New Morphological Evidence for the Phylogeny of Artiodactyla, Cetacea, and Mesonychidae

56 Pages·2001·0.96 MB·English

by  Jonathan H. Geisler

PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY
CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024
Number 3344, 53 pp., 11 figures August 28, 2001
New Morphological Evidence for the Phylogeny of
Artiodactyla, Cetacea, and Mesonychidae
JONATHAN H. GEISLER1
ABSTRACT
Parsimony-based analyses of a data set including 68 taxa coded for 186 morphological
characters corroborate monophyly of Artiodactyla (even-toed ungulates), Suiformes (hippos,
pigs, peccaries), Neoselenodontia (camels, deer, cows), and Acreodi (whales, dolphins, por-
poises, mesonychids). Additional findings include a sister-group relationship between Cain-
otheriidae and Cameloidea (Camelidae (cid:49) Oromerycidae), Elomeryx as the sister group to all
other suiform artiodactyls, Protoceratidae as the basal branch of Neoselenodontia, and para-
phyly of Mesonychidae. The molecule-based groups Whippomorpha (whales, dolphins, hip-
pos),Cetruminantia(whales,deer,cows),andArtiofabula(whales,cows,pigs)arecontradicted
by these data and occur together in trees that are at least 25 steps longer than the most
parsimonious ones. In terms of tree length, the molecule-based topology is contradicted by
morphological data with and without extinct taxa, and unlike previous, morphology-based
analyses, the exclusion of Cetacea from the clade of living artiodactyls is not dependent on
the inclusion of extinct taxa. Artiodactyla is diagnosed in all most parsimonious trees by
severalcharacters,includingashortmastoidprocessofthepetrosal,absenceofanalisphenoid
canal, and presence of an entocingulum on P4. Some previously suggested artiodactyl syna-
pomorphies, such as an enlarged facial exposure of the lacrimal and absence of contact be-
tween the frontal and alisphenoid, are shown to be synapomorphies of more exclusive clades
within Artiodactyla.
INTRODUCTION the most hotly debated issues in mammalian
systematics, as shown by a review of the
The phylogenetic position of Cetacea controversy surrounding cetacean and artio-
(whales, dolphins, and porpoises) is one of dactyl phylogeny (Luo, 2000), a volume on
1Graduate Student, Division of Paleontology, American Museum of Natural History Currently: PostdoctoralRe-
searchAssistant,DivisionofVertebrateZoology(Mammalogy).
Copyright(cid:113)AmericanMuseumofNaturalHistory2001 ISSN0003-0082
2 AMERICAN MUSEUM NOVITATES NO. 40
cetacean origins (Thewissen, 1998), and nu- Geisler (1999). Although these studies have
merous analytical studies (e.g., Gatesy et al., madedetailedcomparisonsbetweenmorpho-
1999a,1999b;Nikaidoetal.,1999;O’Leary, logical and molecular data possible, much of
1999; O’Leary and Geisler, 1999; Shima- thedataconcerningthephylogenywithinAr-
mura et al., 1999). Almost all morphology- tiodactylahaveyettobeincluded.Thisstudy
based studies have found Mesonychidae (or has four primary goals: (1) to add taxa and
one or more mesonychids) to be the sister new characters to previously published mor-
group to Cetacea, and have found Artiodac- phological data sets (Geisler and Luo, 1998;
tyla (even-hoofed ungulates, including cam- O’Leary and Geisler, 1999; Luo and Ginger-
els, pigs, and deer) to be monophyletic (Van ich, 1999); (2) to determine what taxonomic
Valen, 1966; Thewissen, 1994; Geisler and groups these characters support, as well as
Luo, 1998; O’Leary, 1998a; O’Leary and the degree of support for these groups; (3)to
Geisler,1999;LuoandGingerich,1999)(fig. determine if the evidence for the exclusion
1A). By contrast, the vast majority of DNA of Cetacea from the clade of extant artiodac-
sequence-based studies have found strong tyls is restricted to the data for extinct taxa;
evidence for two clades that render Artio- and (4) to test alternative phylogenies, par-
dactyla paraphyletic: (1) Whippomorpha, ticularly those based on molecules.
which includes Hippopotamidae and Ceta-
cea, and (2) Cetruminantia, which includes TAXONOMY
Whippomorpha and Ruminantia (includes
deer, cows, antelope, chevrotain, and many The molecule-based and morphology-
others) (Gatesy et al., 1996, 1999a, 1999b; based hypotheses of artiodactyl and cetacean
Gatesy, 1997, 1998; Montgelard et al., 1997; phylogenynotonlydifferinthephylogenetic
Shimamura et al., 1997, 1999; Ursing and position of extant cetaceans and extant artio-
Arnason, 1998; Nikaido et al., 1999; Klei- dactyls, but they are based on significantly
neidam et al., 1999) (fig. 1B). The incongru- different, yet slightly overlapping, sets of
ence between morphological and molecular taxa. The disparity in topology and in the
data is statistically significant (O’Leary, choice of taxa highlights the confusion
1999), and there are no plausible explana- caused by phylogenetic definitions for taxa.
tions for the conflict between the two classes Someofthetaxadiscussedinthispaperhave
of data. notbeenproperlyorexplicitlydefined,while
Incongruence between different classes of the use of other taxa varies between authors.
data can be objectively measured only if the For instance, Artiodactyla has either not in-
character data have been compiled in the cluded Cetacea (Simpson, 1945; McKenna
form of a character/taxon matrix. The spe- and Bell, 1997), hasincludedCetacea(Graur
cific observations that lead to the incongru- and Higgins, 1994; Xu et al., 1996; Kleinei-
ence can be isolated and reexamined if the dam et al., 1999), or has been replaced by
data are in a matrix form. Although there the taxon Cetartiodactyla, which includes
have been numerous studies on artiodactyl Cetacea (Montgelard et al., 1997; Nikaido et
phylogeny (e.g., Matthew, 1929, 1934; Janis al., 1999). The inclusion of Cetacea within
and Scott, 1987; Gentry and Hooker, 1988; Artiodactyla,asadvocatedbyGraurandHig-
ScottandJanis,1993)andothersoncetacean gins (1994), Xu et al. (1996), and Kleinei-
phylogeny (e.g., Muizon, 1991; Fordyce, dam et al. (1999), can be justified if their
1994; Messenger and McGuire, 1998; Luo molecule-based cladogramsarethemostpar-
and Gingerich, 1999; Uhen, 1999), there simonious hypotheses and if they use a phy-
have been few studies that have made com- logenetic definition for Artiodactyla.
parisons between membersof bothtaxonom- Taxa that have been defined using phylo-
ic groups. Geisler and Luo (1998) presented genetic taxonomy (sensu stricto de Quieroz
the first cladistic analysis of morphological andGauthier,1990)arenotusedinthispaper
data that included basal cetaceans as well as because taxon membership varies signifi-
several artiodactyls. Their work was signifi- cantly with the choice of cladogram. To
cantlyexpandedandimproveduponbyGeis- avoid confusion, only group-based defini-
ler and O’Leary (1997) and O’Leary and tions are used here. The content of each
2001 GEISLER: ARTIODACTYLA, CETACEA, MESONYCHIDAE 3
group follows McKenna and Bell (1997), Geisler and Luo, 1998; O’Leary and Geisler,
with the following exceptions. Suiformes is 1999). McKenna and Bell (1997) did not
redefined here as the group including An- provide a name for the group that includes
thracotheriidae, Entelodontidae, Hippopo- Cetacea, Artiodactyla, Mesonychidae, Hapa-
tamidae,Suidae,andTayassuidaebutexclud- lodectidae, and Andrewsarchus. Following
ing Ruminantia, Camelidae, Oromerycidae, Thewissen (1994) and Geisler and Luo
Cainotheriidae, Oreodontoidea, Xiphodonti- (1998), I use Paraxonia for thisgroup,which
dae,Mixtotherium,Cebochoerus,Gobiohyus, McKenna and Bell listed as a junior syno-
Homacodon, and all species of Diacodexis. nym of Artiodactyla.
Simpson (1945) and McKenna and Bell
(1997) placed all nonselenodont artiodactyls PREVIOUS STUDIES
in Suiformes, which makes the group para-
phyletic with respect to virtually all mor- Molecular and morphological studies on
phology-based hypotheses of artiodactyl the phylogenetic position of Cetacea have
phylogeny (Matthew, 1934; Gentry and been reviewed by Gatesy (1998) and
Hooker, 1988; Geisler and Luo, 1998; O’Leary and Geisler (1999); only recently
O’Leary and Geisler, 1999). The present re- published papers not reviewed by these au-
definition maintains traditional members of thors will be described here. Gatesy (1998)
this group, such as Suidae and Hippopotam- presented new nucleotide sequences for sev-
idae, but excludes former members so that it eral mammalian taxa and performed com-
becomes monophyletic, at least based on binedandpartitionedanalysesofhisdataset,
morphological data. If future parsimony- which included over 4500 alignednucleotide
based phylogenies have a paraphyletic Sui- positions. His analysis with all genes com-
formes, I suggest that this group be aban- bined and most of his partitioned analyses
doned instead of being redefined. supportedasister-grouprelationshipbetween
As in Simpson (1945), but unlike McKen- Hippopotamidae and Cetacea, as well as a
na and Bell (1997), Suina is used to denote larger clade including these two taxa plus
the group including Suidae and Tayassuidae Ruminantia. These controversial clades that
to the exclusion of Hippopotamidae and oth- result in artiodactyl paraphyly received sig-
er suiform artiodactyls. McKenna and Bell nificant branch support and had bootstrap
(1997) did not recognize this clade in their valuesover90%(Gatesy,1998).Luckettand
classificationandinsteadlistedSuinaasaju- Hong (1998) presented an exhaustive analy-
nior synonym of Suinae. Following Viret sis of selected morphological characters and
(1961)andWebbandTaylor(1980),butcon- previously published or available cyto-
trary to McKenna and Bell (1997) and Gen- chrome b sequences. They found that two
try and Hooker (1988), Ruminantia, as used characters, the double-trochleated astragalus
here, does not include the Amphimerycidae. and a trilobed, deciduous, fourth lower pre-
Instead, Amphimerycidae and Xiphodonti- molar, are rare among mammals but occurin
dae are considered as the only two families everyextantandextinctartiodactylgenusfor
inthegroupXiphodontoidea,namedbyViret which these anatomical regions are pre-
(1961). Use of the group Neoselenodontia served. They also determined that most of
follows Webb and Taylor (1980) and in- the nucleotides that supported the Hippopo-
cludes Camelidae, Oromerycidae, Ruminan- tamidae (cid:49) Cetacea clade exhibit some level
tia, Protoceratidae, and Xiphodontoidea but of homoplasy across all mammals. Based on
excludes Oreodontoidea. McKenna and Bell these observations,LuckettandHong(1998)
(1997) elevated Acreodi to subordinal rank concluded that existing molecular data are
and placed triisodontids, mesonychids, and not sufficient to overturn artiodactyl mono-
hapalodectids inside it; however, Ifollowthe phyly; however, other genes that corroborate
use of Acreodi by Prothero et al. (1988) to Whippomorpha and Cetruminantia (e.g., (cid:107)
denote the group including Hapalodectidae, and (cid:98) casein and (cid:103) fibrinogen) were not dis-
Mesonychidae, and Cetacea. Andrewsarchus cussed in much detail.
is excluded from Acreodi based on previous Ursing and Arnason (1998) sequenced the
morphological studies (O’Leary, 1998a; entire mitochondrial genome of Hippopota-
4 AMERICAN MUSEUM NOVITATES NO. 40
mus amphibius and included it in a phylo- monophyletic Whippomorpha, Cetruminan-
genetic analysis with 15 other mammals. tia, and Artiofabula (fig. 1B).
Maximum likelihood, maximum parsimony, O’Leary (1999) presented the first com-
and neighbor-joining methods produced op- bined morphological and molecular analysis
timal trees that supported a hippopotamid that included significant numbers of ceta-
and cetacean clade as well as a hippopotam- ceans and artiodactyls. The morphological
id, cetacean, and ruminant clade. Milinkov- data were based on the matrix of O’Leary
itch et al. (1998) retrieved nucleotide se- and Geisler (1999), and the molecular data
quences of the (cid:97)-lactalbumin protein from cameprimarilyfromGatesyetal.(1996)and
several artiodactyls and cetaceans. Using a Gatesy (1997). O’Leary (1999) found thein-
variety of phylogenetic methods, they found congruence between the neontological (al-
additional support for artiodactyl paraphyly; most entirely molecular) and osteological
however, their taxonomic sampling waspoor partitions to be statistically significant ac-
(only four cetaceans and four artiodactyls). cording to the partition-homogeneity test of
Montgelard et al. (1998) completed the first Farrisetal.(1995).Sequencealignmentsand
phylogenetic analysis of higher level artio- analyses of the combined matrix were per-
formed using nine different combinations of
dactyl phylogeny that combined morpholog-
parameters (e.g., gap cost, transition/trans-
ical and molecular data; however, little new
version ratio), and all resulted in a paraphy-
data were presented, Cetaceawasnotinclud-
letic Artiodactyla. Apparently all most par-
ed, and the ingroup only included six taxa.
simonious trees from all analyses had the
They found substantial support for Suina
(Suidae (cid:49) Tayassuidae)butweaksupportfor HippopotamidaeandCetaceacladetotheex-
a suiform clade of Suina (cid:49) Hippopotamidae. clusionofotherextantartiodactyls(O’Leary,
1999).
Gatesy et al. (1999b) added several pre-
Shimamura et al. (1999) expanded upon
viously published data sets to that of Gatesy
the work of Shimamura et al. (1997) by se-
(1998), resulting in a 64% increase in the
quencing and comparing morenucleotidese-
number of informative characters. They also
quencesforseveraldifferentSINEs(shortin-
defined and implemented several new meth-
terspersedrepetitiveelements)foundinsome
ods of evaluating nodal support, resulting in
artiodactyls and cetaceans. The identification
the discovery of significant amounts of hid-
of related SINEs in Sus (pigs) and Tayassu
den support for the Hippopotamidae (cid:49) Ce-
(peccaries) but not in Camelus (camels) cor-
tacea clade as well as the more inclusive
roborated the phylogeny of Gatesy (1998:
clade including Cetacea, Hippopotamidae,
fig. 16), where Suidae and Tayassuidae are
and Ruminantia (Gatesy et al., 1999b). Four
more closely related to cetaceans than is Ca-
new sequences were added to a growing
melidae. Nikaido et al. (1999) presentednew
body of molecular data by Gatesy et al.
SINE and LINE (long interspersed element)
(1999a). These new sequences plus previ-
data, including the distribution of SINEs at
ously published data were compiled into a
10 new loci. In addition to corroborating the
data set (WHIPPO-1), which resulted in a
phylogeny of Shimamuraetal.(1997,1999),
67% increase in the number of informative
they found four insertions that support the
characters over Gatesy et al. (1999b). The Hippopotamidae and Cetacea clade. Nikaido
most parsimonious trees for the WHIPPO-1 etal.(1999)assertedthatSINEsarevirtually
matrix were the same as those for the matrix homoplasy-free and that their insertions can
analyzed by Gatesy et al. (1999b) but had be treated as irreversible; however, consid-
increasedsupportforthecontroversialclades ering the small number of SINE characters
that group cetaceans withextantartiodactyls. and the large amount of missing data in the
The cost of artiodactyl monophyly was ap- matrix of Nikaido et al. (1999), such claims
proximately 120 steps (Gatesy et al.,1999a). arepremature.Aswithallotherphylogenetic
Gatesy et al. (1999a) also presented and an- data, their only source of validation is con-
alyzed a larger matrix dubbed WHIPPO-2. gruence with preexisting, independent data,
Like many previous molecule-based hypoth- in this case nucleotide distributions.
eses, all most parsimonious trees had a Kleineidam et al. (1999) sequenced pan-
2001 GEISLER: ARTIODACTYLA, CETACEA, MESONYCHIDAE 5
Fig. 1. Previous phylogenetic hypotheses for artiodactyls, cetaceans, and mesonychids. Taxa not
included in this study were pruned from each tree, and taxa shared between the previous two studies
are in boldface. A. The most parsimonious tree for the morphological data analyzed by O’Leary and
Geisler (1999). Note that Artiodactyla, Neoselenodontia, and Suiformes are monophyletic. B. Thestrict
consensus of the shortest trees for the WHIPPO-2 molecular data set of Gatesy et al. (1999a). Unlike
O’Leary and Geisler (1999), Artiodactyla, Neoselenodontia, and Suiformes are paraphyletic, while
Whippomorpha, Cetruminantia, and Artiofabula are monophyletic.
creatic ribonuclease genes for eight artiodac- O’Leary and Geisler (1999) presented a
tyls and cetaceans. A phylogenetic analysis detailed phylogenetic analysis of a matrix of
of these sequences plus previouslypublished 40 taxa scored for 123 morphological char-
data supported a Hippopotamus and Cetacea acters, a significant increase in both charac-
clade;however,unlikeotherrecentmolecular ters and taxa over the data set used by Geis-
studies, Suidae (pigs) instead of Camelidae ler and Luo (1998). Theirmostparsimonious
was the sister group to a clade including all trees included a monophyletic Artiodactyla,
other extant artiodactyls and Cetacea. Wad- Mesonychidae (cid:49) Cetacea, Neoselenodontia,
dell et al.(1999), in asummarypaperforthe and Suiformes (O’Leary and Geisler, 1999)
1998 ‘‘International Symposium on the Ori- (fig. 1A). They found that the recovery of
gin of Mammalian Orders’’, presented no artiodactyl monophyly hinged on the addi-
new data or analyses but did name several tion of extinct taxa to the phylogenetic anal-
controversialcladesofartiodactylssupported ysis. Thewissen and Madar (1999) described
by molecular data. The clade of Cetacea (cid:49) the functional morphology of the ankle in
HippopotamidaewasnamedWhippomorpha, ungulates, listed eight phylogenetically in-
the Whippomorpha (cid:49) Ruminantia clade was formative characters of this region (some
named Cetruminantia, and the Whippomor- new and others previously described), and
pha (cid:49) Suidae (and presumably Tayassuidae) presented a character matrix of ankle char-
was named Artiofabula (Waddell et al., acters scored for a diverse group of mam-
1999). mals.Mostofthenewdatainthematrixwas
6 AMERICAN MUSEUM NOVITATES NO. 40
based on several astragali that were referred ChM PV Charleston Museum vertebrate pale-
to cetaceans by Thewissen et al. (1998); ontology collection, Charleston,
however, O’Leary and Geisler (1999) ques- South Carolina
GSM Georgia Southern Museum, States-
tioned their referral because it is based on
boro, Georgia.
size andfaunalcomponents,notondirectas-
GSP-UM Geological Survey of Pakistan/Uni-
sociation with definitive cetacean remains.
versity of Michigan, Ann Arbor
The matrix of Thewissen and Madar (1999) H-GSP Howard University/ Geological Sur-
was analyzed by calculating the fit of all vey of Pakistan, Washington, D.C.
characterstothetreeofProtheroetal.(1988) IVPP Institute of Vertebrate Paleontology
or to modified versions of this tree. They andPaleoanthropology,Beijing,China
state that tarsal morphologies are ‘‘also con- MAE Mongolian Academy of Sciences–
AmericanMuseumofNaturalHistory
sistent with the inclusion of cetaceans in ar-
Paleontological Expeditions, collec-
tiodactyls, if one assumes that the wide arc
tion to be deposited attheMongolian
of rotation of the trochleated head was lost
Academy of Sciences, Ulaan Bataar
during the origin of Cetacea’’ (Thewissen MCZ Museum of Comparative Zoology,
and Madar, 1999: 28). However, the only HarvardUniversity,Cambridge,Mas-
cladogram in their figure 2 that had Cetacea sachusetts
grouped within Artiodactyla was five steps SMNS Staatliches Museum fu¨r Naturkunde,
longer than alternative topologiesthatplaced Stuttgart, Germany
USNM National Museum of Natural History,
Cetacea outside of, but still the sister group
Smithsonian Institution, Washington,
to, Artiodactyla.
D.C.
Luo and Gingerich (1999) described the
YPM Yale Peabody Museum, New Haven,
basicraniaofseveralbasalcetaceansandme-
Connecticut
sonychids,determinedthehomologsofhigh- YPM-PU Princeton University collection (now
ly derived cetacean basicranial structures in at Yale Peabody Museum)
other terrestrial mammals, and presented a
parsimony-based analysis of 64 basicranial MATERIALS AND METHODS
characters. Their phylogenetic analysis sup-
ported a sister group relationship between TAXON SAMPLING
Cetacea and Mesonychidae, and they listed
In general, taxa werechosentoadequately
several characters that support this clade;
sample the diversity of Artiodactyla,Meson-
however, artiodactyl monophyly was not
ychidae, and Cetacea (O’Leary and Geisler,
tested because only one artiodactyl taxon,
1999; method 3 of Hillis, 1998). MostOTUs
Diacodexis, was included. O’Leary and
(operational taxonomic units) were genera,
Uhen (1999) added the taxon Nalacetus to
leaving monophyly of more inclusive taxato
the matrix of O’Leary and Geisler (1999)
be tested. Extant genera, which were used as
and tested hypotheses concerning the strati-
taxonomic exemplars in the molecular stud-
graphic fit of the most parsimonious trees
ies of Gatesy (1998) and Gatesy et al.
and the relative timing of the evolution of
(1999a), were also included to facilitate a
characters.Theirmostparsimonioustreesare
combined molecule and morphology phylo-
identical to those of O’Leary and Geisler genetic analysis (Geisler, work in progress).
(1999) except that Harpagolestes was the The selection of extinct taxa was based on
sister group to Synoplotherium instead of simulation studies, which show that phylo-
Mesonyx. genetic accuracy can be increased by break-
ing up long branches, where branchlengthis
INSTITUTIONAL ABBREVIATIONS the number of evolutionary events (Gray-
beal, 1998; method 4 of Hillis, 1998). The
AMNH-M Department of Mammalogy, Division
phylogenyofArtiodactylaandCetacealikely
ofVertebrateZoology,AmericanMu-
contains long branches because many of the
seum of Natural History, New York
AMNH-VP Division of Paleontology (vertebrate branching events occurred in the Late Cre-
collection only), American Museum taceous or Paleocene (O’Leary and Geisler,
of Natural History, New York 1999).Atleast89%ofArtiodactyla,Cetacea,
2001 GEISLER: ARTIODACTYLA, CETACEA, MESONYCHIDAE 7
and close relatives are extinct (O’Leary and The ingroup for this study included 10 ce-
Geisler, 1999); therefore, including extinct taceans, 9 mesonychids, 2 hapalodectids, 32
taxa for consideration greatly increases the artiodactyls, 4 perissodactyls (horses, rhinos,
pool of taxa that likely attach near the bases tapirs),and 6 archaicungulates(appendix1).
of long branches. In comparison to O’Leary and Geisler
Several model-based studies have shown (1999), which is the most comprehensive
that long branch attraction is a potential morphological analysis of artiodactyls and
problem for phylogeny reconstruction using cetaceans to date, the present study includes
parsimony, and that taxonomic sampling can 28 additional taxa. Leptictidae and Orycter-
be used to reduce this problem. Felsenstein opus were included as outgroups and were
(1978) demonstrated that, given a model of used to root all mostparsimonioustrees.The
evolution that specifies probabilities of stasis exclusion of Leptictidae from the ingroup
or change between character states, phylog- wassupportedbyNovacek(1986,1992),and
enies that have long terminal branches sep- Orycteropus was outside of the clade includ-
arated by short internal branches will be in- ing artiodactyls and cetaceans in the most
correctly reconstructed using parsimony. parsimonious trees of morphological studies
Hendy and Penny (1989) suggested that this (Novacek, 1986, 1992; Gaudin et al., 1996;
problem could be alleviated by adding taxa Shoshani and McKenna, 1998), molecule-
thatattachtothebaseoflongbranches.Their basedanalyses(Stanhopeetal.,1996;Gatesy
suggestion has been supported by the work et al., 1999b), and one combined analysis
of Hillis (1998) and Graybeal (1998). (Liu and Miyamoto, 1999). Two carnivores
Kim (1996) described apparently counter- (Canis and Vulpavus) and Rattus wereadded
intuitive examples of phylogenies that led to to aid in a project that will integrate the cur-
incorrectreconstructionsusingparsimonyre- rent data set with previously published mo-
gardlessof thenumberandtypeoftaxasam- lecular data (Geisler, in prep.). Diacodexis is
pled. Hisexamplesrequiredthatsamplingbe a critical but problematic early artiodactyl
restricted to subtrees within the entire phy- taxon. It was split into two OTUs: Diacod-
logeny, and he calculated the inconsistency exis pakistanensis and North AmericanWas-
using fixed probabilities for estimating the atchian Diacodexis, with the latter being
correct phylogeny of each subtree. Actual based primarily on specimens referred to D.
studies are not restricted to sampling within metsiacus (Rose, 1985). The allocation of
parts of the phylogeny, except possibly by species to Elomeryx follows MacDonald
extinctionortheabsenceoffossils;therefore, (1956), and the allocation of specimens to
the probabilities of correctly estimating sub- Pakicetus follows Thewissen and Hussain
treesdependonthesamplingoftaxa.Adding (1998). Most taxa were scored from speci-
taxathatbreakuplongbranchescanincrease mens in the vertebrate paleontology and
theprobability of gettingthewrongtreewith mammalogycollectionsattheAmericanMu-
parsimony if the branch lengths of the added seum of Natural History (appendix 1).
taxa are longer than the original inconsistent
branch (Kim, 1996). Both Kim (1996) and CHARACTER DATA
Hulsenbeck (1991) showed that the converse
is also true, that the inconsistency can be re- Each of the 68 ingroup and outgroup taxa
moved if the added taxa have very short ter- were scored for the 186 morphological char-
minal branches. Extinct taxa that are found acters listed in appendix 2, with codings for
in strata near the age of speciation events of each taxon listed in appendix 3. Of the 186
interest (i.e., disputed nodes) are expected to morphological characters, approximately 47
have shorter branch lengths because ‘‘they areoriginaltothiswork,whiletheremaining
had less time to evolve’’ (Gauthier et al., characters are from previous morphological
1988: 193). Many of the taxa used in this studies (Webb and Taylor, 1980; Novacek,
study are from the Paleocene and Eocene 1986; Janis and Scott, 1987; Gentry and
(McKenna and Bell, 1997) and are close in Hooker, 1988; Scott and Janis, 1993; Thew-
time to the estimated origin of the most ex- issen and Domning, 1992; Thewissen, 1994;
clusive clade for which they are members. Geisler and Luo, 1998; O’Leary, 1998a;
8 AMERICAN MUSEUM NOVITATES NO. 40
O’Leary and Geisler, 1999; Luo and Ginger- 1936: pl. 2, figs. 2, 3); therefore, it is a po-
ich, 1999). An attempt was made to include tential synapomorphy of Artiodactyla. In the
all previously published morphological char- ruminants Bos and Ovis and in the suid Sus
acters useful in determining whether or not the alisphenoid canal is absent and the infra-
Cetacea belongs within the clade of living orbitalramusofthemaxillaryarteryislateral
artiodactyls. Considering the diversity of to the alisphenoid (state 1) (Getty, 1975).
taxa that belong within the ingroup, as well The alisphenoid canal is also absent in all
as the volume of previous work on artiodac- extant cetaceans, and as in most artiodactyls
tyl phylogeny, my goal was probably unre- the infraorbital ramus of the maxillaryartery
alistic; however, this matrix does provide a is lateral to the alisphenoid (Fraser and Pur-
useful contribution for those wishing to pur- ves, 1960). Absence of the alisphenoid canal
sue this problem further. In comparsion to also occurs in the most basal cetaceans Pak-
O’Leary and Geisler (1999), the present icetusandAmbulocetus;however,itsabsence
study includes an additional 63 morphologi- in cetaceans may not besynapomorphicwith
cal characters. Subheadings within the char- the morphology of most artiodactylsbecause
acter list in appendix 2 denote groups of the probable sister groups of Cetacea, the
characters that occur in the same anatomical Mesonychidae and Hapalodectidae, have an
region or share a common function. alisphenoid canal (Geisler and Luo, 1998).
Character 96: P4 entocingulum.—Pre-
SELECTED CHARACTER DESCRIPTIONS sent, partially or completely surrounds the
base of the protocone (0); absent or very
Of the 186 morphological characters in small (1). If present, the entocingulum of P4
thisstudy,Ihaveselected11ofthemthatare is on the lingual margin of the tooth. In the
either potential synapomorphies of Artiodac- artiodactylElomeryx,P4hasanentocingulum
tyla or synapomorphies of a more inclusive thatbeginsattheparastyle,wrapsaroundthe
mammalian clade. In cases where descrip- base of the protocone, and ends at the me-
tions are insufficient, I have included illus- tastyle (state 0). The cingulum is separated
trations. For additional descriptions of basi- from adjacent parts of the tooth by a deep
cranial characters, see Geisler and Luo groove except for its lingualmost portion,
(1998) and Luo and Gingerich (1999), and which is appressed to the base of the proto-
for descriptions of dental characters, see cone (fig. 2A: en). Although most basal ar-
Gentry and Hooker (1988) and O’Leary tiodactyls have a well-defined entocingulum,
(1998a). it is absent in most extant artiodactyls in-
Character 49: Alisphenoid canal (alarca- cluding all ruminants except for Hypertra-
nal).—Present (0); absent (1) (Novacek, gulus, camelids, Sus, and Tayassu (state 1).
1986; Thewissen and Domning, 1992). The An entocingulumoccurson theP4 oftheear-
alisphenoid canal transmits the infraorbital ly cetaceans Pakicetus and Georgiacetus, al-
ramus of the maxillary artery (Wible, 1987; though it is absent in Basilosaurus. In con-
Evans, 1993), and if the foramen rotundum trast to basal cetaceans, there is no entocin-
opens into the medial wall ofthealisphenoid gulum on the P4 of all mesonychids, such as
canal, then the anterior half of the canal also Harpagolestes (fig. 2B) (state 1).
carriesthemaxillarybranchofthetrigeminal Character 124: Occipital condyles.—
nerve (Sisson, 1921; Evans, 1993). For the Broadly rounded in lateral view (0); V-
group of taxa studied here, most oftheprim- shaped in lateral view, in posterior view the
itive taxa have an alisphenoid canal, includ- condyle is divided into a dorsalandaventral
ing Leptictidae, Eoconodon, Hyopsodus, half by a transverse ridge (1). The occipital
Phenacodus, and Meniscotherium (state 0). condyles of many mammals, such as in Or-
These observations are consistent with the ycteropus and Phenacodus, are smoothly
view of Thewissen and Domning (1992)that convex and do not have a transverse ridge
presence of the canal is primitive for Euthe- (state0).Bycontrast,inmostartiodactylsthe
ria. occipital condyle has a transverse ridge that
The alisphenoid canal is absent in all ar- dividesitintodorsalandventralhalves(state
tiodactylsexceptforCainotherium(Hu¨rzeler, 1). The ridge begins at thelateraledgeofthe
2001 GEISLER: ARTIODACTYLA, CETACEA, MESONYCHIDAE 9
Fig. 2. Representative morphologies for the lingual margin of P4. Labial is toward the top of the
page,anterioristotheleft,andthescalebarsrepresent10mm.A.Thethirdandfourthupperpremolars
of the artiodactyl Elomeryx armatus (AMNH 582). Notethepresenceofaprominententocingulumthat
nearly encircles the base of the protocone. An entocingulum on P4 is widely distributed among basal
artiodactyltaxa;therefore,itisapotentialsynapomorphyofArtiodactyla.B.Thethirdandfourthupper
premolars of the mesonychid Harpagolestes orientalis (AMNH 26300). Note the complete absence of
an entocingulum on P4. Abbreviations: en, entocingulum; P3, upper third premolar; P4, upper fourth
premolar.
condyle and stretches across its entire pos- because an entepicondylar foramenoccursin
terior aspect. In lateral view the ridge gives most of the archaic taxa surveyed in this
thecondyleaV-shapedprofile.Thevertexof study, including Leptictidae, Orycteropus,
the ‘‘V’’ is the top of the ridge, and in the Vulpavus, Arctocyon, Eoconodon, Hyopso-
artiodactyl Poebrotherium the vertex points dus, Phenacodus, and Meniscotherium. The
ventrally and slightly posteriorly (fig. 3: or). entepicondylar foramen is absent in all artio-
The functional morphology of the ridge is dactyls, and thus its absence is a potential
unknown; however, I suspect it works with synapomorphyofthatgroup.Itisalsoabsent
the alar and lateral atlanto-occipital liga- in all cetaceans, perissodactyls,thecarnivore
ments to temporarily lock the occipital/atlas Canis, and the rodent Rattus (state 1).
joint in the position that most efficiently ori- Character 152: Third trochanter of femur
ents the head for feeding. When the muscles (ordered).—Present (0); highly reduced (1);
that nod the head are relaxed, the morphol- absent(2)(LuckettandHong,1998;O’Leary
ogy of the joint and the tension in the liga- and Geisler, 1999). The third trochanter is a
ments would passively restore the head to its flange that projects from the lateral side of
former position. the humeral shaft. On average, it is situated
Character 135: Entepicondylar fora- atonethirdofthedistancefromtheproximal
men.—Present (0); absent (1) (Thewissen to the distal end of the humerus. The super-
and Domning, 1992). The entepicondylarfo- ficial gluteus muscle, which extends the
ramen transmits the median nerve and the hindlimb at the hip joint (Evans, 1993), in-
brachial artery, as in the carnivore Felis serts on the third trochanter. In ruminants,
(Crouch,1969).Itislocatedonthedistalend which lack a third trochanter, the superficial
of the humerus and perforates the proximal gluteus has fused with the biceps femoris to
half of the medial epicondyle. Shoshani form a gluteobiceps. Instead of inserting on
(1986) hypothesized that presence of an en- thefemur,thegluteobicepsinsertsonthecru-
tepicondylar foramen was primitive for eu- ral facia, lateral patellar ligament, and facia
therian mammals.Hisviewissupportedhere lata (Getty, 1975). Presence of a large,
10 AMERICAN MUSEUM NOVITATES NO. 40
Fig.3. ObliqueposterolateralviewoftherightoccipitalcondyleofPoebrotherium(AMNH42257),
with right and left stereopair views. The occipital condyle is divided into dorsal and ventral halves by
a transverse ridge. The occipital ridge is a potential synapomorphy of Artiodactyla. Scale bar is 10mm
in length. Abbreviations: fm, foramen magnum; or, occipital ridge; tb, tympanic bulla.
square-shaped third trochanter is probably 1947; O’Leary and Geisler, 1999). The most
primitivefortheingroupbecauseitispresent widely recognized character that diagnoses
in the outgroup taxon Orycteropus and the Artiodactyla is the double-pulleyed astraga-
archaic taxa Arctocyon, Hyopsodus, Phena- lus (Schaeffer, 1947). The ‘‘double-pulley’’
codus, and Mesonychidae. refers to the fact that the proximal and distal
The third trochanter is absent in all extant ends of the astragalus are deeply grooved,
artiodactyls, and it is absent or very small in and that each end resembles a pulley. As in
allextinctartiodactyls.Specimensofthebas- previous morphological studies (e.g., Thew-
al artiodactyl Diacodexis from North Amer- issen and Domning, 1992; O’Leary and
ica (Rose, 1985) and from Asia (Thewissen Geisler, 1999), the proximal and distal ends
and Hussain, 1990) have a small rectangular of the astragalus are treated as independent
flange on the femur that is homologous to, characters.
but smaller than, the third trochanter of Arc- The tibial articulation surface of the as-
tocyon, Hyopsodus, perissodactyls, andother tragalus is divided into two parts: (1) a me-
mammals. Thus, reduction of the third tro- dial part that faces medially or proximome-
chanter is a potential synapomorphy of Ar- dially and articulates with the medial malle-
tiodactyla, while complete loss of this struc- olus of the tibia, and (2) a lateral part that
ture is a potential synapomorphy of a higher faces proximally and articulates with therest
level artiodactylcladethatincludestheartio- ofthetibia.Itisthesecondpartthatbecomes
dactyl crown group. The archaic cetacean trochleated in many mammals. In the Creta-
Ambulocetus has a third trochanter (Thewis- ceous eutherians Ukhaatherium (Horovitz,
sen et al., 1996); therefore, its presence in 2000), Asioryctes (Kielan-Jaworowska,
this taxon supports the exclusion of Cetacea 1977),andProtungulatum(SzalayandDeck-
from the clade of all artiodactyls. er,1974),thelateralpartofthetibialarticular
Character 156: Proximal end of astraga- surface is slightly concave (state 0); there-
lus (ordered).—Nearly flat to slightly con- fore, a flat to slightly concave articulating
cave (0); well grooved, but depth of trochlea surfaceontheastragalusforthetibiaisprob-
(cid:44)25% its width (1); deeply grooved, depth ably primitive for Eutheria. In the outgroups
(cid:46)30% its width (2) (derived from Schaeffer, Orycteropus and Leptictidae and the ungu-