsimonious trees Well-known ref-erence taxa even if they are somewhat lessnested are preferable to poorly known taxa(Fig 11) A single poorly known basal taxonis least stable and therefore least desirableas a reference taxon

Use with Recent or extinct ltersmdashldquoRecentrdquo(ldquoextantrdquo or ldquolivingrdquo) or ldquoextinctrdquo may beused in phylogenetic denitions of crownand stem taxa respectively to help to stabi-lize taxonomic content in the face of poorlyresolved basal relationships For a crowntaxon with a stem-based denition a Recentlter can restrict the boundaries of a cladeto living taxa (Meng et al 1994 Wyss andMeng 1996) Rodentia for example could

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 343

be dened as ldquothe least inclusive clade com-posed of Recent mammals more closely re-lated to Mus than to Lepusrdquo For stem taxain contrast a stem-based denition using anldquoextinctrdquo lter can restrict inclusion to ex-tinct taxa when outgroup relationships areuncertain Moas for example could be de-ned as ldquoall extinct species more closelyrelated to Dinornis than to Struthiordquo theostrich

NumbermdashThe number of reference taxathat have been used in phylogenetic deni-tions has varied from two to innity Gau-thier (198614) for example dened Aves asa node-based taxon including ldquoall descen-

FIGURE 11 Nestedness and completeness in the se-lection of reference taxa A nested well-known taxonis more likely to constitute a stable reference taxon (en-circled)

dants of the most recent common ancestorof Ratitae Tinami and Neognathaerdquo Threereference taxa rather than two were chosenpresumably because of unstated uncertaintyin the phylogenetic relationships amongthese three avian subgroups De Queirozand Gauthier (1990310 1992461) suggestedthat phylogenetic denitions should utilizeonly two reference taxamdashbut then erecteddenitions for total taxa that used as refer-ence taxa one crown clade plus ldquoall extinct

taxa more closely related to it than to anyother crown claderdquo (de Queiroz and Gau-thier 1992475)

Listing more than a pair of referencetaxa does not seem to compromise a node-based phylogenetic denition in any dis-cernible way Node-based denitions areconstructed with what may be termed in-ternal inclusion A contradictory relation-ship between multiple ingroup referencetaxa for a node-based denition is im-possible to construct with internal inclu-sion because a clade always exists thatincludes all ingroup reference taxa Multi-ple ingroup reference taxa effectively stabi-lize the taxonomic content of a node-basedtaxon if basal relationships are uncertain(Fig 12a)

For stem-based denitions multiple out-group reference taxa effectively stabilize thetaxonomic content of a taxon if outgroup re-lationships are uncertain (Fig 12b) Multi-ple ingroup reference taxa for stem-baseddenitions can create a contradictory sit-uation if one (or more) ingroup referencetaxa are repositioned closer to the outgroupreference taxon (Fig 12c) or to a positionoutside the clade as originally dened Inthis case the external inclusion fundamen-tal to a stem-based denition creates a con-tradictory relationship among the originalingroup reference taxa which are no longermore closely related to each other than to theoutgroup reference taxon (Fig 12c) In theexample shown here relocation of ingroupreference taxon D creates a contradictory re-lationship with a second ingroup referencetaxon E if taxon A has the stem-based def-inition of ldquoall descendants more closely re-lated to taxon D and E than to taxon Brdquo Aninclusive ingroup reference taxon howevercan manifest the same problems in this cir-cumstance (Fig 12c) If taxon A is denedas ldquoall descendants more closely related totaxon C than to taxon Brdquo and if taxon C isdened on the basis of taxon D a contra-dictory denition results with relocation oftaxon D as shown taxon A now subsumesits original ingroup reference taxon C Suchcontradictory denitions can be avoided ifa nested ingroup reference taxon (or nestedtaxa) are selected such as taxon E its un-

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 14: Sereno, 1999b

342 SYSTEMATIC BIOLOGY VOL 48

and stem-based taxa In the current litera-ture maximally inclusive reference taxa areoften used in phylogenetic denitions Intheir tetrapod classication for example deQueiroz and Gauthier (1992475) used suc-cessive maximally inclusive reference taxafor all denitionsmdasha pattern here termed re-cursive referencing (Fig 10a) Recursive refer-encing can have undesirable consequencesbecause it shifts the burden of a more precisedenition to a slightly less inclusive highertaxon This higher taxon may not be de-ned or it too may be affected by an alter-native conguration of basal relationshipsSalgado et al (1997) for example choseProsauropoda and Sauropoda as maximallyinclusive reference taxa for Sauropodomor-pha If a basal sauropodomorph is reposi-tioned the denitions of Prosauropoda andSauropoda must be determined to assess po-tential taxonomic consequences HoweverSalgado et al (1997) used Prosauropodaas a terminal taxon without denitionSauropoda was dened as a node-basedtaxon on the basis of two maximally in-clusive reference taxa Vulcanodon and Eu-sauropoda (all other sauropods) Thusif the incompletely known basal sauro-pod Vulcanodon is reinterpreted as be-ing more closely related to prosauropodsSauropoda and Sauropodomorpha wouldbecome synonyms

If on the other hand Sauropodomorphais dened on the basis of the nested ref-erence taxamdashthe prosauropod Plateosaurusand sauropod Saltasaurus (Sereno 1998)mdasha change in the phylogenetic positionof any other prosauropod or sauropodhas no effect on the taxonomic con-tents of Sauropodomorpha or SauropodaSauropodomorpha can be dened as ldquoPla-teosaurus Saltasaurus their most recentcommon ancestor and all descendantsrdquoand Sauropoda can be dened as ldquoallsauropodomorphs more closely related toSaltasaurus than to Plateosaurusrdquo The phylo-genetic position of Vulcanodon has no effecton the identication of respective commonancestors

CompletenessmdashPoorly known taxa aremore likely to be phylogenetically unsta-ble than well-known taxa (Wilkinson 1995)

FIGURE 10 Referencing styles (a) Recursive refer-encing selects a succession of the most inclusive ref-erence taxa (b) Nested referencing selects that sameremote reference taxon Reference taxa are encircled

Taxa with numerous missing entries in aphylogenetic analysis usually reduce phylo-genetic resolution by generating numerousequally parsimonious trees Well-known ref-erence taxa even if they are somewhat lessnested are preferable to poorly known taxa(Fig 11) A single poorly known basal taxonis least stable and therefore least desirableas a reference taxon

Use with Recent or extinct ltersmdashldquoRecentrdquo(ldquoextantrdquo or ldquolivingrdquo) or ldquoextinctrdquo may beused in phylogenetic denitions of crownand stem taxa respectively to help to stabi-lize taxonomic content in the face of poorlyresolved basal relationships For a crowntaxon with a stem-based denition a Recentlter can restrict the boundaries of a cladeto living taxa (Meng et al 1994 Wyss andMeng 1996) Rodentia for example could

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 343

be dened as ldquothe least inclusive clade com-posed of Recent mammals more closely re-lated to Mus than to Lepusrdquo For stem taxain contrast a stem-based denition using anldquoextinctrdquo lter can restrict inclusion to ex-tinct taxa when outgroup relationships areuncertain Moas for example could be de-ned as ldquoall extinct species more closelyrelated to Dinornis than to Struthiordquo theostrich

NumbermdashThe number of reference taxathat have been used in phylogenetic deni-tions has varied from two to innity Gau-thier (198614) for example dened Aves asa node-based taxon including ldquoall descen-

FIGURE 11 Nestedness and completeness in the se-lection of reference taxa A nested well-known taxonis more likely to constitute a stable reference taxon (en-circled)

dants of the most recent common ancestorof Ratitae Tinami and Neognathaerdquo Threereference taxa rather than two were chosenpresumably because of unstated uncertaintyin the phylogenetic relationships amongthese three avian subgroups De Queirozand Gauthier (1990310 1992461) suggestedthat phylogenetic denitions should utilizeonly two reference taxamdashbut then erecteddenitions for total taxa that used as refer-ence taxa one crown clade plus ldquoall extinct

taxa more closely related to it than to anyother crown claderdquo (de Queiroz and Gau-thier 1992475)

Listing more than a pair of referencetaxa does not seem to compromise a node-based phylogenetic denition in any dis-cernible way Node-based denitions areconstructed with what may be termed in-ternal inclusion A contradictory relation-ship between multiple ingroup referencetaxa for a node-based denition is im-possible to construct with internal inclu-sion because a clade always exists thatincludes all ingroup reference taxa Multi-ple ingroup reference taxa effectively stabi-lize the taxonomic content of a node-basedtaxon if basal relationships are uncertain(Fig 12a)

For stem-based denitions multiple out-group reference taxa effectively stabilize thetaxonomic content of a taxon if outgroup re-lationships are uncertain (Fig 12b) Multi-ple ingroup reference taxa for stem-baseddenitions can create a contradictory sit-uation if one (or more) ingroup referencetaxa are repositioned closer to the outgroupreference taxon (Fig 12c) or to a positionoutside the clade as originally dened Inthis case the external inclusion fundamen-tal to a stem-based denition creates a con-tradictory relationship among the originalingroup reference taxa which are no longermore closely related to each other than to theoutgroup reference taxon (Fig 12c) In theexample shown here relocation of ingroupreference taxon D creates a contradictory re-lationship with a second ingroup referencetaxon E if taxon A has the stem-based def-inition of ldquoall descendants more closely re-lated to taxon D and E than to taxon Brdquo Aninclusive ingroup reference taxon howevercan manifest the same problems in this cir-cumstance (Fig 12c) If taxon A is denedas ldquoall descendants more closely related totaxon C than to taxon Brdquo and if taxon C isdened on the basis of taxon D a contra-dictory denition results with relocation oftaxon D as shown taxon A now subsumesits original ingroup reference taxon C Suchcontradictory denitions can be avoided ifa nested ingroup reference taxon (or nestedtaxa) are selected such as taxon E its un-

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 15: Sereno, 1999b

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 343

be dened as ldquothe least inclusive clade com-posed of Recent mammals more closely re-lated to Mus than to Lepusrdquo For stem taxain contrast a stem-based denition using anldquoextinctrdquo lter can restrict inclusion to ex-tinct taxa when outgroup relationships areuncertain Moas for example could be de-ned as ldquoall extinct species more closelyrelated to Dinornis than to Struthiordquo theostrich

NumbermdashThe number of reference taxathat have been used in phylogenetic deni-tions has varied from two to innity Gau-thier (198614) for example dened Aves asa node-based taxon including ldquoall descen-

FIGURE 11 Nestedness and completeness in the se-lection of reference taxa A nested well-known taxonis more likely to constitute a stable reference taxon (en-circled)

dants of the most recent common ancestorof Ratitae Tinami and Neognathaerdquo Threereference taxa rather than two were chosenpresumably because of unstated uncertaintyin the phylogenetic relationships amongthese three avian subgroups De Queirozand Gauthier (1990310 1992461) suggestedthat phylogenetic denitions should utilizeonly two reference taxamdashbut then erecteddenitions for total taxa that used as refer-ence taxa one crown clade plus ldquoall extinct

taxa more closely related to it than to anyother crown claderdquo (de Queiroz and Gau-thier 1992475)

Listing more than a pair of referencetaxa does not seem to compromise a node-based phylogenetic denition in any dis-cernible way Node-based denitions areconstructed with what may be termed in-ternal inclusion A contradictory relation-ship between multiple ingroup referencetaxa for a node-based denition is im-possible to construct with internal inclu-sion because a clade always exists thatincludes all ingroup reference taxa Multi-ple ingroup reference taxa effectively stabi-lize the taxonomic content of a node-basedtaxon if basal relationships are uncertain(Fig 12a)

For stem-based denitions multiple out-group reference taxa effectively stabilize thetaxonomic content of a taxon if outgroup re-lationships are uncertain (Fig 12b) Multi-ple ingroup reference taxa for stem-baseddenitions can create a contradictory sit-uation if one (or more) ingroup referencetaxa are repositioned closer to the outgroupreference taxon (Fig 12c) or to a positionoutside the clade as originally dened Inthis case the external inclusion fundamen-tal to a stem-based denition creates a con-tradictory relationship among the originalingroup reference taxa which are no longermore closely related to each other than to theoutgroup reference taxon (Fig 12c) In theexample shown here relocation of ingroupreference taxon D creates a contradictory re-lationship with a second ingroup referencetaxon E if taxon A has the stem-based def-inition of ldquoall descendants more closely re-lated to taxon D and E than to taxon Brdquo Aninclusive ingroup reference taxon howevercan manifest the same problems in this cir-cumstance (Fig 12c) If taxon A is denedas ldquoall descendants more closely related totaxon C than to taxon Brdquo and if taxon C isdened on the basis of taxon D a contra-dictory denition results with relocation oftaxon D as shown taxon A now subsumesits original ingroup reference taxon C Suchcontradictory denitions can be avoided ifa nested ingroup reference taxon (or nestedtaxa) are selected such as taxon E its un-

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 16: Sereno, 1999b

344 SYSTEMATIC BIOLOGY VOL 48

FIGURE 12 The effects of multiple reference taxa (a) Multiple ingroup reference taxa effectively maintain thetaxonomic content of node-based taxon A when basal relationships are unresolved or are unstable as a result ofmissing data (b) Multiple outgroup reference taxa effectively maintain the taxonomic content of stem-based taxonA when outgroup relationships are unresolved or are unstable as a result of missing data (c) Inclusive ingroupreference taxon C or multiple ingroup reference taxa (D and E) can result in denitional ambiguity for stem-basedtaxon A when basal taxon D changes its relationships a single nested ingroup reference taxon such as taxon Eis preferable in stem-based denitions Reference taxa are encircled a dot indicates a node-based denition anarrow indicates a stem-based denition a dashed line indicates a change in position of a basal taxon

labeled sister taxon or both Most contra-dictory denitionsmdashwhich previously havebeen cited as a general weakness of stem-based denitions (Schander and Thollesson1995)mdashare the result of choosing maximallyinclusive reference taxa rather than more-nested relatives Still multiple ingroup ref-erence taxa may best be avoided in stem-based denitions

Node-Stem TripletGiven taxon A and subordinate taxa B and

C (a simple dichotomy) there are only three

resolved positions for relocation or additionof taxa (Fig 8) If taxa A B and C have stem-based denitions the following obtains withintroduction of new taxa as shown Taxa Band C will incorporate an additional taxonwith no change in their relationship withthe basal dichotomy or with taxon A Whentaxon A incorporates a new taxon howeverit is removed from the original basal di-chotomy the taxonomic content of taxon Acan no longer be expressed as taxon B plustaxon C but rather must include an addi-tional taxon If taxa A B and C have node-

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 17: Sereno, 1999b

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 345

based denitions the following obtains withintroduction of new taxa (Fig 8) The taxo-nomic content of taxon A is not affected byan additional outgroup taxon which is ex-cluded by denition Taxa B and C howeverare removed from the basal dichotomy thetaxonomic content of taxon A as a result isno longer taxa B and C and their most re-cent ancestor but rather must include twoadditional taxa

Only one denitional conguration aNST preserves the taxonomic content ofthree taxa about a dichotomy as expressedby the equivalence statement A = B + C ANST is composed of a node-based taxon andtwo subordinate stem-based taxa Additionof taxa to resolved locations around a NSTcannot alter the simplest expression of rela-tive taxonomic content A = B + C

When applying a NST to dichotomies ona cladogram two observations are note-worthy First only a small subset of exist-

ing dichotomies in the history of life arelabeled with a trio of taxon names suit-able for a NST such as Dinosauria = Or-nithischia + Saurischia (Fig 13) Naming allor even most clades is not a heuristic en-deavor that would enhance communicationSecond NSTs cannot be constructed for ad-jacent dichotomies on a cladogram withoutcoining new companion node- or stem-based taxa An existing stem-based subor-dinate group of one NST such as Ornithis-chia cannot simultaneously function as anode-based group of a less-inclusive NST atthe base of Ornithischia A companion node-based taxon (composed of Pisanosaurus andother ornithischians) would need to be de-ned for this less-inclusive NST (Fig 13)Coining numerous companion taxa that dif-fer only in denitional type but not in cur-rent taxonomic content will not enhancecommunication A complete NST networkwould require node- and stem-based taxa at

FIGURE 13 Recognition of NSTs in the higher-level taxonomy of Dinosauria (from Sereno 1998) Stem-basedtaxa are represented by an arrow (a) or regular typeface (b) Node-based taxa are represented by a dot (a) orboldface (b)

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 18: Sereno, 1999b

346 SYSTEMATIC BIOLOGY VOL 48

each branchpoint A more conservative ra-tionale for placement of NSTs at named di-chotomies is considered below

Taxonomic FrameworkTaxon names areapplied toorganismal di-

versity as tags for identication and infor-mation storage and retrieval (Farris 1979)The following kinds of dichotomies havebeen ldquotaggedrdquo with a trio of names andused mostoften by systematists (1) a higher-level crown clade with subordinate crown or

total clades (2) a clade composed of two di-verse subclades (3) a morphologically dis-tinct clade composed of two distinctive sub-clades and (4) a dichotomy with signicanthistorical usage Placement of NSTs is basedon four criteria that mirror these traditionallabeled dichotomies

SurvivorshipmdashCrown stem and totaltaxa are dened on the basis of survivor-ship (Table 1) which can provide the basisfor NSTs (Fig 14) The strength of this crite-rion lies in its unambiguous interpretation

FIGURE 14 Survivorship as the basis for supraspecic taxonomy (a) Indented crown-total taxonomy empha-sizing the association between a total taxon and its less inclusive crown taxon (de Queiroz and Gauthier 1992)(b) Indented crown-total taxonomy emphasizing NST structure (c) Indented crown taxonomy omitting total taxa(d) NST (shaded) form the framework of a crown-total taxonomy Regular typeface (andashc) or an arrow (d) indicatesa stem-based taxon boldface (andashc) or a dot (d) indicates a node-based taxon daggers indicate extinct taxa

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 19: Sereno, 1999b

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 347

It is highly unlikely that a species believed tobe extinct will be discovered alive and viceversa The rapid extinction of the modernbiota is a notable exception (Lucas 1992) se-lection of a particular historical datum (suchas the beginning of the Recent) may at leastpartially side-step this problem NSTs basedon higher-level crown taxa and their asso-ciated subordinate total taxa form the back-bone of the tetrapod classication proposedby de Queiroz and Gauthier (1992) Becausethese authors emphasized the relationshipbetween a crown clade and its more inclu-sive total clade (as shown by their indentedclassication see Fig 14a) the NST struc-ture of their classication was not discussedand is not immediately apparent Their clas-sicatory scheme nevertheless consists ofsubordinate NSTs that can be tabulated asan indented crown-total taxonomy (Fig 14bd) One can list only crown groups (a crownclassication Fig 14c) although the pur-pose of such a classication is not immedi-ately apparent given the inclusion of manyextinct species within crown taxa

DiversitymdashRelative diversity is an impor-tant consideration in the erection and posi-tioning of higher taxon names Some groupsare much more diverse than others and thisdifference is often accorded biological sig-nicance On a cladogram the simplest di-versity comparisons are between sister taxaon either side of a dichotomy (Fig 15a) Adiversity-based NST stabilizes the associa-tion of taxon names around a dichotomy thatinvolves diverse sister groups and severallow-diversity outgroups A diversity-basedNST ties taxa to diversity despite relocationor addition of taxa near the basal dichotomyThere is only one way in which a diversity-based NST can be altered to dissociate thenamed trio of taxa and high diversity relo-cation of most of the diversity of one of thesister groups but not the included referencetaxon Alternatively one could view thisdissociation as the relocation of a referencetaxon butnotof themajority of its previouslyassociated diversity Such phylogenetic rear-rangements are unlikely (with nested refer-ence taxa) and diversity remains one of theleast ambiguous and most stable criteria forestablishing NSTs

A diversity-based NST recognizing Di-nosauria and its two diverse subclades Or-nithischia and Saurischia ties these threetaxa to a major dichotomy in the evolu-tion of archosaurs (Fig 13a) Traditional sys-tematists have maintained this dichotomyby interpreting newly discovered basal di-nosaurs as either basal ornithischians orbasal saurischians and by excluding out-groups from Dinosauria The only way todissociate diversity and a diversity-basedNST positioned at Dinosauria is to relocatethe deeply nested reference taxa Triceratopsand Neornithes outside of Ornithischia andSaurischia respectively (Sereno 1998)

MorphologymdashSome taxa are more distinc-tive morphologically than others Phyloge-netic analysis of morphological data rarelydistributes apomorphies evenly acrossnodes and those nodes characterized byprofound transformations are often the sub-ject of biological discourse A morphology-based NST recognizes a dichotomy boundedby numerous apomorphies (Fig 15b) Thenumber of apomorphies (or the size of themorphologic ldquogaprdquo) however is less sta-ble than diversity as a criterion for distin-guishing some branchpoints from othersDiscovery of new taxa inevitably entailsthe splitting of morphologic ldquogapsrdquo andalternative optimization of character datawith many missing entries may also dra-matically shift the locus of character-statechange

Some morphologic ldquogapsrdquo nonethelesshave remained remarkably stable partic-ularly those among extant species asso-ciated with a poor or nonexistent fos-sil record Gap-bounded dichotomies alsooccur among extinct taxa The distinctiveclade Sauropodomorpha for example iscomposed of the distinctive subcladesProsauropoda and Sauropoda Despitemore than a century of discovery morpho-logically intermediate species have yet tobe uncovered that signicantly reduce thedistances at this dichotomy A morphology-based NST at this dichotomy reects his-torical experiencemdashthat newly discoveredsauropodomorphs have been positionedamong known prosauropod or sauropodtaxa whether this is the result of punctu-

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 20: Sereno, 1999b

348 SYSTEMATIC BIOLOGY VOL 48

FIGURE 15 Criteria for positioning NST (a) Diversity-based NST in which two high-diversity taxa have succes-sive low-diversity outgroups (b) Morphology-based NST in which a distinctive (particularly apomorphic) taxonis composed of two distinctive subordinate taxa (c) Tradition-based NST in which a taxon and its subordinatetaxa have a long historical association

ated evolution temporal gaps in the fossilrecord or chance

TraditionmdashSome taxonomic names havea long-standing association with a particu-lar dichotomy Most often this association isbased on a ldquokeyrdquo character or set of charac-ters and involves a taxon that is transitionalin form (ie a ldquomissing linkrdquo bounded bymorphologic ldquogapsrdquo) For the past centuryfor example ldquoAvesrdquo has been associatedwith the dichotomy between Archaeopteryxand Ornithurae a taxon comprising all otherbirds (Fig 15c) Archaeopteryx preservesevidence of feathers wings a perching footand other striking adaptations reminiscentof those of living birds Surely however

a ldquofeatherrdquo or ldquowingrdquo will be scored andinterpreted in different ways by differentsystematists Likewise the long-standingassociation of several of these ldquokeyrdquo synapo-morphies with Archaeopteryx and Ornithu-rae clearly is oversimplied (Chen et al1998 Ji et al 1998) Still the name Aves maybe better applied to its traditional dichotomy(Fig 15c Lee 1996 Sereno 1998) than re-located to a distant node such as crown-group avians (Gauthier 1986) even if only toavoid confusion on the part of the majorityof biologists A tradition-based NST placedat the dichotomy between Archaeopteryx andOrnithurae stabilizes the long-standing his-torical interpretation of Aves and maintains

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 21: Sereno, 1999b

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 349

aclose albeit imprecise association betweenAves and synapomorphies involving mod-ern feather wing and pedal design

The four criteria discussed above are nei-ther exact nor mutually exclusive The num-ber of taxa or apomorphies characterizingldquohigh diversityrdquo or ldquomorphologically dis-tinctrdquo clades respectively is relative and im-precise Of the four criteria morphology-based dichotomies are least useful becausethey are rare and relatively unstable overtime and can exist solely as artifacts of miss-ing data

The best case for location of a NST is madewhen several of the criteria described abovepoint to the same dichotomy as is the casewith the nodeDinosauria Within dinosaursthe ceratopsid subgroups Ceratopsinae andCentrosaurinae provide another exampleeach more diverse and distinctive than theirnearestoutgroups (Fig 16) When criteria forplacement of a NST disagree or when theproximity of another NST would necessitatecoining new companion taxa a NST neednot be established (eg the dichotomy in-volving Saurischia Sauropodomorpha andTheropoda Fig 13a)

Subdivision of one group into twosubordinate groups is the simplest form ofnonredundant ordination in taxonomy andreects the mostprevalentphylogenetic pat-tern The principal reasons that these di-chotomies were named in the rst place wasto acknowledge survivorship to the Recentto reference large-scale patterns in diversityand to record major changes in morphologyMaintaining the integrity of such named di-chotomies is a heuristic endeavor in con-cert with the aims of phylogenetic taxonomyand can be achieved by judicious choice ofnested reference taxa and the NST congu-ration of phylogenetic denitions

CONCLUSIONS

The foregoing critique of taxonomic def-initions has resulted in the following majorconclusions

1 Currently there is no clear rationale forformulation and placement of phyloge-netic denitions

2 Apomorphy-based denitions unlike

FIGURE 16 NST among ceratopsian dinosaursbased on the criteria outlined in Figure 15 Num-ber of genera for each terminal taxon are shownwithin circles which highlights the diversity of cer-atopsid subclades as compared with the outgroupsThe thickened clade lines are scaled according tothe number of synapomorphies that diagnose eachclade which underscores the morphologic dispar-ity of Ceratopsidae and its two subgroups com-pared with adjacent more inclusive nodes (scalebar equals 20 apomorphies Sereno 1998) The in-dented classication reveals a century-long tradi-tion that recognizes Ceratopsidae and its two sub-groups Node-based taxa are represented by a dot(or boldface) and stem-based taxa are representedby an arrow (or regular typeface)

node-based and stem-based denitionsmanifest many of the same ambiguitiesthat have long been associated with tra-ditional character-based taxa

3 Crown taxa are dened with living ref-erence taxa but they require an imme-diate outgroup that is extinct Crowntaxa are not demonstrably more stableor informative than stem or total taxaand ldquowidely usedrdquo higher taxon nameshavenotbeen historically associated withcrown taxa

4 Stem taxa are restricted to extinct clades(or species) whose immediate outgroupincludes at least one living species De-

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 22: Sereno, 1999b

350 SYSTEMATIC BIOLOGY VOL 48

ned in this way a total taxon is com-posed of a crown taxon plus all stem taxamore closely related to it than to anothercrown taxon

5 Well-known nested reference taxa ratherthan maximally inclusive reference taxastabilize taxonomic content Use of mul-tiple reference taxa in addition does notopen phylogenetic denitions to internalconict except as ingroup reference taxain a stem-based denition

6 A NST conguration of complementarytaxonomic denitions stabilizes the as-sociation of a trio of taxa (A B C) at adichotomous branchpoint such that thetaxonomic content of taxon A will alwaysequal that of taxa B and C plus their mostrecent common ancestor

7 Survivorship diversity morphology andtradition are heuristic criteria for place-ment of node-stem triplets

ACKNOWLEDGMENTS

I am indebted to C Abraczinskas for executing thenished illustrations and to A Beck J Gauthier J Hop-son H Larsson M Lee and J Wilson for their com-ments on earlier drafts of this article This research wassupported by grants from the Pritzker Foundation andThe David amp Lucile Packard Foundation

REFERENCES

AX P 1987 The phylogenetic system The systemati-zation of organisms on the basis of their phylogene-sis Wiley and Sons New York

BROCHU C A 1997 A review of ldquoLeidyosuchusrdquo(Crocodyliformes Eusuchia) from the Cretaceousthrough Eocene of North America J Vertebr Pale-ontol 17679ndash697

BRYANT H N 1994 Comments on the phylogeneticdenition of taxon names and conventions regardingthe naming of crown clades Syst Biol 43124ndash130

BRYANT H N 1996 Explicitness stability and uni-versality in the phylogenetic denition and usage oftaxon names A case study of the phylogenetic taxon-omy of the Carnivora (Mammalia) Syst Biol 45174ndash189

BUCK R AND D L HULL 1966 The logical structureof the Linnaean hierarchy Syst Zool 1597ndash111

CHEN P Z DONG AND S ZHEN 1998 An exception-ally well-preserved theropod dinosaur from the Yix-ian Formation of China Nature 391147ndash152

CHIAPPE L M 1991 Cretaceous avian remains fromPatagonia shed new light on the early radiation ofbirds Alcheringa 15333ndash338

CHIAPPE L M 1996 Late Cretaceous birds of south-ern South America Anatomy and systematics of

Enantiornithes and Patagopteryx deferrariisi MunchGeowiss Abh 30203ndash244

DAESCHLER E B AND N SHUBIN 1998 Fish with n-gers Nature 391133

DARWIN C 1859 On the origins of species by naturalselection John Murray London

DE QUEIROZ K 1992 Phylogeneticdenitions and tax-onomic philosophy Biol Philos 7295ndash313

DE QUEIROZ K 1994 Replacement of an essentialisticperspective on taxonomic denitions as exempliedby the denition of ldquoMammaliardquo Syst Biol 43497ndash510

DE QUEIROZ K AND J A GAUTHIER 1990 Phylogenyas a central principle in taxonomy Phylogenetic def-initions of taxon names Syst Zool 39307ndash322

DE QUEIROZ K AND J A GAUTHIER 1992 Phyloge-netic taxonomy Annu Rev Ecol Syst 23449ndash480

FARRIS J S 1976 Phylogenetic classication of fossilswith recent species Syst Zool 25271ndash282

FARRIS J S 1979 The information content of the phy-logenetic system System Zool 28483ndash519

FOREY P C 1992 Fossils and cladistic analysis Pages124ndash136 in Cladistics A practical course in system-atics (P L Forey C J Humphries I C Kitching RW Scoltand D J Siebert and R M Williams eds)Oxford University Press Oxford

FORSTER C A S D SAMPSON L M CHIAPPE AND D WKRAUSE 1998 The theropod ancestry of birds Newevidence from the Late Cretaceous of MadagascarScience 2791915ndash1919

FURBRINGER M 1888 Untersuchungen zur Mor-phologie und Systematik der Vogel Holkema Am-sterdam

FUTUYMA D J 1986 Evolutionary biology SinauerSunderland Massachusetts

GADOW H 1893 Vogel II Systematischer Theil Dr JG Bronnrsquos Klassen and Ordnungen des Their-ReichsC F Winterrsquosche Verlagshandlung Leipzig

GAUTHIER J A 1986 Saurischian monophyly and theorigin of birds Mem Calif Acad Sci 81ndash55

GHISELIN M T 1984 ldquoDenitionrdquo ldquocharacterrdquo andother equivocal terms Syst Zool 33104ndash110

GRIFFITHS G C D 1973 Some fundamental problemsin biological classication Syst Zool 22338ndash343

GRIFFITHS G C D 1974 On the foundations of bio-logical systematics Acta Biotheor 2385ndash131

HAECKEL E 1866 Generelle Morphologie der Or-ganismen Allgemeine Grundzuge der organischenFormen Wissenschaft mechanisch begrundet durchdie von Charles Darwin reformierte Deszendenz-Theorie Georg Reimer Berlin

HENNIG W 1969 Die Stammesgeschichte der Insek-ten Kramer Frankfurt

HENNIG W 1983 Stammesgeschichte der ChordatenFortschr Zool Syst Evolutionsforsch 21ndash208

JEFFERIES R P S 1979 The origin of chordatesmdashamethodological essay Pages 443ndash477 in The origin ofmajor invertebrate groups (M R House ed) Aca-demic Press London

JI Q P J CURRIE M A NORELL AND S-A JI Twofeathered dinosaurs from northeastern China Na-ture 393753ndash761

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett

Page 23: Sereno, 1999b

1999 SERENOmdashRATIONALE FOR PHYLOGENETIC DEFINITIONS 351

JOLLIE M T 1973 Chordate morphology RobertKrieger Huntington New York

LEE M S Y 1996 Stability in meaning and content oftaxon names An evaluation of crown-clade deni-tions Proc R Soc London B 2631103ndash1109

LEE M S Y 1998 Ancestors and taxonomy TrendsEcol Evol 1326

LUCAS S G 1992 Extinction and denition of theClass Mammalia Syst Biol 41370ndash371

MARSH O C 1872a Discovery of a remarkable fossilbird Am J Sci 356ndash57

MARSH O C 1872b Notice of a new remarkable fossilbird Am J Sci 4344

MCKENNA M C AND S K BELL 1997 Classicationof mammals above the species level Columbia Uni-versity Press New York

MENG J A R WYSS M R DAWSON AND R-J ZHAL 1994 Primitive fossil rodent from InnerMongolia and its implications for mammalian phy-logeny Nature 370134ndash136

PATTERSON C 1981 Signicance of fossils in deter-mining evolutionary relationships Annu Rev EcolSyst 12195ndash223

PENNY D AND M HASEGAWA 1997 The platypus putin its place Nature 387549ndash550

ROWE T 1987 Denition and diagnosis in the phylo-genetic system Syst Zool 36208ndash211

ROWE T 1988 Denition diagnosis and origin ofMammalia J Vertebr Paleontol 8241ndash264

ROWE T AND J A GAUTHIER 1992 Ancestry paleon-tology and denition of the name Mammalia SystBiol 41372ndash378

SALGADO L R A CORIA AND J O CALVO 1997 Evo-lution of titanosaurid sauropods I Phylogenetic

analysis based on the postcranial evidence Amegh-iniana 343ndash32

SCHANDER C AND M THOLLESSON 1995 Phyloge-netic taxonomymdashsome comments Zool Scripta24263ndash268

SERENO P C 1997 The origin and evolution of di-nosaurs Annu Rev Earth Planet Sci 1997435ndash489

SERENO P C 1998 A rationale for phylogenetic def-initions with application to the higher-level taxon-omy of Dinosauria Neues Jahrb Geol Palaont Abh21041ndash83

SMITH A B 1994 Systematics and the fossil recordDocumenting evolutionary patterns Blackwell Sci-entic Oxford England

WALKER E P 1975 Mammals of the world 3rd edi-tion Johns Hopkins Univ Press Baltimore

WILEY E O 1981 Phylogenetics The theory andpractice of phylogenetic systematics Wiley and SonsNew York

WILKINSON M 1995 Coping with abundant missingentries in phylogenetic inference using parsimonySyst Biol 44501ndash514

WILSON D E AND D M REEDER 1993 Mammalspecies of the world 2nd ed Smithsonian InstitutePress Washington D C

WYSS A AND J MENG 1996 Application of phylo-genetic taxonomy to poorly resolved crown cladesA stem-modied node-based denition of RodentiaSyst Biol 45559ndash568

Received 8 June 1998 accepted 27 September 1998Associate Editor D Hibbett