Inferring ‘weak spots’ in phylogenetic trees: …phylogenetic relationships by reevaluation of a recent version of that data set, published by Simıes et al. (2017), that represents

(the latter being non-monophyletic) outside the lsquotraditionalrsquomosasauridswhenhalisaurinesare reconstructed as the sister taxon to mosasaurines (Figs 2 3D and 3F) There isalso an apparent lack of resolution within the more advanced mosasaurines (the cladeformed by Globidens the species attributed to Prognathodon Mosasaurus EremiasaurusPlesiotylosaurus and Plotosaurus) which are nevertheless inferred monophyletic by allanalyses (Figs 1ndash7 see also above for the support of this grouping) The most strikingis the non-monophyly of Prognathodon (inferred also by other authors eg LeblancCaldwell amp Bardet 2012 Simotildees et al 2017) Some analyses unite certain taxa assignedto Prognathodon but only the monophyly of P solvayi and P currii is reconstructedconsistently (Figs 1ndash7 except for Fig 3A) though still poorly supported (DI lt 2 bootstraplt050 pp= 061)

Further the monophyly of Clidastes is supported only by lsquoweighted-unorderedrsquoparsimony analyses regardless of the value of K (Figs 3A 3C 3E) All other analysesincluding the Bayesian inference keepClidastes paraphyletic relative to othermosasaurines

Phylogenetic nomenclatureInferred phylogenetic relationships are further discussed within the context of mosasauroidsystematics and used as the primary basis for nomenclatural revision of the mainmosasauroid clades

The recommended phylogenetic definitions applied for the taxon names follow theInternational Code of Phylogenetic Nomenclature or PhyloCode hereafter ICPN (Cantinoamp De Queiroz 2010) They are summarized in Table 1 Likewise the taxon names areattributed to the authors that introduced them (following the ICPN Art 98 Note 98A2)and not according to the Principle of Coordination (ICZN 1999 Art 36) This approachis preferred due to its more transparent account of the original literature

Even though the majority of the preferred phylogenetic definitions is labeled as lsquonewrsquo(see Table 1) most of themmerely represent modified versions of the definitions proposedby other authors We attempted to provide only the necessary changes to maintain thetraditional meaning of the clade names and to maximize their stability given the inferredlsquoweak spotsrsquo in the mosasauroid phylogenetic tree

Mosasauroidea Camp 1923Preferred phylogenetic definitionThemost inclusive clade containingMosasaurus hoffmanniiMantell 1829 andAigialosaurusdalmaticus Kramberger 1892 but not Dolichosaurus longicollis Owen 1850 Adriosaurussuessi Seeley 1881 or Pontosaurus lesinensis Kornhuber 1873 This definition is branch-based

RemarksMosasauroidea traditionally includes mosasaurids and lsquoaigialosaursrsquo (eg Bell 1997 Bell ampPolcyn 2005 Conrad 2008) Proper delimitation of the extent of the name Mosasauroideahowever requires adequate knowledge of the early evolution of Mosasauria and reappraisalof the phylogenetic positions of potential non-mosasauroid mosasaurs (eg the speciesbelonging toAdriosaurusPontosaurusDolichosaurus) These taxa or their subset have been

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 1340

Table 1 Recommended phylogenetic definitions applied to mosasauroid taxon names

Clade name Internal specifier(s) External specifier(s) Type of phylogeneticdefinition

Authorship

Mosasauroidea Mosasaurus hoffmanniiAigialosaurus dalmaticus

Dolichosaurus longicollisAdriosaurus suessiPontosaurus lesinensis

Branch-based New

Aigialosauridae Aigialosaurus dalmaticusOpetiosaurus bucchichi

Dolichosaurus longicollisAdriosaurus suessiPontosaurus lesinensisMosasauridae = (Mosasaurushoffmannii Halisaurusplatyspondylus Tylosaurusproriger)

Branch-based New

Mosasauridae Mosasaurus hoffmanniiHalisaurus platyspondylusTylosaurus proriger

Node-based Madzia amp Conrad (in press)

Halisaurinae Halisaurus platyspondylus Mosasaurus hoffmanniiTylosaurus prorigerTethysaurus nopcsaiYaguarasaurus columbianus

Branch-based New

Natantia Mosasaurus hoffmanniiTylosaurus prorigerPlioplatecarpus marshii

Halisaurus platyspondylus Branch-based Conrad (2008)

Mosasaurinae Mosasaurus hoffmannii Tylosaurus prorigerPlioplatecarpus marshiiHalisaurus platyspondylusTethysaurus nopcsaiYaguarasaurus columbianus

Branch-based New

Mosasaurini Mosasaurus hoffmannii Globidens alabamaensis Branch-based NewGlobidensini Globidens alabamaensis Mosasaurus hoffmannii Branch-based NewRussellosaurina Russellosaurus coheni

Tylosaurus prorigerPlioplatecarpus marshii

Mosasaurus hoffmannii Node-based New

Tethysaurinae Tethysaurus nopcsaiPannoniasaurus inexpectatus

Halisaurus platyspondylusMosasaurus hoffmanniiTylosaurus prorigerPlioplatecarpus marshiiYaguarasaurus columbianus

Node-based New

Yaguarasaurinae Yaguarasaurus columbianusRussellosaurus coheniRomeosaurus fumanensis

Tethysaurus nopcsaiHalisaurus platyspondylusTylosaurus prorigerPlioplatecarpus marshiiMosasaurus hoffmannii

Node-based New

Plioplatecarpinae Plioplatecarpus marshii Mosasaurus hoffmanniiTylosaurus prorigerTethysaurus nopcsaiYaguarasaurus columbianus

Branch-based New

Tylosaurinae Tylosaurus proriger Plioplatecarpus marshiiMosasaurus hoffmannii

Branch-based Conrad (2008)

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 1440

hypothesized to be either more closely related to snakes (see eg Palci amp Caldwell 2007Caldwell amp Palci 2010 Palci amp Caldwell 2010) or to mosasaurids (eg Reeder et al 2015)Considering that (1) the lsquodolichosaursrsquo are traditionally regarded as non-mosasauroidsand (2) lsquoaigialosaursrsquo and mosasaurids are frequently inferred more closely related toeach other than either is to the lsquodolichosaursrsquo we propose a new definition that seems toadhere to the traditional use of Mosasauroidea (ie lsquoaigialosaursrsquo plus mosasaurids butnot lsquodolichosaursrsquo) and reflects the uncertainties surrounding the phylogenetic placementsof near-mosasaurids and early mosasaurids as inferred among others in the present study(see Figs 1ndash7)

Aigialosauridae Kramberger 1892Preferred phylogenetic definitionThe most inclusive clade containing Aigialosaurus dalmaticus Kramberger 1892 andOpetiosaurus bucchichi Kornhuber 1901 but not Dolichosaurus longicollis Owen 1850Adriosaurus suessi Seeley 1881 Pontosaurus lesinensis Kornhuber 1873 or the cladeoriginating with the most recent common ancestor of Halisaurus platyspondylus Marsh1869 Mosasaurus hoffmannii Mantell 1829 and Tylosaurus proriger (Cope 1869) Thisdefinition is branch-based

RemarksAigialosauridae has a long and problematic history The last thorough review of theinterrelationships of early Mosasauria ie those species associated with the evolutionarytransition to aquatic lifestyle was published by Dutchak (2005) who concluded thatlsquolsquoredescriptions of the key taxa (Aigialosaurus dalmaticus Opetiosaurus bucchichi andlsquothe Trieste aigialosaurrsquo) are essential to further investigations into re-testing the mostrecent hypothesesrsquorsquo (p 228) Although A dalmaticus and O bucchichi have since beenredescribed (Dutchak amp Caldwell 2006 Dutchak amp Caldwell 2009 respectively) and lsquotheTrieste aigialosaurrsquo was assessed and given the nameKomensaurus carrolli (Caldwell amp Palci2007) the status of Aigialosauridae did not change Indeed Dutchak amp Caldwell (2009)argued that O bucchichi should be assigned to Aigialosaurus (as A bucchichi) suggestingclose relationships of the two taxa Still their analysis does not necessarily support thisconclusion (see Dutchak amp Caldwell 2009 Fig 4)

While it is certainly possible that A dalmaticus and O bucchichi are more closely relatedto one another than either is to other mosasauroids such a result is currently not stronglysupported statistically The lsquofullrsquo parsimony analyses (with all lsquodolichosaursrsquo includedand A suessi selected as outgroup) reconstruct the taxa in a basal polytomy with othermosasauroid subclades (Fig 1) or as successively more closely related to mosasaurids withA dalmaticus being the more basal of the two (Fig 2) The Bayesian inference majorityof the weighted parsimony analyses (except for Figs 3D and 3F) and parsimony analysesusing different lsquodolichosaursrsquo as outgroups nevertheless reconstruct a clade formed byboth these species (Figs 3ndash5) though their position on the mosasauroid tree is unstable

Considering the problematic nature of mosasauroid origins we admit thatAigialosauridae might be of use in the future In this case however we strongly encourageusing a complex self-destructive phylogenetic definition to reflect the history of the name

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 1540

as well as its unstable contents (see ICPN Art 119) The self-destructive branch-baseddefinition that is proposed here keeps Aigialosauridae in use only if A dalmaticus andO bucchichi are more closely related to each other than either is to lsquodolichosaursrsquo orMosasauridae sensu Madzia amp Conrad (in press) Also it does not allow the use of the namein the cases when A dalmaticus and O bucchichi are reconstructed within Mosasauridae

Mosasauridae Gervais 1853Preferred phylogenetic definitionThe least inclusive clade containing Mosasaurus hoffmannii Mantell 1829 Halisaurusplatyspondylus Marsh 1869 and Tylosaurus proriger This definition is node-based

RemarksThe history of the nameMosasauridae its approximate synonyms and its application werediscussed byMadzia amp Conrad (in press) who also provided the phylogenetic definition forthe clade name as will be recognized by the ICPN

The Bayesian analysis and parsimony analyses using different lsquodolichosaursrsquo as theoutgroup maintain the monophyly of mosasaurines plioplatecarpines tylosaurinestethysaurines yaguarasaurines and the two halisaurine species The lsquounweighted-orderedrsquoparsimony analysis however reconstructs tethysaurines and yaguarasaurines outsideMosasauridae with Romeosaurus being inferred as the sister taxon to Komensaurus carrolli+mosasaurids outside tethysaurines+ a clade formed byYaguarasaurus andRussellosaurus(Fig 2) Thus it makes Yaguarasaurinae polyphyletic

The mutual relationships of particular mosasaurid clades are unsettled and highlydependent on the tree-search strategies used (Figs 1ndash7) Still even though the hypothesesof mosasaurid interrelationships are differing the definition proposed by Madzia ampConrad (in press) does not require modifications It covers all lsquotraditionalrsquo mosasaurid taxaincluding the plioplatecarpines Though not represented in the phylogenetic definitionPlioplatecarpus and its kin are kept within Mosasauridae under all inferred topologies

Halisaurinae Bardet et al 2005Preferred phylogenetic definitionThe most inclusive clade containing Halisaurus platyspondylus Marsh 1869 but notMosasaurus hoffmannii Mantell 1829 Tylosaurus proriger (Cope 1869) Tethysaurusnopcsai Bardet Suberbiola amp Jalil 2003 or Yaguarasaurus columbianus Paacuteramo 1994This definition is branch-based

RemarksBardet et al (2005) defined Halisaurinae as lsquolsquoMosasauridae more closely related toHalisaurus than to Mosasaurusrsquorsquo (p 464) Later Conrad (2008) used equivalent branch-based definition with type species as specifiers lsquolsquoAll taxa sharing a more recent commonancestor with Halisaurus platyspondylus than Mosasaurus hoffmanniirsquorsquo (p 127) Becausethe position of the species for which the name Halisaurinae was proposed is not very stablewithin Mosasauroidea (see the results of the present analysis and the Natantia paragraphbelow) we consider the proposed branch-based definition including additional externalspecifiers representing other inferred clades to be the most appropriate one

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 1640

Nevertheless the current data set is not fully suitable for testing the phylogeneticposition of Halisaurinae within Mosasauridae as the clade is represented by only two taxa(H platyspondylus and Eonatator sternbergii)

Natantia Owen 1851Preferred phylogenetic definitionThe most inclusive clade containing Mosasaurus hoffmannii Mantell 1829 Tylosaurusproriger (Cope 1869) and Plioplatecarpus marshii Dollo 1882 but not Halisaurusplatyspondylus Marsh 1869 This definition is branch-based

RemarksBell (1997) resurrected the name Natantia from the mid-nineteenth century (Owen 1851)It was used to unite Bellrsquos (1997) lsquoRussellosaurinaersquo (see the Russellosaurina paragraph) andMosasaurinae exclusive of theHalisaurus species and the lsquoaigialosaursrsquo Conrad (2008 128)proposed the following branch-based definition lsquolsquoAll taxa sharing a more recent commonancestor with Mosasaurus hoffmanni Tylosaurus proriger and Plioplatecarpus marshi thanwith Halisaurus platyspondylusrsquorsquo When applied on some recent phylogenetic hypothesesbased on the data set initially published by Bell amp Polcyn (2005) that infer halisaurines tobe nested within the smallest clade containingMosasaurus Tylosaurus and PlioplatecarpusNatantia self-destructs

Our analyses do not support the concept of Natantia either (Figs 1ndash7) In thelsquounweighted-orderedrsquo parsimony analysis (Fig 2) some weighted parsimony analyses(Figs 3D and 3F) parsimony analysis with Pontosaurus as the outgroup (Fig 4C) andBayesian analysis (Fig 5) halisaurines form the sister taxon to mosasaurines WhenAdriosaurus is used as outgroup and other lsquodolichosaursrsquo are excluded and under someweighted parsimony analyses halisaurines are more closely related to the clade formed bytethysaurines yaguarasaurines tylosaurines and plioplatecarpines than to mosasaurines(Figs 3Andash 3C 3E and 4A)

It is worth noting that Boas (1880) used the name Natantia for a subgroup of decapodcrustaceans Although Owenrsquos (1851) Natantia was published earlier the priority issueis problematic The ICZN (1999) does not govern the names above the family groupand Natantia approximately corresponding to the concept of Owen (1851) had not beenin use until Bell (1997) Similarly the use of Boas (1880) is outdated (WoRMS 2015)though it was of importance in the past (see for example the discussion in Felgenhauser ampAbele 1983)

We refrain from providing a lengthy discussion of the nomenclatural issue or a solutionto it but since the name Natantia Owen (1851) was published earlier we provisionally keepit as the name for the potential grouping as discussed above

Mosasaurinae Williston 1897Preferred phylogenetic definitionThe most inclusive clade containing Mosasaurus hoffmannii (Mantell 1829) butnot Tylosaurus proriger (Cope 1869) Plioplatecarpus marshii Dollo 1882 Halisaurus

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 1740

platyspondylus Marsh 1869 Tethysaurus nopcsai Bardet Suberbiola amp Jalil 2003 orYaguarasaurus columbianus Paacuteramo 1994 This definition is branch-based

RemarksMosasaurinae is traditionally considered to represent a species-rich clade withsubstantial morphological and ecological diversity (eg Bell 1997 Bell amp Polcyn 2005Bardet et al 2015)

The first published phylogenetic definition is the following lsquolsquoAll taxa sharing a morerecent common ancestor with Mosasaurus hoffmanni than with Tylosaurus proriger orPlioplatecarpus marshirsquorsquo (Conrad 2008 128) This branch-based definition keeps thetraditional contents of Mosasaurinae intact when applied to the majority of recent analysesWe added additional external specifiers Halisaurus platyspondylus Tethysaurus nopcsaiand Yaguarasaurus columbianus to reflect the traditional contents of Mosasaurinae andthe inferred overall instability in the mosasaurid interrelationships The monophyly ofmosasaurines however is inferred by all our analyses (Figs 1ndash7)

Mosasaurini Russell 1967Preferred phylogenetic definitionThe most inclusive clade containing Mosasaurus hoffmannii Mantell 1829 but notGlobidens alabamaensis Gilmore 1912 This definition is branch-based

RemarksBell (1997 322) abandoned Mosasaurini on the basis of the supposed paraphyly ofMosasaurus and lsquolsquoexpanded [Plotosaurini] to include basic taxa previously referredto Mosasaurusrsquorsquo Both taxon names Mosasaurini and Plotosaurini were introducedin the same publication (Russell 1967) However it seems that the former has gainedmore attention (eg Leblanc Caldwell amp Bardet 2012 Fanti Cau amp Negri 2014) LeblancCaldwell amp Bardet (2012 101) argued to replace Plotosaurini with Mosasaurini which theyused for lsquolsquothe group consisting of (Eremiasaurus (Mosasaurus + Plotosaurus))rsquorsquo Althoughthe close connection of these taxa is generally supported by recent phylogenetic studies(egGrigoriev 2013 Palci Caldwell amp Papazzoni 2013 Fanti Cau amp Negri 2014 Jimeacutenez-Huidobro amp Caldwell 2016) analyses using multiple tree-search strategies show conflictingresults (Simotildees et al 2017) The grouping is maintained in the lsquounweighted-unorderedrsquoparsimony analysis under one lsquoweighted-unorderedrsquo parsimony analysis (Fig 3E) andwhen only one of the lsquodolichosaurrsquo taxa is included (Fig 4) Still lsquounweighted-orderedrsquoparsimony other weighted parsimony analyses and the Bayesian inference fail to supportsuch topology

Globidensini Russell 1967Preferred phylogenetic definitionThe most inclusive clade containing Globidens alabamaensis (Gilmore 1912) but notMosasaurus hoffmannii Mantell 1829 This definition is branch-based

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 1840

RemarksBell (1997) used Russellrsquos (1967) Globidensini to unite Globidens Prognathodon andPlesiotylosaurus Although such close connection of these taxa is not necessarily supportedby current studies (eg Palci Caldwell amp Papazzoni 2013 Fanti Cau amp Negri 2014Jimeacutenez-Huidobro amp Caldwell 2016) there is indeed a tendency to keep them togetherunder the name Globidensini (eg Schulp et al 2008 Leblanc Caldwell amp Bardet 2012)Nevertheless forcing Prognathodon solvayi the type species of Prognathodon to be aglobidensin (by selecting it as an internal specifier) would be potentially ineffectiveconsidering the likely para- or even polyphyletic nature of the taxa attributed toPrognathodon

All our analyses fail to reconstruct Globidensini with more than only the two speciesof Globidens included (Figs 1ndash7) Nevertheless the clade name may still be useful fordiscussions related to mosasaurid ecology (due to the specialized dentition of Globidensand Carinodens its potential close relative (Schulp Jagt amp Fonken 2004))

Russellosaurina Polcyn amp Bell 2005Preferred phylogenetic definitionThe least inclusive clade containing Russellosaurus coheni Polcyn amp Bell 2005 Tylosaurusproriger (Cope 1869) and Plioplatecarpus marshii Dollo 1882 but not Mosasaurushoffmannii Mantell 1829 This definition is node-based

RemarksDue to its problematic history the name Russellosaurina is discussed here in detail Inhis PhD thesis Bell (1993) proposed a new name Russellosaurinae to link tylosaurinesand plioplatecarpines together and provided the following node-based definition lsquolsquoThemost recent common ancestor of Tylosaurus Ectenosaurus and Plioplatecarpus and allof its descendantsrsquorsquo (p 183) He noted that Russellosaurinae consists of lsquolsquoTylosaurus andPlioplatecarpinirsquorsquo (p viii) which matched his definition Bellrsquos PhD thesis was publishedfour years later (Bell 1997) Until that time lsquoRussellosaurinaersquo was in use in an informalsense as a node-based name for a clade consisting of lsquotylosaurinesrsquo and lsquoplioplatecarpinesrsquo(Caldwell 1996) Because the paper by Bell (1997) was originally intended to simply be thepublished version of his PhD thesis Bell (1997) again introduced lsquoRussellosaurinaersquo as anew taxon name However its extent seems to be different as the name was introduced lsquolsquoinanticipation of formally designating the taxon and describing a new taxon Russellosaurusfrom new Turonian material from Texasrsquorsquo (p 322) Although there was no explicitinformation about how closely related Russellosaurus was to lsquorussellosaurinesrsquo (sensu Bell1993) and in the lsquoSummaryrsquo paragraph of Bell (1997 324) lsquoRussellosaurinaersquo is again listedas consisting of lsquolsquoTylosaurus and Plioplatecarpinirsquorsquo only it is clear that Bell (1997) intendedto anchor lsquoRussellosaurinaersquo on the taxon Russellosaurus Until Polcyn amp Bell (2005) wherelsquoRussellosaurinaersquo was officially replaced with Russellosaurina authors used the name inthe traditional informal way and always as a node-based name for a clade containingTylosaurus and Plioplatecarpini (Christiansen amp Bonde 2002) or Plioplatecarpinae (Bardetet al 2005) the latter two names referring to the same content

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 1940

Polcyn amp Bell (2005) introduced the name Russellosaurina lsquolsquoto give identity to themonophyletic grouping of Tylosaurinae plus Plioplatecarpinae and closely related formsrsquorsquo(Polcyn amp Bell 2005 323)What the lsquolsquoclosely related formsrsquorsquo are is clear from the lsquoSystematicpalaeontologyrsquo paragraph (p 322) according to which the only non-mosasaurinemosasaurid taxa listed there as Russellosaurina are lsquolsquo[t]he subfamilies Tylosasaurinae[sic] and Plioplatecarpinae and their sister-clade containing the genera TethysaurusRussellosaurus and Yaguarasaurusrsquorsquo Unfortunately the composition of Russellosaurina isnot that transparent in other parts of that paper According to the abstract Russellosaurinalsquolsquoincludes Plioplatecarpinae Tylosaurinae their [most recent] common ancestor and all [ofits] descendantsrsquorsquo (p 321) and according to the phylogenetic definition Russellosaurinaconsists of lsquolsquo[a]ll mosasaurs more closely related to Tylosaurinae and Plioplatecarpinaethe genus Tethysaurus their common ancestor and all descendants than to Mosasaurinaersquorsquo(p 322) This definition is clearly branch-based with lsquolsquoTylosaurinae and Plioplatecarpinaethe genus Tethysaurus their common ancestor and all descendantsrsquorsquo being a node-basedclade and an internal specifier of the definition This wording is therefore inconsistent withall previously cited statements

When Polcyn amp Bell (2005) established the name they gave it the rank of lsquoparafamilyrsquo aterm introduced by Olshevsky (1991) for lsquoparaphyletic familyrsquo (the prefix lsquopara-rsquo indicateslsquoparaphylyrsquo) and not recognized by the ICZN Therefore it is of the same level as lsquofamilyrsquoHowever the suffix lsquo-inarsquo typically indicates a subtribe in zoological nomenclature sowhen assigning the name Russellosaurina a rank the taxon should be contained within atribe and a subfamily Here Russellosaurina is considered an unranked clade name withthe node-based definition provided above In our definition M hoffmannii is used as aqualifying clause (ICPN Art 119) The suggested compilation is preferred for variousreasons First it should lsquolsquo[supersede] previous references to lsquoRussellosaurinaersquorsquorsquo (Polcynamp Bell 2005 323) thus applying to the clade originating with the most recent commonancestor of Tylosaurinae Plioplatecarpinae and R coheni Further Russellosaurina hasalways been understood as a node-based name Although Conrad (2008) lsquolsquotentativelyrsquorsquofollowed the original branch-based definition he simultaneously noted that lsquolsquothe definitionPolcyn amp Bell (2005) intended for Russellosaurina is frustratingly ambiguousrsquorsquo (Conrad2008 129) Since R coheni was omitted from the specifiers the original definition violatedthe ICPN (Art 117)

According to the new definition Russellosaurina contains the species R coheniY columbianus T nopcsai the clade Plioplatecarpinae and the clade Tylosaurinae (asinferred eg in Bell amp Polcyn 2005 Dutchak amp Caldwell 2006 Cuthbertson et al 2007) Itmay also contain Halisaurinae as reconstructed in Caldwell amp Palci (2007) or self-destructunder the hypothesis from Bardet et al (2005) Russellosaurina may also contain onlyPlioplatecarpinae and Tylosaurinae if R coheni and Y columbianus are basal members ofPlioplatecarpinae as it was suggested by Polcyn amp Bell (2005 332) and inferred in Dutchakamp Caldwell (2009 Fig 5) Russellosaurina self-destructs if R coheni Y columbianus andT nopcsai form the sister taxon to the least inclusive clade including M hoffmannii andT proriger as reconstructed in Dutchak amp Caldwell (2009 Fig 4)

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 2040

The lsquounweighted-unorderedrsquo parsimony analysis (Fig 1) some weighted parsimonyanalyses (Figs 3Andash3C and 3E) parsimony analyses with Adriosaurus and Pontosaurus usedas outgroups (Figs 4A and 4C) and Bayesian analysis (Fig 5) support RussellosaurinaUnder all other topologies Russellosaurina self-destructs (Figs 2 3D 3F and 4B)

Tethysaurinae Makaacutedi Caldwell amp Oumlsi 2012Preferred phylogenetic definitionThe least inclusive clade containing Tethysaurus nopcsai Bardet Suberbiola amp Jalil 2003and Pannoniasaurus inexpectatus Makaacutedi Caldwell amp Oumlsi 2012 but not Halisaurusplatyspondylus Marsh 1869 Mosasaurus hoffmannii (Mantell 1829) Tylosaurus proriger(Cope 1869) Plioplatecarpus marshii Dollo 1882 or Yaguarasaurus columbianus Paacuteramo1994 This definition is node-based

RemarksMakaacutedi Caldwell amp Oumlsi (2012) introduced the name Tethysaurinae for lsquolsquo[t]he most recentcommon ancestor of Pannoniasaurus inexpectatus and Russellosaurus coheni Polcyn amp Bell2005 [ ] and all its descendantsrsquorsquo Following the results of their phylogenetic analysisthe clade Tethysaurinae was formed by P inexpectatus R coheni Tethysaurus nopcsai andYaguarasaurus columbianus However by omitting T nopcsai from the internal specifiersthe phylogenetic definition violates the ICPN (Art 117) Later Palci Caldwell amp Papazzoni(2013) introduced the name Yaguarasaurinae and defined it as lsquolsquo[t]he most recent commonancestor of Romeosaurus gen nov Russellosaurus and Yaguarasaurus and all of itsdescendantsrsquorsquo Tethysaurinae was kept only for Pannoniasaurus and Tethysaurus thatformed the sister clade to the Yaguarasaurinae (see below for comments on this name)

We follow the node-based concept of Tethysaurinae as delimited by Palci Caldwellamp Papazzoni (2013) but considering the unstable position of the two tethysaurines onthe mosasauroid tree (see Figs 1ndash7) we added five external specifiers to maintain thelsquotraditionalrsquo contents

All our analyses reconstruct monophyletic tethysaurines (Figs 1ndash7)

Yaguarasaurinae Palci Caldwell amp Papazzoni 2013Preferred phylogenetic definitionThe least inclusive clade containing Yaguarasaurus columbianus Paacuteramo 1994Russellosaurus coheni Polcyn amp Bell 2005 and Romeosaurus fumanensis Palci Caldwellamp Papazzoni 2013 but not Tethysaurus nopcsai Bardet Suberbiola amp Jalil 2003 HalisaurusplatyspondylusMarsh 1869Tylosaurus proriger (Cope 1869) Plioplatecarpus marshii Dollo1882 orMosasaurus hoffmannii Mantell 1829 This definition is node-based

RemarksAs noted above Yaguarasaurinae was introduced by Palci Caldwell amp Papazzoni (2013)who defined it as lsquolsquo[t]he most recent common ancestor of Romeosaurus gen novRussellosaurus andYaguarasaurus and all of its descendantsrsquorsquoWe follow such definition butconsidering the weak support for the connection of Yaguarasaurinae and Tethysaurinae(Figs 1 2 5 and 7) we added five external specifiers to prevent the name to cover anunintended clade

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 2140

The Bayesian analysis and majority of the parsimony analyses support the monophylyof the yaguarasaurines as delimited by Palci Caldwell amp Papazzoni (2013) Only under thetopology resulting from the lsquounweighted-orderedrsquo parsimony analysis and two lsquoweighted-orderedrsquo parsimony analyses Yaguarasaurinae self-destructs (Figs 2 3D and 3F)

Plioplatecarpinae Dollo 1884Preferred phylogenetic definitionThemost inclusive clade containingPlioplatecarpus marshii Dollo 1882 but notMosasaurushoffmannii Mantell 1829 Tylosaurus proriger (Cope 1869) Tethysaurus nopcsai BardetSuberbiola amp Jalil 2003 or Yaguarasaurus columbianus Paacuteramo 1994 This definition isbranch-based

RemarksConrad (2008 130) defined Plioplatecarpinae as lsquolsquo[a]ll taxa sharing a more recent commonancestor with Plioplatecarpus marshi[i ] than with Tylosaurus proriger or Mosasaurushoffmanniirsquorsquo Such definition matches the published hypotheses Plioplatecarpinae assister taxon to Tylosaurinae or to Mosasaurinae (eg Bell 1997 Bardet et al 2005Bell amp Polcyn 2005 Leblanc Caldwell amp Bardet 2012 Palci Caldwell amp Papazzoni 2013Jimeacutenez-Huidobro amp Caldwell 2016) but does not reflect the possible close connectionof plioplatecarpines with yaguarasaurines (as suggested by Polcyn amp Bell [2005 332] andthen inferred together with Tethysaurus by Dutchak amp Caldwell [2009 Fig 5]) Thuswe included two additional external specifiers Tethysaurus nopcsai and Yaguarasauruscolumbianus that assure the adherence of the name Plioplatecarpinae to the traditionalcontents under alternative hypotheses

The topologies inferred through our parsimony and Bayesian analyses support themonophyly of the traditional plioplatecarpines as delimited by Konishi amp Caldwell (2011)(Figs 1ndash7)

Tylosaurinae Williston 1897Preferred phylogenetic definitionThe most inclusive clade containing Tylosaurus proriger (Cope 1869) but notPlioplatecarpus marshii Dollo 1882 or Mosasaurus hoffmannii Mantell 1829 Thisdefinition is branch-based

RemarksThe tylosaurine interrelationships have been intensively studied during the past decade(eg Bullard 2006 Martin amp Fernaacutendez 2007 Caldwell et al 2008 Bullard amp Caldwell2010 Jimeacutenez-Huidobro amp Caldwell 2016 Otero et al 2017) resulting among othersin numerous changes in binomial nomenclature The monophyly of Tylosaurinaenevertheless has not been put into question

Conrad (2008 130) defined Tylosaurinae as lsquolsquo[a]ll taxa sharing a more recent commonancestor with Tylosaurus proriger than with Mosasaurus hoffmannii or Plioplatecarpusmarshi[i ]rsquorsquo This definition adheres to the traditional contents of Tylosaurinae underall current topologies including these inferred by our parsimony and Bayesian analyses(Figs 1ndash7)

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 2240

Figure 8 Frequency of sampled ancestors among the alternative topologies produced by the Bayesiananalysis using the FBDSAmodelGrey area indicates the 95 confidence interval of sampled trees

DISCUSSIONInferences using the Fossilized BirthndashDeath model with sampledancestors (FBDSA)The FBDSA model that discriminates between cladogenetic and anagenetic patterns inmacroevolution (Gavryushkina et al 2014 Gavryushkina et al 2017) inferred severalancestral-descendent relationships a subset of which is shown in the MCCT (see Fig 5)Nevertheless all of them were weakly supported and therefore are not discussed furtherInstead of focusing on the consensus topologies (like the MCCT) a more accurate way forestimating the frequency of ancestor-descendant relationships obtained by the Bayesiananalysis is by considering all the post-burnin topologies inferred (see Cau 2017) Inthe 95 of the sampled trees using the data set of Simotildees et al (2017) the number ofsampled ancestors inferred ranges between 0 and 5 (Fig 8) which suggests that up to11 of the included mosasauroid taxa are potential direct ancestors of one or more othermosasauroids included Nevertheless these values probably overestimate the frequency ofsampled ancestors It should be remarked that in these analyses the character list a prioriexcludes invariant characters (in particular the autapomorphies of terminal units) as iscommon practice in parsimony analyses sampling exclusively potential synapomorphiesThis methodological bias thus may inflate the frequency of the sampled ancestors sinceit does not discriminate between actual ancestors along anagenetic lineages (that have a

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 2340

null terminal branch length) from spurious zero-length terminal branches due to omissionof autapomorphies In conclusion taking into account the methodological bias due toomission of invariant characters from the morphological features included this analysissuggests that no more than one-tenth of the inferred relationships among the actualphylogenetic tree of Mosasauroidea could be tentatively interpreted as anagenetic (directancestor-descendant) patterns

Potential issues resulting from application of the Implied WeightingfunctionAs shown by Simotildees et al (2017) and our parsimony and Bayesian analyses the structureof the mosasauroid phylogenetic tree is highly dependent on the applied tree-searchstrategies Use of some phylogenetic methods may currently lead to prefer insufficientlysupported phylogenetic hypotheses For example Simotildees et al (2017) performed a singletest of parsimony analysis using the Implied Weighting (IW) function keeping thedefault value for the K parameter (K = 3) Compared to their unweighted parsimonyanalyses which show polytomies near the base of Mosasauroidea and within Mosasaurinae(Simotildees et al 2017 Figs 1A 1B) the topology inferred from the parsimony analysis withIW function was fully resolved (Simotildees et al 2017 Fig 1C) and represented the onlyunambiguous support for a single origin of the hydropedal and hydropelvic conditionsthat are related to the transition from semi- to a fully aquatic lifestyle (with a reversalwithin Tethysaurinae to plesiopelvic condition) However the evolutionary meaning ofthe K parameter is currently hotly debated (eg OrsquoReilly et al 2016 Congreve amp Lamsdell2016 Goloboff Torres amp Arias 2017) and a recent investigation of the effects of impliedweighting on modeled phylogenetic data revealed particularly poor abilities of the methodto resolve data sets with large amounts of conflicts or polytomies (Congreve amp Lamsdell2016) Goloboff Torres amp Arias (2017) criticized some aspect of the studies by OrsquoReilly etal (2016) and Congreve amp Lamsdell (2016) but repeated the necessity for the investigationof proper values of K relative to the numbers of analyzed taxa (Goloboff 1993 Goloboff1995) and evaluation of more than a single concavity parameter (Goloboff et al 2008)

It is far beyond the scope of the present paper to contribute to the debate but given thatconcerns regarding the lsquoproperrsquo use of weighted parsimony still exist we suggest that theresults of parsimony analyses with the IW function are generally treated lsquoconservativelyrsquoThat is rather than preferring a single inferred topology with a particular value of Kthat seems to fit best for the analyzed data trees produced by different runs shouldbe compared in order to spot and prioritize the groupings that are consistently beingreconstructed For example all weighted parsimony analyses reconstruct monophyleticHalisaurinae (Halisaurus + Eonatator) but the position of this clade on the mosasauroidtree is unstable They are either the sister taxon to the clade formed by tethysaurinesyaguarasaurines tylosaurines and plioplatecarpines (Figs 3Andash3C and 3E) or the sistertaxon to mosasaurines (Figs 3D 3F) We suggest that regardless of which of the twohypotheses is inferred following the use of the best-fitting value(s) of K the position ofhalisaurines should be regarded as unstable and ideally compared to the results producedby other methods of phylogenetic inference Therefore in the case of the present data set

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 2440

the position of halisaurines should be treated as ambiguous The only method that infers astrong support for either hypothesis is the Bayesian analysis that reconstructs halisaurinesas the sister taxon to mosasaurines (pp= 096)

Data samplingFollowing the results of the phylogenetic analyses using multiple tree-search strategies wediscuss the factors in the data sampling that might influence the differing hypotheses ofmosasauroid phylogenetic relationships and their statistical support and suggest furtherchanges to the explored data set that might improve the resolution of the mosasauroidphylogenetic relationships

Outgroup selectionIn the initial version of the data set introduced by Bell (1993) and Bell (1997) the outgroupwas constructed following the algorithm described by Maddison Donoghue amp Maddison(1984) The final outgroup OTU was based on the characters present in eight modernsquamates (Aspidoscelis sexlineata Crotaphytus collaris Dipsosaurus dorsalis Gekko geckoGerrhonotus liocephalus Plestiodon laticeps Shinisaurus crocodilurus andVaranus niloticus)and two extinct squamates (Estesia mongoliensis and Gilmoreteius chulsanensis) Suchlsquocompositersquo operational taxonomic unit was used by most later authors (eg Bell amp Polcyn2005 Caldwell amp Palci 2007 Leblanc Caldwell amp Bardet 2012) More recently howeversome studies preferred to use only the character states present in Varanus as the outgroup(eg Palci Caldwell amp Papazzoni 2013 Jimeacutenez-Huidobro amp Caldwell 2016) lsquolsquobecauseboth taxa [ie Mosasauroidea and Varanus] are large-bodied anguimorphs that share anumber of symplesiomorphic featuresrsquorsquo (Palci Caldwell amp Papazzoni 2013 608)

The outgroup sampling is known to have a great effect on the structure of phylogenetictrees (eg Graham Olmstead amp Barrett 2002 Spaulding OrsquoLeary amp Gatesy 2009Kirchberger et al 2014Wilberg 2015) Given the alternative placements of Mosasauroideaamong different phylogenies published (eg Conrad 2008 Gauthier et al 2012 Reederet al 2015) it is not universally agreed which squamates may represent the closest sistergroup of mosasauroids Therefore outgroup selection among extant squamates may bebiased by preference among the alternative placement of Mosasauroidea

The problems with the use of the lsquocompositersquo OTU then was already commented on byPalci Caldwell amp Papazzoni (2013 608) who noted that the lsquolsquooutgroup is problematic forseveral reasons (1) it does not reflect the character state composition of a real organism(2) it can produce paradoxical combinations of character states where a feature codedas absent in one character is further defined in a second character [] and (3) lack ofrepeatability of the process that produced such codingsrsquorsquo noting that Bell (1997) lsquolsquowas notvery explicit on how he obtained the character states for his outgrouprsquorsquo The third point(lack of repeatability of the process) however does not seem to be entirely fair Eventhough Palci Caldwell amp Papazzoni (2013) are certainly correct that Bell (1997) was notparticularly specific regarding the scores of his lsquocompositersquo OTU that paper was supposedbe the published version of his PhD thesis (Bell 1993) which is explicitly referred to by Bell(1997 294) and includes information on where the scores come from (Bell 1993 9ndash16251 265ndash268)

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 2540

To solve the issues with outgroup selection Simotildees et al (2017) expanded the data setby adding three lsquodolichosaur-gradersquo taxa Adriosaurus suessi Seeley 1881 Dolichosauruslongicollis Owen 1850 and Pontosaurus kornhuberi Caldwell 2006 and designed A suessias the basalmost outgroup Even though A suessi constitutes a much better outgroup thanthe lsquocompositersquo OTU and Varanus because its age and morphology more closely reflectthose of the last common ancestor of all mosasauroids such approach forcesDolichosaurusand Pontosaurus to be inferred more closely to mosasaurids than to Adriosaurus Thisoutgroup setting may thus lead to the construction of an artificial lsquodolichosaur gradersquo asthe basalmost mosasauroid condition (ie due to the outgroup setting in TNT used bySimotildees et al 2017 lsquodolichosaursrsquo are constrained to form a paraphyletic series leading toMosasauroidea) which may lead to spurious relationships among the ingroup taxa merelybased on squamate symplesiomorphies that are absent among the lsquodolichosaurrsquo taxaAs Simotildees et al (2017) noted some studies reconstruct these lsquodolichosaursrsquo to representsnake-branch pythonomorphs (see eg Palci amp Caldwell 2007 Caldwell amp Palci 2010Palci amp Caldwell 2010) Thus all these three OTUs may be lsquoequallyrsquo distantly related toMosasauridae It is noteworthy that the latter hypothesis is supported by the Bayesiananalysis using the FBDSA model which reconstructed all lsquodolichosaurrsquo taxa as forming aclade excluding all other OTUs

To avoid any bias due to a priori assumptions on character state transformation (becauseof the alternative extant squamate outgroup used and potentially incorrect outgroupbasalingroup designation) we suggest to perform analyses using different outgroup selection orto consider the use of a lsquoremote outgrouprsquo Perhaps the well preserved Early Cretaceous(Aptian) squamate Huehuecuetzpalli mixtecus Reynoso 1998 might serve as the root in aseparate analysis That taxon is universally recognized as more basal than any alternativemosasauroid outgroup used previously (Conrad 2008 Gauthier et al 2012) and mayrepresent the ancestral squamate morphology regardless of the preferred closest relativesof mosasauroids However see also Graham Olmstead amp Barrett (2002) and Kirchberger etal (2014) for independent tests regarding the effects of the use of phylogenetically distantoutgroups in molecular studies

Taxon samplingAs discussed above the outgroup selection has a substantial impact on the structure of theinferred tree topology including the statistical support of the basal branching near the rootof Mosasauroidea Still the resolution of the rootward mosasauroids might not necessarilyimprovewithout an increased number of earlymosasaurids andnear-mosasaurids analyzedThe most recent version of the data set was expanded with the addition of AdriosaurussuessiDolichosaurus longicollis and Pontosaurus kornhuberi and separation ofOpetiosaurusbucchichi from the AigialosaurusOTU (even if it is assigned to Aigialosaurus as A bucchichiDutchak amp Caldwell 2009 Simotildees et al 2017) Still it could also benefit for instance fromaddition of Acteosaurus tommasinii (Palci amp Caldwell 2010) Adriosaurus microbrachis(Palci amp Caldwell 2007) Adriosaurus skrbinensis (Caldwell amp Palci 2010) Aphanizocnemuslibanensis (Dal Sasso amp Pinna 1997) Carsosaurus marchesettii (eg Caldwell Carroll ampKaiser 1995 Caldwell amp Palci 2007) Coniasaurus crassidens (Caldwell amp Cooper 1999)

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 2640

Eidolosaurus trauthi (Nopcsa 1923) and Pontosaurus lesinensis (Pierce amp Caldwell 2004)The fact that some or most of these taxa can be more closely related to snakes than tomosasaurids (see eg Palci amp Caldwell 2007 Caldwell amp Palci 2010 Palci amp Caldwell2010) is not a problem as their morphology approximates to that of the mosasauridancestor and therefore supplements the knowledge of early pythonomorph evolution

The data set of Simotildees et al (2017) contains members of all well-recognizedmosasauroid subclades the taxa traditionally containedwithinHalisaurinaeMosasaurinaePlioplatecarpinae and Tylosaurinae It also contains all tethysaurines and yaguarasaurines(except Romeosaurus sorbinii Palci Caldwell amp Papazzoni 2013) as these two clades wereinferred in studies using recent versions of the data set (Makaacutedi Caldwell amp Oumlsi 2012Palci Caldwell amp Papazzoni 2013 respectively) Still some of the clades are substantiallyunderrepresented even though detailed descriptions of their members have been publishedand some of those taxa have been scored for characters in older versions of the samedata set For example the current version of the data set includes only two halisaurineOTUs (Halisaurus platyspondylus and Eonatator sternbergii with the latter being labeled aslsquoHalisaurus sternbergirsquo) even though detailed studies have also been published for examplefor Halisaurus arambourgi (Bardet et al 2005 Polcyn et al 2012) or Phosphorosaurusortliebi (Lingham-Soliar 1996 Holmes amp Sues 2000 Bardet et al 2005) Likewise the dataset could be supplemented by recently described Eonatator coellensis (Paacuteramo-Fonseca2013) and Phosphorosaurus ponpetelegans (Konishi et al 2016) Such sampling couldtest some of the implied relationships (the connection of E coellensis to E sternbergiiH arambrourgi toH platyspondylus P ponpetelegans to P ortliebi) A phylogenetic analysisof Halisaurinae was recently published by Konishi et al (2016) The analysis did notreconstruct monophyletic Halisaurus nor Eonatator but inferred sister-taxon relationshipsbetween P ortliebi and P ponpetelegans a taxon described by these authors However theanalysis was based on only 21 cranial characters and rooted on Platecarpus tympaniticus aderived plioplatecarpine that might not serve best as the outgroup for such analysis due toits placement and age Considering the unsettled relationships within Halisaurinae and thediffering position of the clade within Mosasauridae an expansion of the data set by usingmore halisaurines (and modification of the characters to better reflect their morphology)might result in improving the resolution of the mosasauroid tree topology

New reappraisals of certain tylosaurine species have also been published recentlyFor example Hainosaurus pembinensis and H bernardi the latter being the type speciesof Hainosaurus have been assigned to Tylosaurus (Bullard amp Caldwell 2010 Jimeacutenez-Huidobro amp Caldwell 2016 respectively) and Tylosaurus kansasensis was proposed to be ajuvenile of T nepaeolicus and thus removed from the data set (Jimeacutenez-Huidobro Simotildeesamp Caldwell 2016) However T pembinensis is not included in the recent version of the dataset which does not enable to further test the newly proposed hypotheses Interestinglythe ordered-unweighted parsimony analysis and the Bayesian analysis do not supportthe monophyly of Tylosaurus (represented by T proriger T bernardi and T nepaeolicus)exclusive of Taniwhasaurus (Figs 2 and 5) When only one lsquodolichosaurrsquo is in the data setand used as the outgroup regardless of which one it is Tylosaurus is monophyletic (Fig 4)The resolution might improve with a more appropriate outgroup selection and addition

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 2740

of T pembinensis and possibly lsquoHainosaurusrsquo neumilleri (Martin 2007) AdditionallyTylosaurus lsquosaskatchewanensisrsquo (Bullard 2006) and lsquoHainosaurusrsquo lsquokenbrownirsquo (Thompson2005 Thompson 2011) can also be considered pending their formal descriptions

The understanding of the plioplatecarpines in turn may improve by separation of thePlioplatecarpus OTU into several terminal units Such sampling could test the monophylyof Plioplatecarpus (a taxon consisting of a few species including P marshii P houzeauiP primaevus and the recently describedP peckensisCuthbertson amp Holmes 2015) estimatethe support for the tree topology obtained by Konishi amp Caldwell (2011) and Cuthbertsonamp Holmes (2015) test the connection of lsquoLatoplatecarpusrsquo nichollsae and L willistoni orprovide additional support for the separation of Plesioplatecarpus planifrons (labeled aslsquoPlatecarpus planifronsrsquo in the data set of Simotildees et al 2017) from Platecarpus tympaniticus(Konishi amp Caldwell 2011)

Mosasaurines are problematic as is apparent from differing and often poorly resolvedtree topologies The inference of the structure of the mosasaurine phylogenetic tree appearsto be difficult especially due to the unstable positions of the taxa attributed to Prognathodon(eg Leblanc Caldwell amp Bardet 2012 Simotildees et al 2017 our study) Neverthelessnumerous derived mosasaurines are currently under revision as is apparent from Street ampCaldwell (2017) that provided detailed reappraisal of Mosasaurus hoffmannii preliminarydiscussion of some other taxa traditionally assigned to Mosasaurus and reported on anongoing research Together with reconsideration of some species traditionally attributedto Prognathodon the resolution of the mosasaurines might benefit from addition of somepresumably rootward mosasaurine taxa that have not been included in previous lsquocompletersquoversions of the Bellrsquos data set (ie when the aim was to assess the interrelationships withinall major clades of mosasauroids) These include for example Kourisodon puntledgensis(Nicholls amp Meckert 2002) This taxon which has previously been used as an outgroup insome analyses (Konishi amp Caldwell 2011Cuthbertson amp Holmes 2015) originates from theupper Santonian of BritishColumbia Canada and is one of the oldest knownmosasaurinesIts inclusion might have an impact on the resolution of Mosasaurinae

Character samplingWe suggest that character statements are redefined from those used in recent versionsof Bellrsquos (1997) data set following the recommendations in Sereno (2007) and Brazeau(2011) In particular compound characters are suggested to be atomized ie neomorphicand transformational features should be considered as distinct characters and not asalternative states of a single character Therefore when not resulting in loss of informationcharacters are suggested to be defined as binary When multistate character statementsare included and the states form unambiguous morphoclines that describe a nested set ofalternative states (eg marginal tooth numbers vertebral numbers phalangeal formulas)the corresponding character statements should be set as ordered to avoid a priori exclusionof potential synapomorphies represented by the subset of states representing a derivedcondition (egWilkinson 1992 Sereno 2007 Brazeau 2011) Such states however shouldbe formulated to avoidmarked polymorphism For example the current version of the dataset (Simotildees et al 2017) includes a six-state character dealing with the dentary tooth count

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 2840

lsquolsquo(53) Dentary tooth number 20ndash24 (0) 17ndash19 (1) 15ndash16 (2) 14 (3) 13 (4) 12 (5)rsquorsquo Yetsuch defined states insufficiently reflect differences in taxa where the dentary tooth countis one of the few distinguishing characters Furthermore once set as ordered to reflect thehomology among nested state-transitions the character defined this way leads to inflatingthe phylogenetic importance of a feature that may be merely size-related and individuallyvariable among the same taxon For instance Mosasaurus hoffmannii is often reported ashaving 14 dentary teeth (eg Street amp Caldwell 2017) However some specimens have 15dentary teeth (eg CAMSM F22228 IRSNB R 0303 D Madzia pers obs 2017 MulderCornelissen amp Verding 2004) or only 13 (NHMM 009002 Everhart et al 2016) Thus Mhoffmannii can be scored for states 2 3 and 4 At the same time Mosasaurus lemonnieriwhich is currently considered to be distinct from M hoffmannii (Street amp Caldwell 2017D Madzia 2017 unpublished data) has always 16 dentary teeth Still it would be coveredunder the same state (2)

This example demonstrates that character definitions and among-state transition settingsmay significantly influence relationships and must be discussed prior to phylogeneticanalyses

lsquoData handlingrsquoAs we have expressed above we consider the current versions of the Bellrsquos (1997) data setto be insufficient for accurate inferences of mosasauroid phylogenetic relationships Wesuggest to (1) reconsider the outgroup selection (2) increase the number of analyzed taxaand named some of those that we think might improve the resolution of the mosasauroidphylogenetic tree and (3) revise the morphological characters and their states Naturallyit is essential to note that the steps should be undertaken after careful considerationsand simultaneously Specifically increasing the number of analyzed taxa could have anentirely opposite effect and cause more instability if the additions do not sufficiently reflectthe differing morphologies of the proposed OTUs and their character evolution Alsowe suggest to consider even those taxa that might be regarded as too incomplete to beincluded in the data matrix (see eg Wiens 2003a Wiens 2003b Wiens amp Morrill 2011)The relevance of all additions might be tested for example following the principle ofsafe taxonomic reduction (Wilkinson 1995) using TAXEQ3 (Wilkinson 2001) or throughlsquoconcatabominationsrsquo (Siu-Ting et al 2015) However it has also been argued that lsquolsquothereis no justificationmdasheither a priori or a posteriorimdashto definitively exclude unstable taxafrom the data matrix as this involves the deletion of phylogenetic information that can berelevant (or even critical) for understanding the relationships of the entire grouprsquorsquo (Polamp Escapa 2009 13) Therefore Pol amp Escapa (2009) offered to use a TNT script IterPCRthat provides a list of characters related to the instability of each unstable taxon This scripthas already been implemented in TNT (Goloboff amp Szumik 2015)

CONCLUSIONSThroughout the last two decades the phylogenetic relationships within Mosasauroideahave been inferred using modified versions of a single data set originally published byBell (1997) In order to estimate the robustness in our understanding of mosasauroid

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 2940

phylogenetic relationships we used a recent version of that data set (published by Simotildeeset al 2017) and focused on the effects of tree-search strategy selection

Parsimony and Bayesian analyses of the same data set showed considerable differencesin tree topologies near the base of Mosasauroidea suggesting that an increased number ofthe basal taxa and morphological characters phylogenetically informative for large-scalerelationships need to be taken into account Furthermore the different topologies obtainedby the alternative tree-search strategies suggest that one particular phylogenetic hypothesismay be significantly biased by the phylogenetic method used as suggested by Simotildees et al(2017) We thus suggest to perform different analyses of the same data using alternativetree-search strategies and tree models and to consider as supported only those hypothesesshared consistently by the majority of analyses Following the results of the present studythe monophyly of the traditional mosasauroid groups (Halisaurinae TethysaurinaePlioplatecarpinae Tylosaurinae Mosasaurinae and possibly also Yaguarasaurinae) canbe currently considered supported Yet their mutual relationships as well as the relationswithin these groups are still largely unsettled

From the nomenclatural perspective we see little or no support for the use of somebinomial combinations Specifically our analyses often failed to reconstruct monophylyfor the mosasaurine taxon Prognathodon Although the Bayesian analysis infers somesupport albeit extremely poor for a clade formed by all taxa attributed to Prognathodon(and including Eremiasaurus) lsquoPrognathodonrsquo requires complex reassessment and sometaxa will have to be removed from it (see also eg Leblanc Caldwell amp Bardet 2012 Simotildeeset al 2017)

We recommend that future implementations of the mosasauroid data set will discussthe combined effects of taxon sampling character construction and tree-search strategysettings For instance in phylogenetic analysis using parsimony and where all charactersare set as having equal weight the splitting of the multistate characters into distinct binarycharacters does not bias the reconstruction of the state transitions On the contraryin phylogenetic analysis using parsimony as tree-search strategy and with the ImpliedWeighting function multistate or compound characters once subdivided into binarycharacters are analyzed with different weighting settings Furthermore in Bayesianphylogenetic analyses where rate variation across morphological characters are modeledusing the gamma parameter different state transitions of the samemorphocline may evolveat different rates

We conclude that until the data set is significantly improved by a more appropriatetaxon sampling and revision of characters the currently inferred phylogenetic relationshipsof mosasauroids should be seen as tentative and subject to change

Institutional abbreviations

CAMSM Sedgwick Museum of Earth Sciences University of Cambridge CambridgeUK

IRSNB Royal Belgian Institute of Natural Sciences Brussels BelgiumNHMM Natuurhistorisch Museum Maastricht Maastricht the Netherlands

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 3040

ACKNOWLEDGEMENTSDM would like to thank Annelise Folie Alain Dregraveze and Ceacutecilia Cousin (all RoyalBelgian Institute of Natural Sciences Belgium) and John W M Jagt (NatuurhistorischMuseum Maastricht the Netherlands) for access to specimens in their care Hallie PStreet (University of Alberta Canada) Valentin Fischer (University of Liegravege Belgium)and three anonymous reviewers provided thorough reviews and valuable suggestions thatsubstantially improved the manuscript The program TNT is made available with thesponsorship of the Willi Hennig Society

ADDITIONAL INFORMATION AND DECLARATIONS

FundingDaniel Madzia is supported by the National Science Centre (Poland) grant No201519NST1001628 The funders had no role in study design data collection andanalysis decision to publish or preparation of the manuscript

Grant DisclosuresThe following grant information was disclosed by the authorsNational Science Centre (Poland) 201519NST1001628

Competing InterestsThe authors declare there are no competing interests

Author Contributionsbull Daniel Madzia and Andrea Cau conceived and designed the experiments performed theexperiments analyzed the data contributed reagentsmaterialsanalysis tools wrote thepaper prepared figures andor tables reviewed drafts of the paper

Data AvailabilityThe following information was supplied regarding data availability

The raw data has been supplied as a Supplementary File

Supplemental InformationSupplemental information for this article can be found online at httpdxdoiorg107717peerj3782supplemental-information

REFERENCESBardet N Houssaye A Vincent P Suberbiola XP AmaghzazM Jourani E Meslouh

S 2015Mosasaurids (Squamata) from the Maastrichtian Phosphates of Moroccobiodiversity palaeobiogeography and palaeoecology based on tooth morphoguildsGondwana Research 27(3)1068ndash1078 DOI 101016jgr201408014

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 3140

Bardet N Suberbiola XP IarocheneM Bouyahyaoui F Bouya B AmaghzazM 2005 A new species of Halisaurus from the Late Cretaceous phosphatesof Morocco and the phylogenetical relationships of the Halisaurinae (Squa-mata Mosasauridae) Zoological Journal of the Linnean Society 143(3)447ndash472DOI 101111j1096-3642200500152x

Bardet N Suberbiola XP Jalil N-E 2003 A new mosasauroid (Squamata) from theLate Cretaceous (Turonian) of Morocco Comptes Rendus Palevol 2607ndash616DOI 101016jcrpv200309006

Bell GL 1993 A phylogenetic revision of Mosasauroidea (Squamata) PhD thesisUniversity of Texas Austin

Bell GL 1997 A phylogenetic revision of North American and Adriatic MosasauroideaIn Callaway JM Nicholls EL eds Ancient marine reptiles San Diego AcademicPress 293ndash332

Bell GL PolcynMJ 2005 Dallasaurus turneri a new primitive mosasauroid fromthe Middle Turonian of Texas and comments on the phylogeny of Mosasauridae(Squamata) Netherlands Journal of Geosciences 84(3)177ndash194DOI 101017S0016774600020965

Boas JEV 1880 Studier over Decapodernes Slaegtskabsforhold Dansk VidenskabernesSeksjeab Copenhagen Skrifter Naturvidenskabelig og Matematisek Afdeling 123ndash210

Bouckaert RR Heled J Kuehnert D Vaughan TGWu C-H Xie D SuchardMA Rambaut A Drummond AJ 2014 BEAST 2 a software platform forBayesian evolutionary analysis PLOS Computational Biology 10(4)e1003537DOI 101371journalpcbi1003537

BrazeauMD 2011 Problematic character coding methods in morphology and theireffects Biological Journal of the Linnean Society 104489ndash498DOI 101111j1095-8312201101755x

Bullard TS 2006 Anatomy and systematics of North American tylosaurine mosasaursMSc thesis University of Alberta Canada

Bullard TS Caldwell MW 2010 Redescription and rediagnosis of the tylosaurinemosasaur Hainosaurus pembinensis Nicholls 1988 as Tylosaurus pembi-nensis (Nicholls 1988) Journal of Vertebrate Paleontology 30(2)416ndash426DOI 10108002724631003621870

Caldwell MW 1996 Ontogeny and phylogeny of the mesopodial skeleton inmosasauroid reptiles Zoological Journal of the Linnean Society 116407ndash436DOI 101111j1096-36421996tb00131x

Caldwell MW 2006 A new species of Pontosaurus (Squamata Pythonomorpha) fromthe Upper Cretaceous of Lebanon and a phylogenetic analysis of PythonomorphaMemorie della Societagrave Italiana di Scienze Naturali e del Museo Civico di StoriaNaturale di Milano 341ndash42

Caldwell MW Carroll RL Kaiser H 1995 The pectoral girdle and forelimb of Car-sosaurus marchesetti (Aigialosauridae) with a preliminary phylogenetic analysisof mosasauroids and varanoids Journal of Vertebrate Paleontology 15(3)516ndash531DOI 10108002724634199510011245

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 3240

Caldwell MW Cooper JA 1999 Redescription palaeobiogeography and palaeoecologyof Coniasaurus crassidens Owen 1850 (Squamata) from the Lower Chalk (Cre-taceous Cenomanian) of SE England Zoological Journal of the Linnean Society127(4)423ndash452 DOI 101111j1096-36421999tb01380x

Caldwell MW Konishi T Obata I Muramoto K 2008 New species of Taniwhasaurus(Mosasauridae Tylosaurinae) from the upper Santonian-lower Campanian (UpperCretaceous) of Hokkaido Japan Journal of Vertebrate Paleontology 28(2)339ndash348DOI 1016710272-4634(2008)28[339ANSOTM]20CO2

Caldwell MW Palci A 2007 A new basal mosasauroid from the Cenomanian (UCretaceous) of Slovenia with a review of mosasauroid phylogeny and evolutionJournal of Vertebrate Paleontology 27(4)863ndash880DOI 1016710272-4634(2007)27[863ANBMFT]20CO2

Caldwell MW Palci A 2010 A new species of marine ophidiomorph lizard Adriosaurusskrbinensis from the Upper Cretaceous of Slovenia Journal of Vertebrate Paleontology30(3)747ndash755 DOI 10108002724631003762963

Camp CL 1923 Classification of the lizards Bulletin of the American Museum of NaturalHistory 48(11)289ndash480

Cantino PD De Queiroz K 2010 International code of phylogenetic nomenclatureVersion 4c Available at httpwwwohioeduphylocodePhyloCode4cpdf (accessedon 15 February 2017)

Cau A 2017 Specimen-level phylogenetics in paleontology using the Fossilized BirthndashDeath model with Sampled Ancestors PeerJ 5e3055 DOI 107717peerj3055

Christiansen P Bonde N 2002 A new species of gigantic mosasaur from theLate Cretaceous of Israel Journal of Vertebrate Paleontology 22(3)629ndash644DOI 1016710272-4634(2002)022[0629ANSOGM]20CO2

Congreve CR Lamsdell JC 2016 Implied weighting and its utility in palaeonto-logical data sets a study using modelled phylogenetic matrices Palaeontology59(3)447ndash462 DOI 101111pala12236

Conrad JL 2008 Phylogeny and systematics of Squamata (Reptilia) based onmorphology Bulletin of the American Museum of Natural History 3101ndash182DOI 1012063101

Conrad JL Ast JC Montanari S Norell MA 2011 A combined evidence phyloge-netic analysis of Anguimorpha (Reptilia Squamata) Cladistics 27(3)230ndash277DOI 101111j1096-0031201000330x

Cope ED 1869 Remarks on Holops brevispinus Ornithotarsus immanis andMacrosaurusproriger Proceedings of the Academy of Natural Sciences Philadelphia 211ndash123

Cuthbertson RS Holmes RB 2015 A new species of Plioplatecarpus (Mosasauri-dae Plioplatecarpinae) from the Bearpaw Formation(Campanian Upper Cre-taceous) of Montana USA Journal of Vertebrate Paleontology 35(3)e922980DOI 101080027246342014922980

Cuthbertson RS Mallon JC Campione NE Holmes RB 2007 A new species ofmosasaur (Squamata Mosasauridae) from the Pierre Shale (lower Campanian) ofManitoba Canadian Journal of Earth Sciencies 44593ndash606 DOI 101139e07-006

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 3340

Dal Sasso C Pinna G 1997 Aphanizocnemus libanensis n gen n sp a new dolichosaur(Reptilia Varanoidea) from the Upper Cretaceous of Lebanon PaleontologiaLombarda 71ndash31

Dollo L 1882 Note sur lrsquoosteacuteologie des Mosasauridaelig Bulletin du Museacutee Royal drsquoHistoireNaturelle de Belgique 155ndash80

Dollo L 1884 Le mosasaure Revue des Questions Scientifiques 16648ndash653Dortangs RW Schulp AS Mulder EWA Jagt JWM Peeters HHG Graaf DT 2002 A

large new mosasaur from the Upper Cretaceous of the Netherlands NetherlandsJournal of Geosciences 81(1)1ndash8 DOI 101017S0016774600020515

Drummond AJ SuchardMA Xie D Rambaut A 2012 Bayesian phylogenetics withBEAUti and the BEAST 17Molecular Biology and Evolution 291969ndash1973DOI 101093molbevmss075

Dutchak AR 2005 A review of the taxonomy and systematics of aigialosaurs Nether-lands Journal of Geosciences 84(3)221ndash222 DOI 101017S0016774600021004

Dutchak AR Caldwell MW 2006 Redescription of Aigialosaurus dalmaticus Kram-berger 1892 a Cenomanian mosasauroid lizard from Hvar Island Croatia Cana-dian Journal of Earth Sciences 431821ndash1834 DOI 101139e06-086

Dutchak AR Caldwell MW 2009 A redescription of Aigialosaurus (= Opetiosaurus)bucchichi Kornhuber 1901 (Squamata Aigialosauridae) with comments onmosasauroid systematics Journal of Vertebrate Paleontology 29(2)437ndash452DOI 1016710390290206

Everhart M Jagt JWMMulder EWA Schulp AS 2016Mosasaursmdashhow large did theyreally get In Kear BP Lindgren J Sachs S eds 5th triennial Mosasaur meetingmdashaglobal perspective on Mesozoic marine amniotes Uppsala 16ndash20 May 2016 Programand Abstracts Museum of Evolution Uppsala Uppsala University 8ndash10

Fanti F Cau A Negri A 2014 A giant mosasaur (Reptilia Squamata) with an unusuallytwisted dentition from the Argille Scagliose Complex (late Campanian) of NorthernItaly Cretaceous Research 4991ndash104 DOI 101016jcretres201401003

Felgenhauser BE Abele LG 1983 Phylogenetic relationships among shrimp-likedecapods In Schram F ed Crustacean issues 1 Crustacean phylogeny RotterdamA A Balkema 291ndash311

FernandezMMartin JE 2009 Description and phylogenetic relationships ofTaniwhasaurus antarcticus (Mosasauridae Tylosaurinae) from the upperCampanian (Cretaceous) of Antarctica Cretaceous Research 30717ndash726DOI 101016jcretres200812012

Gauthier JA KearneyMMaisano JA Rieppel O Behlke ADB 2012 Assembling thesquamate tree of life perspectives from the phenotype and the fossil record Bulletinof the Peabody Museum of Natural History 53(1)3ndash308 DOI 1033740140530101

Gavryushkina A Heath TA Ksepka DT Stadler TWelch D Drummond AJ 2017Bayesian total evidence dating reveals the recent crown radiation of penguinsSystematic Biology 6657ndash73

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 3440

Gavryushkina AWelch D Stadler T Drummond AJ 2014 Bayesian inference ofsampled ancestor trees for epidemiology and fossil calibration PLOS ComputationalBiology 10(12)e1003919 DOI 101371journalpcbi1003919

Gervais P 1853 Observations relatives aux Reptiles fossiles de France (deuxiegraveme partie)Comptes Rendus Hebdomadaires des Seacuteances de lrsquoacadeacutemie des Sciences 36470ndash474

Gilmore CW 1912 A new mosasauroid reptile from the Cretaceous of AlabamaProceedings of the United States National Museum 40(1870)489ndash484

Goloboff PA 1993 Estimating character weights during tree search Cladistics 983ndash91DOI 101111j1096-00311993tb00209x

Goloboff PA 1995 Parsimony and weighting a reply to Turner and Zandee Cladistics1191ndash104 DOI 101111j1096-00311995tb00006x

Goloboff PA Carpenter JM Arias JS Esquivel DFM 2008Weighting against ho-moplasy improves phylogenetic analysis of morphological data sets Cladistics24758ndash773 DOI 101111j1096-0031200800209x

Goloboff PA Farris J Nixon K 2008 TNT a free program for phylogenetic analysisCladistics 24774ndash786 DOI 101111j1096-0031200800217x

Goloboff PA Szumik C 2015 Identifying unstable taxa efficient implemen-tation of triplet-based measures of stability and comparison with Phyu-tility and RogueNaRokMolecular Phylogenetics and Evolution 8893ndash104DOI 101016jympev201504003

Goloboff PA Torres A Arias JS 2017Weighted parsimony outperforms other methodsof phylogenetic inference under models appropriate for morphology Cladistics Epubahead of print June 4 2017 DOI 101111cla12205

Graham SW Olmstead RG Barrett SCH 2002 Rooting phylogenetic trees with distantoutgroups a case study from the commelinoid monocotsMolecular Biology andEvolution 191769ndash1781 DOI 101093oxfordjournalsmolbeva003999

Grigoriev D 2013 Redescription of Prognathodon lutugini (Squamata Mosasauridae)Proceedings of the Zoological Institute RAS 317(3)246ndash261

Holmes RB Sues H-D 2000 A partial skeleton of the basal mosasaur Halisaurusplatyspondylus from the Severn Formation (Upper Cretaceous Maastrichtian) ofMaryland Journal of Paleontology 74(2)309ndash316 DOI 101017S0022336000031516

International Commission on Zoological Nomenclature (ICZN) 1999 Internationalcode of zoological nomenclature In The international trust for zoological nomencla-ture Fourth Edition London ICZN 306 pp

Jimeacutenez-Huidobro P Caldwell MW 2016 Reassessment and reassignment ofthe early Maastrichtian mosasaur Hainosaurus bernardi Dollo 1885 to Ty-losaurusMarsh 1872 Journal of Vertebrate Paleontology 36(3)e1096275DOI 1010800272463420161096275

Jimeacutenez-Huidobro P Simotildees TR Caldwell MW 2016 Re-characterization of Ty-losaurus nepaeolicus (Cope 1874) and Tylosaurus kansasensis Everhart 2005ontogeny or sympatry Cretaceous Research 6568ndash81DOI 101016jcretres201604008

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 3540

Jones MEH Anderson CL Hipsley CA Muumlller J Evans SE Schoch RR 2013Integration of molecules and new fossils supports a Triassic origin for Lep-idosauria (lizards snakes and tuatara) BMC Evolutionary Biology 13208DOI 1011861471-2148-13-208

Kirchberger PC Sefc KM Sturmbauer C Koblmuumlller S 2014 Outgroup effectson root position and tree topology in the AFLP phylogeny of a rapidly radi-ating lineage of cichlid fishMolecular Phylogenetics and Evolution 7057ndash62DOI 101016jympev201309005

Konishi T Caldwell MW 2011 Two new plioplatecarpine (Squamata Mosasauridae)genera from the Upper Cretaceous of North America and a global phylogeneticanalysis of plioplatecarpines Journal of Vertebrate Paleontology 31(4)754ndash783DOI 101080027246342011579023

Konishi T Caldwell MW Nishimura T Sakurai K Tanoue K 2016 A new hal-isaurine mosasaur (Squamata Halisaurinae) from Japan the first record inthe western Pacific realm and the first documented insights into binocularvision in mosasaurs Journal of Systematic Palaeontology 14(10)809ndash839DOI 1010801477201920151113447

Kornhuber A 1873 Uumlber einen neuen fossilen saurier aus Lesina Herausgegeben VonDer K K Geologischen Reichsanstalt 575ndash90

Kornhuber A 1901 Opetiosaurus bucchichi eine neue fossile Eidechse aus der unterenKreide von Lesina in Dalmatien AbhandLungender Kaiserlich-Koumlniglichen Geologis-chen Reichsanstalt zu Wien 17(5)1ndash24

Kramberger KG 1892 Aigialosaurus eine neue Eidechse aus den Kreideschiefern derInsel Lesina mit Ruumlcksicht auf die bereits beschriebenen Lacertiden von Comen undLesina Glasnik Hrvatskoga Naravoslovnoga Društva (Societas Historico-NaturalisCroatica) u Zagrebu 774ndash106

Leblanc ARH Caldwell MW Bardet N 2012 A new mosasaurine from the Maas-trichtian (Upper Cretaceous) phosphates of Morocco and its implications formosasaurine systematics Journal of Vertebrate Paleontology 32(1)82ndash104DOI 101080027246342012624145

LeeMSY 1998 Convergent evolution and character correlation in burrowing reptilestowards a resolution of squamate relationships Biological Journal of the LinneanSociety 65369ndash453 DOI 101111j1095-83121998tb01148x

LeeMSY Cau A Naish D Dyke GJ 2014aMorphological clocks in palaeontologyand a mid-Cretaceous origin of crown Aves Systematic Biology 63442ndash449DOI 101093sysbiosyt110

LeeMSY Cau A Naish D Dyke GJ 2014b Sustained miniaturization and anatomicalinnovation in the dinosaurian ancestors of birds Science 345(6196)562ndash566DOI 101126science1252243

Lewis PO 2001 A likelihood approach to estimating phylogeny from discrete morpho-logical character data Systematic Biology 50(6)913ndash925DOI 101080106351501753462876

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 3640

Lingham-Soliar T 1996 The first description of Halisaurus (Reptilia Mosasauridae)from Europe from the Upper Cretaceous of Belgium Bulletin de lrsquoInstitut Royal desSciences Naturelles de Belqique Sciences de la Terre 66129ndash136

MaddisonWP DonoghueMJ Maddison DR 1984 Outgroup analysis and parsimonySystematic Zoology 3383ndash103 DOI 1023072413134

Madzia D Conrad JL Mosasauridae In De Queiroz K Cantino PD Gauthier JA edsPhylonyms a companion to the PhyloCode Berkeley University of California Press(In Press)

Makaacutedi LS Caldwell MW Oumlsi A 2012 The first freshwater mosasauroid (Upper Creta-ceous Hungary) and a new clade of basal mosasauroids PLOS ONE 7(12)e51781DOI 101371journalpone0051781

Mantell GA 1829 A tabular arrangement of the organic remains of the county of SussexTransactions of the Geological Society 2201ndash216

Marsh OC 1869 Notice of some new mosasauroid reptiles from the Greensand of NewJersey American Journal of Science 48392ndash397

Martin JE 2007 A North American Hainosaunts (Squamata Mosasauridae) from theLate Cretaceous of southern South Dakota In Martin JE Parris DC eds The geologyand paleontology of the Late Cretaceous marine deposits of the dakotas GeologicalSociety of America Special Paper vol 427 199ndash207

Martin JE FernaacutendezM 2007 The synonymy of the Late Cretaceous mosasaur (Squa-mata) genus Lakumasaurus from Antarctica with Taniwhasaurus from New Zealandand its bearing upon faunal similarity within the Weddellian Province GeologicalJournal 42(2)203ndash211 DOI 101002gj1066

Mulder EWA Cornelissen D Verding L 2004 IsMosasaurus lemonnieri a juvenileMosasaurus hoffmanni A discussion In Schulp AS Jagt JWM eds First mosasaurmeeting Maastricht 8ndash12 May 2004 abstract book and field guide MaastrichtNatuurhistorisch MuseumMaastricht 2ndash66

Nicholls EL Meckert D 2002Marine reptiles from the Nanaimo Group (Upper Creta-ceous) of Vancouver Island Canadian Journal of Earth Science 39(11)1591ndash1603DOI 101139e02-075

Nopcsa F 1923 Eidolosaurus und Pachyophis Zwei neue Neocom-Reptilien Palaeonto-graphica 5597ndash154

Olshevsky G 1991 A revision of the parainfraclass Archosauria Cope 1869 excluding theadvanced CrocodyliaMesozoic Meanderings 2 196

OrsquoReilly J Puttick M Parry L Tanner A Tarver J Fleming J Pisani D Donoghue P2016 Bayesian methods outperform parsimony but at the expense of precisionin the estimation of phylogeny from discrete morphological data Biology Letters1220160081 DOI 101098rsbl20160081

Otero RA Soto-Acuntildea S Rubilar-Rogers D Gutstein CS 2017 Kaikaifilu herveigen et sp nov a new large mosasaur (Squamata Mosasauridae) from the upperMaastrichtian of Antarctica Cretaceous Research 70209ndash225DOI 101016jcretres201611002

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 3740

Owen R 1850 Description of the fossil reptiles of the chalk formation In Dixon F edThe geology and fossils of the tertiary and cretaceous formations of sussex LondonLongman Brown Green and Longmans 378ndash404

Owen R 1851 A history of British fossil reptiles In Section II the fossil Reptilia of theCretaceous period London Cassell amp Company Limited 155ndash210

Palci A Caldwell MW 2007 Vestigial forelimbs and axial elongation in a 95-million-year-old non-snake squamate Journal of Vertebrate Paleontology 27(1)1ndash7

Palci A Caldwell MW 2010 Redescription of Acteosaurus tommasinii von Meyer 1860and a discussion of evolutionary trends within the clade Ophidiomorpha Journal ofVertebrate Paleontology 3094ndash108 DOI 10108002724630903409139

Palci A Caldwell MW Papazzoni CA 2013 A new genus and subfamily of mosasaursfrom the Upper Cretaceous of northern Italy Journal of Vertebrate Paleontology33(3)599ndash612 DOI 101080027246342013731024

PaacuteramoME 1994 Posicioacuten sistemaacutetica de un reptil marino con base en los restos foacutesilesencontrados en capas del Cretaacutecico Superior en Yaguaraacute (Huila) Revista de laAcademia Colombiana de Ciencias Exactas Fiacutesicas y Naturales 1963ndash80

Paacuteramo-Fonseca ME 2013 Eonatator coellensis nov sp (Squamata Mosasauridae)nueva especie del Cretaacutecico Superior de Colombia Revista de la Academia Colom-biana de Ciencias 37(145)499ndash518

Pierce SE Caldwell MW 2004 Redescription and phylogenetic position of the Adriatic(Upper Cretaceous Cenomanian) dolichosaur Pontosaurus lesinensis Kornhuber1873 Journal of Vertebrate Paleontology 24(2)373ndash386 DOI 1016711960

Pol P Escapa IH 2009 Unstable taxa in cladistic analysis identification and the assess-ment of relevant characters Cladistics 251ndash13 DOI 101111j1096-0031200800233x

PolcynMJ Bell GL 2005 Russellosaurus coheni n gen n sp a 92 million-year-oldmosasaur from Texas (USA) and the definition of the parafamily RussellosaurinaNetherlands Journal of Geosciences 84321ndash333 DOI 101017S0016774600021107

PolcynMJ Everhart MJ 2008 Description and phylogenetic analysis of a new species ofSelmasaurus (Mosasauridae Plioplatecarpinae) from the Niobrara Chalk of westernKansas Proceedings of the Second Mosasaur Meeting 13ndash28

PolcynMJ Jacobs LL Arauacutejo R Schulp AS Mateus O 2014 Physical drivers ofmosasaur evolution Palaeogeography Palaeoclimatology Palaeoecology 40017ndash27DOI 101016jpalaeo201305018

PolcynMJ Lindgren J Bardet N Cornelissen D Verding L Schulp AS 2012 Descrip-tion of new specimens of Halisaurus arambourgi Bardet amp Pereda Suberbiola 2005and the relationships of Halisaurinae Bulletin de la Socieacuteteacute Geacuteologique de France183(2)123ndash136 DOI 102113gssgfbull1832123

Rambaut A Drummond AJ 2009 Tracer MCMC trace analysis tool v15 Available athttp beastbioedacuk

Reeder TW Townsend TMMulcahy DG Noonan BPWood Jr PL Sites JWWiensJJ 2015 Integrated analyses resolve conflicts over squamate reptile phylogenyand reveal unexpected placements for fossil taxa PLOS ONE 10(3)e0118199DOI 101371journalpone0118199

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 3840

Reynoso V-H 1998 Huehuecuetzpalli mixtecus gen sp Nov a basal squamate (Reptilia)from the Early Cretaceous of Tepexi De Rodriacuteguez Central Meacutexico PhilosophicalTransactions B Biological Sciences 353477ndash500

Russell DA 1967 Systematics and morphology of American mosasaurs Bulletin of thePeabody Museum of Natural History 231ndash241

Schulp AS 2006 A comparative description of Prognathodon saturator (MosasauridaeSquamata) with notes on its phylogeny In Schulp AS ed On maastricht mosasaurspublicaties van het natuurhistorisch genootschap in limburg 45(1) MaastrichtNatuurhistorisch Genootschap in Limburg 19ndash56

Schulp AS Jagt JWM Fonken F 2004 New material of the mosasaur Carinodens belgicusfrom the Upper Cretaceous of The Netherlands Journal of Vertebrate Paleontology24744ndash747 DOI 1016710272-4634(2004)024[0744NMOTMC]20CO2

Schulp AS PolcynMJ Mateus O Jacobs LL Morais ML 2008 A new species ofPrognathodon (Squamata Mosasauridae) from the Maastrichtian of Angola and theaffinities of the mosasaur genus Liodon Proceedings of the Second Mosasaur Meeting1ndash12

Schulp AS PolcynMJ Mateus O Jacobs LL Morais ML Da Silva Tavares T 2006 Newmosasaur material from the Maastrichtian of Angola with notes on the phylogenydistribution and palaeoecology of the genus Prognathodon In Schulp AS ed OnMaastricht Mosasaurs Publicaties van het Natuurhistorisch Genootschap in Limburg45(1) 57ndash67

Seeley HG 1881 On Remains of a small Lizard from the Neocomian Rocks of Comeacutennear Trieste preserved in the Geological Museum of the University of ViennaQuarterly Journal of the Geological Society 3752ndash56DOI 101144GSLJGS188103701-0407

Sereno PC 2007 Logical basis for morphological characters in phylogenetics Cladistics23565ndash587

Simotildees TR Vernygora O Paparella I Jimenez-Huidobro P Caldwell MW 2017Mosasauroid phylogeny under multiple phylogenetic methods provides new insightson the evolution of aquatic adaptations in the group PLOS ONE 12(5)e0176773DOI 101371journalpone0176773

Siu-Ting K Pisani D Creevey CJ WilkinsonM 2015 Concatabominations identifyingunstable taxa in morphological phylogenetics using a heuristic extension to safetaxonomic reduction Systematic Biology 64137ndash143 DOI 101093sysbiosyu066

SpauldingM OrsquoLeary MA Gatesy J 2009 Relationships of Cetacea (Artiodactyla)among mammals Increased taxon sampling alters interpretations of key fossils andcharacter evolution PLOS ONE 4(9)e7062 DOI 101371journalpone0007062

Street HP Caldwell MW 2017 Rediagnosis and redescription ofMosasaurus hoffmannii(Squamata Mosasauridae) and an assessment of species assigned to the genusMosasaurus Geological Magazine 154(3)521ndash557 DOI 101017S0016756816000236

Strganac C Salminen J Jacobs LL PolcynMJ Ferguson KMMateus O Schulp ASMorais ML Da Silva Tavares T Goncalves AO 2014 Carbon isotope stratig-raphy magnetostratigraphy and 40Ar39Ar age of the Cretaceous South Atlantic

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 3940

coast Namibe Basin Angola Journal of African Earth Sciences 99(2)452ndash462DOI 101016jjafrearsci201403003

ThompsonWA 2005 The first record of Hainosaurus (Reptilia Mosasauridae) fromthe Pierre Shale of South Dakota and implications for differentiating between theTylosaurine Genera Tylosaurus and Hainosauras MSc thesis South Dakota School ofMines and Technology Rapid City

ThompsonWA 2011 The Phylogeny and Biostratigraphy of the Tylosaurine Mosasauri-dae (Reptilia Squamata) PhD thesis South Dakota School of Mines and Technol-ogy Rapid City

Wiens JJ 2003a Incomplete taxa incomplete characters and phylogenetic accuracyIs there a missing data problem Journal of Vertebrate Paleontology 23297ndash310DOI 1016710272-4634(2003)023[0297ITICAP]20CO2

Wiens JJ 2003bMissing data incomplete taxa and phylogenetic accuracy SystematicBiology 52528ndash538 DOI 10108010635150390218330

Wiens JJ Morrill MC 2011Missing data in phylogenetic analysis reconcilingresults from simulations and empirical data Systematic Biology 60719ndash731DOI 101093sysbiosyr025

Wilberg EW 2015Whatrsquos in an outgroup the impact of outgroup choice on thephylogenetic position of thalattosuchia (crocodylomorpha) and the origin ofcrocodyliformes Systematic Biology 64(4)621ndash637 DOI 101093sysbiosyv020

WilkinsonM 1992 Ordered versus unordered characters Cladistics 8375ndash385DOI 101111j1096-00311992tb00079x

WilkinsonM 1995 Coping with abundant missing entries in phylogenetic inferenceusing parsimony Systematic Biology 44501ndash514 DOI 101093sysbio444501

WilkinsonM 2001 TAXEQ3 software and documentation In Department of ZoologyLondon The Natural History Museum

Williston SW 1897 Range and distribution of the mosasaurs Kansas UniversityQuarterly 6177ndash189

World Register of Marine Species (WoRMS) 2015 Natantia Available at httpwwwmarinespeciesorgaphiaphpp =taxdetailsampid=181484 (accessed on 02 February2017)

Madzia and Cau (2017) PeerJ DOI 107717peerj3782 4040