Phylogenetic relationships among sloths (Mammalia ...web2.utc.edu/~gvv824/Gaudin 2004.pdf · Pholidota, or pangolins. The results of the analysis strongly corroborate the diphyly
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Zoological Journal of the Linnean Society, 2004, 140, 255–305. With 11 figures
Blackwell Science, LtdOxford, UKZOJZoological Journal of the Linnean Society0024-4082The Lin-nean Society of London, 2004? 2004140?255305Original Article
Phylogenetic relationships among sloths (Mammalia, Xenarthra, Tardigrada): the craniodental evidence
TIMOTHY J. GAUDIN*
Department of Biological and Environmental Sciences (Department 2653), University of Tennessee at Chattanooga, 615 McCallie Avenue, Chattanooga, TN 37403-2598, USA
Received November 2002; accepted for publication October 2003
The diphyly of the two living tree sloth genera,Bradypus (for the three species of three-toed sloths)and Choloepus (including two species of two-toedsloths), has been widely accepted in the mammaliantaxonomic literature. Nearly all recently publishedtaxonomic reviews of mammals as a whole (Barlow,1984; Gardner, 1993; McKenna & Bell, 1997;Feldhamer et al., 1999; Nowak, 1999; Vaughan, Ryan& Czaplewski, 2000) or Neotropical mammals in par-
ticular (Eisenberg & Redford, 1999) place Choloepusin the family Megalonychidae, a family composedlargely of extinct ‘ground sloths’ from North America,South America and the West Indies, whereas Brady-pus is assigned to the monotypic family Bradypo-didae. As noted by Gaudin (1995), the tree slothsshare a number of superficial similarities that are notevident in extinct sloth taxa, most prominently theirpeculiar suspensory mode of locomotion (although seeWhite, 1993, 1997). If substantiated, this taxonomicarrangement would surely present one of the moststriking examples of convergent evolution knownamong mammals.
Surprisingly, the evidence that supports this phylo-genetic hypothesis is not overwhelming despite itswide degree of acceptance. As reviewed by Gaudin(1995), the hypothesis of tree sloth diphyly was ini-tially proposed by Bryan Patterson and colleagues(Patterson & Pascual, 1968, 1972) based upon Patter-son’s until recently unpublished studies of the xenar-thran ear region (Patterson, Segall & Turnbull, 1989;Patterson et al., 1992). Patterson and his co-workers(Patterson & Pascual, 1968, 1972; Patterson et al.,1992) suggested that Choloepus was derived fromwithin the Megalonychidae, whereas Bradypus wasallied with nothrotheres, a group of extinct slothsplaced in the family Megatheriidae by these authors.Guth (1961) had earlier come to similar conclusions inhis study of the temporal and auditory region of fossiland extant xenarthrans, allying Bradypus with mega-theriids but linking Choloepus with the extinct groundsloth family Mylodontidae, based largely on the mor-phology of the ectotympanic. Guth (1961) also notedsimilarities in the squamosal and entotympanic ofCholoepus and mylodontids. Guth’s (1961) conclusionsare particularly noteworthy, because an oft cited studyby Webb (1985) affirming tree sloth diphyly employedmylodontids as an outgroup to the tree sloths and theremaining extinct sloths. Webb (1985) proposed fivemorphological synapomorphies to unite Bradypus andmegatheriids, and only a single derived skull traituniting Choloepus with megalonychids. Unfortu-nately, the character polarization for Webb’s (1985)analysis depended on the weakly supported notion of amonophyletic Megalonychoidea, a clade encompassingthe two extant tree sloths and the extinct megalony-chids and megatheriids exclusive of the mylodontids.The a priori assumption of megalonychoid monophylyin Webb’s (1985) study rendered Guth’s (1961) hypoth-esis of a linkage between Choloepus and mylodontidsuntestable. Gaudin (1995) asserted that the questionof tree sloth diphyly or monophyly is inextricablylinked to the question of the interrelationships amongthe various families of extinct sloths. The former can-not properly be addressed outside the context of thelatter.
Gaudin’s (1995) analysis was the first cladisticstudy to examine the relationships among the two treesloth genera and the various families of extinct groundsloths simultaneously. Gaudin (1995) analysed 85morphological characters drawn from the auditoryregion and basicranium in 21 extinct and extant slothtaxa, polarizing these features via comparisons withsuccessive outgroups outside of the Tardigrada (theclade comprising extinct and extant sloths) – namely,the Vermilingua, comprising living and extinct anteat-ers, and the Cingulata, comprising living and extinctarmadillos and glyptodonts. As in the studies byPatterson and his co-workers (Patterson & Pascual,
1968, 1972; Patterson et al., 1992) and Webb (1985),Gaudin’s (1995) cladogram joined Choloepus to theMegalonychidae. However, Bradypus was placed asthe sister taxon to all remaining sloths. There was nosupport for a monophyletic Megalonychoidea. Instead,megalonychids and mylodontids formed a monophyl-etic clade, with megatheriids as a sister taxon to thisclade and nothrotheres forming a paraphyleticstem group for a clade including megatheriids,megalonychids + Choloepus, and mylodontids.
Gaudin’s (1995) study is significant as the first toincorporate the two extant tree sloth genera and taxafrom all the major clades of extinct sloths in a singlecladistic analysis. However, confidence in the study’sphylogenetic conclusions is limited by the restrictedcharacter base from which characters for the studywere drawn. Several subsequent studies haveexpanded the character base that can be brought tobear on the questions of sloth interrelationships.Remarkably, three molecular phylogenetic studieshave been published. Höss et al. (1996) obtained rDNAsequences from a preserved skin sample of the extinctmylodontid Mylodon darwinii, and compared thesesequences with homologous samples of rDNA inBradypus variegatus, Choloepus didactylus, theextant anteater species Tamandua tetradactyla andthe extant armadillo species Cabassous unicinctus.The phylogenetic results were consistent with those ofGaudin (1995), allying Mylodon and Choloepus to theexclusion of Bradypus. A similar study by Poinar et al.(1998) sequenced a portion of the 12S rRNA gene inDNA extracted from a coprolite attributed to theextinct nothrothere Nothrotheriops shastense. Thissequence was compared with homologous sequencesobtained from Mylodon, Bradypus, Choloepus,Tamandua and Cabassous. In this instance, the phy-logenetic results were inconsistent with those of Gau-din (1995). Mylodon and Nothrotheriops were found tobe sister taxa, whereas Choloepus and Bradypusformed an unresolved trichotomy with the extinctforms at the basal sloth node. The most recent study(Greenwood et al., 2001) combined new mitochondrialcytochrome b gene sequences with the 12S rDNA datafor Mylodon, Nothrotheriops and the two extant slothgenera. Their results were consistent with those ofGaudin (1995) and Höss et al. (1996) in linking Cho-loepus and Mylodon, but contradicted both Gaudin(1995) and Poinar et al. (1998) in placing Bradypusand Nothrotheriops in a monophyletic group.
A recently published cladistic morphological analy-sis of sloth phylogeny by White & MacPhee (2001)incorporated postcranial, dental and cranial charac-ters, including some auditory region characters fromGaudin (1995). This study also came to conclusionsdiscordant from those of Gaudin (1995), with Brady-pus forming the sister taxon to a clade including Cho-
loepus and a subgroup of extinct West Indianmegalonychids. Although drawn from a broader char-acter base than Gaudin (1995), its taxonomic sample,like those of the molecular studies, was limited. White& MacPhee (2001) scored their 75 characters inBradypus, Choloepus and in 13 species of extinct meg-alonychid sloths from the West Indies. The only othersloths considered were the Miocene South Americannothrothere Hapalops longiceps and the PleistoceneNorth American mylodontid Paramylodon harlani.
Although none of the aforementioned studies cor-roborates a monophyletic grouping of the living treesloths, the evidence for tree sloth diphyly and for aclose linkage between Choloepus and megalonychidsis not completely satisfactory. To this point there isstill no comprehensive cladistic analysis of sloth inter-relationships that employs a broad character base andincorporates a comprehensive taxonomic sample ofextinct and extant sloths. Published phylogeneticstudies either encompass a restricted subset of slothtaxa (Webb & Perrigo, 1985; Webb, 1989; De Iuliis,1994; de Muizon & McDonald, 1995; Höss et al., 1996;Poinar et al., 1998; Greenwood et al., 2001; White &MacPhee, 2001; McDonald & de Muizon, 2002;McDonald & Perea, 2002) or employ a restricted sub-set of characters (Gaudin, 1995).
The goal of the present study is to re-analyse thephylogenetic relationships among sloths, using abroader base of osteological characters scored in asomewhat larger number of sloth taxa than in my ownpreviously published work (Gaudin, 1995). I have com-bined the matrix from Gaudin (1995) with a matrixincorporating 12 additional extinct sloth taxa and 201additional morphological characters derived from theskull (apart from the ear region) and dentition. The 12additional sloth taxa have also been scored for the 85auditory and basicranial characters employed in Gau-din (1995).
Cranial and dental characters have traditionallybeen emphasized in diagnosing the order Xenarthraand its three suborders Cingulata, Vermilingua andTardigrada (Weber, 1928; Scott, 1937; Winge, 1941;Grassé, 1955a; Hoffstetter, 1958, 1982; Romer, 1966;Engelmann, 1978, 1985; Paula Couto, 1979; Ferigolo,1985; Wetzel, 1985; Carroll, 1988), as well as in exam-ining the higher level relationships of Xenarthra (seeRose & Emry, 1993; Novacek, 1994; Gaudin et al.,1996; Szalay & Schrenk, 1998, and referencestherein). Dental characters have been of particularimportance in dividing the diverse assemblage ofextinct ground sloths into its traditional constituentfamilies: the Megatheriidae, typified by large squareteeth with parallel transverse crests and a thickenedexternal layer of cementum; the Mylodontidae, diag-nosed by their peculiar lobate dentition; the Megalo-nychidae, identified by their quadrate or ovate
molariform teeth with subparallel crests, and by theirenlarged caniniform or incisiform first upper andlower teeth; and the nothrotheres, which resemble themegatheriids in their dental morphology, with theexception of their more rectangular posterior teethand reduced first upper and lower teeth (Stock, 1925;Winge, 1941; Hoffstetter, 1958, 1982; Engelmann,1978, 1985; Paula Couto, 1979; Ferigolo, 1985).
Given the past importance of craniodental charac-ters in phylogenetic analyses of the Xenarthra, Ibelieve that the present study, combining as it doescharacters of the auditory region with those fromother areas of the skull and dentition, will yield a morerobust understanding of various unresolved questionsin the realm of sloth phylogeny. These questions wouldinclude not only the diphyly or monophyly of livingtree sloths and their relationship to various extincttaxa, but also the interrelationships among the vari-ous sloth families, the purported monophyly of theMegalonychoidea (Webb, 1985; Patterson et al., 1992;but see Gaudin, 1995), the possible family-level statusof late Miocene – Pleistocene nothrotheres (Gaudin,1994; McDonald, 1994; but see McKenna & Bell,1997), and the uncertain family-level affinities of var-ious early Miocene planopsines and nothrotheres (DeIuliis, 1994; Gaudin, 1995).
MATERIALS AND METHODS
The 201 osteological characters of the skull, lower jaw,hyoid apparatus and dentition (and their respectivecharacters states) utilized in the present study aredescribed in detail in Appendix 2. These characterswere primarily derived from two sources: (1) the gen-eral systematic works of Scott (1937), Winge (1941),Grassé (1955a), Hoffstetter (1958), Engelmann (1978,1985), Paula Couto (1979) and Naples (1982); and (2)personal observation of relevant specimens of livingand extinct edentates. In those instances in which agiven character was derived from another source, anappropriate reference is provided in Appendix 2. Ofthe 201 characters, 97 are multistate, and of these 39are unordered. The character states for the orderedmultistate characters were ordered along positional,orientational or structural morphoclines (Appendix 2).As stated previously, these 201 characters were joinedwith the 85 auditory region characters (32 multistate,11 of these unordered) from Gaudin (1995) to form acombined data matrix (Appendix 3).
Character state assignments for the 46 edentatetaxa in the craniodental portion of this study werebased upon direct observation of specimens and uponinformation obtained from the primary literature. Thespecimens examined are listed in Appendix 1. Thosetaxa not included in Gaudin (1995) but considered inthe present study were coded for the 85 characters uti-
lized in Gaudin (1995) based upon direct observationof specimens and upon information obtained from theprimary literature. The following sources served asgeneral references: Flower (1882, 1885), Weber (1928),Scott (1937), Winge (1941), Grassé (1955a), Hoffstet-ter (1958, 1982), Guth (1961), Engelmann (1978,1985), Paula Couto (1979), Moore (1981), Wetzel(1985) and Patterson et al. (1992). Additional sourcesfor individual taxa are listed below.
Pholidota: Grassé (1955b), Jollie (1968). Palaean-odon: Matthew (1918a), Simpson (1931).Metacheiromys: Osborn (1904), Matthew (1918a), Sim-pson (1931). Holmesina: Edmund (1985). Glyptodonts:Scott (1903–4), Gillette & Ray (1981). Prozaedyus:Scott (1903–4). Palaeomyrmidon: Rovereto (1914).Octodontotherium: Hoffstetter (1956). Octomylodon:Scillato-Yané (1977). Pseudoprepotherium: Hirschfeld(1985). Thinobadistes: Webb (1989). Lestodon: Lydek-ker (1894). Pleurolestodon: Rovereto (1914). Glossothe-rium: Owen (1842), Lydekker (1894), Kraglievich(1928), Hoffstetter (1952). Paramylodon: Allen (1913),Stock (1925), Kraglievich (1928). Nematherium: Scott(1903–4), Simpson (1941). Scelidotherium: Owen(1857), Lydekker (1886, 1894), McDonald (1987). Cat-onyx: McDonald (1987). Mylodon: Reinhardt (1879),Burmeister (1881), Lydekker (1894), Roth (1899),Woodward (1900), Kraglievich (1928). Choloepus:Wegner (1950), Sicher (1944), Naples (1982, 1985,1986, 1987), Webb (1985). Acratocnus: Anthony (1918,1926), Matthew (1931), Matthew & Paula Couto(1959), Paula Couto (1967). Neocnus: Matthew &Paula Couto (1959), Paula Couto (1967), Fischer(1971). Parocnus (= Mesocnus): Matthew & PaulaCouto (1959), Paula Couto (1967), Fischer (1971).Megalocnus: Matthew & Paula Couto (1959), PaulaCouto (1967), Fischer (1971). Pliomorphus: Kra-glievich (1923). Pliometanastes: Hirschfeld & Webb(1968), Hirschfeld (1981). Megalonyx: Leidy (1855),Stock (1925), Hirschfeld & Webb (1968), McDonald(1977). Bradypus: Sicher (1944), Schneider (1955),Naples (1982, 1985, 1986, 1987), Webb (1985). Prono-throtherium: Rovereto (1914). Nothropus: Frailey(1986). Nothrotherium: Reinhardt (1878), Paula Couto(1959, 1971, 1980), Cartelle & Fonseca (1982), Cartelle& Bohórquez (1986). Nothrotheriops: Stock (1925),Lull (1929), Paula Couto (1971), Naples (1987). Eucho-loeops: Scott (1903–4). Hapalops: Scott (1903–4). Plan-ops: Scott (1903–4), Hoffstetter (1961). Eremotherium:Hoffstetter (1952), Paula Couto (1954), Gazin (1956),Cartelle & Bohórquez (1982, 1986). Megatherium:Owen (1856), Burmeister (1864), Lydekker (1894),Hoffstetter (1952). Schismotherium: Scott (1903–4).Pelecyodon: Scott (1903–4). Analcimorphus: Scott(1903–4).
The data matrix was analysed using the computerprogram PAUP [Version 4.0b10 (Swofford, 2002)]. The
data were analysed using a heuristic search with ran-dom-addition sequence and 1000 repetitions in orderto find the most-parsimonious trees and avoid localminima. As in Gaudin (1995), two different weightingschemes were applied to multistate characters inorder to assess their effect on the analysis: (1) all char-acter state changes weighted equally; and (2) charac-ter state changes scaled so that all characters areweighted equally regardless of the number of charac-ter states. Characters were optimized using PAUP’sDELTRANS option in all analyses, for reasons dis-cussed in Gaudin (1995). In instances in which termi-nal taxa were found to possess more than one state ofa single character, that variation was coded into thedata matrix and treated as polymorphism. Severalcharacters proved to be parsimony uninformative inthe final analyses, but all values reported for consis-tency index exclude uninformative characters.
Characters were polarized using outgroup compari-sons to at least two successive outgroups (followingMaddison, Donoghue & Maddison, 1984). The mostproximate outgroup to the Tardigrada is the Vermilin-gua (Flower, 1882; Engelmann, 1978, 1985; Delsucet al., 2001; Greenwood et al., 2001). In the presentstudy, the monophyly of the Pilosa, the sloths and ant-eaters, was employed as a constraint in all analyses.In addition, the monophyly of both Tardigrada andVermilingua were constrained a priori, with the inter-nal relationships among the various anteater taxaconstrained according to the phylogenetic hypothesesof Gaudin & Branham (1998). The anteaters were rep-resented in the present study by five of the six knownextinct and extant genera (Gaudin & Branham, 1998):the modern genera Cyclopes, Tamandua and Myrme-cophaga, and the extinct genera Palaeomyrmidon andProtamandua.
The second outgroup to the sloths is the Cingulata(Engelmann, 1985; Delsuc et al., 2001). This groupwas constrained to form a monophyletic outgroup tothe Pilosa in all analyses. The cingulates were repre-sented by five taxa in the present study: the Pleis-tocene pampathere genus Holmesina, the Miocenearmadillo Prozaedyus, the extant armadillos Euphra-ctus and Tolypeutes, and the extinct glyptodonts.Character state assignments for the glyptodonts werebased primarily on the morphology of the Santacru-cian (early to mid-Miocene) members of the group(Scott, 1903–4; Appendix 1). However, in thoseinstances in which character information was unavail-able for the Miocene taxa, more derived Pleistoceneforms were used as the basis for character stateassignments (Guth, 1961; Gillette & Ray, 1981). Aswas the case in Gaudin (1995), in certain instanceswhere characters were highly variable within the liv-ing armadillo genera Euphractus and Tolypeutes,additional closely related extant taxa were examined
to assist in character coding (Zaedyus and Chaeto-phractus in the case of Euphractus; Priodontes, Cabas-sous and Dasypus in the case of Tolypeutes). Thephylogenetic relationships among the five cingulatetaxa were constrained a priori in all analyses accord-ing to the phylogenetic hypotheses of Engelmann(1978, 1985).
Although the primary focus of this study, like itspredecessor (Gaudin, 1995), is to elucidate relation-ships among Tardigrada, it is of general interest toknow something about the distribution of the cranio-dental features utilized in this study across the CohortEdentata (sensu Novacek, 1986; Novacek & Wyss,1986; Novacek, Wyss & McKenna, 1988) as a whole.Therefore, as in Gaudin (1995), the data matrix wasanalysed using only anteaters and cingulates as out-groups, and again using two additional edentate out-group taxa. The most proximate of these outgroups tothe Xenarthra was assumed to be the Palaeanodonta(following Simpson, 1931, 1945; Szalay, 1977; Patter-son et al., 1992; Szalay & Schrenk, 1998), as repre-sented by the relatively well-preserved andunspecialized metacheiromyid genera Palaeanodonand Metacheiromys. The second outgroup to theXenarthra was assumed to be the Pholidota (followingNovacek, 1986, 1992; Novacek & Wyss, 1986; Novaceket al., 1988).
Robustness of results was assessed using severaldifferent methods. The relative support for variousgroupings was assessed using a bootstrap analysis(Hillis & Bull, 1993) and by determining branch sup-port, i.e., the number of additional steps required tocollapse each node (Bremer, 1994). The bootstrap anal-ysis employed heuristic methods, using ten random-addition sequences per replicate for 1000 bootstrapreplicates, with character states weighted equally, andthe Pholidota and Palaeanodonta included as out-groups. Other PAUP settings were identical to thosedescribed in the preceding paragraphs. Branch sup-port was calculated by instructing PAUP to save treesprogressively longer than the MPT, in increments ofone step. At each incremental step, a strict consensustree was generated. Again, all character states wereweighted equally, the Pholidota and Palaeanodontawere included as outgroups, and the PAUP settingswere otherwise identical to those described in the pre-ceding paragraphs. Finally, the results of the presentstudy were compared with previous phylogenetichypotheses of sloth relationships. PAUP was con-strained to produce the shortest tree(s) consistentwith a given phylogenetic hypothesis [e.g., the mono-phyly of living tree sloths as in Simpson (1945) andothers, the alliance of Bradypus and megatheriids asin Webb (1985) and others]. These constrained alter-native trees were then compared with the MPT result-ing from the present study using non-parametric
statistical tests, following the procedure outlined inLuo, Kielan-Jaworowska & Cifelli (2002 – see alsoTempleton, 1983). As in Luo et al. (2002), comparisonswere made between raw trees and strict consensustrees. PAUP settings were identical to those used indetermining branch support. Results of these tests aresummarized in Appendix 5.
Institutional abbreviations: AMNH, AmericanMuseum of Natural History, New York; ANSP, Acad-emy of Natural Sciences, Philadelphia; F:AM, Frickcollection, American Museum of Natural History, NewYork; FMNH, Field Museum of Natural History, Chi-cago; IMNH, Idaho Museum of Natural History, IdahoState University, Pocatello, Idaho; LACM, NaturalHistory Museum of Los Angeles County, Los Angeles;LACMHC, Hancock collection of Los Angeles CountyMuseum housed at George C. Page Museum, LosAngeles; ROM, Royal Ontario Museum, Toronto, Can-ada; UCMP, Museum of Paleontology, University ofCalifornia, Berkeley; USGS, United States GeologicalSurvey, Washington, DC; YPM-PU, Princeton Univer-sity collection housed at Peabody Museum, Yale Uni-versity, New Haven.
Other abbreviations: BNL, basonasal length, mea-sured from the posterior edge of the occipital condylesto the anterior edge of the nasal bone; C1/c1, canini-form tooth or first upper and lower tooth in sloths –according to Grassé (1955a), this tooth corresponds tothe true canine of the upper toothrow and the firstlower premolar and will be homologized as such inpalaeanodonts, whereas in cingulates these teeth willbe homologized with the first maxillary tooth and thelower tooth which occludes posteriorly with the firstupper maxillary tooth; CI, consistency index; M1–4,m1–3, the four upper and three lower tooth positionsin sloths that follow C1 and c1 consecutively and aretypically molariform in morphology; LMA, land mam-mal age; MML, maximum mandibular length, mea-sured along a single ramus from the anterior tip of thesymphysis to the posterior tip of the angular or condy-loid process (whichever extended further posteriorly);MPT, most parsimonious tree(s); RI, retention index;TL, tree length.
RESULTS
The results of the four PAUP analyses performed inthe present study are remarkably consistent. Theanalysis in which all taxa, including palaeanodontsand pangolins, were included and all character statechanges are weighted equally yielded two MPT(TL = 1936 steps; CI = 0.370; RI = 0.645). A strict con-sensus of these two MPT is illustrated in Figure 1. Ifcharacter state changes are scaled to the number ofstates per character so that each character isweighted equally, a single MPT (TL = 1169.1 steps;
CI = 0.360; RI = 0.657) is produced that is identical tothat shown in Figure 1 with four exceptions. Two ofthese exceptions involve pairs of Santacrucian genera:Schismotherium and Pelecyodon are allied as sistertaxa at the base of the Megatherioidea (Node 19,
Fig. 1), and Analcimorphus and Hapalops form suc-cessive sister taxa to the family Megalonychidae(Node 29, Fig. 1), rather that the clade includingnothrotheriids and megatheriids (Node 23, Fig. 1).The third distinction lies within the family
Nothrotheriidae, where the Pliocene genus Prono-throtherium is allied as the closest sister taxon to thecrown group clade of Nothrotherium + Nothrotheriops,rather than the late Miocene (?)/Pleistocene genusNothropus. Finally, within Mylodontidae the Pleis-tocene genera Paramylodon and Mylodon are groupedas sister taxa. This clade is then the sister group toNode 17 (Fig. 1, including Glossotherium, Lestodonand Thinobadistes), rather than the late Miocenegenus Pleurolestodon.
Both character weighting schemes have also beenemployed in analyses in which the non-xenarthranoutgroups Palaeanodonta and Pholidota are omitted.A single MPT (TL = 1790; CI = 0.385; RI = 0.652) isproduced when all character states changes areweighted equally. This MPT differs from that shown inFigure 1 only in the position of Pelecyodon and Schis-motherium. These genera form successive sister taxato the remainder of the Megatherioidea (Node 20,Fig. 1). When character state changes are scaled sothat all characters receive equal weight, a single MPT(TL = 1081.3 steps; CI = 0.377; RI = 0.665) resultsthat closely resembles the tree produced under thesame character weighting scheme described above,except for the arrangement of taxa within the familyMegalonychidae. The extant genus Choloepus is alliedwith Acratocnus rather than Neocnus. The Choloepus/Acratocnus clade in turn represents the sister group toParocnus + Megalocnus, with Neocnus as the sistertaxon to the clade comprising these four taxa.
A strict consensus tree of all four analyses is shownin Figure 2, and probably represents the best esti-mate of sloth phylogenetic relationships based uponthe data set in the present study. However, becausethe settings used to produce the tree illustrated inFigure 1 were also used in all assessments of therobusticity of these results, the discussion below willconcentrate largely on this tree. The tree in Figure 1differs in only minor points from the overall consen-sus tree shown in Figure 2, but some of these differ-ences will be addressed in ensuing discussions. Acomplete list of apomorphies for all the numberednodes in Figure 1 is provided in Appendix 4. The num-bers used to refer to these apomorphies in Appendix 4
and in the following discussions are based on thenumeration provided in Appendix 2. Those charactersderived from Gaudin (1995) are referred to by an ‘E’followed by a number. This number is based on thenumeration of characters provided in Appendix 1 ofGaudin (1995).
COHORT EDENTATA (sensu Novacek, 1986). Thereare 24 unambiguous characters that appear at thebase of the cladogram in Figure 1 (Appendix 4). Thesecharacters cannot be polarized without comparisonwith even more remote outgroups. The six ear regioncharacters (E18, E23, E29, E57, E58 and E82) opti-mized at this node were discussed at length in Gaudin(1995). As for the other craniodental features, compar-isons with primitive eutherians (Kielan-Jaworowska,1980, 1981, 1984; Novacek, 1986; Qiang et al., 2002)suggest that most of these features are autapomor-phies of the Pholidota attributable to the extremelymodified masticatory apparatus of these animals, e.g.,the loss of teeth [1(0)], the loss of the mandibularcoronoid process [47(3)] and presence of a fused man-dibular symphysis [61(1)] with a ventral symphysialkeel [66(1)], the extreme reduction of the temporalfossa [97(2)], the zygomatic process of the squamosal[168(1)] and the nuchal crest [182(0)], as well as theloss of the postorbital process of the frontal [175(0)].Additional pholidotan autapomorphies would includethe reflexed basicranial/basifacial angle [90(2)], theelongate pterygoid hamulus [134(0)], and the fusion ofthe alisphenoid and pterygoid early in ontogeny[166(1)]. However, two of the features undergoingstate changes at this basal node may in fact representsynapomorphies of the Xenarthra plus the Palaean-odonta: the presence of a bony ridge extendinganteriorly from the dorsolateral margin of the sphe-norbital fissure/optic foramen, forming a groove intowhich those openings emerge anteriorly [162(1)]; anda reduced exposure of the orbitosphenoid [163(0)]. Inaddition, there are two features in which pholidotans,palaeanodonts and xenarthrans all show derived con-ditions that might represent synapomorphies of theentire Cohort Edentata. All edentates possess a tooth-less symphysial spout, which is fairly short in palae-
Figure 1. Phylogeny of the Tardigrada based on PAUP analysis of 286 craniodental characters, including the 85 auditoryregion characters from Gaudin (1995), in 33 extinct and extant sloth genera. Characters are polarized via comparison with13 genera from the successive outgroups Vermilingua, Cingulata, Palaeanodonta and Pholidota. All character state changesare weighted equally in this analysis. The tree represents a strict consensus of two MPT (TL = 1936 steps; CI = 0.370;RI = 0.645). Nodes are numbered in bold type. See text and Appendix 4 for listing of characters at each node. Only thosenodes included in Appendix 4 are labelled. The first number immediately underneath each node label represents a boot-strap value; the second number a branch support value. Calculation of these values is discussed in Materials and Methods.Extant taxa are written in all-capital letters. The clade illustrated with dark grey lines represents the family Megalony-chidae. The clade illustrated with single-dashed black lines represents the family Nothrotheriidae; that with single-dasheddark grey lines the family Megatheriidae; that with double-dashed black lines the family Mylodontidae.
anodonts and at the base of Xenarthra [68(0)] butmoderately elongate in pangolins [68(1)]. Additionally,the frontal/parietal suture, which in primitive euthe-rians lies well anterior to the glenoid (Kielan-Jaworowska, 1981; Novacek, 1986), is displaced pos-teriorly in edentates. It is situated at the level of the
anterior edge of the glenoid in palaeanodonts and atthe base of Xenarthra [172(1)], and well posterior tothe anterior glenoid edge in Pholidota [172(2)].
Node 2. ORDER XENARTHRA. The xenarthrans arediagnosed by ten craniodental characters on the tree
Figure 2. Phylogeny of the Tardigrada based on PAUP analysis of 286 craniodental characters, including the 85 auditoryregion characters from Gaudin (1995), in 33 extinct and extant sloth genera. This tree represents a strict consensus of allMPT obtained in the present study under various weighting and outgroup schemes (see Materials and Methods and Resultsfor a discussion). Extant taxa are written in all-capital letters. The clade illustrated with dark grey lines represents thefamily Megalonychidae. The clade illustrated with single-dashed black lines represents the family Nothrotheriidae; thatwith single-dashed dark grey lines the family Megatheriidae; that with double-dashed black lines the family Mylodontidae.
in Figure 1, only three of which are unambiguouslyassigned to this node: septomaxilla present [118(1)],facial exposure of lacrimal larger than orbital expo-sure [140(1)] and a character from Gaudin (1995) – thepresence of the pterygoid in the bony wall of the tym-panic cavity [E40(1)]. However, two of the sevenambiguous characters are unique to this node. I defineunique binary characters as those having a CI = 1.0.In the case of multistate characters, I employ a restric-tive definition. Unique character states are those thatare synapomorphies of a particular clade, occur in allmembers of that clade and are not found in taxa out-side of that clade. The two unique xenarthran charac-ters are a toothrow in which the more posterior upperteeth slant labially and the posterior lower teeth areinclined lingually [5(1)], and an ossified larynx [80(1)].
The order Xenarthra has traditionally been diag-nosed primarily on the basis of postcranial and dentalfeatures (Weber, 1928; Scott, 1937; Winge, 1941; Hoff-stetter, 1958, 1982). As both the present study andthat of Gaudin (1995) clearly attest, it is difficult tofind cranial features that are unique to this group.
Node 3. CINGULATA and Node 4. PILOSA. Thedichotomous division of xenarthrans into an armouredclade Cingulata and a ‘hairy’ clade Pilosa is an a prioriconstraint of all PAUP analyses performed in thepresent study (following Flower, 1882; Engelmann,1985; most recent molecular studies, including VanDijk et al., 1999; Delsuc et al., 2001; Greenwood et al.,2001). However, the results of the present study pro-vide additional morphological characters to diagnosethese clades. Thirteen unambiguous character statechanges are assigned to the base of Cingulata(Appendix 4), including four characters of the auditoryregion [E12(1), E35(2), E41(2) and E77(0)] discussedin Gaudin (1995). Several of the features optimized atthis node, e.g. the presence of a sagittal crest [93(2)], afirmly sutured premaxilla [113(0)] and a well-devel-oped postorbital constriction [177(1)], are in all likeli-hood primitive features of cingulate cranial anatomy,but are optimized here because of the use of palaean-odonts and pangolins as outgroups. Both palaean-odonts and pangolins are fairly derived in theircranial anatomy relative to basal eutherians. Never-theless, a number of features appearing at this nodeare probably derived diagnostic features of Cingulata.These would include a non-recurved C1 [26(0)], poste-rior teeth covered laterally by the ascending ramus ofthe mandible [40(2)], nearly vertical posterior edge ofcondylar process of mandible [52(1)], anterior edge ofnasal evenly convex [102(1)] and occipital condylespositioned immediately posterior to condyloid foram-ina [194(0)].
There are ten unequivocal craniodental featuresthat unite anteaters and sloths to the exclusion of
cingulates, with eight additional features equivo-cally assigned to this node (Appendix 4). Among theunequivocal characters are four auditory regioncharacters [E50(1), E63(1), E79(1) and E84(0)]discussed by Gaudin (1995). The remaining sixunequivocal features comprise an angular process ofthe mandible with a medially inflected tip [50(1)],lateral edges of the mandibular spout that are con-vergent anteriorly in dorsal view [72(1)], a nasotur-binal and maxilloturbinal of subequal length[119(1)], a loosely attached jugal [155(1)], a separateforamen rotundum [159(1), Fig. 3], and a nuchalcrest that does not overhang the occiput posteriorly[184(0), Fig. 3].
Node 5. VERMILINGUA. As was the case with thePilosa and Cingulata, the monophyly of the Vermilin-gua is an a priori constraint of all PAUP analysesconducted in the present study. Moreover, the rela-tionships among the various anteater genera havebeen constrained according to the phylogeneticresults of Gaudin & Branham (1998), who consideredthe matter of anteater relationships in some detail.However, the results of the present study highlightthe distinctiveness of the anteater skull, and providesome additional potential synapomorphies for one ofits more weakly supported subclades. Vermilinguanmonophyly is corroborated by at least 27 unequivocalcraniodental synapomorphies, among which five arefeatures of the auditory region. Of these, four of theear region characters and 13 of the remaining char-acters are unique to anteaters, with another seven[1(0), 54(2), 97(2), 99(0), 162(0), 166(1), 175(0)] foundonly in anteaters and pangolins. The unique anteatercharacters are: inferior edge of mandible concave inlateral view [38(0)], mandibular condyle hooks later-ally in dorsal view [57(0)], mandibular symphysisvery short, <10% of MML [62(0)], mandibular sym-physis anteroventrally inclined [64(1)], basicranial/basifacial axis markedly concave in lateral view[90(0)], palate elongate and narrow, widened at baseof zygomatic processes of maxilla [122(0)], palateelongated posteriorly to level of tympanic cavity[124(2) and 127(3)], pterygoids exposed in hard pal-ate [128(1)], no free-standing descending lamina ofpterygoid [135(0)], infraorbital foramina exposed inventral view [157(0)] and zygomatic process of squa-mosal strongly reduced, £5% BNL [168(0)]. The fourunique characters of the anteater auditory region[E40(2), E60(1), E62(0) and E69(0)] are discussed byGaudin (1995).
As noted by Gaudin & Branham (1998), the pur-ported sister-group relationship between the extantpygmy anteater Cyclopes and the extinct Pliocenegenus Palaeomyrmidon in a subfamily Cyclopinae isthe most weakly supported of their phylogenetic
hypotheses concerning relationships within Vermilin-gua. However, the present study does provide at leasttwo unambiguous characters not employed by Gaudin& Branham (1998) that might serve to diagnose the
Cyclopinae. These are a short and wide skull, withmaximum braincase width ≥40% of BNL [82(4)], and ahorizontal or slightly dorsally inclined zygomatic pro-cess of the squamosal [169(1)].
Node 6. TARDIGRADA. The monophyly of the livingand extinct sloths is an a priori assumption of allPAUP analyses conducted in the present study.Despite this, the number of characters that serve todiagnose this clade is impressive, highlighting the dis-tinctive nature of sloth craniodental anatomy. Node 6is diagnosed by 60 characters, including 31 unambig-uous and 29 ambiguous characters (Appendix 4). Atleast ten of these 60 features are completely unique tosloths. Moreover, eight of the 29 ambiguous charactersare dental characters, and are ambiguous onlybecause the absence of teeth in the Vermilinguarequires that these characters be coded as unknown inthe closest outgroup to Tardigrada. Seven of thenine unambiguous auditory synapomorphies [E4(1),E16(2), E23(1), E32(1), E33(0), E68(1), E78(1)] of thisnode are discussed in Gaudin (1995). Two additionalauditory features that were equivocal synapomorphiesof the Tardigrada in Gaudin (1995) are unequivocallyassigned to this group in the present study: the max-imum ventral extent of the entotympanic and ecto-tympanic roughly equivalent [E17(1)], and anentotympanic that forms the lateral wall and roof ofthe sulcus for the internal carotid artery, and has amedial ridge forming at least part of the medial wall ofthe sulcus [E26(2), Fig. 3]. Features of the skull anddentition that are unique to sloths include: teeth char-acterized by a large core of well-vascularized modifiedorthodentine [9(2), Fig. 3], presence of a posteriorexternal opening of the mandibular canal near thejunction of the ascending and horizontal rami of themandible [74(1), Fig. 3], presence of a large posteriorlyor posteroventrally directed process on the proximalend of the stylohyal [79(1)], presence of a rugose pal-ate, marked by numerous pits and grooves [123(1)],presence of a large pterygoid exposure in the roof ofthe nasopharynx [131(1), Fig. 3], presence of a broad,deep descending lamina of the pterygoid, typicallywith a semicircular ventral margin [135(2), Fig. 3],
optic foramen opening into sphenorbital canal, the twoforamina sharing a common external aperture [160(1),Fig. 3], and three of the ear region characters [E19(1),E51(2) and E68(1)] described in Gaudin (1995).
BRADYPUS. As was the case in Gaudin (1995), Brady-pus (Fig. 3) is assigned a position as the sister-group toall other sloths in the present study. The node (Node 7)linking all ground sloths plus Choloepus, designatedhere the Eutardigrada [a modification of a term coinedby Gaudin (1995) to refer to a clade including mylodon-tids, megatheriids and megalonychids + Choloepusbut excluding nothrotheres and Bradypus] is sup-ported by as many as 28 characters, of which 23 areassigned to this node unequivocally (Appendix 4;Figs 4, 7). Ear region characters account for 12 of theseunequivocal synapomorphies. Gaudin (1995) arguedthat many of the ear region features that characterizeall sloths except Bradypus may be lacking in the lattergenus owing to its neotenic retention of primitive con-ditions. However, it seems unlikely that the sameargument can be made for most of the remaining 11craniodental synapomorphies that serve to diagnoseNode 7.
Of the 51 unambiguous and ambiguous charactersthat are autapomorphic for Bradypus in Appendix 4,22 are convergent on Planops, Megatherium, Eremoth-erium or some combination of these genera. This isconsistent with the suggestion of Guth (1961) andWebb (1985) that Bradypus may be closely related tomegatheriids.
Node 8. MYLODONTIDAE. The monophyly of themylodontids is much less strongly corroborated by thepresent study than was the case in Gaudin (1995).Node 8 is diagnosed by only six characters, five ofwhich are unambiguously assigned to this node(Appendix 4). Unambiguous mylodontid synapomor-phies include loss of the ridge on the anteromedial
Figure 3. Skull and lower jaw of Bradypus. A, skull and lower jaw shown in left lateral view. B, skull shown in ventralview. Abbreviations: ap, angular process; as, alisphenoid; bo, basioccipital; bs, basisphenoid; C1, first upper ‘caniniform’tooth; cd, condyloid process; cf, condyloid foramen; cr, coronoid process; e, entotympanic; et, opening for the eustachian tube;f, frontal; fm, foramen magnum; fo, foramen ovale; fr, foramen rotundum; g, glenoid fossa; ic, internal carotid artery fora-men; if, infraorbital foramen; j, jugal; l, lacrimal; lf, lacrimal foramen; m, maxilla; M1; first upper ‘molariform’ tooth; n,nasal; nc, nuchal crest; o, occiput; oc, occipital condyle; of-sf, confluent optic foramen and sphenorbital fissure; p, parietal;pa, palatine; plf, posterior lacerate foramen; pm, premaxilla; pmc, posterior external opening of mandibular canal; pt, ptery-goid; so, supraoccipital; spf, sphenopalatine and posterior palatine foramina; sq, squamosal; t, tympanic; v, vomer. Char-acters and states illustrated: 9(2), teeth with large, well-vascularized modified orthodentine core; 42(2), condyle closer tocoronoid process than angle; 74(1), posterior external opening of mandibular canal present; 127(0), palate ends at level ofsphenopalatine foramen; 131(1), pterygoid with large exposure in roof of nasopharynx; 135(2), large, broad descending lam-ina of pterygoid present; 145(2), jugal with elongate ascending and descending processes; 159(1), separate foramen rotun-dum; 160(1), optic foramen shares common external aperture with sphenorbital fissure; 184(0), nuchal crest in line withposterior surface of occiput, does not overhang occiput posteriorly; E26(2), entotympanic forms roof and lateral and medialwalls of internal carotid artery sulcus. [Modified from Naples (1982).]
edge of the mandibular coronoid process [46(0)], thepresence of a low mandibular condyle that lies at orjust above the level of the toothrow [53(1), Fig. 4], thepresence of a parallel-sided mandibular spout [72(0)]
and a reduction in the length of the zygomatic processof the squamosal [168(1), Fig. 4]. The fact that thisnode is not more strongly supported may be in largepart attributable to the peculiar morphology of the
enigmatic Miocene mylodontid Octomylodon, whichlacks a number of the morphological features thathave traditionally been considered characteristic ofmylodontids. Some of these traditional mylodontidfeatures do appear at the next higher node, Node 9(Appendix 4), and would include an M4 with a T-shaped cross-section [35(7), a unique though equivo-cally assigned character state; Fig. 4], a dorsoven-trally short, anteroposteriorly broad mandibularcoronoid process [47(2), Fig. 4], and an anteroposteri-orly short, dorsoventrally deep angular process of themandible [48(0), Fig. 4]. Moreover, seven of the 13auditory synapomorphies of Gaudin (1995) alsoappear at Node 9, although all are equivocal featuresof this node owing to the lack of information about theauditory anatomy of Octomylodon. Node 9 is a some-what stronger node than Node 8. It is supported by25 characters, but only seven of these are assignedunambiguously.
The Santacrucian mylodontid Nematherium (Fig. 4)is placed as the sister group to all remaining mylodon-tids (Node 10), a position contrary to numerous recentphylogenetic studies (Patterson & Pascual, 1968,1972; Engelmann, 1985; Hirschfeld, 1985; McDonald,1987; McKenna & Bell, 1997; McDonald & Perea,2002) that include Nematherium in the subfamilyScelidotheriinae. There are a number of dental conver-gences between Nematherium and the scelidotheri-ines [13(2), 16(2), 30(4), and also the shape of m3,character 36(5), which is optimized here as a synapo-morphy of Node 9 that changes state again at Node12]. However, the two scelidotheriine genera share 13unambiguous synapomorphies [51(0), 82(0), 87(1),117(1), 137(1), 152(0), 172(2) and 174(3) – see Node 10,Appendix 4; Fig. 5] with members of the subfamilyMylodontinae exclusive of Nematherium, includingthree of the mylodontid auditory synapomorphies[E33(1), E58(3) and E62(2), as well as the equivocalfeature E45(0)] listed by Gaudin (1995).
The monophyly of the subfamily Scelidotheriinae(Node 11), represented by the Plio-Pleistocene taxaCatonyx and Scelidotherium (Fig. 5), is robustly sup-ported by as many as 24 unambiguous synapomor-phies (Appendix 4), six of which are unique to thisclade. Unique scelidothere characters include charac-
ters relating to the shape of the molariform teeth, withM1 lobate, its transverse width greater than itsanteroposterior length [31(3), Fig. 5], m1 irregularlylobate and elongate anterolabially to posterolingually,compressed perpendicular to its long axis [32(4)], M2and M3 lobate, their transverse width greater thantheir anteroposterior length [33(5), Fig. 5], and m2irregularly lobate and elongate anterolabially to pos-terolingually, compressed perpendicular to its longaxis [34(5)]. In addition, the maxilla is elevated for thedental alveoli only in its middle section, coincidentwith the molariform row [105(1), Fig. 5], and theanteroposterior profile of the palate is ventrally con-vex [121(5), Fig. 5].
The monophyly of the subfamily Mylodontinae(Node 12) receives less robust support. It is diagnosedby as many as 11 characters, ten of which are unequiv-ocally assigned to this node (Appendix 4; Fig. 6). Aswas the case in Gaudin (1995), the La Ventan slothPseudoprepotherium is the most basal mylodontine,separated from the remaining taxa by eight unambig-uous synapomorphies (Node 13, Appendix 4; Fig. 6).The Deseadan genus Octodontotherium, the oldestsloth known from reasonably complete skeletalremains, is the next most basal taxon, again as inGaudin (1995). The remaining mylodontines areunited by as many as 21 characters (Node 14). Onlyeight of these characters are assigned to this nodeunequivocally owing to the large amount of missingdata for Octodontotherium, but two of these eightunambiguous synapomorphies are unique to thisclade: long axis of posterior molariform teeth obliqueto long axis of skull [16(1), Fig. 6], and free end of thezygomatic process of the squamosal broad and some-what flattened [171(1), Fig. 6]. Among the remainingmylodontines, the North American Miocene genusThinobadistes and the gigantic South American Pleis-tocene genus Lestodon form a crown clade. This groupmay be termed Lestodontini following Webb (1989),who has suggested joining these two genera into a dis-tinct mylodontid subfamily Lestodontinae. The mono-phyly of lestodontines is strongly supported in thepresent study by 11 unambiguous synapomorphies(Node 18, Appendix 4). The remaining mylodontinesform a series of sister taxa that are progressively fur-
Figure 4. Skull and lower jaw of Nematherium. A, skull and lower jaw shown in left lateral view. B, skull shown in ventralview. Characters and states illustrated: 35(7), M4 with T-shaped cross-section; 47(2), coronoid process short and broad;48(0), angular process short and deep; 52(1), posterior edge of condyloid process nearly vertical; 53(1), condyle positioned ator just above the level of the toothrow; 56(2), condylar surface inclined posteroventrally in lateral view; 68(1), symphysealspout of moderate length; 124(1), postpalatine shelf extends along inner edge of descending laminae of pterygoids to roughlytheir midpoint; 140(0), orbital portion of lacrimal larger than facial exposure; 145(3), jugal with ascending, descending andmiddle processes; 168(1), zygomatic process of squamosal of moderate length; 169(1), zygomatic process of squamosal hor-izontal or inclined slightly dorsad in lateral view. Drawings based upon the skull of Nematherium sp. (YPM-PU 18009) andthe mandibles of Nematherium angulatum (YPM-PU 15521 & YPM-PU 15530).
Figure 5. Skull and lower jaw of Scelidotherium. A, skull and lower jaw shown in left lateral view. B, skull shown in ven-tral view. Characters and states illustrated: 14(2), c1 neither smallest nor largest tooth; 17(3), molariforms with flatocclusal surface; 20(4), C1/c1 with flat occlusal surface; 31(3), M1 lobate, its transverse width greater than its anteropos-terior length; 33(5), M2 & M3 lobate, their transverse width greater than their anteroposterior length; 51(0), short condy-loid process; 85(1), snout moderately elongate, <40%, ≥27% of BNL; 87(1), snout elevated anteriorly; 105(1), maxillaelevated for dental alveoli only in the middle, coincident with molariform row; 107(0), dorsal contact of maxilla and frontalexcluded by nasal/lacrimal contact; 111(0), medial palatal process of maxilla anterior to lateral palatal process; 113(0), pre-maxilla tightly sutured to skull; 117(1), incisive foramen slit-like, hidden in ventral view by medial palatal process of max-illa; 121(5), palatal profile evenly convex in lateral view; 137(1), pterygoid inflated at base; 152(0), descending process ofjugal wide; 172(2), frontal/parietal suture well posterior to front of glenoid; 189(1), posterior edge of occipital condyles at oranterior to posterior edge of foramen magnum. [Modified from Owen (1857).]
Figure 6. Skull and lower jaw of Paramylodon harlani. A, skull and lower jaw shown in left lateral view. B, skull shown inventral view. Characters and states illustrated: 16(1), long axis of posterior molariform teeth oblique to long axis of skull;65(0), mandibular symphysis with convex profile in lateral view; 91(1), profile of nasal region and braincase roughly hor-izontal in lateral view, but nasal region depressed relative to braincase; 95(1), complete zygomatic arch; 111(0), medial pal-atal process of maxilla anterior to lateral palatal process; 115(1), palatal process of premaxilla V-shaped, wide; 149(0), wideascending process of jugal; 171(1), free end of zygomatic process of squamosal broad and somewhat flattened. [Modifiedfrom Stock (1925).]
ther removed from the Lestodontini. The South Amer-ican Plio-Pleistocene genus Glossotherium is theclosest relative of the lestodontines, followed by thelate Miocene South American genus Pleurolestodon,the North American Pleistocene genus Paramylodon(Fig. 6) and the South American Pleistocene genusMylodon (Figs 1, 2). This scheme of mylodontine rela-tionships is consistent with that proposed by Gaudin(1995), but differs significantly from those proposed inother recent works (e.g. Engelmann, 1985; Hirschfeld,1985; McDonald, 1987; Webb, 1989; McKenna & Bell,1997).
Node 19. MEGATHERIOIDEA. Members of the fami-lies Megatheriidae, Megalonychidae, Nothrotheri-idae as well as a number of enigmatic, basal taxafrom Santacrucian age deposits of Patagonia areunited in a single monophyletic clade Megatherioideain the present study. Patterson & Pascual (1968,1972) and Patterson et al. (1992) previously proposeduniting these three families in a superfamily Megalo-nychoidea. However, their Megalonychoidea alsoincluded the extant three-toed sloth Bradypus as amember of the Megatheriidae. Node 19 in the presentstudy more closely resembles the superfamily Mega-therioidea of McKenna & Bell (1997) in that itexcludes Bradypus. In addition, the term ‘Megatheri-oidea’ apparently enjoys priority (McKenna & Bell,1997). There are as many as 31 characters that maysupport the monophyly of megatherioids, althoughonly 15 are assigned to this node unambiguously(Appendix 4). These unambiguous synapomorphiesinclude six dental characters [thin layer of orthoden-tine, orthodentine thinner than outer layer of cemen-tum, 10(1), Fig. 7; occlusal surface of the molariformswith strong transverse crests, 17(2), Fig. 7; M4 curvesanteriorly in lateral view, 27(1), Fig. 7; M1 circular incross-section, 31(0); m2 rectangular in cross-section,34(1); and m3 circular in cross-section, 36(1)], andeight non-auditory skull characters [weak fossa onlateral surface of mandible posterior to c1, 76(1),Fig. 7; orbit displaced ventrally, lying at or belowlevel of toothrow, 84(1); fossa behind root of zygoma,anterodorsal to the mastoid process, 96(1); weaklydeveloped buccinator fossa, 106(1), Fig. 7; maxillawith a fossa on its lateral surface immediately poste-rior to the root of M4, 112(1); palate short, uniformlywide, 122(4), Fig. 7; lacrimal eminence present143(1), Fig. 7; and jugal and lacrimal overlap facialportion of maxilla anteriorly, 147(1), Fig. 7]. There isonly one unambiguous synapomorphy of this nodefrom the auditory region [the presence of ananteroventral process on the entotympanic, E21(1),Fig. 7], which is perhaps not surprising, as Pattersonet al. (1992) found only a single feature that mightdiagnose megatherioids (= megalonychoids), and in
Gaudin (1995) the Nothrotheriidae, Megatheriidaeand Megalonychidae formed successive sister groupsto the Mylodontidae.
At the base of the Megatherioidea are four San-tacrucian genera of uncertain affinities. The moreprimitive of these genera are two small bodied forms,both characterized by a poorly developed C1/M1diastema, Pelecyodon and Schismotherium. As notedin the preceding discussion, the relationship of thesetwo genera to one another and to other megatherioidsvaries according the outgroup and character weight-ing scheme employed (Figs 1, 2), yet they are sepa-rated from the remaining megatherioids in all trees.The node including all megatherioids except Pelecyo-don and Schismotherium is diagnosed by at least nineunambiguous synapomorphies (Node 20, Appendix 4).The Santacrucian genera Analcimorphus and Hapal-ops (Fig. 7) are shown in Fig. 1 as successive sistertaxa to a clade uniting the families Megatheriidae andNothrotheriidae. However, as noted in preceding dis-cussions, under different character weighting schemesthese same two Santacrucian genera form successivesister taxa to the Megalonychidae. Their relationshipto other megatherioids is probably best representedas uncertain, as illustrated in Figure 2, althoughthey are clearly more derived than Pelecyodon andSchismotherium.
Node 23. MEGATHERIA. The results of the presentstudy consistently link late Miocene – Pleistocenenothrotheres to megatheriids. In most previous stud-ies in which these groups have been allied (Patterson& Pascual, 1968, 1972; Paula Couto, 1971, 1979;Engelmann, 1985; Perea, 1988; Patterson et al., 1992),the nothrotheres have been considered a subfamily ofthe family Megatheriidae, although McKenna & Bell(1997) consider the late Miocene – Pleistocene generaas a tribe within a subfamily Megatheriinae. Otherworkers have suggested, however, that the Plio-Pleis-tocene nothrotheres warrant recognition as a separatefamily (Gaudin, 1994; McDonald, 1994; Gaudin & DeIuliis, 1999). Hence I propose modifying McKenna &Bell’s term ‘Megatheria’ to refer to the clade encom-passing both families. The evidence for a close rela-tionship between the two megatherian families is notoverwhelming. Node 23 is supported by four unambig-uous synapomorphies (Appendix 4): snout narrow,width at midpoint <20% of BNL [86(0), Figs 8, 9], eth-moid unexposed in roof of nasopharynx, covered ven-trally by fusion of vomerine wings [199(1), Fig. 9(B)],medial expansion of entotympanic dorsal to floor ofbasicranium [E30(0)] and stylomastoid foramen con-nected to nearby ventral opening of canal for occipitalartery by a strong groove [E59(3)]. Within Megatheria,the monophyly of both the families Megatheriidae andNothrotheriidae is supported.
Figure 7. Skull and lower jaw of Hapalops. A, skull and lower jaw shown in left lateral view. B, skull shown in ventralview. Characters and states illustrated: 3(2), C1 & c1 slightly depressed relative to molariforms in lateral view; 6(1), elon-gate diastema present; 10(1), orthodentine forms thin layer, thinner than outer layer of cementum; 17(2), occlusal surfaceof molariforms with strong transverse crests; 27(1), M4 curved anteriorly in lateral view; 31(2), M1 rectangular in cross-sec-tion; 56(2), condylar surface inclined posteroventrally in lateral view; 65(2), mandibular symphysis with concave profile inlateral view; 68(1), symphyseal spout of moderate length; 73(0), symphyseal spout horizontal in lateral view; 76(1), man-dible with weak fossa posterior to c1; 106(1), buccinator fossa weakly developed; 122(4), palate short, uniformly wide;140(0), orbital portion of lacrimal larger than facial exposure; 143(1), lacrimal eminence present; 147(1), jugal and lacrimaloverlap facial portion of maxilla in lateral view; 169(1), zygomatic process of squamosal horizontal or inclined slightly dor-sad in lateral view; 184(1), nuchal crest overhangs occiput posteriorly; 195(0), occipital condyles elongated anteroposteri-orly in ventral view; E21(1), anteroventral process of entotympanic present. [Modified from Scott (1903–4).]
Node 24. MEGATHERIIDAE. The Megatheriidaeincludes not only the Plio-Pleistocene taxa Megathe-rium and Eremotherium (Fig. 8) but also the Santacru-cian taxon Planops, a result concordant with thehypotheses of Scott (1903–4), De Iuliis (1994), Gaudin(1995) and others. The Megatheriidae is diagnosed byeight unequivocal synapomorphies (Appendix 4).These include an elongate mandibular symphysis andspout [62(3) and 68(2), Fig. 8], the lack of a cleardemarcation between the posterior end of the symphy-sis and the ventral edge of the horizontal ramus of themandible in lateral view [69(1), Fig. 8], the anteriordisplacement of the frontal postorbital process so thatit lies at the same level as the maxillary foramen[178(1)] and a condylar foramen that is reduced in size[187(0), Fig. 8]. In addition, the three auditory regionsynapomorphies of the Megatheriidae identified byGaudin (1995) also serve to diagnose this node. The sis-ter-group relationship between Eremotherium andMegatherium (subfamily Megatheriinae, Node 25) isone of the most robust nodes on the cladogram. It isdiagnosed by 48 unambiguous synapomorphies, amongthem nine unique to the megatheriines (Appendix 4).Of these nine, seven are dental characters [15(0), 29(4),30(5), 32(2), 33(3), 34(3), 36(2); Fig. 8]. The other twoinclude a posterior external opening of the mandibularcanal that faces anteromedial, lying on the internalsurface of the ascending ramus [75(2)], and a lacrimalforamen that opens into a ventrally directed canal onthe surface of the lacrimal bone [144(2), Fig. 8].
Node 26. NOTHROTHERIIDAE. The monophyly ofthe Nothrotheriidae is diagnosed by as many as 14synapomorphies, nine of which are assigned to thisnode unambiguously (Appendix 4). Two of theseunambiguous synapomorphies are completely uniqueto nothrotheriids: the presence of an elongate, asym-metrical ventral keel on the vomer that extends intothe nasopharynx [200(1); Fig. 9], and the expansion ofthe vomerine wings themselves into the roof of thenasopharynx, so that they cover the presphenoid andmuch of the basisphenoid ventrally [201(1), Fig. 9].
Additional nothrotheriid synapomorphies include theloss of the ridge on the anteromedial edge of the man-dibular coronoid process [46(0) – a convergence on theMylodontidae], absence of a sagittal crest or closelyconvergent temporal lines [93(0)], contact between thepterygoid and vomer [132(1), Fig. 9], and occipitalcondyles that lie at or anterior to the posterior edge ofthe foramen magnum [189(1), Fig. 9]. Two of the threeauditory synapomorphies used by Gaudin (1995) tolink the nothrotheriids Pronothrotherium andNothrotheriops also appear at this node, one as anunequivocal character [E14(1)] and one as an equivo-cal synapomorphy [E85(0)]. A feature touted by Gau-din (1995) as unique to Pronothrotherium andNothrotheriops [E69(4)] also appears in Nothropus butnot Nothrotherium, and is assigned to Node 26 as anambiguous synapomorphy.
The interrelationships among nothrotheriids arenot completely resolved by the present study (Fig. 2).When all character state changes are weightedequally, the late Miocene/Pleistocene (Burmeister,1882; Frailey, 1986, 1995) taxon Nothropus is the sis-ter taxon to the crown clade Nothrotheriops +Nothrotherium, regardless of the outgroup schemeemployed. These results are consistent with the phy-logenetic hypotheses of de Muizon & McDonald (1995)and Gaudin & De Iuliis (1999). However, if charactersstate changes are scaled such that all characters areweighted equally, the Pliocene genus Pronothrothe-rium forms the sister taxon to the crown clade, regard-less of the outgroups employed. The crown clade itselfcomprising two Pleistocene genera, one from NorthAmerica, Nothrotheriops, and one from South Amer-ica, Nothrotherium, is robustly supported by 12 unam-biguous synapomorphies (Node 28, Appendix 4), one ofwhich, the presence of a fenestrated pterygoid bulla[138(2)], is unique to these two taxa.
Node 29. MEGALONYCHIDAE. As suggested by Hir-schfeld & Webb (1968), Frailey (1988) and McKenna &Bell (1997), the Santacrucian genus Eucholoeopsgroups not with other Santacrucian ‘nothrotheres’, but
Figure 8. Skull and lower jaw of Eremotherium. A, skull and lower jaw shown in left lateral view. B, skull shown in ventralview. Characters and states illustrated: 3(0), toothrow horizontal in lateral view; 6(0), diastema absent; 15(0), m3 smallestlower molariform; 19(0), C1/c1 with molariform morphology; 20(5), C1/c1 with transverse crests on occlusal surface; 26(0),C1 straight in lateral view; 29(4), C1 with square cross-section; 33(3), M2 & M3 with square cross-section; 48(1), angularprocess of intermediate development, ratio of length to depth >1.0, <1.25; 68(2), symphyseal spout elongate; 69(1), no cleardemarcation between symphysis and horizontal ramus of mandible in lateral view; 86(0), snout narrow; 95(1), completezygomatic arch; 136(1), posterior edge of pterygoid straight, nearly vertical; 144(2), lacrimal foramen opens into ventrallydirected canal; 159(0), foramen rotundum confluent with sphenorbital fissure; 161(1), sphenopalatine foramen situated incommon fossa with orbital foramina; 178(2), postorbital process lies anterior to maxillary foramen; 187(0), small condyloidforamen; 188(1), occipital condyles situated dorsal to toothrow in lateral view. Skull drawings based on Paula Couto (1954),Cartelle & Bohórquez (1982) and Patterson et al. (1992). Lower jaw based on De Iuliis (1994) and specimen of Eremothe-rium laurillardi (FMNH P26962).
Figure 9. Skull and lower jaw of Nothrotheriops and Pronothrotherium. A, skull and lower jaw of Nothrotheriops shown inleft lateral view. B, skull of Pronothrotherium shown in ventral view. Characters and states illustrated: 2(4), 4/3 dental for-mula; 3(0), toothrow horizontal in lateral view; 37(2), mandible of moderate depth, >20%, £22.5% of MML; 56(1), condylarsurface nearly horizontal in lateral view; 68(2), symphyseal spout elongate; 86(0), snout narrow; 132(1), pterygoid/vomercontact; 137(2), large pterygoid sinus present; 189(1), posterior edge of occipital condyles at or anterior to posterior edge offoramen magnum; 199(1), ethmoid covered by vomer in roof of nasopharynx; 200(1), vomer with elongate, asymmetricalkeel extending posteriorly into nasopharynx; 201(1), vomer with large exposure in roof of nasopharynx, covering presphe-noid and much of basisphenoid. Drawing A modified from Stock (1925); drawing B based upon specimen of Pronothrothe-rium typicum (FMNH P14467).
rather is allied with the extant two-toed sloth Choloe-pus (Fig. 11) and extinct genera classically placed inthe family Megalonychidae. This more inclusive fam-ily Megalonychidae is diagnosed by 20 unequivocalsynapomorphies in the present study (Appendix 4),including one unique feature, the presence of a fossaon the palatal surface of the maxilla immediately pos-terior to C1 [23(1), Figs 10, 11]. The family is unitedlargely by features associated with the caniniformfirst upper and lower teeth [13(1), 14(1), 20(0), 24(1),25(1), 29(1), 30(1); Figs 10, 11] and distinctive charac-teristics of the snout [4(1), 86(2), 106(2), 122(3);Figs 10, 11]. The node uniting all megalonychidsexclusive of Eucholoeops (Node 30) is even morerobust. It is diagnosed by 26 unambiguous synapomor-phies (Appendix 4), including one unique feature [C1/c1 strongly depressed relative to the molariform teeth,
3(3); Fig. 10] and four of the seven unambiguous meg-alonychid auditory synapomorphies [E64(0), E75(1),E76(2), E85(2)] recognized by Gaudin (1995). In addi-tion, one of Gaudin’s (1995) unequivocal megalonychidsynapomorphies [E14(1)] is equivocally assigned toNode 30, whereas one of Gaudin’s (1995) equivocalmegalonychid synapomorphies [E81(0)] is an unequiv-ocal synapomorphy of Node 30.
Within the Megalonychidae, the South Americangenus Pliomorphus, the North American generaPliometanastes and Megalonyx, and Eucholoeops formsuccessive sister-taxa to a crown group including theextant genus Choloepus and the extinct sloths derivedfrom an endemic radiation of megalonychids in theWest Indies (Figs 1, 2). The crown group is diagnosedby ten unambiguous synapomorphies and an equalnumber of ambiguous features (Node 33, Appendix 4;
Figure 10. Skull and lower jaw of Acratocnus odontrigonus in left lateral view. Characters and states illustrated: 3(3), C1& c1 strongly depressed relative to molariforms in lateral view; 13(1), C1 largest upper tooth; 14(1), c1 largest lower tooth;20(0), C1/c1 with oblique, nearly vertical wear facet; 23(1), fossa on palatal surface of maxilla posterior to C1 present; 25(1),alveolus of C1/c1 projects anteriorly; 37(4), mandibular depth >25%, £27.5% of MML; 40(2), ascending ramus of mandiblecovers posterior teeth in lateral view; 48(1), angular process of intermediate development, ratio of length to depth >1.0,<1.25; 84(0), orbit in typical mammalian position in lateral view; 106(2), well-developed buccinator fossa; 152(1), descendingprocess of jugal wide at base, tapers strongly toward tip; 170(2), zygomatic process of squamosal deep; 178(2), postorbitalprocess lies anterior to maxillary foramen. Drawings based on skull (AMNH 17722) and mandibles (AMNH 17710 & AMNH17719) of Acratocnus odontrigonus.
Figure 11. Skull and lower jaw of Choloepus. A, skull and lower jaw shown in left lateral view. B, skull shown in ven-tral view. Characters and states illustrated: 4(1), left and right toothrows anteriorly divergent; 13(1), C1 largest uppertooth; 14(1), c1 largest lower tooth; 20(0), C1/c1 with oblique, nearly vertical wear facet; 23(1), fossa on palatal surfaceof maxilla posterior to C1 present; 24(1), C1/c1 displaced laterally relative to molariform toothrow; 29(1), C1 with trigo-nal cross-section; 76(2), mandible with strong fossa posterior to c1; 84(0), orbit in typical mammalian position in lateralview; 85(1), snout relatively short, <40%, ≥27% of BNL; 114(3), dorsal process of premaxilla absent; 122(3), palate elon-gate, strongly widened anteriorly; 146(1), postorbital process of jugal weak; 152(1), descending process of jugal wide atbase, tapers strongly toward tip; 169(1), zygomatic process of squamosal horizontal or inclined slightly dorsad in lat-eral view; 170(1), zygomatic process of squamosal of moderate depth; 187(0), small condyloid foramen. [Modified fromNaples (1982).]
Figs 10, 11). However, within this crown group rela-tionships are not consistently resolved. In all analy-ses, the large bodied genera Megalocnus and Parocnusare allied as sister taxa. This relationship, whichaccords with the results of Engelmann (1985), Webb& Perrigo (1985) and White & MacPhee (2001), isstrongly supported by 12 unambiguous synapomor-phies (Node 36, Appendix 4). In three of the fourweighting and ordering schemes, the genus Acrato-cnus (Fig. 10) is the sister taxon to Megalocnus +Parocnus, whereas the extant Choloepus is allied tothe diminutive extinct taxon Neocnus. However, ifpangolins and palaeanodonts are excluded from theoutgroup and characters state changes are scaled suchthat all characters are weighted equally, Choloepus isthe sister taxon to Acratocnus, and these two togetherform a sister clade to Megalocnus + Parocnus. Neocnuswould then represent the sister group to a cladeincluding the other four taxa. Choloepus and Neocnusshare six unequivocal synapomorphies (Node 34,Appendix 4). These include an anteroposteriorly elon-gate, ovate m1 [32(0)], a relatively elongate snout[85(1), Fig. 11], a horizontal or slightly dorsallyinclined zygomatic process of the squamosal [169(1) –a reversal to the primitive condition; Fig. 11], a circu-lar ectotympanic [E6(1)] and a porus acousticussituated immediately behind the glenoid [E82(0)]. Bycontrast, Choloepus and Acratocnus share only fiveunambiguous synapomorphies: anterodorsally in-clined symphyseal spout [73(1)], snout broad trans-versely, >30% BNL [86(3)], anterior palatine foraminaand grooves absent [125(0)], strongly developed pos-torbital constriction [177(1)] and dorsal edge of ento-tympanic flat [E25(2)]. Nevertheless, one of thesecharacters [125(0)] is nearly unique among sloths,found elsewhere only in Scelidotherium, and one isunique among xenarthrans in general [E25(2)]. Fur-thermore, there are seven auditory synapomorphiesshared by Acratocnus and Choloepus that cannot beassigned unequivocally to this node because of missingdata in other megalonychids. An alliance of Choloepusand Acratocnus is in accordance with previous phylo-genetic hypotheses of Patterson et al. (1992), Gaudin(1995) and White & MacPhee (2001).
DISCUSSION
The goal of the present study has been to improveunderstanding of sloth phylogenetic relationships bylooking at a broad array of characters in a wide varietyof sloth taxa, including representatives of all themajor families and subfamilies within Tardigrada.Questions of particular interest have included themonophyly or diphyly of the living tree sloths and theclosely related question of how the various families ofsloths are related to one another.
The results of the present study (Figs 1, 2) providerobust support for the diphyletic origin of the twoextant tree sloth genera. As in Gaudin (1995), thethree-toed sloth Bradypus (Fig. 3) is placed as the sis-ter-taxon to all remaining sloths, a clade designatedEutardigrada. Such a placement is consistent with theassignment of Bradypus to its own monotypic family,as is the case in most recent classifications (Barlow,1984; Gardner, 1993; McKenna & Bell, 1997; Eisen-berg & Redford, 1999; Feldhamer et al., 1999; Nowak,1999; Vaughan et al., 2000). The two-toed sloth Cho-loepus (Fig. 11) is allied with extinct West Indiansloths in the family Megalonychidae (Fig. 10), inaccord with the phylogenetic hypotheses of numerousprevious workers (Patterson & Pascual, 1968, 1972;Webb, 1985; Patterson et al., 1992; Gaudin, 1995;White & MacPhee, 2001) as well as most recent clas-sifications. Although none of the recent molecularanalyses of sloth relationships has managed to extractany DNA from an extinct megalonychid, it is worthnoting that none has supported a sister group rela-tionship between Bradypus and Choloepus either(Höss et al., 1996; Poinar et al., 1998; Greenwoodet al., 2001). To impose monophyly of the two extantsloth genera on the present data set as the tradition-ally defined Bradypodidae of older mammalian classi-fications (e.g. Weber, 1928; Winge, 1941; Simpson,1945; Hoffstetter, 1958) requires the addition of asmany as 25 steps to the MPT illustrated in Figure 1(Appendix 5). The results of pair-wise non-parametricTempleton tests suggest that the hypothesis of extanttree sloth monophyly can be rejected on statisticalgrounds (Appendix 5). The split between these twogenera appears to be an ancient one, dating back tothe very base of the Tardigrada, at least to the Casa-mayoran LMA (Pascual et al., 1985) at roughly 40 Myr(Kay et al., 1999; Kay et al., 2002). This is a slightlyolder divergence date than that suggested by theimmunological studies of Sarich (1985; ~35 Myr), andmuch older than the estimated divergence date in arecent molecular study (~ 18 Myr, Delsuc et al., 2001).Such an ancient divergence makes their convergentadaptations for folivory, suspensory locomotion, andeven the retention of water and growth of algae in thehair (Grassé, 1955a; Naples, 1982, 1985; Aiello, 1985;White, 1993, 1997) all the more remarkable.
Although I think the case is now clear and convinc-ing, based on the results of this and other recent stud-ies, that the tree sloths do not share a close commonancestry, the case for the remote position of Bradypuswithin sloths and the alliance of Choloepus with WestIndian megalonychids is strong but not yet ironclad.The present study corroborates Gaudin’s (1995) place-ment of Bradypus as the sister group to the groundsloths plus Choloepus, the Eutardigrada. The newcraniodental characters that support this relationship
are particularly significant. Gaudin (1995) suggestedthat the basal position of Bradypus might be an arte-fact attributable to the neotenic retention of primitivefeatures in this genus. Such an argument is muchmore difficult to make in the case of the craniodentalfeatures that separate Bradypus from Eutardigrada(Appendix 4). There is no evidence that any of thesenon-auditory characters are absent from Bradypusowing to developmental heterochronies. Moreover, thebasal node for Eutardigrada (Node 7) has high boot-strap and branch support values (Fig. 1). In both Gau-din (1995) and the present study, roughly half theautapomorphic features that characterize Bradypusare convergent on megatheriids, suggesting that thetwo might be closely related, as asserted by Guth(1961) and Webb (1985). Although such an arrange-ment is as many as 12 steps longer than the MPT inthe present study, statistical tests do not support theoutright rejection of this hypothesis (Appendix 5).Additional postcranial evidence might serve to clarifythis issue. Nevertheless, given the available evidence,the placement of Bradypus outside of the remainingsloths seems by far the most plausible allocation.
Choloepus shares with other megalonychids charac-teristic morphologies of the first upper and lower(‘caniniform’) teeth, the snout, the glenoid and man-dibular condyle, and the overall structure of the skull(see preceding discussion, also Nodes 29 & 30,Appendix 4; Figs 10, 11). The node representing thecommon ancestor of all megalonychids (Node 29) andthe node including all megalonychids except for thebasal Santacrucian taxon Eucholoeops (Node 30) areboth quite robust, with high branch support and boot-strap values (Fig. 1). Choloepus shows a number ofstriking resemblances to West Indian megalonychidsin its auditory anatomy (Node 33, Appendix 4), asnoted by Patterson et al. (1992) and Gaudin (1995),although this node has neither high branch supportnor high bootstrap values (Fig. 1). Yet Choloepus is inmany ways an atypical megalonychid, as evidenced bythe fact that it is diagnosed by as many as 59 autapo-morphies in the present study, including 30 unambig-uously assigned features. Of these 59 unambiguousand ambiguous autapomorphies, ten are convergenteither on Mylodontidae as a whole (Node 8) or on oneof the basal nodes of that clade [Node 9, Node 10 orNode 12 (= Mylodontinae)], including a straight ven-tral edge on the mandible [38(1)], a mandibularcondyle that lies at the level of the toothrow [53(1)]and a relatively short zygomatic process of the squa-mosal [168(1)]. The hypothesis of Guth (1961) andGreenwood et al. (2001) that Choloepus is closelyrelated to the Mylodontidae cannot be rejected on sta-tistical grounds (Appendix 5). However, the bulk ofevidence currently available weighs heavily againstsuch an arrangement.
The position of Choloepus within the Megalony-chidae is not unambiguously resolved by the presentanalysis. It is joined to members of the Antilleanground sloth radiation in all MPT (Figs 1, 2). Withinthis group it is allied either to the genus Neocnus, thesmallest of the extinct West Indian sloths, or to thePuerto Rican and Hispaniolan genus Acratocnus. Theformer relationship is supported by more unambigu-ous synapomorphies (six vs. five, see preceding discus-sion), and is preferred under three of four outgroupand character weighting schemes employed in thisstudy. However, a close common ancestry of Choloepusand Acratocnus accords better with other recent anal-yses of megalonychid relationships (Patterson et al.,1992; Gaudin, 1995; White & MacPhee, 2001), and issupported by a large number of auditory charactersthat are equivocal synapomorphies because of missingdata in other Antillean megalonychids, includingNeocnus.
The results of the present study provide a consistentresolution to the question of how the major families ofsloths are related to one another. In all MPT gener-ated under various outgroup and character weightingschemes, the family Mylodontidae is the sister groupto a monophyletic clade including all nothrotheriid,megatheriid and megalonychid sloths, the Mega-therioidea. Within Megatherioidea, the Nothrotheri-idae and Megatheriidae form a clade that in turnshares a common ancestor with a monophyleticMegalonychidae.
This fundamental dichotomy among extinct slothsinto mylodontids on the one hand and a clade includ-ing nothrotheriids, megatheriids and megalonychidson the other is a feature of many previous analyses ofsloth relationships (e.g., Winge, 1941; Patterson &Pascual, 1968, 1972; Webb, 1985; Patterson et al.,1992; McKenna & Bell, 1997). Patterson & Pascual(1968, 1972) refer to the latter grouping as ‘Megalony-choidea’, but the contents of the clade in the presentstudy match almost exactly that of the ‘Megatherio-idea’ of McKenna & Bell (1997), and hence the latterterm is preferred. Megatherioid monophyly, which wasnot corroborated by Gaudin (1995), receives fairlyrobust support in the present analysis. The node(Node 19) has a high branch support value and is diag-nosed by at least 15 unequivocal synapomorphies(Appendix 4; Fig. 7), although it receives relativelylow bootstrap support (Fig. 1). The rigorous establish-ment of megatherioid monophyly suggests that dentalresemblances long noted among the nothrotheriids,megatheriids and megalonychids, including molari-form teeth with quadrate or ovate cross-sections andocclusal surfaces characterized by transverse crests(Hoffstetter, 1958; Figs 7–10), are in fact derived. Thepresent study provides additional cranial featuresthat diagnose megatherioids (Appendix 4), including a
ventrally displaced orbit [84(1)], a large lacrimal emi-nence [143(1), Fig. 7], a jugal and lacrimal that over-lap the facial process of the maxilla laterally [147(1),Fig. 7], and an anteroventral process of the entotym-panic [E21(1), Fig. 7].
Perhaps no phylogenetic problem within Tardigradahas posed more difficulty than the proper allocation ofnothrotheres. As traditionally defined (e.g. Simpson,1945; Hoffstetter, 1958; Patterson & Pascual, 1968,1972; Paula Couto, 1979), the nothrotheres haveencompassed two distinct groupings: (1) relativelysmall bodied, unspecialized sloths from the early tomiddle Miocene (Colhuehuapian and SantacrucianLMA), including the well-known genus Hapalops andits allies (Scott, 1903–4), and (2) somewhat larger, lateMiocene to Pleistocene taxa including the well-knownPleistocene genera Nothrotherium and Nothrotheriops(the ‘Shasta ground sloth’ of North America) and theirclose relatives (de Muizon & McDonald, 1995;McDonald & de Muizon, 2002). These two groupingshave been allied on the basis of characteristics, that, inthe words of De Iuliis (1994: 582) “have been inter-preted consistently as plesiomorphic for sloths”,although De Iuliis (1994) later states thatnothrotheres may be united by the shared possessionof a derived, y-shaped premaxilla (see Fig. 7B).Nothrotheres have been variously considered a sub-family of the family Megatheriidae (Patterson & Pas-cual, 1968, 1972; Paula Couto, 1971, 1979; Engelmann,1985; Perea, 1988; Patterson et al., 1992; McKenna &Bell, 1997) or a subfamily of the family Megalony-chidae (Winge, 1941; Simpson, 1945; Hoffstetter, 1958,1982; de Muizon & McDonald, 1995; McDonald & deMuizon, 2002). However, in Gaudin’s (1995) phylogenynothrotheres formed a paraphyletic stem groupbetween Bradypus and a clade including mylodontids,megalonychids and megatheriids sensu stricto The twoPlio-Pleistocene nothrotheres included in Gaudin’s(1995) analysis formed a discrete clade separate fromthe more basal Santacrucian forms, prompting Gaudin(1995: 685) to state that “a familial distinction for[Plio-Pleistocene nothrotheres] might be . . . appro-priate.” Indeed, several published abstracts (Gaudin,1994; McDonald, 1994; Gaudin & De Iuliis, 1999) for-mally advocated recognition of a separate familyNothrotheriidae to include only the late Miocene toPleistocene nothrotheres. These authors cite both themorphological distinctiveness of the group and the factthat it encompasses as much taxonomic diversity asother tardigrade families [e.g., in McKenna & Bell(1997), the Tribe Megatheriini (= Megatheriidae ofpresent study) has 12 genera, the Tribe Nothrotheriini(= Nothrotheriidae of present study) has ten genera].
The results of the present study support the desig-nation of a family Nothrotheriidae to include the lateMiocene – Pleistocene taxa Pronothrotherium, Nothro-
pus, Nothrotherium and Nothrotheriops. This clade(Node 26) is diagnosed by nine unambiguous synapo-morphies (Appendix 4) including several unique (andbizarre!) features of the vomer (Fig. 9B), and has highbootstrap and branch support values (Fig. 1). TheNothrotheriidae so defined is in turn the sister groupto the family Megatheriidae, including the Santacru-cian genus Planops and the giant Plio-Pleistocenesloths Eremotherium and Megatherium. The cladeencompassing nothrotheriids and megatheriids, herelabelled the Megatheria [a modified usage of McKenna& Bell’s (1997) term – see preceding discussion], isonly weakly supported, however. It is diagnosed byonly four unequivocal synapomorphies (Appendix 4),and the node (Node 23) is marked by weak bootstrapand branch support values (Fig. 1).
The allocation of the early to middle Miocenenothrotheres within Megatherioidea is not fullyresolved in the present study. The genera Pelecyodonand Schismotherium are the most basal megatherio-ids in all MPT (Fig. 2), but their relationship to oneanother and to the more derived taxa is not clear. Thetwo do share some derived resemblances [177(1),180(1), E26(1)], and are sister taxa in several of theMPT as suggested by Patterson et al. (1992). However,Pelecyodon is characterized by several derived fea-tures [93(2), 121(3)] not present in Schismotherium,and hence in some MPT is located one node higherthan Schismotherium. The genera Analcimorphus andHapalops are more derived than Schismotherium andPelecyodon in all MPT (Fig. 2), but whether they arepositioned as basal relatives of megalonychids ormegatherians (Fig. 1) depends upon the manner inwhich characters state changes are weighted. Ineither case, the placement of these four genera at thebase of Megatherioidea represents a novel phyloge-netic arrangement, although it is reminiscent of theresults of Gaudin (1995), in which Pelecyodon, Schis-motherium and Hapalops are part of a paraphyleticstem group for all sloths except Bradypus.
Because none of these early to middle Miocenenothrotheres is allied with the Nothrotheriidae asdefined above, the continued use of the term‘nothrotheres’ to refer to them is misleading. It istherefore suggested that they be referred to simply asbasal megatherioids. The designation of these taxa asbasal megatherioids accords well not only with theirgeneralized, plesiomorphic anatomy (De Iuliis, 1994),but also with the purportedly primitive nature of theSantacrucian xenarthran fauna (Scott, 1903–4; Gau-din, 1995).
An improved understanding of higher-level relation-ships among sloths has been the primary goal of thepresent study. However, the results of the analysisalso provide a fairly consistent resolution of relation-ships within each of the sloth families (Figs 1, 2).
The monophyly of the Mylodontidae, as noted pre-viously, does not receive overly strong support, despitethe long recognized morphological distinctiveness ofthis clade (Stock, 1925; Winge, 1941; see also com-ments of Engelmann, 1985). The analysis identifiesonly five unambiguous mylodontid synapomorphies(Appendix 4; Fig. 4), and the basal mylodontid nodehas only modest branch support and bootstrap sup-port values (Fig. 1). The weakness of this node isattributable in large part to the atypical morphologyof the Chasicoan (late Miocene) genus Octomylodon.This genus, known from a single poorly preservedskull and mandible, possesses the lobate dentitiontypical of mylodonts, but has a mandibular morphol-ogy that resembles megatherioids in a number ofrespects, e.g., in its elongate slender coronoid andangular processes and its ventrally bulging mandibu-lar ramus (Scillato-Yané, 1977). The node above Octo-mylodon (Node 9, Fig. 1) is more robust. Althoughdiagnosed by only seven unambiguous synapomor-phies, there are as many as 18 equivocal featuresassigned to this node (Appendix 4), most of which can-not be coded in Octomylodon. Furthermore, the nodehas good branch support and a very high bootstrapvalue (Fig. 1).
Apart from the relationship of ‘nothrotheres’, noaspect of tardigrade phylogeny is as controversial asthe relationships within the Mylodontidae. It seemsthat every author who has examined the matter hascome up with a different taxonomic or phylogeneticscheme (Simpson, 1945; Hoffstetter, 1958; Scillato-Yané, 1977; Paula Couto, 1979; Engelmann, 1985; Hir-schfeld, 1985; McDonald, 1987; Webb, 1989; Pattersonet al., 1992; Gaudin, 1995; McKenna & Bell, 1997;McDonald & Perea, 2002). Most authors agree thatthe family can be broadly divided into two subunits,the subfamilies Scelidotheriinae and Mylodontinae(although these groups are not always assignedsubfamily status; see, e.g., McKenna & Bell, 1997).However, various authors have added additional sub-families [e.g., the Octomylodontinae of Scillato-Yané(1977) or the Lestodontinae of Webb (1989)], and thecontents of the two primary subfamilies vary fromauthor to author. In addition, some mylodontid generaare excluded from both subfamilies by a variety ofauthors [e.g., Nematherium in Hoffstetter (1958) andScillato-Yané (1977); Octomylodon, Octodontotheriumand Pseudoprepotherium in Engelmann (1985) andMcKenna & Bell (1997); Octomylodon and Octodon-totherium in Hirschfeld (1985)], and the genus Octod-ontotherium is excluded from the Mylodontidaeentirely by a number of authors (Hoffstetter, 1958;Paula Couto, 1979; McDonald, 1987).
As might be expected from the preceding discussion,the phylogenetic scheme for mylodontids resultingfrom the present analysis conforms to certain aspects
of those proposed by previous authors, but except forthat of Gaudin (1995) it is not fully consistent withany previous scheme. The unusual Chasicoan genusOctomylodon (Scillato-Yané, 1977) is the sister taxonto all other mylodontids (Figs 1, 2), as noted previ-ously. Such a position is consistent with Engelmann’s(1985) phylogeny. The Santacrucian genus Nemathe-rium (Scott, 1903–4; Fig. 4) is one step more derivedthan Octomylodon (Figs 1, 2). It too occupies a positionoutside the two traditional mylodontid subclades, theScelidotheriinae and Mylodontinae, a placement con-sistent with that proposed by Hoffstetter (1958) andScillato-Yané (1977). However, most recent authorsconsider Nematherium a basal member of Scelido-theriinae (Patterson & Pascual, 1968, 1972; PaulaCouto, 1979; Engelmann, 1985; Hirschfeld, 1985;McDonald, 1987; McKenna & Bell, 1997; McDonald &Perea, 2002). As noted above, there are dental resem-blances between Nematherium and scelidotheriines.Nevertheless, the node (Node 10) separating Nemathe-rium from scelidotheriines and mylodontines is fairlyrobust, diagnosed by 13 unambiguous synapomor-phies (Appendix 4) and characterized by high boot-strap and branch support values (Fig. 1).
The monophyly of the Scelidotheriinae (Node 11,Fig. 1) is one of the most strongly corroborated resultsof the present analysis. The two Plio-Pleistocene sceli-dothere genera included in the analysis, Catonyx andScelidotherium (Fig. 5), are united by 24 unambiguoussynapomorphies (Appendix 4), six of which are uniqueto this clade. Node 11 has a bootstrap value of 100 andthe highest branch support value possible in the cur-rent analysis (Fig. 1). Because of the absence of suit-ably complete material from other scelidotheriine taxain the museum collections utilized in this study, onlytwo scelidotheriine genera were included in the anal-ysis, precluding any statements about scelidotheriineinterrelationships. However, the small number ofscelidotheriines considered also reflects the low taxo-nomic diversity in the group relative to mylodontines(McKenna & Bell, 1997).
The basal node (Node 12) of the Mylodontinae isless strongly supported than that of the scelidotheri-ines, although both the bootstrap and branch supportvalues are relatively robust (Fig. 1). The La Ventan(mid to late Miocene; Flynn & Swisher, 1995) genusPseudoprepotherium is placed as the most primitivemylodontine (Figs 1, 2), a position it also occupies inthe studies of Hirschfeld (1985), Webb (1989) andGaudin (1995). Engelmann (1985) excludes this taxonfrom both Scelidotheriinae and Mylodontinae basedon the absence of an odontoid process of the astraga-lus, the presence of a separate foramen rotundumand some differences in the morphology of m3. How-ever, Pseudoprepotherium shares at least ten cranio-dental synapomorphies with other mylodontines
(Appendix 4). It should also be noted that the odon-toid process of the astragalus evolves independentlyat least twice, and perhaps more, in sloths (De Iuliis,1994), and that the morphology of m3 in Pseudopre-potherium resembles that of other mylodontines moreclosely than does that of Octodontotherium, a morederived taxon in Engelmann’s (1985) phylogeny.
Octodontotherium, a genus from the OligoceneDeseadan LMA, is the oldest described sloth knownfrom reasonably complete skeletal remains (Hoffstet-ter, 1956; Patterson et al., 1992). It is one step morederived than Pseudoprepotherium in the presentstudy (Figs 1, 2), an arrangement identical to that pos-tulated by Gaudin (1995). Previous authors haveplaced Octodontotherium outside the Mylodontinae(Scillato-Yané, 1977; Engelmann, 1985; Hirschfeld,1985; McKenna & Bell, 1997). Others have even sug-gested that the genus is not a true mylodontid, basedon some differences in dental histology and its pur-ported association with scutes resembling those of thecingulate Palaeopeltis (Hoffstetter, 1956, 1958; PaulaCouto, 1979; McDonald, 1987). However, as noted byPatterson & Pascual (1972), the cranial anatomy ofthis genus strongly resembles that of other mylodon-tines, and this taxon shares eight unambiguous syna-pomorphies with other mylodontines exclusive ofPseudoprepotherium (Node 13, Appendix 4). Node 13is only slightly less robust than the basal mylodontinenode, with lower branch support but a slightly higherbootstrap value. Pseudoprepotherium and Octodon-totherium are separated from post-La Ventan myl-odontines by a node (Node 14) with modest bootstrapsupport but low branch support (Fig. 1). The lateMiocene – Pleistocene mylodontines are united by asmany as 21 synapomorphies (Appendix 4), althoughonly eight can be unequivocally assigned to this cladebecause of missing data in the earlier taxa.
Like Webb (1989) the results of the present analysisstrongly support the close common ancestry of the lateMiocene North American genus Thinobadistes and thegiant South American Pleistocene genus Lestodon in amonophyletic group termed the Lestodontini (Figs 1,2). The group is diagnosed by 11 unequivocal synapo-morphies (Appendix 4; Fig. 6), and has strong boot-strap and branch support indices (Fig. 1). As inGaudin (1995), the South American Plio-Pleistocenegenus Glossotherium is the closest relative of the Lest-odontini, with the contemporaneous South Americangenus Mylodon a more distant relative. However, theNorth American Pleistocene genus Paramylodon(Fig. 6), which is sometimes considered a congener ofGlossotherium (e.g., Hoffstetter, 1958; Paula Couto,1979; Naples, 1989; Webb, 1989), is not closely alliedwith Glossotherium or the clade including Glossothe-rium and the Lestodontini in the present analysis.Rather, it is either separated from the latter clade by
the Huayquerian (late Miocene) genus Pleurolestodon(Fig. 1), or it is the sister taxon to Mylodon, dependingupon the character weighting scheme employed. Botharrangements are quite different from previous sys-tematic treatments of the post-La Ventan mylodon-tines (Hoffstetter, 1958; Engelmann, 1985; Hirschfeld,1985; McDonald, 1987; Webb, 1989; McKenna & Bell,1997). Engelmann (1985: 57–8) has pointed out the“exceedingly wide morphological range” encompassedby Paramylodon and Glossotherium, making a resolu-tion of their relationships to one another and to othermylodontines difficult, as evidenced by the lack of res-olution in this section of the tree in the present study(Fig. 2). I would suggest that Engelmann’s (1985)statement could apply equally well to virtually theentire late Miocene – Pleistocene assemblage, with thepossible exception of the lestodontines. A moredetailed examination of these taxa than was possiblein the present analysis would be needed to assess theirinterrelationships more definitively. Such a studywould greatly benefit from a thorough revision of Glos-sotherium and Paramylodon.
The results of the present study offer little insightinto relationships within the family Megatheriidae, asthe group is represented by only three taxa. The fam-ily, like the subfamily Scelidotheriinae, encompasseslimited taxonomic diversity. De Iuliis (1996) has pro-vided a thorough review of the Megatheriidae. Never-theless, the results are noteworthy in providingadditional confirmation of the close relationshipbetween the Santacrucian Planopsines, representedby the genus Planops, and later megatheriids.Although the relationship is supported by relativelyweak bootstrap and branch support values, it is diag-nosed by eight unequivocal synapomorphies (Node 24,Appendix 4). These results thus provide perhaps thestrongest support to date for this relationship, whichhas been advocated by numerous previous authors(Scott, 1903–4; Simpson, 1945; Hoffstetter, 1958;Paula Couto, 1979; De Iuliis, 1994; Gaudin, 1995;McKenna & Bell, 1997).
As noted in the preceding discussions, the relation-ships among nothrotheriids are not fully resolved bythe present analysis. The analysis provides verystrong support for a clade including the Pleistocenegenera Nothrotherium from South America, andNothrotheriops (Fig. 9A) from North America. Thisnode (Node 28) is diagnosed by 12 unambiguous syn-apomorphies. It has a bootstrap value of 98, and thehighest possible branch support value (Fig. 1). Theclose alliance of Nothrotherium and Nothrotheriopscorroborates the phylogeny of de Muizon & McDonald(1995) and McDonald & de Muizon (2002), and contra-dicts previous claims that Nothropus (Frailey, 1986) orNothropus and Pronothrotherium (Paula Couto, 1971)are more closely related to Nothrotheriops than to
Nothrotherium. Unfortunately, the present study didnot succeed in unambiguously resolving the relation-ships among the South American genera Nothropus(late Miocene – Pleistocene) and Pronothrotherium(Pliocene, Fig. 9B) and the crown group. Depending onthe character weighting scheme employed, eitherNothropus or Pronothrotherium is placed as the sistertaxon to Nothrotherium + Nothrotheriops. The closealliance of Nothropus with the crown group would beconsistent with the phylogenetic hypotheses of deMuizon & McDonald (1995) and Gaudin & De Iuliis(1999), but contradicts the linkage of Pronothrothe-rium and Nothropus as sister taxa by McDonald & deMuizon (2002). This lack of consensus concerningbasal nothrotheriid relationships points to the needfor additional detailed studies of the phylogenetic his-tory of this family.
The inclusion of the Santacrucian genus Eucho-loeops in the family Megalonychidae as a basal taxondiffers from traditional classifications and severalmore recent studies that ally this taxon with the San-tacrucian ‘nothrotheres’ (Simpson, 1945; Hoffstetter,1958; Paula Couto, 1979; Patterson et al., 1992; Gau-din, 1995). However, several authors have noted thedental resemblances between Eucholoeops and megal-onychids and suggested that the two are related (Mat-thew, 1918b; Hirschfeld & Webb, 1968; Frailey, 1988;McKenna & Bell, 1997). The monophyly of a Megalo-nychidae including Eucholoeops is very robustly sup-ported in the present analysis. Node 29 is diagnosedby 20 unambiguous synapomorphies (Appendix 4),including a number of non-dental features. It is wellsupported by the bootstrap analysis, and has the high-est possible branch support value (Fig. 1).
As noted in preceding sections, the node (Node 30)including all post-Santacrucian megalonychids is alsoquite robust. It is diagnosed by 26 unequivocal syna-pomorphies, with higher bootstrap values and abranch support value only one step lower than Node29. None of the internal nodes in this clade receivesvery robust support, however, with the exception of anode linking the Antillean genera Megalocnus andParocnus (Fig. 1). The relationships among the basalmembers of this clade, including the South American,late Miocene genus Pliomorphus and the North Amer-ican genera Pliometanastes (late Miocene) and Megal-onyx (late Miocene – Pleistocene) differ significantlyfrom previous hypotheses of megalonychid interrela-tionships (Figs 1, 2). Engelmann (1985) and Webb &Perrigo (1985) hypothesize that Pliometanastes andMegalonyx are sister taxa. Pliometanastes sharesseven unambiguous synapomorphies with morederived megalonychids (Appendix 4). However, thenode has extremely weak bootstrap and branch sup-port (Fig. 1), and four of the eight autapomorphiesassigned to Pliometanastes are convergent on Megal-
onyx [34(2), 199(1), E34(1), E35(2)]. This leaves openthe possibility that the two are closely related. Plio-morphus has been closely associated with a subgroupof Antillean megalonychids by several previousauthors (Kraglievich, 1923; Simpson, 1945; Hoffstet-ter, 1958; Paula Couto, 1979). Hirschfeld & Webb(1968) assert that the genus is ancestral to the entireAntillean radiation, as well as Pliometanastes andMegalonyx, and McKenna & Bell (1997) suggest thatit is a basal megalonychid closely related to Eucho-loeops. The results of the present analysis most closelyresemble that of Hirschfeld & Webb (1968) in thatPliomorphus is the sister taxon to a monophyleticgroup of Antillean sloths. Unfortunately, the nodelinking Pliomorphus to the Antillean radiation isweak, supported by only three unambiguous synapo-morphies (Node 32, Appendix 4) and characterized byweak bootstrap and branch support values (Fig. 1).
As noted in the preceding discussions, the results ofthe present study unite the extant two-toed sloth Cho-loepus and the West Indian megalonychid radiation ina monophyletic group. The clade is diagnosed by a rel-atively large number of synapomorphies (20, including10 unambiguously assigned characters; see Node 33,Appendix 4), although it is characterized by low boot-strap and branch support values (Fig. 1), and theinternal relations are poorly resolved (see precedingdiscussions). The monophyly of the Antillean sloths isalso advocated by Hirschfeld & Webb (1968), Engel-mann (1985), Webb & Perrigo (1985) and Gaudin(1995), although none but the last includes Choloepusin this clade. A number of previous authors have splitthe Antillean radiation up into at least two subgroups(Kraglievich, 1923; Simpson, 1945; Hoffstetter, 1958;Paula Couto, 1979). Perhaps the most radical sugges-tion of Antillean diphyly comes from White &MacPhee (2001), who place Choloepus and the Anti-llean genera Acratocnus, Paulocnus and Neocnus in amonophyletic group with the extant Bradypus and thebasal megatherioid Hapalops as successive sistertaxa. They then join the large bodied Antillean generaMegalocnus and Parocnus in a clade with the myl-odontid Paramylodon as its sister taxon. I suspect thatsuch a radical rearrangement of the relationshipsamong West Indian sloths, and indeed of tardigraderelationships as a whole, results from the inclusion ofso few non-Antillean taxa in their analysis. Theyincorporate no non-Antillean megalonychids in theiranalysis except for the extant Choloepus, and onlythree non-megalonychid sloth taxa are considered.Nevertheless, given the large number of postcranialcharacters included in their analysis, their results pro-vide an intriguing counterpoint to the results of thepresent study. It would be interesting indeed to testtheir results using the same character base analysedover a much broader array of tardigrade taxa.
The purpose of the present study was to expand uponmy previous analysis of interrelationships amongsloths (Gaudin, 1995). Gaudin (1995) relied solely uponosteological characters of the ear region. The presentstudy draws upon a larger character base of cranio-dental characters analysed in a broader array of slothtaxa. Questions of particular interest concerned higherlevel relationships within the Tardigrada, specificallythe purported diphyly of the two extant tree sloth gen-era and the relationships among the various families ofsloths. The results of this study have supported someof the phylogenetic conclusions of Gaudin (1995), pro-viding new characters that increase the confidence thatcan be placed in these hypotheses. In addition, theresults have suggested schemes of phylogenetic rela-tionship that contradict those of Gaudin (1995).
Because of the inclusion of a wide variety of out-group taxa (whose relationships to Tardigrada wereconstrained a priori), the results of the present study,like those of Gaudin (1995), identify a large body ofnew characters that can serve to diagnose the orderXenarthra and several of its higher level clades,including the Cingulata and Pilosa, and, within thePilosa, the Vermilingua and the Tardigrada itself.
The present study, perhaps more so than its prede-cessor (Gaudin, 1995), puts forth strong and convinc-ing evidence for the diphyletic origin of the tree sloths.In combination with a series of other recently pub-lished studies (Webb, 1985; Patterson et al., 1992;Gaudin, 1995; Höss et al., 1996; Poinar et al., 1998;Greenwood et al., 2001; White & MacPhee, 2001), Ithink there can be little doubt that the tree sloths donot share a recent common ancestor, and that thenumerous superficial similarities shared by these twogenera represent a remarkable case of convergent evo-lution, perhaps the most remarkable example in all ofMammalia. The results of the present study also con-firm Gaudin’s (1995) placement of the living tree slothBradypus as the sister group to all other sloths, aclade termed the Eutardigrada. This in turn suggeststhat the split between Bradypus and Choloepus is anancient one, dating back perhaps 40 million years. Thepresent study also firmly supports the alliance of theextant sloth Choloepus with the extinct members ofthe family Megalonychidae.
As in Gaudin (1995), the present study supportedthe monophyly of the traditional sloth families Myl-odontidae, Megalonychidae and Megatheriidae. It alsosupported the assertion of Gaudin (1995) that thesloths traditionally grouped together as ‘nothrotheres’can be divided into a paraphyletic group of early tomiddle Miocene forms, the basal megatherioids, and amonophyletic clade of late Miocene – Pleistoceneforms. The latter group is formally recognized as mer-
iting family level status and is designated the familyNothrotheriidae. Like Gaudin (1995), the presentstudy recognizes a monophyletic Megatheriidaeincluding the Santacrucian genus Planops. Lastly, it iscongruent with Gaudin’s (1995) study in its arrange-ment of relationships within the families Megalony-chidae and Mylodontidae.
Conversely, the results of the present study contra-dict those of Gaudin (1995) on several points. Unlikethe previous study, the results of the present analysissupport the monophyly of the Megatherioidea, a cladeuniting the Megatheriidae, Nothrotheriidae, Megalo-nychidae and basal megatherioids to the exclusion ofthe Mylodontidae, as suggested by several previousauthors (Patterson & Pascual, 1968, 1972; Webb,1985; Patterson et al., 1992; McKenna & Bell, 1997).Likewise, as in a range of previous studies (Patterson& Pascual, 1968, 1972; Paula Couto, 1971, 1979;Engelmann, 1985; Perea, 1988; Patterson et al., 1992;McKenna & Bell, 1997) the results of the presentstudy support a close relationship betweennothrotheriid and megatheriid sloths in a clade desig-nated Megatheria. Also noteworthy is the allocation ofthe Santacrucian ‘nothrothere’ Eucholoeops to thefamily Megalonychidae as a basal member.
This study and its predecessor (Gaudin, 1995)together represent the most comprehensive anddetailed cladistic investigation of tardigrade relation-ships undertaken to date. The taxonomic coverage ofthese two studies is more extensive than that of mostrecent investigations of sloth phylogeny (Webb & Per-rigo, 1985; Webb, 1989; De Iuliis, 1994; de Muizon &McDonald, 1995; Höss et al., 1996; Poinar et al., 1998;Greenwood et al., 2001; White & MacPhee, 2001), andthey examine cranial morphology in much greaterdetail, including many more cranial characters thaneither Engelmann (1985) or Webb (1985). Neverthe-less, both the present study and that of Gaudin (1995)fail to include relevant phylogenetic information onthe postcranial anatomy of sloths. The decision toexclude these characters was based on time andresource limitations, rather than any belief that post-cranial characters are somehow inherently less infor-mative about phylogenetic relationship than cranialfeatures. I do not doubt that an analysis similar to thisone in taxonomic scope but incorporating informationon tardigrade postcrania would only improve ourunderstanding of sloth interrelationships. The inclu-sion of postcranial characters should also prove espe-cially useful in elucidating how the two extant treesloth genera convergently evolved such remarkablesimilarities in their locomotory structure and function.
Finally, although the present study provides a con-sistent resolution of relationships within the twolargest sloth families, the Megalonychidae and Myl-odontidae, a number of these relationships are
weakly supported, and differ substantially fromprevious hypotheses of relationships within thesegroups. The results of this study point out the needfor further detailed study of the relationships withinthe various sloth families. Such studies will of coursebe enhanced by further improvements in our under-standing of the fossil record of these clades, and byimprovements in our knowledge of the anatomy oftheir members.
ACKNOWLEDGEMENTS
I am indebted to the following institutions and indi-viduals for access to the specimens that formed thebasis of this study: Ted Daeschler, Academy of NaturalSciences, Philadelphia, PA; Richard Tedford, MalcolmMcKenna and John Alexander, Department of Verte-brate Paleontology, American Museum of Natural His-tory, New York; Larry Heany, Bruce Patterson and BillStanley, Division of Mammals, and John Flynn andBill Simpson, Department of Geology, Field Museumof Natural History, Chicago; George Jefferson andChristopher Shaw, George C. Page Museum, LosAngeles, CA; Larry Barnes and Sam MacCleod, Nat-ural History Museum of Los Angeles County, LosAngeles, CA; Rufus Churcher and Kevin Seymour,Royal Ontario Museum, Toronto, Ontario, Canada;and Mary Ann Turner, Peabody Museum, Yale Univer-sity, New Haven, CT. For the loan of specimens thatwere incorporated in the present study, I thank TonyBarnosky and Howard Hutchinson, Museum of Palae-ontology, University of California, Berkeley, CA; BillAkersten, Idaho Museum of Natural History, IdahoState University, Pocatello, ID; and Ken Rose, JohnsHopkins University, Baltimore, MD. Earlier versionsof this manuscript benefited greatly from the com-ments of Andy Biewener, John Flynn, James Hopson,Bill Turnbull and John Wible. In addition, I thankGreg McDonald for his thoughtful review of this work.Finally, I am grateful to Julia Morgan Scott for herassistance in the preparation of illustrations for thisstudy. This project was initiated as part of my PhD dis-sertation at the University of Chicago (1993). Work tocomplete the study was supported in part by NSF RUIGrant DEB 0107922.
REFERENCES
Aiello A. 1985. Sloth hair: unanswered questions. In: Mont-gomery GG, ed. The ecology and evolution of armadillos,sloths, and vermilinguas. Washington, DC: SmithsonianInstitution Press, 213–218.
Allen GM. 1913. A new Mylodon. Memoirs of the Museum ofComparative Zoology 40: 318–346.
Anthony HE. 1918. The indigenous land mammals of Porto
Rico, living and extinct. Memoirs of the American Museum ofNatural History 2: 331–435.
Anthony HE. 1926. Mammals of Porto Rico, living and extinct– Rodentia and Edentata. New York Academy of Sciences, Sci-entific Survey of Porto Rico and the Virgin Islands 9: 97–243.
Barlow JC. 1984. Xenarthrans and pholidotes. In: Anderson S,Knox Jones J Jr, eds. Orders and families of recent mammalsof the world. John Wiley & Sons, Inc, New York, 219–239.
Bremer K. 1994. Branch support and tree stability. Cladistics10: 295–304.
Burmeister G. 1864. Listo de los mamíferos fosiles del terrenodiluviano. Anales Del Museo Público de Buenos Aires. 1: 121–300.
Burmeister G. 1881. Atlas de la description physique de laRépublique Argentine. Deuxieme section. Mammifères. Paris:Paul-Émile.
Burmeister G. 1882. Nothropus priscus, ein bisher unbe-kanntes fossiles Faulthier. Sitzungsberichte der (Königlich-Preussischen) Akademie der Wissenschaften, Berlin 1882:613–620.
Carroll RL. 1988. Vertebrate paleontology and evolution. NewYork: W.H. Freeman.
Cartelle C, Bohórquez GA. 1982. Eremotherium laurillardiLund 1842. Parte I. Deteminação específica e dimorfismosexual. Iheringia, Séria Géologica, Porto Alegre 7: 45–63.
Cartelle C, Bohórquez GA. 1986. Descrição das pré-maxilasde Nothrotherium maquinense (Lund) Lydekker, 1889(Edentata, Megalonychidae) e de Eremotherium laurillardi(Lund) Cartelle & Bohórquez, 1982 (Edentata, Megatheri-idae). Iheringia, Séria Géologica, Porto Alegre 11: 9–14.
Cartelle C, Fonseca JS. 1982. Contribuição ao melhor con-hecimento da pequena preguiça terrícola Nothrotheriummaquinense (Lund) Lydekker, 1889. Lundiana 2: 127–182.
De Iuliis G. 1994. Relationships of the Megatheriidae,Nothrotheriinae and Planopsinae: some skeletal character-istics and their importance for phylogeny. Journal of Verte-brate Paleontology 14: 577–591.
De Iuliis G. 1996. A systematic review of the Megatheriinae(Mammalia: Xenarthra: Megatheriidae). Unpublished DPhilthesis, University of Toronto.
Delsuc F, Catzeflis FM, Stanhope MJ, Douzery EJP. 2001.The evolution of armadillos, anteaters and sloths depicted bynuclear and mitochondrial phylogenies: implications for thestatus of the enigmatic fossil Eurotamandua. Proceedings ofthe Royal Society of London B 268: 1605–1615.
Edmund G. 1985. The fossil giant armadillos of North Amer-ica (Pampatheriinae, Xenarthra = Edentata). In: Montgom-ery GG, ed. The ecology and evolution of armadillos, sloths,and vermilinguas. Washington, DC: Smithsonian InstitutionPress, 83–93.
Eisenberg JF, Redford KH. 1999. Mammals of the Neotro-pics, Vol. 3. The Central Neotropics: Ecuador, Peru, Bolivia,Brazil. Chicago, IL: University of Chicago Press.
Engelmann G. 1978. The logic of phylogenetic analysis andthe phylogeny of the Xenarthra. Unpublished DPhil Thesis,Columbia University.
Engelmann G. 1985. The phylogeny of the Xenarthra. In:Montgomery GG, ed. The ecology and evolution of armadil-
Ferigolo J. 1985. Evolutionary trends of the histological pat-tern in the teeth of Edentata (Xenarthra). Archives of OralBiology 30: 71–82.
Fischer K. 1971. Riesenfaultiere (Megalonychidae, Edentata,Mammalia) aus dem Pleistozän der Plio-Domingo-Höhle inKuba. Wissenschaftliche Zeitschrift, Humboldt-Universität,Berlin, Mathematisch-Naturwissenschaftliche Reihe 20:609–673.
Flower WH. 1882. On the mutual affinities of the animalscomposing the order Edentata. Proceedings of the ZoologicalSociety of London 358–367.
Flower WH. 1885. An introduction to the osteology of theMammalia. London: Macmillen.
Flynn JJ, Swisher CC. 1995. Cenozoic South American landmammal ages: correlation to global geochronologies. Geo-chronology Time Scales and Global Stratigraphic Correla-tion, SEPM Special Publication 54: 317–333.
Frailey CD. 1986. Late Miocene and Holocene mammals,exclusive of the Notoungulata, of the Río Acre region, west-ern Amazonia. Contributions in Science, Natural HistoryMuseum of Los Angeles County 374: 1–74.
Frailey CD. 1988. The Miocene vertebrates of QuebradaHonda, Bolivia. Part II. Edentata. Occasional Papers of theMuseum of Natural History, University of Kansas 123: 1–13.
Frailey CD. 1995. Nothrotheriine sloths of the Amazon basin.Journal of Vertebrate Paleontology 15 (Suppl. 3): 30A.
Gardner AL. 1993. Order Xenarthra. In: Wilson DE, ReederDM, eds. Mammal species of the world, 2nd edn. Washing-ton, DC: Smithsonian Institution Press, 63–68.
Gaudin TJ. 1994. Phylogeny of the Tardigrada (Xenarthra,Mammalia): the cranio-dental evidence. Journal of Verte-brate Paleontology 14 (Suppl. 3): 27A.
Gaudin TJ. 1995. The ear region of edentates and the phy-logeny of the Tardigrada (Mammalia, Xenarthra). Journal ofVertebrate Paleontology 15: 672–705.
Gaudin TJ, Branham DG. 1998. The phylogeny of theMyrmecophagidae (Mammalia, Xenarthra, Vermilingua)and relationship of Eurotamandua to the Vermilingua. Jour-nal of Mammalian Evolution 5: 237–265.
Gaudin TJ, De Iuliis G. 1999. The late Miocene sloth Nothro-pus priscus, with comments on the phylogeny of theNothrotheriidae. Journal of Vertebrate Paleontology 19(Suppl. 3): 46A.
Gaudin TJ, Wible JR, Hopson JA, Turnbull WD. 1996.Reexamination of the morphological evidence for the CohortEpitheria (Mammalia, Eutheria). Journal of MammalianEvolution 3: 31–79.
Gazin CL. 1956. Exploration for the remains of giant groundsloths in Panama. Annual Reports of the Smithsonian Insti-tution 4272: 341–354.
Gillette DD, Ray CE. 1981. Glyptodonts of North America.Smithsonian Contributions to Paleobiology 40: 1–255.
Grassé P-P. 1955a. Ordre ses Édentés. In: Grassé P-P, ed.
Traité de Zoologie, vol. 17 Mammifères. Paris: Masson et Cie,1182–1266.
Grassé P-P. 1955b. Ordre des Pholidotes. In: Grassé P-P, ed.Traité de Zoologie, vol. 17 Mammifères. Paris: Masson et Cie,1267–1282.
Greenwood AD, Castresana J, Feldmaier-Fuchs G,Pääbo S. 2001. A molecular phylogeny of two extinct sloths.Molecular Phylogenetics and Evolution 18: 94–103.
Guth C. 1961. La région temporale des Édentés. UnpublishedDPhil Thesis, L’Université de Paris.
Hillis DM, Bull JJ. 1993. An empirical test of bootstrappingas a method for assessing confidence in phylogenetic analy-sis. Systematic Biology 42: 182–192.
Hirschfeld SE. 1981. Pliometanastes protistus (Edentata,Megalonychidae) from Knight’s Ferry, California with dis-cussion of Early Hemphillian megalonychids. PaleoBios 36:1–16.
Hirschfeld SE. 1985. Ground sloths from the Friasian LaVenta Fauna, with additions to the pre-Friasian CoyaimaFauna of Colombia, South America. University of CaliforniaPublications, Geological Sciences 128: 1–91.
Hirschfeld SE, Webb SD. 1968. Plio-Pleistocene megalony-chid sloths of North America. Bulletin of the Florida StateMuseum 12: 213–296.
Hoffstetter R. 1952. Les mammifères Pléistocènes de laRépublique de L’Équateur. Mémoires, Société Géologique deFrance 31: 375–488.
Hoffstetter R. 1956. Contribution a l’étude des Orophodon-toidea, gravigrades cuirassés de la Patagonie. Annales dePaléontologie 42: 27–64.
Hoffstetter R. 1958. Xenarthra. In: Piveteau P, ed. Traité dePaléontologie, vol. 2, no. 6, Mammifères Évolution. Paris:Masson et Cie, 535–636.
Hoffstetter R. 1961. Description d’un squelette de Planops(Gravigrade du Miocène de Patagonie). Mammalia 25: 57–96.
Hoffstetter R. 1982. Les édentés xénarthres, un groupe sin-gulier de la faune Neotropical. In: Gallitelli EM, ed. Paleon-tology, essential of historical geology. Modena, Italy: STEMMocchi Modena Press, 385–443.
Höss M, Dilling A, Currant A, Pääbo S. 1996. Molecularphylogeny of the extinct ground sloth Mylodon darwinii.Proceedings of the National Academy of Sciences, USA 93:181–185.
Jollie M. 1968. The head skeleton of a new-born Manis jav-anica with comments on the ontogeny and phylogeny of themammal head skeleton. Acta Zoologica 49: 227–305.
Kay RF, Madden RH, Carlini AA, Vucetich MG, Bond M,Bellosi E, Heizler M, Re GH, Vilas JF. 2002. The Mus-tersan interval at Gran Barranca. Journal of Vertebrate Pale-ontology 22 (Suppl. 3): 73A.
Kay RF, Madden RH, Vucetich MG, Carlini AA, MazzoniMM, Re GH, Heizler M, Sandeman H. 1999. Revised geo-chronology of the Casamayoran South American Land Mam-mal Age: climatic and biotic implications. Proceedings of theNational Academy of Sciences, USA 96: 13235–13240.
Kielan-Jaworowska Z. 1980. Cranial morphology of the Cre-taceous eutherian mammal Barunlestes. Acta Palaeontolog-ica Polonica 25: 167–185.
Kielan-Jaworowska Z. 1981. Results of the Polish-Mongo-lian palaeontological expeditions- Part IX. Evolution of thetherian mammals in the Late Cretaceous of Asia. Part IV.Skull structure in Kennalestes and Asioryctes. Palaeontolog-ica Polonica 42: 25–78.
Kielan-Jaworowska Z. 1984. Results of the Polish-Mongo-lian palaeontological expeditions- Part X. Evolution of thetherian mammals in the Late Cretaceous of Asia. Part V.Skull structure in the zalambdalestidae. PalaeontologicaPolonica 46: 107–117.
Kraglievich L. 1923. Descripción de dos cráneos y otros restosdel género Pliomorphus Ameghino procedentes de la For-mación Entrerriana de las barrancas del río Paraná. Analesdel Museo Nacional de Historia Natural de Buenos Aires 23:1–56.
Kraglievich L. 1928. Mylodon darwinii Owen es la especiegenotipo de Mylodon Owen. Retificación de la nomenclaturagenérica de los milodontes. Physis 9: 169–185.
Leidy J. 1855. A memoir on the extinct sloth tribe of NorthAmerica. Smithsonian Contributions to Knowledge 7: 1–68.
Lull RS. 1929. A remarkable ground sloth. Memoirs of the Pea-body Museum of Yale University 3: 1–39.
Luo Z, Kielan-Jaworowska Z, Cifelli RL. 2002. In quest fora phylogeny of Mesozoic mammals. Acta PalaeontologicaPolonica 47: 1–78.
Lydekker R. 1886. Description of three species of Scelidothe-rium. Proceedings of the Zoological Society of London 491–498.
Lydekker R. 1894. Contributions to a knowledge of the fossilvertebrates of Argentina. Anales del Museo de la Plata, Pale-ontologia Argentina 3: 1–103.
Maddison WP, Donoghue MJ, Maddison DR. 1984. Out-group analysis and parsimony. Systematic Zoology 33: 83–103.
Matthew WD. 1918a. Edentata. A revision of the LowerEocene Wasatch and Wind River Faunas. Part V- Insectivora(continued), Glires, Edentata. Bulletin of the AmericanMuseum of Natural History 38: 565–657.
Matthew WD. 1918b. Affinities and origin of the Antilleanmammals. Bulletin of the Geological Society of America 29:657–666.
Matthew WD. 1931. Genera and new species of ground slothsfrom the Pleistocene of Cuba. American Museum Novitates511: 1–5.
Matthew WD, Paula Couto C de. 1959. The Cuban eden-tates. Bulletin of the American Museum of Natural History117: 1–56.
McDonald HG. 1977. Description of the osteology of theextinct gravigrade edentate Megalonyx with observations onits ontogeny, phylogeny and functional anatomy. Unpub-lished M.Sci. Thesis, University of Florida.
McDonald HG. 1987. A systematic review of the Plio-Pleis-tocene scelidotheriine ground sloths (Mammalia: Xenarthra:Mylodontidae). Unpublished DPhil Thesis, University ofToronto.
McDonald HG. 1994. A reexamination of the relationships ofmegalonychids, nothrotheres, and megatheres. Journal ofVertebrate Paleontology 14 (Suppl. 3): 37A.
McDonald HG, de Muizon C. 2002. The cranial anatomy ofThalassocnus (Xenarthra, Mammalia), a derivednothrothere from the Neogene of the Pisco Formation (Peru).Journal of Vertebrate Paleontology 22: 349–365.
McDonald HG, Perea D. 2002. The large scelidothere Cat-onyx tarijensis (Xenarthra, Mylodontidae), from the Pleis-tocene of Uruguay. Journal of Vertebrate Paleontology 22:677–683.
McKenna MC, Bell SK. 1997. Classification of mammalsabove the species level. New York: Columbia UniversityPress.
Moore WJ. 1981. The mammalian skull. Cambridge, UK:Cambridge University Press.
de Muizon C, McDonald HG. 1995. An aquatic sloth fromthe Pliocene of Peru. Nature 375: 224–227.
Naples VL. 1982. Cranial osteology and function in the treesloths, Bradypus and Choloepus. American Museum Novi-tates 2739: 1–41.
Naples VL. 1985. Form and function of the masticatory mus-culature in the tree sloths, Bradypus and Choloepus. Journalof Morphology 183: 25–50.
Naples VL. 1986. The morphology and function of the hyoidregion in the tree sloths, Bradypus and Choloepus. Journalof Mammalogy 67: 712–724.
Naples VL. 1987. Reconstruction of cranial morphology andanalysis of function in the Pleistocene ground slothNothrotheriops shastense (Mammalia, Megatheriidae). Con-tributions in Science, Natural History Museum of Los Ange-les County 389: 1–21.
Naples VL. 1989. The feeding mechanism in the Pleistoceneground sloth, Glossotherium. Contributions in Science, Nat-ural History Museum of Los Angeles County 415: 1–23.
Novacek MJ. 1986. The skull of lepticitid insectivorans andthe higher-level classification of eutherian mammals. Bulle-tin of the American Museum of Natural History 183: 1–111.
Novacek MJ. 1992. Mammalian phylogeny: shaking the tree.Nature 356: 121–125.
Novacek MJ. 1994. The radiation of placental mammals. In:Prothero DR, Schoch RM, eds. Major features of vertebrateevolution. Short courses in paleontology, no. 7. Knoxville, TN:University of Tennessee Press, 220–237.
Owen R. 1856. On the Megatherium. Part III. The skull. Philo-sophical Transactions of the Royal Society of London 146:571–589.
Owen R. 1857. On the scelidothere (Scelidotherium leptoceph-alum Owen). Philosophical Transactions of the Royal Societyof London 147: 101–110.
Pascual R, Vucetich MG, Scillato-Yané GJ, Bond M. 1985.Main pathways of mammalian diversification in SouthAmerica. In: Stehli FG, Webb SD, eds. The great Americanbiotic interchange. New York: Plenum Press, 219–247.
Patterson B, Pascual R. 1968. Evolution of mammals onsouthern continents. Quarterly Review of Biology 43: 409–451.
Patterson B, Pascual R. 1972. The fossil mammal fauna ofSouth America. In: Keast A, Erk FC, Glass B, eds. Evolution,mammals, and southern continents. Albany, NY: State Uni-versity of New York Press, 247–309.
Patterson B, Segall W, Turnbull WD. 1989. The ear regionin xenarthrans (= Edentata, Mammalia). Part I. Cingulates.Fieldiana, Geology n.s. 18: 1–46.
Patterson B, Segall W, Turnbull WD, Gaudin TJ. 1992.The ear region in xenarthrans (= Edentata, Mammalia). PartII. Pilosa (sloths, anteaters), palaeanodonts, and a miscel-lany. Fieldiana, Geology n.s. 24: 1–79.
Paula Couto C de. 1954. Megatérios intertropicais do Pleis-toceno. Anais, Academia Brasileira de Ciências 26: 447–463.
Paula Couto C de. 1959. Uma pequena preguiça terrícola deSåo Paulo. Anais, Academia Brasileira de Ciências 31: 91–108.
Paula Couto C de. 1967. Pleistocene edentates of the WestIndies. American Museum Novitates 2304: 1–55.
Paula Couto C de. 1971. On two small Pleistocene groundsloths. Anais, Academia Brasileira de Ciências 43: 499–513.
Paula Couto C de. 1979. Tratado de Paleomastozoologia. Riode Janeiro, Brazil: Academia Brasileira de Ciências.
Paula Couto C de. 1980. Pleistocene mammals from MinasGerais and Bahia, Brazil. Actas II Congreso Argentino dePaleontología y Bioestratigrafía y I Congreso Latinoameri-cano de Paleontología 3: 193–209.
Perea D. 1988. Dos Nothrotheriinae (Tardigrada, Megatheri-idae) del Mio-Plioceno de Uruguay. Ameghiniana 25: 381–388.
Poinar HN, Hofreiter M, Spaulding WG, Martin PS,Stankiewicz BA, Bland H, Evershed RP, Possnert G,Pääbo S. 1998. Molecular coproscopy: dung and diet of theextinct ground sloth Nothrotheriops shastensis. Science 281:402–406.
Qiang J, Luo Z, Yuan C, Wible JR, Zhang J, Georgi JA.2002. The earliest known eutherian mammal. Nature 416:816–822.
Reed CA. 1954. Some fossorial mammals from the Tertiary ofwestern North America. Journal of Paleontology 28: 102–111.
Reinhardt J. 1879. Beskrivelse af Hovedskallen af et Kæmpe-dovendyr, Grypotherium darwinii, fra La Plata- Landenes
plejstocene Dannelser. Danske Videnskabernes SelskabSkrifter, Naturvidenskabelig Og Mathematisk Afdeling, ser.5 12: 355–380.
Romer AS. 1966. Vertebrate paleontology. Chicago, IL: Uni-versity of Chicago Press.
Rose KD, Emry R. 1993. Relationships of Xenarthra, Pholi-dota, and fossil ‘edentates’. In: Szalay FS, Novacek MJ,McKenna MC, eds. Mammal phylogeny: placentals. NewYork: Springer-Verlag, 81–102.
Roth S. 1899. Descripción de los restos encontrados en la cav-erna de Ultima Esperanza. Revista del Museo de la Plata 9:421–453.
Rovereto C. 1914. Los estratos araucanos y sus fósiles. Analesdel Museo Nacional de Buenos Aires 25: 1–249.
Sarich VM. 1985. Xenarthran systematics: albumin immuno-logical evidence. In: Montgomery GG, ed. The ecology andevolution of armadillos, sloths, and vermilinguas. Washing-ton, DC: Smithsonian Institution Press, 77–81.
Schneider R. 1955. Zur Entwicklung des Chondrocraniumsder Gattung Bradypus. Morphologisches Jahrbuch 95: 209–301.
Scillato-Yané GJ. 1977. Octomylodontinae: nueva subfamiliade Mylodontidae (Edentata, Tardigrada). Descripción delcráneo y mandíbula de Octomylodon robertoscagliai n. sp.,procedentes de la Formación Arroyo Chasicó (Edad Cha-siquense, Plioceno Temprano) del sur de la Provincia de Bue-nos Aires (Argentina). Algunas consideraciones filogenéticasy sistemáticas sobre los Mylodontoidea. Publicaciones delMuseo de Ciencias Naturales de Mar del Plata ‘Lorenzo Sca-glia’ 2: 123–140.
Scott WB. 1903-4. Mammalia of the Santa Cruz Beds. Part 1:Edentata. Reports of the Princeton Expeditions to Patagonia5: 1–364.
Scott WB. 1937. A history of the land mammals of the westernhemisphere. New York: The Macmillen Company.
Sicher H. 1944. Masticatory apparatus of the sloths. Fieldi-ana, Zoology 29: 161–168.
Simpson GG. 1931. Metacheiromys and the relationships ofthe Edentata. Bulletin of the American Museum of NaturalHistory 59: 295–381.
Simpson GG. 1941. A Miocene sloth from Southern Chile.American Museum Novitates 1156: 1–6.
Simpson GG. 1945. The principles of classification and a clas-sification of mammals. Bulletin of the American Museum ofNatural History 85: 1–350.
Stock C. 1913. Nothrotherium and Meglonyx from the Pleis-tocene of southern California. University of California Pub-lications, Bulletin of the Department of Geology 10: 137–164.
Stock C. 1925. Cenozoic gravigrade edentates of westernNorth America. Carnegie Institute of Washington Publica-tions 331: 1–206.
Swofford DL. 2002. PAUP: phylogenetic analysis using par-simony, Version 4.0b10. Champaign, IL: Illinois Natural His-tory Survey.
Szalay FS. 1977. Phylogenetic relationships and a classifica-tion of the eutherian mammals. In: Hecht MK, Goody PC,Hecht BM, eds. Major patterns in vertebrate evolution. NewYork: Plenum Press, 315–374.
Szalay FS, Schrenk F. 1998. The middle Eocene Eurota-mandua and a Darwinian phylogenetic analysis of ‘eden-tates’ Kaupia. Darmstädter Beiträge zur Naturgeschichte,Heft 7: 97–186.
Templeton AR. 1983. Phylogenetic inference from restrictionendonuclease cleavage site maps with particular reference tothe evolution of humans and apes. Evolution 37: 221–244.
Van Dijk MAM, Paradis E, Catzefelis F, De Jong WW.1999. The virtues of gaps: Xenarthran (Edentate) mono-phyly supported by a unique deletion in a aA-crystallin. Sys-tematic Biology 48: 94–106.
Vaughan TA, Ryan JM, Czaplewski NJ. 2000. Mammalogy,4th edn. New York: Saunders College Publishing.
Webb SD. 1985. The interrelationships of tree sloths andground sloths. In: Montgomery GG, ed. The ecology and evo-lution of armadillos, sloths, and vermilinguas. Washington,DC: Smithsonian Institution Press, 105–112.
Webb SD. 1989. Osteology and relationships of Thinoba-distes segnis, the first mylodont sloth in North America.In: Redford KH, Eisenberg JF, eds. Advances in neotropi-cal mammalogy. Gainesville, FL: Sandhill Crane Press,469–532.
Webb SD, Perrigo S. 1985. New megalonychid sloths from ElSalvador. In: Montgomery GG, ed. The ecology and evolutionof armadillos, sloths, and vermilinguas. Washington, DC:Smithsonian Institution Press, 113–120.
Weber M. 1928. Die Säugetiere. Jena: Gustav Fischer.
Wegner G. 1950. Unterschiede der Nasenlochgestaltung unddes Os nariale bei den Säugetieren (Choloepus) und denBauriamorphen. Verhandlungen der Anatomische Gesell-schaft 28: 104–111.
Wetzel R. 1985. The identification and distribution of recentXenarthra (= Edentata). In: Montgomery GG, ed. The ecologyand evolution of armadillos, sloths, and vermilinguas. Wash-ington, DC: Smithsonian Institution Press, 5–21.
White JL. 1993. Indicators of locomotor habits in xenar-thrans: evidence for locomotor heterogeneity among fossilsloths. Journal of Vertebrate Paleontology 13: 230–242.
White JL. 1997. Locomotor adaptations in Miocene xenar-thrans. In: Kay RF, Madden RH, Cifelli RL, Flynn JJ, eds.Vertebrate paleontology in the neotropics: the Miocene faunaof la Venta, Columbia. Washington, DC: Smithsonian Insti-tution Press, 246–264.
White JL, MacPhee RDE. 2001. The sloths of the WestIndies: a systematic and phylogenetic review. In: Woods CA,ed. Biogeography of the West Indies: patterns and perspec-tives. New York: CRC Press, 201–235.
Winge H. 1941. The interrelationships of the mammalian gen-era, vol. 1: Monotremata, Marsupialia, Insectivora, Chi-roptera, Edentata. Copenhagen: C.A. Reitzels Forlag.
Woodward AS. 1900. On some remains of Grypotherium(Neomylodon) listai and associated mammals from a cavernnear Consuelo Cove, Last Hope Inlet, Patagonia. Proceed-ings of the Zoological Society of London 64–79.
APPENDIX 1
List of specimens examined for this study. See Materials and Methods for a list of abbreviations.
Listing of characters and character states. Charactersmarked with a single asterisk (*) are multistate, thosemarked with two asterisks (**) are multistate andordered.
(1) present, small, typically avascular; (2)present, large, typically well-vascularized(Ferigolo, 1985).
10. Thickness of orthodentine: (0) thick layer,thickness greater than or equal to the thick-ness of outer layer of enamel or cementum; (1)thin layer, thickness less than outer layer ofcementum.
**11. Outer layer of cementum: (0) absent; (1) formsthin layer around outside of tooth; (2) formsthick layer around outside of tooth; (3) greatlyhypertrophied, nearly as thick as core of mod-ified orthodentine. (Ferigolo, 1985)
*14. Size of c1: (0) smallest tooth; (1) greatlyenlarged; (2) neither the smallest norenlarged.
*15. Size of m3: (0) smallest molariform; (1) largestmolariform, or equivalent in size to the largest;(2) neither the smallest nor largest molari-form.
*16. Long axis of molariform teeth: (0) parallel ororthogonal to long axis of the toothrow; (1)oblique to long axis in posterior portion oftoothrow; (2) oblique along entire length of
toothrow; (3) oblique to long axis in anteriorportion of toothrow.
*17. Occlusal surface of molariforms: (0) beveled, attimes worn flat; (1) with large anterior andposterior step-like facets (Hoffstetter, 1956);(2) with strong transverse crests; (3) flat.
18. Anterior extent of upper and lower too-throw: (0) lower toothrow extends anterior toupper; (1) upper toothrow extends anterior tolower.
*19. Morphology of C1/c1: (0) molariform; (2) canin-iform; (3) incisiform.
**21. Position of C1 relative to the anterior edge ofthe maxilla: (0) right at the edge [separation <3% BNL]; (1) near the edge [separation >3%,<10% BNL]; (2) well-separated from the ante-rior edge [separation >10% BNL].
22. Fossa anterior to C1: (0) absent; (1) present.23. Fossa on palatal surface of maxilla posterior to
C1: (0) absent; (1) present.24. Alignment of C1/c1: (0) in line with other
**37. Depth of mandible: (0) shallow & elongate,maximum depth of horizontal ramus £17.5% ofMML; (1) >17.5%, £20% of MML; (2) >20%,£22.5% of MML; (3) >22.5%, £25% of MML; (4)>25%, £27.5% of MML; (5) short and deep,maximum depth of horizontal ramus >27.5% ofMML.
**38. Inferior edge of mandible: (0) concave in lat-eral view; (1) straight, horizontal; (2) weakly,uniformly convex; (3) with strong convex ven-tral bulge.
39. Horizontal ramus of mandible bulgesmediolaterally at toothrow: (0) absent; (1)present (Scott, 1903–4).
*40. Ascending ramus of mandible covers posteriorteeth in lateral view: (0) no; (1) partially; (2)yes.
**41. Relative position of processes of ascendingramus: (0) condyle posterior to coronoid andangle; (1) condyle and angle subequal, bothposterior to coronoid; (2) angle posterior tocondyle posterior to coronoid.
**42. Distance between processes of ascendingramus: (0) condyle closer to angle than coro-noid; (1) three processes equidistant; (2)condyle closer to coronoid.
43. Junction between ascending and horizontalramus of mandible: (0) horizontal ramusblends into ascending ramus; (1) distinctconstriction at junction, ascending ramusindented below anterior to base of angular pro-cess, joins horizontal ramus well dorsal to ven-tral margin of horizontal ramus.
44. Ascending ramus with internal ridge runningobliquely vertically from ventral edge, near thebase of the angle, toward the last tooth: (0)absent; (1) present.
45. Coronoid process hooked posteriorly: (0)absent; (1) present.
46. Coronoid process with medial ridge runningalong anterior edge: (0) absent or rudimentary;(1) present.
**47. Shape of coronoid process: (0) elongate & nar-row, ratio of maximum height to anteroposte-rior length measured at mid-height >1.25; (1)intermediate development, ratio of height tolength £1.25, >1.0; (2) short and broad, ratio ofheight to length £1.0 (Scott, 1903–4); (3) rudi-mentary or absent.
**48. Shape of angular process: (0) short and deep,ratio of maximum length to depth measured atmidlength < 1.0; (1) intermediate development,ratio of length to depth >1.0, <1.25; (2) elongateand narrow, ratio of length to depth ≥1.5.
that the lateral end of the facet faces moreanterior or dorsal than the medial end: (0)absent; (1) present.
61. Mandibular symphysis: (0) unfused; (1) fused.**62. Length of symphysis: (0) very short, <10% of
MML; (1) short, ≥10%, <20% of MML; (2) mod-erate length, ≥20%, <27% of MML; (3) elon-gate, >28% of MML.
63. Position of posterior end of symphysis vs. den-tition: (0) symphysis ends anterior to firstlower tooth; (1) symphysis extends posterior tofirst lower tooth.
**65. Profile of anterior edge of symphysis in lateralview: (0) convex; (1) straight; (2) concave.
*66. Symphyseal keel: (0) absent; (1) present alongwhole length of symphysis; (2) present on sym-physeal spout only.
**67. Width of symphysis at midpoint: (0) narrow,£15% of MML; (1) moderately wide, >16%,<19% of MML; (2) very wide, >20% of MML.
**68. Length of symphyseal spout: (0) rudimentaryor very short, <10% of MML; (1) moderatelydeveloped, >10%, <30% of MML; (2) elongate,>30% of MML.
69. Junction of symphysis and lower edge of hori-zontal ramus: (0) forms sharp or roundedangle; (1) no clear demarcation between sym-physis and horizontal ramus.
70. Profile of anterior edge of symphysis in dorsalview: (0) flat; (1) rounded or pointed.
73. Orientation of spout in lateral view: (0) hori-zontal; (1) inclined anterodorsally.
74. Posterior external opening of mandibularcanal: (0) absent; (1) present.
**75. Position of posterior external opening of mand-ibular canal: (0) canal opens laterally onhorizontal ramus; (1) canal opens anterolat-erally, on ascending ramus; (2) canal opensanteromedially, on internal side of ascendingramus.
**76. Mandible with fossa posterior to c1: (0) absent;(1) weakly developed; (2) strongly developed(Scott, 1903–4).
77. Length of stylohyal: (0) short, roughly equiva-lent in length to epihyal or less than 20% ofBNL; (1) elongate, longer than epihyal orgreater than 20% of BNL (Flower, 1885;Naples, 1986).
*78. Shape of stylohyal shaft in lateral view: (0)curved, concave dorsally; (1) curved, concavo-
88. Shape of snout in dorsal view: (0) uniform, orslightly tapered anteriorly; (1) widened anteri-orly.
89. Depth of nasopharynx: (0) shallow, depth £10%of BNL; (1) deep, depth >10% of BNL.
*90. Basicranial/basifacial angle: (0) parallel, butwhole cranial base concave in lateral view; (1)parallel, cranial base roughly horizontal; (2)reflexed (Webb, 1985).
*91. Profile of dorsal surface of the skull in lateralview: (0) horizontal or irregular; (1) profile ofnasal region and braincase relatively horizon-tal, but nasal region depressed relative tobraincase; (2) evenly convex (Patterson et al.,1992).
*97. Temporal lines: (0) are confluent with sagittalcrest or with nuchal crest posteriorly; (1) donot meet, curve ventrally and run anterior butparallel to nuchal crest; (2) temporal fossareduced, temporal lines lie far forward ofnuchal crest.
**99. Inclination of lateral wall of external nares: (0)anteroventral; (1) vertical; (2) anterodorsal(Flower, 1885).
**100. Length and width of nasal: (0) short and wide,ratio of maximum length to width measured atmidpoint < 3.0; (1) ratio of length to width >3.0,<4.0; (2) elongate and narrow, ratio of length towidth >4.0.
*102. Anterior edge of nasal: (0) with lateral processand medial process separated by distinctnotch; (1) evenly convex; (2) straight or con-cave (Scott, 1903–4).
103. Anterior edge of maxilla with fossa lateral toexternal nares: (0) absent; (1) present.
104. Anterior edge of palatal process of maxillaextends under external nares: (0) absent; (1)present (Kraglievich, 1928).
*105. Maxilla elevated for dental alveoli: (0) not ele-vated; (1) elevated in the middle, along thelength of the molariform row; (2) elevated pos-teriorly only; (3) elevated anteriorly only; (4)elevated anteriorly and posteriorly.
**106. Antorbital or buccinator fossa of maxilla: (0)absent; (1) weak; (2) well-developed.
108. Maxilla with orbital exposure: (0) absent orrudimentary; (1) present (Novacek, 1986).
109. Maxilla contacts lacrimal within orbit: (0)present; (1) excluded by orbital exposure ofjugal.
110. Jugal participation in rim of maxillary fora-men: (0) absent; (1) present.
**111. Anterior extent of lateral and medial palatalprocesses of maxilla: (0) medial process ante-rior; (1) two processes of equivalent length; (2)lateral process anterior (Scott, 1903–4).
112. Maxilla with fossa behind last upper tooth: (0)absent; (1) present.
113. Attachment of premaxilla to skull: (0) tightlysutured; (1) loosely attached.
**114. Dorsal process of premaxilla: (0) very large; (1)
narrow anteroposteriorly, but contacts nasaldorsally; (2) reduced in height, does not contactnasal; (3) absent.
*115. Shape of palatal process of premaxilla: (0) V-shaped, narrow mediolaterally; (1) V-shaped,wide; (2) rectangular plate, left and righthalves separate, converge anteriorly; (3) ovalplate, left and right halves sutured in midline;(4) Y-shaped, with elongate anterior processand medial and lateral rami posteriorly; (5)with elongate anterior process and posteriormedial and lateral rami, but squared, thick-ened mediolaterally and dorsoventrally; (6)wide elongate flat surface.
116. Relative size of medial and lateral rami of pre-maxilla: (0) lateral ramus much larger; (1) lat-eral and medial ramus of nearly equivalentsize.
117. Shape of incisive foramen: (0) ovate or trian-gular; (1) slit-like, hidden in ventral view bymedial palatal process of maxilla.
**119. Length of nasoturbinal vs. maxilloturbinal: (0)nasoturbinal shorter; (1) equal length; (2)nasoturbinal longer.
120. Mediolateral contour of palate: (0) concavebetween toothrows; (1) flat to convex betweentoothrows (Paula Couto, 1971).
*121. Anteroposterior contour of palate: (0) evenlyconcave; (1) flat; (2) flat posteriorly, concaveanteriorly; (3) convex posterior to dentition,concave anteriorly; (4) convex along length oftoothrow, concave anteriorly; (5) evenly con-vex.
*122. Length and width of palate: (0) elongate andnarrow, widened at zygomatic processes ofmaxilla; (1) elongate and narrow; (2) elongate,slightly widened anteriorly; (3) elongate,strongly widened anteriorly; (4) short, uni-formly wide.
123. Palate rugose, with many pits and grooves: (0)absent; (1) present.
**124. Palate extends posteriorly and dorsally as ashelf that runs alongside the inner edge ofdescending laminae of the pterygoids: (0)absent; (1) present, shelf ends at midpoint ofdescending lamina; (2) present, shelf extendsposteriorly all the way back to the level of thetympanic cavity.
125. Palate with paired anterior foramina thatopen into distinct grooves that run anteriorlytoward the incisive foramina: (0) absent; (1)present.
126. Postpalatine foramina: (0) small to absent; (1)enlarged (Stock, 1913).
**145. Shape of jugal: (0) simple, no processes; (1)with large descending process; (2) with largeascending and descending processes; (3) withascending, descending, and middle processes.
**146. Postorbital process of zygomatic arch (jugal orsquamosal): (0) absent; (1) weak; (2) present.
147. Jugal and lacrimal overlap facial portion ofmaxilla anteriorly in lateral view: (0) absent;(1) present.
148. Middle process of jugal: (0) elongate, triangu-lar; (1) short, deep dorsoventrally.
**149. Width of ascending process of jugal: (0) wide;(1) narrow, slender; (2) rod-like; (3) rod-likeproximally, with large, flat distal expansion(Scott, 1903–4).
150. Orientation of ascending process of jugal inlateral view: (0) oblique to nearly horizontal;(1) nearly vertical (Webb, 1985).
151. Relative lengths of ascending and descendingprocesses of jugal: (0) ascending process lessthan or equal to descending process; (1)ascending process longer.
**152. Width of descending process of jugal: (0) wide;(1) wide at base, tapers strongly toward tip; (2)narrow.
153. Descending process of jugal hooked posteri-orly: (0) absent; (1) present (Scott, 1903–4).
154. Number of posteriorly projecting points on dis-tal portion of descending process of jugal: (0)one; (1) two.
155. Attachment of jugal to skull: (0) firmlysutured; (1) loosely attached (Webb, 1985).
**156. Position of infraorbital canal: (0) canal short,ventrally situated; (1) canal elongate and ven-tral; (2) canal elongate and displaced dorsally
157. Infraorbital foramen exposure in ventral view:(0) unexposed; (1) exposed.
*158. Relationship of foramen ovale to orbital bones:(0) foramen surrounded by the alisphenoid; (1)foramen between the alisphenoid and squamo-sal; (2) foramen between alisphenoid, ptery-goid and squamosal, or between squamosaland pterygoid externally, with alisphenoid sur-rounding the opening internally; (3) foramenbetween alisphenoid and pterygoid.
159. Foramen rotundum: (0) confluent with thesphenorbital fissure; (1) separate.
160. Optic foramen vs. sphenorbital fissure: (0) twoforamina clearly separate, with distinct exter-nal openings; (1) optic foramen empties intosphenorbital canal, two foramina share com-mon external aperture.
*161. Position of sphenopalatine foramen relative tosphenorbital fissure/optic foramen: (0) situ-ated well anterior and ventral to these open-ings; (1) just anteroventral to orbitalforamina, situated in common fossa; (2) dis-placed posteriorly, lies between optic foramenand foramen ovale.
**162. Bony ridge lateral to orbital foramina: (0)absent; (1) anterior ridge extending from wallof sphenorbital fissure/optic foramen anteri-
orly, foramina open into anterior groove; (2)ridge continues posteriorly from sphenorbitalfissure/optic foramen toward glenoid, oftenwith large muscular process.
163. Orbital exposure of orbitosphenoid: (0) small toabsent; (1) well-developed.
*164. Orbital exposure of palatine: (0) low, elongateanteroposteriorly; (1) higher, more rectangularor square; (2) L-shaped, with tall anterior por-tion, low long posterior portion; (3) very tall,narrow anteroposteriorly.
166. Alisphenoid and pterygoid: (0) unfused, orfused only in adults; (1) fuse very early inontogeny.
167. Squamosal with lateral bulge at root of zygomafor epitympanic sinus: (0) absent or rudimen-tary; (1) present.
**168. Length of zygomatic process of squamosal: (0)reduced, length £5% of BNL; (1) moderate,length >5%, £10% of BNL; (2) elongate, length>10%, £15% of BNL; (3) greatly elongate,length >15% of BNL.
169. Inclination of zygomatic process in lateralview: (0) ventral; (1) horizontal or slightly dor-sal.
**170. Depth of zygomatic process: (0) narrow dors-oventrally, depth measured at midpoint < 5%of BNL; (1) moderately deep, depth ≥5%, <10%of BNL; (2) deep, depth ≥10% of BNL.
*171. Shape of free end of zygomatic process: (0)rounded; (1) broad and somewhat flattened; (2)pointed.
**172. Position of frontal/parietal suture: (0) anteriorto glenoid fossa; (1) at anterior edge of glenoid;(2) well posterior to front of glenoid (Naples,1982).
*173. Frontal and parietal dorsal shape: (0) convexanteroposteriorly and mediolaterally; (1) flat-tened anteroposteriorly and mediolaterally; (2)flattened mediolaterally, though strongly con-vex anteroposteriorly.
*174. Frontal sinus: (0) confluent with maxillarysinus and nasal cavity; (1) absent; (2) small,restricted to frontal; (3) large, extends intoparietal and nasal.
**175. Postorbital process of frontal: (0) absent (1)weakly developed; (2) strongly developed.
183. Nuchal crest: (0) uniform width; (1) splits dor-sally into anterior and posterior occipitalcrests, which together outline a raised trian-gular area in the dorsal surface of the skullroof.
184. Nuchal crest position vs. occiput: (0) in linewith the posterior surface of the occiput; (1)overhangs occiput posteriorly.
185. Median ridge of occiput: (0) extends from fora-men magnum dorsally to the nuchal crest; (1)extends dorsally onto the roof of the skull(Scott, 1903–4).
**186. Distance between occipital condyles: (0) widelyseparate, minimum distance between condyles(in ventral view) >10% of BNL; (1) moderatelywell separated, distance between condyles£10%, >5% of BNL; (2) close to one another, dis-tance between condyles £5% of BNL (Scott,1903–4).
188. Position of occipital condyles relative to den-tition: (0) at nearly the same level as thedentition; (1) situated well dorsal to thedentition.
189. Posterior edge of occipital condyles: (0) pro-trudes posterior to posterior edge of foramenmagnum; (1) ends at or anterior to posteriorforamen magnum.
190. Exoccipital crest vs. occipital condyles: (0)crest separated from lateral edge of condyles;(1) crest abuts lateral edge of condyles.
**191. Occipital condyle proportions in posteriorview: (0) mediolaterally elongate, ratio of max-imum width to maximum height ≥1.0; (1) ratioof width to height <1.0, ≥0.75; (2) mediolater-ally compressed, dorsoventrally elongate, ratioof width to height <0.75.
*192. Occipital condyle shape in posterior view: (0)rhomboid, quadrangular; (1) roughly triangu-lar, with straight or slightly concave medialedge, strongly convex lateral margin; (2)
roughly triangular but extended far medioven-trally; (3) roughly triangular but extended lat-erally; (4) irregularly shaped.
193. Occipital condyles: (0) sessile; (1) with distinctneck (Scott, 1903–4).
**194. Position of occipital condyles vs. condyloidforamina: (0) condyles lie just posterior toforamina, minimum distance betweencondyles and foramina <1.0% of BNL; (1) dis-tance between condyles and foramina >1.0,<2.5% of BNL; (2) condyles well-separatedfrom foramina, distance >2.5% of BNL.
195. Occipital condyle shape in ventral view: (0)condyles not conspicuously elongated antero-posteriorly; (1) condyles elongated anteropos-teriorly.
196. Rectus capitis fossae: (0) absent; (1) present(Scott, 1903–4).
197. Shape of basioccipital: (0) wide and flat; (1)narrow and convex mediolaterally.
198. Shape of basisphenoid: (0) uniformly narrow;(1) triangular, narrows anteriorly; (2) butterflyshaped, with two posterior processes and threeanterior processes, two extending laterally andone in the middle.
*199. Ethmoid exposure in nasopharynx: (0) vomer-ine wings separate exposing intervening eth-moid; (1) vomerine wings fused, leavingoverlying ethmoid unexposed; (2) ethmoidunexposed, covered by posterior extension ofhard palate.
200. Vomer: (0) with short, straight ventral keel, orwith keel lacking altogether; (1) with elongateasymmetrical ventral keel extending posteri-orly into nasopharynx (Lull, 1929; Pattersonet al., 1992).
201. Exposure of vomer in nasopharynx: (0) small,presphenoid and basisphenoid broadlyexposed; (1) very large, covers presphenoid andmuch of basisphenoid.
Distribution of apomorphies on tree shown inFigure 1. Characters and character states numberedaccording to character list provided in Appendix 2.Those characters and character states shown in boldtype are optimized as unambiguous synapomorphies,those in plain type are ambiguous synapomorphies.
Node 1. Pholidota apomorphies [Polarization ofcharacters at Node 1 based on comparisonwith a series of primitive fossil eutherians(Kielan-Jaworowska, 1980, 1981, 1984;Novacek, 1986; Qiang et al., 2002)]: 1(0),47(3), 61(1), 66(1), 90(2), 97(2), 134 (0)166(1), 168(1), 175(0), 182(0), E58(1),E82(2).Xenarthra + Palaeanodonta apomor-phies: 162(1), 163(0), E18(1), E23(0),E29(1), E57(1).Polarities indeterminate, or derived forboth Pholidota and Xenarthra +Palaeanodonta (Ph. = Order Pholidota;Xe. = Order Xenarthra; Pa. = Palaean-odonta): 54(2 = Ph., 1 = Xe. + Pa.), 68(1= Ph., 0 = Xe. + Pa.), 99(0 = Ph., 1 = Xe.+ Pa.), 121(24 = Ph., 0 = Xe. + Pa.),172(2 = Ph., 1 = Xe. + Pa.).
Pair-wise non-parametric Templeton tests (Templeton, 1983; Luo et al., 2002; – as implemented by PAUP 4.0b10,Swofford, 2002) on alternative hypotheses of sloth interrelationships. PAUP has been constrained to produce aset of MPT consistent with each of the alternative hypotheses listed below. These MPT are compared with the treeillustrated in Figure 1. Comparisons are carried out both on fundamental trees, i.e. sets of MPT, and on the strictconsensus trees derived from these MPT. Significance levels (P) are generated from the tree score function ofPAUP 4.0b10 (Swofford, 2002). Those hypotheses that are significantly different at P < 0.05 are marked by asingle asterisk (*); those significantly different at P < 0.01 are marked by a double asterisk (**).
Constraint hypothesis tested
A. Comparison of fundamental treesaB. Comparison of consensus treesb
# MPT TLc P TLd P
Bradypodid monophyly (as in Simpson, 1945; Hoffstetter, 1958)
3 1961(25)
0.0172–0.0202* 1963(24)
0.0214*
Bradypus + Megatheriidae (as in Guth, 1961; Webb, 1985)
1 1948(12)
0.2261 1948(9)
0.3679
Megalonychoidea, i.e. a clade including all sloths except Mylodontidae (as in Patterson & Pascual, 1972; Patterson et al., 1992; Webb, 1985)
3 1949(13)
0.0280* 1950(11)
0.0705
Choloepus + Mylodontidae (as in Guth,1961; Greenwood et al., 2001)
3 1952(16)
0.1397–0.1449 1957(18)
0.0908
Patterson et al. (1992), figure 24 1 1992 0.0002**(56)
1992 0.0005**(53)
Gaudin (1995), figure 2 1 1971(35)
0.0016** 1971(32)
0.0053**
McKenna & Bell (1997) 3 2003 <0.0001** 2004 <0.0001**(67) (65)
aIn order to simplify the testing procedure, the MPT from each of the alternative hypotheses is being tested against only oneof the two MPT used to construct the strict consensus tree in Figure 1. In the tree used for testing, the Santacrucian taxaPelecyodon and Schismotherium are sister-taxa. This clade in turn forms the sister group to remaining megatherioids.bThe TL of the strict consensus tree in Figure 1 is slightly greater than the TL of the two MPT from which it is constructed.The TL of the fundamental trees is 1936; the TL of the strict consensus tree is 1939.cThe number in parentheses represents the number of steps by which the MPT from each of the alternative hypothesesexceed the TL of the MPT from the present study (TL = 1936).dThe number in parentheses represents the number of steps by which the strict consensus tree from each of the alternativehypotheses exceeds the TL of the strict consensus tree from Figure 1 (TL = 1939).