The Identity and Homology of the Postprotocrista and its Role in Molarization of Upper Premolars of Perissodactyla (Mammalia)

1 23

1 23

Your article is protected by copyright and allrights are held exclusively by Springer Science+Business Media New York. This e-offprint isfor personal use only and shall not be self-archived in electronic repositories. If you wishto self-archive your article, please use theaccepted manuscript version for posting onyour own website. You may further depositthe accepted manuscript version in anyrepository, provided it is only made publiclyavailable 12 months after official publicationor later and provided acknowledgement isgiven to the original source of publicationand a link is inserted to the published articleon Springer's website. The link must beaccompanied by the following text: "The finalpublication is available at link.springer.com”.

ORIGINAL PAPER

The Identity and Homology of the Postprotocrista and its Rolein Molarization of Upper Premolarsof Perissodactyla (Mammalia)

Luke Holbrook

Published online: 9 September 2014# Springer Science+Business Media New York 2014

Abstract Two different crests have been identified in theupper cheek teeth of perissodactyls as the postprotocrista.Both occur in the upper premolars of some early perissodac-tyls, demonstrating that they are not homologous. The truepostprotocrista extends distolabially from the protocone, oftenconnecting the protocone and the metaconule, at least primi-tively. The other crest, here termed the endoprotocrista, ex-tends less labially and more distally than the postprotocristaand does not connect to the metaconule. In some taxa, thehypocone arises from the distal end of the endoprotocrista.Thus, the endoprotocrista plays an important role inmolarization of some perissodactyl upper premolars.Molarization is completed by separation of the hypocone fromthe protocone and the metaconule connecting to the hypocone,involving either migration or loss of the postprotocrista. Theendoprotocrista is similar to the “Nannopithex fold” of NorthAmerican adapid primates, but it is otherwise currently knownonly in some early perissodactyls. Perissodactyl upper premo-lars become molarized by at least two modes. One modeinvolves the development of the hypocone from theendoprotocrista, as described here. A second involves theenlargement and lingual migration of the paraconule, as in-ferred for the third upper premolar of early equids. In theupper molars, the hypocone appears to be derived from the

cingulum. The present study highlights the difficulties inmaking inferences regarding the mode of premolarmolarization from static morphology. This further reflectsthe uncertainty inherent in using modes of premolarmolarization in phylogenetic analysis and concomitant prob-lems in determining primary homology.

Keywords Perissodactyla . Dentalmorphology .Homology .

Premolar . Molarization

Introduction

Perissodactyls exhibit a number of interesting trends in theirevolution. Multiple lineages have evolved larger body size,increased lophodonty, and molarized premolars (e.g., Granger1908; Radinsky 1963; Savage et al. 1965; Fig. 1). The modesof achieving greater lophodonty or greater premolarmolarization can differ among perissodactyl lineages, as wellas among different premolar loci within the same taxon. Thus,patterns of evolution of lophodonty and premolar molarizationcharacterize specific perissodactyl lineages, and dental char-acters related to these phenomena figure prominently in stud-ies of early perissodactyl phylogeny (Hooker 1989, 1994,2005, 2010; Froehlich 1999, 2002; Hooker and Dashzeveg2003, 2004; Holbrook and LaPergola 2011).

Although the precise mode of molarization may differamong lineages as well as among premolar loci, molarizationof upper premolars in perissodactyls always involves achiev-ing or approaching an occlusal surface with two major labialcusps (the paracone and metacone) and two major lingualcusps (the protocone and hypocone). In more lophodont taxa,crests may connect the cusps, typically an ectoloph joining thelabial cusps, a protoloph joining the anterior (mesial) labialand lingual cusps, and a metaloph joining the posterior (distal)labial and lingual cusps (Fig. 1b). In less lophodont taxa, a

L. Holbrook (*)Department of Biological Sciences, Rowan University, 201 MullicaHill Rd., Glassboro, NJ 08028, USAe-mail: [email protected]

L. HolbrookDivision of Paleontology, American Museum of Natural History,Central Park West and 79th St., New York, NY 10024, USA

L. HolbrookVertebrate Zoology, Academy of Natural Sciences of DrexelUniversity, 1900 Benjamin Franklin Parkway, Philadelphia,PA 19103, USA

J Mammal Evol (2015) 22:259–269DOI 10.1007/s10914-014-9276-3

Author's personal copy

mesial paraconule and a distal metaconule may be presentbetween the labial and lingual cusps (Figs. 1a and 2b).

Even in the most primitive perissodactyls, P3 and P4always exhibit a paracone and at least some development ofa metacone, as well as at least one lingual cusp, the protocone(Figs. 1a and 2a). Thus, the key distinction of molarized thirdand fourth upper premolars is the development of a secondlingual cusp (Figs. 1b and 2b). While these two lingual cusps

resemble (and are serial homologs of) the protocone andhypocone of the molars, the exact historical homologies ofthe cusps and crests depend on the mode of molarization.

One possible mode of molarization involves the metaloph.The metaloph, in tritubercular teeth like the unmolarizedpremolars of perissodactyls, consists of a crest connectingthe metaconule to the ectoloph (typically the premetaconulecrista), and the postprotocrista, connecting the metaconule to

Fig. 1 Comparison of a bunodont perissodactyl with non-molariformpremolars and a lophodont perissodactyl with molarized premolars. aNHMUK M16336, holotype of Hyracotherium leporinum, cast of rightmaxilla with P2 to M3. Note the presence of a distinct paraconule andmetaconule and only one lingual cusp on P3 and P4, versus two lingual

cusps on the molars. b ANSP 19160, Tapirus pinchaque, right maxillarydentition with P1 to M2 (M3 in crypt). Note the lophodont occlusalsurfaces of all teeth with no paraconules or metaconules visible, and thesimilarities between the molars and the second to fourth premolars,including presence of two lingual cusps. Scale bars equal 1 cm

Fig. 2 Diagrammatic representations of an upper molar of a primitiveeutherian mammal (a), based on Szalay (1969: fig. 1), and a perissodactylupper molar (or a molarized upper premolar) (b). Note that (a) is alsotypical of the morphology of a posterior upper premolar (e.g., P4) of aperissodactyl with little premolar molarization. Abbreviations: ect,ectoloph; hy, hypocone; mcl, metaconule; me, metacone; mes, mesostyle;

met, metastyle; pa, paracone; pas, parastyle; pcl, paraconule; postprc,postprotocrista; pr, protocone; premclc, premetaconule crista; prepclc,preparaconule crista; preprc, preprotocrista; psmclc, postmetaconule cris-ta; pspclc, postparaconule crista. Asterisk (*) indicates a crest without aname that might be homologous with the postprotocrista

260 J Mammal Evol (2015) 22:259–269


the protocone (Fig. 2a). Thus, in perissodactyls withunmolarized premolars, the lingual ends of both the protolophand metaloph attach to the protocone. The labial portions ofthe protoloph and metaloph generally derive from elaborationof the preparaconule and premetaconule cristae, and thepostparaconule and postmetaconule cristae are greatlyreduced in perissodactyls (Fig. 4).

In molarized premolars of perissodactyls, the metaconule(if it is visible and not merged into the loph) is attached to themore distal of the two lingual cusps, serially homologous tothe molar hypocone, resulting in a gap between the lingualcusps that opens into the central valley of the molarized tooth(Fig. 2b). An important part of this transformation is the fate ofthe postprotocrista. This study examines the role of thepostprotocrista in molarization and concludes that there areactually two non-homologous crests that have been consid-ered to be the postprotocrista, each of which has a role inmolarization of perissodactyl premolars.

Institutional Abbreviations— AMNH, AmericanMuseum of Natural History, New York; ANSP, Academy ofNatural Sciences of Drexel University, Philadelphia; CM,Carnegie Museum of Natural History, Pittsburgh; GMH,Geiseltalmuseum, Halle; NHMUK, Natural HistoryMuseum, London; NMB, Naturhistorisches Museum, Basel;UCMP, Museum of Paleontology, University of California,Berkeley; UMMP, Museum of Paleontology, University ofMichigan, Ann Arbor; USNM, U.S. National Museum ofNatural History, Smithsonian Institution, Washington, D.C.;YPM(PU), Yale Peabody Museum (Princeton UniversityCollection), New Haven.

Identity of the Postprotocrista in Perissodactyls: PreviousAuthors

In diagramming the various structures of mammalian molarsbased on the terminology of Van Valen (1966), Szalay (1969:fig. 1 and Fig 2a illustrated the postprotocrista as a crestrunning distolabially from the protocone and connecting theprotocone to the metaconule. Thus, the postprotocrista formsthe lingual portion of the primitive metaloph (or theposterocrista of Van Valen 1966). The only crests Szalaydescribed as extending from the protocone are themesiolabially-directed preprotocrista, running toward andpossibly connecting to the paraconule, and the distolabially-directed postprotocrista, running toward and possiblyconnecting to the metaconule. (These crests can be presentwithout connecting to their respective conules.)

A crest similar to the one identified as the postprotocristaby Szalay (1969) is found in the premolars of many perisso-dactyls. However, some of the crests identified as thepostprotocrista show some discrepancies with what Szalaydescribed. For instance, Froehlich’s (2002) diagrams of

hypothetical perissodactyl fourth upper premolars indicateda postprotocrista that primitively runs essentially distally, asopposed to distolabially, and the metaconule lies on this crestand therefore lies almost directly distal to the protocone(Froehlich 2002: figs. 25—28; Fig. 3a and b). Froehlich(2002: fig. 28; Fig. 3c) also illustrated a derived conditionwhere the postprotocrista extends more labially and joins theectoloph, thus forming the metaloph. This diagram does notinclude the metaconule, so there is no explicit incorporation ofthe metaconule in the metaloph.

Froehlich (2002: fig. 26 and character 34; Fig. 3d) alsoillustrated a transformation where the P4 metaconule loses itsconnection to the protocone (via the postprotocrista) andinstead connects only to the ectoloph. This forms a metalophthat is separate from the protocone and whose lingualend will ultimately form a hypocone. In essence, hehypothesized that loss of the postprotocrista was a keystep in the molarization of P4.

In actual specimens, the metaconule, when present, isalways more labially positioned than depicted by Froehlich(2002: fig. 26; Fig. 3b), e.g., as illustrated in Fig. 1a, andtherefore the postprotocrista runs more labially than suggestedin his diagrams, and is therefore more similar to the crestidentified by Szalay (1969). Froehlich’s (2002) depiction ofthe postprotocrista might simply be criticized as showing thiscrest in the wrong orientation, but there is in fact a crestextending distally from the protocone on P4 (and in somecases on P3 or P2) in a number of early perissodactyls(Fig. 4a—c, and e) that closely resembles what he illustrated

Fig. 3 Representations of various morphologies of the left P4 in peris-sodactyls based on Froehlich (2002, modified from figs. 26 and 28).Abbreviations: mcl, metaconule; ppc, postprotocrista

J Mammal Evol (2015) 22:259–269 261


(Froehlich 2002: fig. 27; Fig. 3a). This crest does not connectto the metaconule. If we interpret this as the postprotocrista ofFroehlich (2002), this crest would indicate that perissodactylshave a postprotocrista that has a fundamentally different ori-entation than in other mammals. Alternatively, this crest is notactually the postprotocrista.

Identity of the Postprotocrista in Perissodactyls:New Observations

Taxa or specimens exhibiting intermediate stages of premolarmolarization are typically required for inferring or for testing

hypotheses of cusp homologies and transformations involvedin the molarization process. Some specimens illustrate that thetwo very different orientations of the postprotocrista asinterpreted by Froehlich (2002) are not homologous. Table 1lists taxa and specimens examined for this study.

Figure 4a, b, c, and e depict upper fourth premolars ofperissodactyls that exhibit two different crests that match thetwo different conditions of the putative postprotocrista. Onecrest matches the definition of the postprotocrista as illustratedby Szalay (1969: fig. 2a), where the crest runs distolabiallyand connects to the metaconule. Additionally, a second crestextends more distally from the protocone but does not connectto the metaconule. Clearly, if both crests exist in the same

Fig. 4 Perissodactyl dentitionsillustrating morphologiesdiscussed in this paper. AMNH4831, Xenicohippus craspedotus,casts of right P3 toM1 (a) and leftP4 to M1 (b). AMNH 14911,Lambdotherium popoagicum,cast of right P4 to M1 (c). UMMP95793, Sifrhippus sandrae, castof right M1 to M3 (d). NMB Cst256, Anchilophus sp., cast of rightP2 to M1 (e). AMNH 11687,Palaeosyops robustus, cast of leftM3 (f). AMNH 74026,Mesohippus bairdi, cast of rightP1 to M1 (g). Abbreviations: epc,endoprotocrista; hy, hypocone;mcl, metaconule; pcl, paraconule;ppc, postprotocrista. Scale barsequal 1 cm

262 J Mammal Evol (2015) 22:259–269


specimen, they cannot both be the postprotocrista, and there-fore the two types of postprotocristae illustrated by Froehlich(2002: fig. 28; Fig. 3a and b) are not the same structure.

Given that there are two crests that have been called thepostprotocrista, one should bear that name and the other shouldbe given a different name. The crest running distolabially to

the metaconule matches the description of the postprotocristaof Szalay (1969) as well as that of Van Valen (1966), andretaining the term postprotocrista for this crest would be mostconsistent with the use of this term for other groups of mam-mals. For the crest extending distally from the protocone, Ipropose the term endoprotocrista, to reflect that this crest is

Table 1 List of taxa examined for this study, including specimens and appropriate references, as well as comments on premolar molarization and thestate of the endoprotocrista

Taxa Specimens and references Comments on premolars and endoprotocrista

Early equoids

Hyracotherium leporinum NHMUK M16336 Short endoprotocrista on P3, P4

Pliolophus vulpiceps NHMUK 44115 Endoprotocrista present as small but distinct distal projection from P3protocone, absent on P4

Sifrhippus sandrae UMMP 82385, 83473, 94908, 95793 Short endoprotocrista on P3, absent on P4

Protorohippus venticolus AMNH 4832 P3 paraconule enlarged and lingually shifted; no endoprotocrista visible

Xenicohippus osborni AMNH 55108, 96284, 96374, 96376 Endoprotocrista present or absent on P3, faint to absent on P4

Xenicohippus craspedotus AMNH 4831, 12819, 55100; USNM522232, 522260

Endoprotocrista varies from faint to present with incipient hypocone on P4and from absent to well-developed on P3

Orohippus pumilus USNM 363520; Granger (1908) P3 paraconule enlarged and lingually shifted; no endoprotocrista visible

Epihippus parvus USNM 21095; Granger (1908) P3 and P4 completely molariform

Mesohippus bairdi AMNH 74026 P2-4 completely molariform

Anchilophus sp. NMB Cst 256, StH 2601 P2-4 with endoprotocrista and attached hypocone

Lophiotherium pygmaeum Savage et al. (1965: fig. 37) Small hypocone variably present on P4, seems to be derived fromendoprotocrista

Lophiotherium cervulum NMB StH 257; AMNH 10662 P2-4 completely molariform to premolariform

Hallensia matthesi GMH XIV/3106 Endoprotocrista faint at best on P4, absent on P3

Propachynolophus gaudryi NMB Ts 83 Endoprotocrista faint at best on P3-4

Eurohippus parvulus NMB Eb 108, Eb 117, Eh 336, Eh 793,El 792

P4 endoprotocrista varies from short and faint to strong and extending todistal cingulum

Plagiolophus annectens NMB StH 404 Endoprotocrista present on P3-4 with small attached hypocone

Palaeotherium crassum UCMP 62704 P2-4 completely molariform

Brontotherioids

Lambdotherium popoagicum AMNH 14911; CM 67403 Endoprotocrista present without hypocone

Eotitanops borealis AMNH 14887 No molarization or endoprotocrista

Palaeosyops robustus AMNH 11687 No molarization or endoprotocrista

Tapiromorphs (including chalicotheres)

Homogalax protapirinus AMNH 16859; UMMP 68548;Radinsky (1963)

No molarization or endoprotocrista

Cardiolophus radinskyi UMMP 78915 No molarization or endoprotocrista

Isectolophus latidens AMNH 12221; Radinsky (1963) No molarization or endoprotocrista on AMNH 12221, but Radinsky (1963)noted a small hypocone on P3 in some specimens

Heptodon calciculus AMNH 294; Radinsky (1963) No molarization or endoprotocrista

Helaletes intermedius YPM(PU) 10166; Radinsky (1963) P3-4 with hypocone attached to protocone via short endoprotocrista

Helaletes nanus AMNH 11635; Radinsky (1963) No molarization or endoprotocrista

Hyrachyus modestus AMNH 12664; Radinsky (1967);Wood (1934)

Some specimens show evidence of molarization of various premolars, withhypocone differentiating from protocone via short endoprotocrista

Litolophus gobiensis AMNH 26645 No molarization or endoprotocrista

Eomoropus amarorum AMNH 22568 No molarization or endoprotocrista

Schlosseria magister AMNH 81754 No molarization or endoprotocrista

Tapirus pinchaque ANSP 19160 P2-4 completely molariform

J Mammal Evol (2015) 22:259–269 263


“inner” (i.e., more lingual) to the other protocone cristae.Figure 5 illustrates the relations of these crests to each otherand to other tooth structures.

Although the endoprotocrista does not connect to themetaconule, it does have a role in molarization of upperpremolars. In some taxa, such as some species of theEuropean Eocene perissodactyl Anchilophus , theendoprotocrista terminates in a cusp at its distal end(Fig. 4e). Separation of this cusp from the protocone (i.e., lossof the endoprotocrista) is one way to form a hypocone.Complete molarization can then be achieved with the devel-opment of a metaloph connected to the hypocone.

This transformation of the metaloph occurs when the con-nection between the metaconule and protocone is lost and anew connection is made with the hypocone. This could meanthat the crest connecting the metaconule and protocone (thepostprotocrista) is lost, and a new crest forms the metaconule-hypocone connection; or it could mean that the postprotocristadetaches from the hypocone at its lingual end and forms a new,more distal connection with the hypocone. As seen inAnchilophus, an intermediate stage involves a connectionbetween the metaconule and the middle of the endoprotocrista(Fig. 4e). This, as well as the lack of specimens displaying twoseparate crests from the metaconule that separately connect tothe protocone and hypocone, is suggestive of the migration ofthe lingual end of the postprotocrista. However, the formerexplanation might be true for other taxa and tooth loci. Forinstance, the molars of Sifrhippus sandrae possess adistolabial ridge on the protocone that does not reach the

metaconule but otherwise appears to be a postprotocrista(Fig. 4d). At the same time, there is a complete metalophincluding a connection between the metaconule and thehypocone. A similar condition is visible in the M1 ofTapirus pinchaque (Fig. 1b), where a distolabial extension isvisible on the slightly worn protocone, suggesting apostprotocrista, yet a complete metaloph is also present.While this does not occur on the premolars of these taxa, itdoes illustrate a pattern of transformation that could possiblyoccur on premolars. A similar situation appears to apply to thepremolars of Lophiotherium cervulum , where thepostprotocrista is apparent on P3 and P4 but does not contributeto the metaloph; instead, a labial crest from the hypocone formsthe lingual portion of the metaloph (Depéret 1917: pl. 14).

The two crests that are at the center of this discussion canbe identified as follows. The postprotocrista primitively ex-tends distolabially or labially from the protocone and typicallyconnects to the metaconule; it is possible that the connectionbetween the metaconule and hypocone in taxa withmolariform represents the postprotocrista, which has discon-nected from the protocone and connected more distally to thehypocone. The endoprotocrista extends distally and less labi-ally from the protocone, it does not connect to the metaconule,and the hypocone may form at its distal end.

Variation and Distribution of the Endoprotocristain Perissodactyla

Table 1 summarizes the distribution of the endoprotocrista invarious perissodactyl taxa. Tapiromorphs, including tapiroids,rhinocerotoids, and chalicotherioids, generally lack anendoprotocrista on all upper premolars. In those taxa thatexhibit molarization, the hypocone appears to be derived fromthe distal aspect of the protocone by an endoprotocrista, butthe connection is so close that the crest is very short. Theendoprotocrista is also absent in brontotheriids, although it ispresent in the putative brontothere relative Lambdotherium.

The endoprotocrista is more commonly found on the pre-molars of early equoids, although they are absent or no morethan faint swellings in Hallensia, Propachynolophus,Protorohippus, and Orohippus. Pliolophus and Sifrhippuspossess a short endoprotocrista on P3, but not on P4. BothP3 and P4 possess an endoprotocrista in Lambdotherium,Hyracotherium, Plagiolophus, and Anchilophus, with incipi-ent hypocones visible in the last two taxa. Xenicohippus andEurohippus show considerable variation in the developmentof the endoprotocrista, ranging from absent to well developedin the P4 of both, with an incipient hypocone in one specimenof Xenicohippus craspedotum, and from absent to well devel-oped in the P3 of Xenicohippus. Taxa with molariform pre-molars do not exhibit an endoprotocrista, but at least in some

Fig. 5 Diagrammatic representation of a perissodactyl upper premolarillustrating the relations of the postprotocrista and the endoprotocrista.Abbreviations: ect, ectoloph; endprc, endoprotocrista; hy, hypocone; mcl,metaconule; me, metacone; mes, mesostyle; met, metastyle; pa, paracone;pas, parastyle; pcl, paraconule; postprc, postprotocrista; pr, protocone;premclc, premetaconule crista; prepclc, preparaconule crista; preprc,preprotocrista

264 J Mammal Evol (2015) 22:259–269


cases the crest was likely to have been lost in the course of theevolution of molarization.

Comparisons with Other Mammals

The endoprotocrista, or an equivalent crest, has not beenreported in mammals other than perissodactyls, with onepossible exception in primates. North American adapid pri-mates possess a similar structure, variously termed thepostprotocrista, Nannopithex fold, postprotocone fold, orpostprotocingulum (Gingerich and Simons 1977; Gingerichand Haskin 1981; Anemone et al. 2012). This structure is acrest extending distolingually from the protocone, and inmany cases it gives rise to a hypocone, sometimes referredto as a “pseudohypocone." Unlike perissodactyls, the crestfound in North American adapids is found in both upperpremolars and molars and attaches to the upper molarhypocone. Anemone et al. (2012) reviewed the contro-versies surrounding this crest and the homologies of theupper molar hypocones and “pseudohypocones” ofEuropean adapines and North American notharctines,respectively. They concluded that these cusps are devel-opmentally distinct and evolved convergently in thesetwo lineages, although they retained the term “hypocone”for both structures, arguing that this term is already usedto identify a variety of independently evolved cusps inmammal dentitions. While the similarities between thecrest in notharctines and the endoprotocrista are striking,these structures have certainly evolved independentlywithin primates and perissodactyls.

One could argue that, like the independently evolvedhypocones of various mammal groups, the endoprotocristaof perissodactyls is essentially the same structure as thatdescribed in notharctine primates and should therefore bearthe same name. The term endoprotocrista is preferred here forthe perissodactyl crest for two reasons. First, the terms appliedto the notharctine crest are problematic. Postprotocrista, as hasbeen discussed above, is most appropriately used for anotherstructure that is present in perissodactyls. Postprotocingulumimplies a structure that is connected to the cingulum (or aportion of the cingulum itself), which is not always the casefor the perissodactyl crest. Nannopithex fold refers to a pri-mate taxon and makes no reference to the cusp from which itoriginates, and postprotocone fold is so general that it could beconfused with the postprotocrista proper. Second, thecrests in question are somewhat different in notharctinesand perissodactyls. The crest in notharctines is confluentwith the distal cingulum, which is not always the casein perissodactyls. The use of the term endoprotocrista isdescriptive of the disposition of the crest and establishesit as distinct from the notharctine crest.

Modes of Molarization in Perissodactyl Upper CheekTeeth

The manner in which premolars become molarized differsamong perissodactyl lineages and premolar loci. A brief com-parison of the transition from a tritubercular tooth (Figs. 2aand 6a) to a quadritubercular tooth at different loci amongdifferent taxa is given below, specifically emphasizing thedevelopment of a second lingual cusp in P3, P4, and themolars.

P4 The pattern described in this study, where the hypoconearises from the endoprotocrista, appears to be common for theP4 in perissodactyl lineages (Fig. 6b). Specifically, the asso-ciation of the P4 hypocone with a distal crest from theprotocone is present in the European equoids Lophiotheriumpygmaeum (Savage et al. 1965: fig. 37) and Anchilophus sp.(Fig. 4e), the tapiroid Helaletes intermedius (Radinsky 1963:pl. 2, Fig. 4), and the early rhinocerotoidHyrachyus modestus(Radinsky 1967; Wood 1934: fig. 42). These intermediatespecies allow us to make inferences about the homologies ofcusps of later members of these genera. For instance,Lophiotherium cervulum has fully molariform P3 and P4,including a hypocone that is completely separate from theprotocone. The hypocone presumably is historically homolo-gous with the cusp attached to the endoprotocrista inL. pygmaeum. This pattern is also observed in notharctineprimates (Anemone et al. 2012).

One problem with making inferences regarding premolarmolarization is that we are attempting to describe a dynamicprocess of evolutionary transformation through the investiga-tion of static morphology in specimens. Hypotheses regardingthe mode of molarization cannot be tested without specimensexhibiting an intermediate morphology. For instance, Grangernoted that early equids acquire a second lingual cusp on P4 byadding it to the “postero-internal angle” (Granger 1908:261).Whether this cusp arises from the endoprotocrista or from thecingulum, or in some other manner, is not clear, because nointermediate stages preserve an endoprotocrista or a vestigialhypocone on the cingulum.

P3 In many perissodactyl taxa, molarization of P3 follows thesame pattern as described for P4, where the hypocone of P3 isderived from the distal end of the endoprotocrista (Fig. 6b; seealso Fig. 4e). Granger (1908; see also Van Valen 1982) dem-onstrated that the anterior lingual cusp of molarized P3s ofearly equids, such as Epihippus, is not the historical homologof the protocone in earlier taxa but actually a linguallydisplaced historical homolog of the ancestral paraconule(Fig. 6c). The distal lingual cusp of the P3 of Epihippus istherefore actually historically homologous to the single lin-gual cusp (the protocone) observed in earlier equids likeHyracotherium (Fig. 1a). Thus, the P3 protocone of

J Mammal Evol (2015) 22:259–269 265


Epihippus is serially homologous with the hypocone of theupper molars, a very different condition from those taxa thatderive a hypocone from the endoprotocrista.

Molars In the upper molars of mammals, the hypocone oftenappears to have developed from the posterolingual cingulum(Szalay 1969; Hunter and Jernvall 1995; Fig. 6d). In artiodac-tyls, the posterolingual cusp is typically an enlarged andlingually displaced metaconule (Hunter and Jernvall 1995).However, some early artiodactyl taxa, such as Homacodon,retain small molar hypocones derived from the cingulum inaddition to enlarged, lingually displaced metaconules(Theodor et al. 2007). While almost all perissodactyls havewell-developed molar hypocones that do not indicate theorigin of the cusp, in the early brontotheriid Palaeosyops

(Fig. 4f) the M3 hypocone is small and appears to be derivedfrom the cingulum.

Cusp Homologies, Phylogeny, and Modes of PremolarMolarization

Van Valen (1982, 1994) distinguished serial homology fromhistorical homology, defining serial homologs as repeatedstructures in the same organism, and historical homologs ascorresponding parts on different individuals that reflect shareddevelopmental information inherited from a common ances-tor. Thus, for Van Valen, historically homologous structurescan arise independently by independent activation of the same

Fig. 6 Comparison of ahypothetical tritubercular tooth(a) with diagrams ofquadritubercular teeth resultingfrom different modes oftransformation, including: ahypocone derived from theendoprotocrista (b); enlargementand lingual migration of theparaconule (c); and a hypoconederived from the cingulum (d).The endoprotocrista is dashed in(e), to indicate its loss after thehypocone connects to themetaloph. In (b) and (d), theconnection between themetaconule and hypocone mightbe homologous with thepostprotocrista, or it could be anew crest. Abbreviations: ect,ectoloph; endprc,endoprotocrista; hy, hypocone;mcl, metaconule; me, metacone;mes, mesostyle; met, metastyle;pa, paracone; pas, parastyle; pcl,paraconule; postprc,postprotocrista; pr, protocone;premclc, premetaconule crista;prepclc, preparaconule crista;preprc, preprotocrista

266 J Mammal Evol (2015) 22:259–269


developmental program. For instance, molar hypocones haveevolved multiple times in mammals (Hunter and Jernvall1995), but Van Valen would consider them to be historicallyhomologous if they developed in the same manner. Thisconcept of homology is very different from those that equatehomology with synapomorphy (Patterson 1982; see alsodiscussion in Nixon and Carpenter 2012) and I apply it herefor its utility and convenience in discussing tooth morphology,where strongly corresponding structures often evolved morethan once.

Based on the results described above, it is clear that themolariform morphology observed in molars and premolars ofmany perissodactyls is the result of different modes of trans-formation, and that cusps that are serially homologous are notalways historically homologous. For instance, in equids, theserially homologous hypocones of P3, P4, and M1 are likelynot historically homologous, representing the ancestralprotocone in P3, possibly a derivative of the endoprotocristain P4, and a derivative of the cingulum in the molars.Likewise, the serially homologous protocones of P3 and P4in equids are historically homologous to the ancestral P3paraconule and the ancestral P4 protocone, respectively.

Because so many perissodactyl taxa are known from teeth,dental characters are prominent in perissodactyl phylogenyand taxonomy, including many recent phylogenetic analyses(Froehlich 1999, 2002; Hooker 1989, 1994, 2005, 2010;Hooker and Dashzeveg 2003, 2004; Holbrook 2009;Holbrook and LaPergola 2011). Premolar molarization isoften important in delimiting fossil perissodactyl species(e.g., Granger 1908; Wood 1934; Kitts 1957; but seeRadinsky 1963, 1967), and characters related to premolarmolarization have figured prominently in many of the phylo-genetic analyses cited above. At the same time, it is recog-nized that premolar molarization has evolved multiple times inperissodactyl evolution (Butler 1952). As a result, identifyinglineage-specific patterns of premolar molarization can help toreduce the confounding effects of homoplasy when all pre-molar molarization is treated as the same. For instance, thedistinctive mode of molarization of P3 in equids has been usedto support monophyly of Equidae (Hooker 1994) or of someclade within the family (Froehlich 2002).

An important point here is that modes of molarization ofpremolars cannot serve as synapomorphies in and of them-selves, because mode of molarization is not an observablestate for a given taxon. What can be observed are states thatmight provide evidence for a given mode of molarization,such as a large and lingually positioned P3 paraconule, butthe mode of transformation is inferred a posteriori.Determining the transformation sequence that resulted inmolarized premolars of a particular taxon requires a phyloge-ny that includes intermediate morphologies. For instance, thepremolars of both Tapirus and the late Eocene to middleOligocene equid Mesohippus are completely molarized

(Figs. 1b and 4g), but inferring that the molarized P3 ofMesohippus is the result of the distinctive equid pattern ofP3 molarization (and that the molarized P3 of Tapirus is notthe result of this) is based on a close relationship betweenMesohippus and earlier equids with intermediate morphol-ogies (Granger 1908: fig. 4; Van Valen 1982: fig. 3). Thus,based on observations of the morphology alone, one could notscore the molarization of P3 of Mesohippus as both differentin nature from that of Tapirus and also sharing historicalhomologies with earlier equids that are not shared withTapirus (e.g., a lingually displaced P3 paraconule that inMesohippus is serially homologous with the protocone ofmore distal teeth). Therefore, there is no a priori basis forsaying that Mesohippus and other equids share a derivedpattern of P3 molarization.

The case of the endoprotocrista also illustrates the impor-tance of intermediate morphologies in establishing historicalhomology. In many cases, it is straightforward to comparemorphology across taxa, such as comparing the size andplacement of the P3 paraconule among various early perisso-dactyls. However, it is not always the case that historicalhomology can be established from these comparisons withminimal assumptions. In the present case, the distinctionbetween the postprotocrista and the endoprotocrista couldonly be made with the observation that both crests are presentin the premolars of some taxa.

Mesohippus provides another more striking illustration ofthe effect of a priori assumptions about homology. As notedabove, Mesohippus possesses completely molarized upperpremolars, including fully developed protoloph and metalophending in separate lingual cusps (Fig. 4g). As an equid, thecusp of P3 that is serially homologous with the molarprotocone would be interpreted as historically homologouswith the paraconule of earlier equids. However, the similarityof the upper premolars and upper molars extends to the pres-ence of small but distinct cuspules between the labial andlingual cusps, which are serial homologs of the paraconulesand metaconules of the molars, and presumably of P4. If weinterpret the mesial lingual cusp of P3 as the historical homo-log of the ancestral paraconule, then the cuspule labial to it onthe protoloph cannot also be the paraconule and must beinterpreted as a new cusp. Note that this interpretation ofhistorical homologies is a consequence of assumptions regard-ing the mode of molarization. In the absence of a prioriinformation on equid relationships, comparisons ofMesohippus with other perissodactyls would likely lead tothe interpretation that the serial homologs of the protoconeand paraconule on P3 are the historical homologs of the P3protocone and paraconule in other taxa, rather than a homologof the P3 paraconule and a neomorphic cuspule, respectively.

The molarization of P3 in equids illustrates a perplexingproblem in how one employs observation and inference incoding the data for phylogenetic analysis, and how this might

J Mammal Evol (2015) 22:259–269 267


affect determination of homology. In the scenario where thefewest assumptions are made, Mesohippus can be scoredbased entirely on what is observed, and therefore, based oncomparisons to other taxa and to other tooth loci, the mesiallingual cusp of P3 is interpreted as a protocone historicalhomolog rather than a paraconule historical homolog, andthe cuspule labial to the protocone is interpreted as aparaconule historical homolog and not a new cusp. Thus, theP3 paraconule ofMesohippus is scored as neither enlarged noras lingually placed. If the analysis supports the inclusion ofMesohippus within Equidae, can we now reinterpret the P3lingual cusps as paraconule and protocone, rather thanprotocone and hypocone? In other words, do we reinterpretthese homologies a posteriori to be in conflict with how weoriginally scored them a priori? The answer is not simple, butthe best course would seem to be the one that minimizesassumptions made a priori, allowing for new inferences ofhomology a posteriori.

Tooth Morphology, Diversity, and Evolution

The preceding discussion of the endoprotocrista, premolarmolarization and lophodonty, and homology demonstratesthe potential complexity of trends in perissodactyl evolutionthat otherwise might seem to be fairly straightforward.Variation in how perissodactyl lineages achieve premolarmolarization and lophodonty have been important forallowing us to recognize that these trends have occurredmultiple times in perissodactyl history. Other questions, how-ever, arise from this inference. If perissodactyls are so prone toconvergence in these traits, is it possible that these trends haveevolved more often than is evident from phylogeny and mor-phology?What is the nature of the causes of these trends? Thefact that the global perissodactyl fauna ultimately comes toconsist only of large-bodied, lophodont taxa with molariformpremolars suggests that these trends have been important forthe evolution of perissodactyl diversity. Addressing thesequestions and issues is beyond the scope of this paper, but Iinclude some discussion of some previous work on toothmorphology and mammalian diversity below.

Tooth morphology is important for understanding patternsand causes of mammalian diversification, particularly forunderstanding the diversity of mammalian herbivores(Hunter and Jernvall 1995; Jernvall et al. 1996). Jernvallet al. (1996) examined patterns of taxonomic diversity andmorphological diversity and disparity in Cenozoic ungulatesand observed a general decrease in morphological and taxo-nomic diversity but an increase in morphological disparityfrom the Eocene onward. They explained these patterns interms of the shift from the adaptive radiation of Eocene formsin warmer, more productive environments to specializationwithin a cooler, less productive environment, where there

were fewer niches to support intermediate occlusal mor-phologies. In particular, they documented an increase inlophodont molar crown types over time, which theyassociated with an increased emphasis on low-qualityfibrous plants as food sources. The pattern of perisso-dactyl diversity is consistent with this, although peris-sodactyls show a shift only toward lophodonty, ratherthan the artiodactyl pattern that shifts toward bothlophodont and bunodont taxa, with intermediate formsdisappearing (Jernvall et al. 1996).

Hunter and Jernvall (1995) documented a general increasein diversity of mammal taxa with hypocones during theCenozoic that they attributed to the hypocone (or its analog)serving as a “key innovation” that allowed hypocone-bearing taxa to exploit new herbivorous niches, espe-cially consuming fibrous plant parts. Molar hypoconesare present in all perissodactyls, although they might bereduced or absent in the M3 of early brontotheriids(Gunnell and Yarborough 2000; Fig. 4f).

Both Hunter and Jernvall (1995) and Jernvall et al.(1996) identify occlusal features that increase efficiencyof processing plants—namely lophodonty and increasingmolar occlusal area by adding a hypocone—as importantfor determining the radiation and extinction of Cenozoicmammalian herbivores. Both studies focused on molarmorphology, without reference to premolar morphology.Premolar molarization may be important for understand-ing the patterns of perissodactyl diversity during theCenozoic. All extant perissodactyls have premolars withtwo lingual cusps, but each lineage acquired this stateindependently and at different times. The Uintan equidEpihippus has essentially molariform premolars (Granger1908), whereas the earliest rhinocerotids and tapiridswith molarized premolars occur in the Miocene (Heissig1989; Albright 1998) with multiple rhinocerotid lineagesevolving this condition. This suggests that molarizationof premolars influenced post-Eocene perissodactyldiversification and extinction. As Hunter and Jernvall(1995) and Jernvall et al. (1996) argued for acquisitionof molar hypocones and lophodonty, respectively, premo-lar molarization may have been an important strategy foradapting to herbivorous niches involving low-qualityfibrous plant matter.

Acknowledgments The following individuals provided access to spec-imens examined for this study: J. Meng, J. Galkin, AMNH; E. Daeschler,N. Gilmore, ANSP; Z.-X. Luo, A. Tabrum, CM; J. Hooker, NHMUK; B.Engesser, A. Ziems, NMB; P. Holroyd, UCMP; P. Gingerich, G. Gunnell,UMMP; M. Carrano, R. Emry, R. Purdy, M. Brett-Surman, USNM; M.Fox, YPM. J. Hooker also provided casts of NHMUK M16336 and44115, as well as of GMH XIV/3106. K. Rose and two anonymousreviewers provided valuable and insightful comments on an earlier draftof this paper. This paper is the result of research supported by a grant fromthe National Science Foundation (DEB-0211976).

268 J Mammal Evol (2015) 22:259–269


References

Albright LB (1998) New genus of tapir (Mammalia: Tapiridae) from theArikareean (earliest Miocene) of the Texas Coastal Plain. J VertebrPaleontol 18:200—217

Anemone RL, Skinner MM, Dirks W (2012) Are there two distinct typesof hypocones in Eocene primates? The ‘pseudohypocone’ ofnotharctines revisited. Palaeontol Electr 15:26A

Butler PM (1952) Molarization of premolars in Perissodactyla. Proc ZoolSoc Lond 121:819—843

Depéret C (1917) Monographie de la faune de mammiferes fossiles duLudien inferieur d’Euzet-les-Bains (Gard). Ann Univ Lyon (NS) 1Sci Med 40:1–288

Froehlich DJ (1999) Phylogenetic systematics of basal perissodactyls. JVertebr Paleontol 19:140–159

Froehlich DJ (2002) Quo vadis eohippus? The systematics and taxonomy ofthe early Eocene equids (Perissodactyla). Zool J Linn Soc 134:141–256

Gingerich PD, Haskin RA (1981) Dentition of early Eocene Pelycodusjarrovi (Mammalia, Primates) and the generic attribution of speciesformerly referred to Pelycodus. Contrib Mus Paleontol Univ Mich25:327—337

Gingerich PD, Simons EL (1977) Systematics, phylogeny, and evolutionof early Eocene Adapidae (Mammalia, Primates) in North America.Contrib Mus Paleontol Univ Mich 24:245—279

Granger W (1908) A revision of the American Eocene horses. Bull AmMus Nat Hist 24:221—264

Gunnell GF, Yarborough VL (2000) Brontotheriidae (Perissodactyla)from the late early and middle Eocene (Bridgerian), Wasatch andBridger formations, southern Green River Basin, southwesternWyoming. J Vertebr Paleontol 20:349—368

Heissig K (1989) The Rhinocerotidae. In: Prothero DR, Schoch RM (eds)The Evolution of Perissodactyls. Oxford University Press, NewYork, pp 399—417

Holbrook LT (2009) Osteology of Lophiodon Cuvier, 1822 (Mammalia,Perissodactyla) and its phylogenetic implications. J VertebrPaleontol 29:212—230

Holbrook LT, LaPergola J (2011) A new genus of perissodactyl(Mammalia) from the Bridgerian of Wyoming with comments onbasal perissodactyl phylogeny. J Vertebr Paleontol 31:895—901

Hooker JJ (1989) Character polarities in early perissodactyls and theirsignificance for Hyracotherium and infraordinal relationships. In:Prothero DR, Schoch RM (eds) The Evolution of Perissodactyls.Oxford University Press, New York, pp 79—101

Hooker JJ (1994) The beginning of the equoid radiation. Zool J Linn Soc112:29–63

Hooker JJ (2005) Perissodactyla. In: Rose KD, Archibald JD (eds) TheRise of Placental Mammals. Johns Hopkins University Press,Baltimore, pp 199—214

Hooker JJ (2010) The mammal fauna of the early Eocene BlackheathFormation of AbbeyWood, London.Monogr Palaeontogr Soc Lond164:1—162

Hooker JJ, Dashzeveg D (2003) Evidence for direct mammalian faunalinterchange between Europe and Asia near the Paleocene-Eoceneboundary. Geol Soc Am Sp Pap 369:479—500

Hooker JJ, Dashzeveg D (2004) The origin of chalicotheres(Perissodactyla, Mammalia). Palaeontology 47:1363–1386

Hunter JP, Jernvall J (1995) The hypocone as a key innovation inmammalian evolution. Proc Natl Acad Sci USA 92:10718—10722

Jernvall J, Hunter JP, Fortelius M (1996) Molar tooth diversity, disparity,and ecology in Cenozoic ungulate radiations. Science 274:1489—1492

Kitts DB (1957) A revision of the genus Orohippus (Perissodactyla,Equidae). Am Mus Novitates 1864:1—40

Nixon KC, Carpenter JM (2012) On homology. Cladistics 28:160—169Patterson C (1982) Morphological characters and homology. In: Joysey

KA, Friday AE (eds) Problems of Phylogenetic Reconstruction.Academic Press, London, pp 21—74

Radinsky LB (1963) Origin and early evolution of North AmericanTapiroidea. Bull Peabody Mus Nat Hist Yale Univ 17:1—106

Radinsky LB (1967)Hyrachyus,Chasmotherium, and the early evolutionof helaletid tapiroids. Am Mus Novitates 2313:1—23

Savage DE, Russell DE, Louis P (1965) European Eocene Equidae. UnivCalif Publ Geol Sci 56:1–94

Szalay FS (1969) Mixodectidae, Microsyopidae, and the insectivore-primate transition. Bull Am Mus Nat Hist 140:193—330

Theodor JM, Erfurt J, Métais G (2007) The earliest artiodactyls.Diacodexeidae, Dichobunidae, Homacodontidae, Leptochoeridae,and Raoellidae. In: Prothero DR, Foss SE (eds) TheEvolution of Artiodactyls. Johns Hopkins University Press,Baltimore, pp 32—58

Van Valen LM (1966) Deltatheridia, a new order of mammals. Bull AmMus Nat Hist 132:1—126

Van Valen LM (1982) Homology and causes. J Morphol 173:305—312Van Valen LM (1994) Serial homology: the crests and cusps of mamma-

lian teeth Acta Palaeontol Pol 38:145—158Wood HE II (1934) Revision of the Hyrachyidae. Bull AmMus Nat Hist

67:181–295

J Mammal Evol (2015) 22:259–269 269


The Identity and Homology of the Postprotocrista and its Role in Molarization of Upper Premolars of Perissodactyla (Mammalia)

Documents