Part III: Postcranial skeleton Larry G. Marshall & Denise Sigoc.neau-Russell** * * Institute of Human Origins, 1288, 9th Street, Berkeley California 94710, USA **Museum national d'Histoire naturelle Laboratoire de Paleontologie, URA 12 CNRS 8 rue Buffon. F-75005 Paris, France ABSTRACT The earliest and most complete articulated skeletons of fossil metatherians yet known arc represented by four specimens of Pucadelphys andinus from the Santa Lucia Formation (early Paleocene) at Tiupampa in southcentral Bolivia. Two sets of what arc interpreted to be male-female pairs were found in a three dimensional, life-like, snout-rump position in burrow-nests that were apparently dug in a bank along a meandering river. The animals probablydie d as the result of a flood which entrapped them in their burrows and tilled the latter with water and sediment. A detailed comparative study of the postcranial bones reveals that the vast majority of character states in Pucadelphys are regarded as mammalian, tribosphcnic ormetatherian plesiomorphies (e.g. atlas not perforated by transverse canal and with a persisting suture between the ossified intercentrum and atlantal arch; absence of transverse canal on axis, with possible unfused rib; absence of enclosed transverse canal onCV7; robust fibula; presence of ossified os marsupium; etc.). Character states of uncertain polarity include the presence of only one vertebra articulating with the ilium (fulcraI vertebra), and a long non-prehensilc tail. The tarsus has a bicontact upper ankle joint (UAJ) as in living Didclphidae; moreover the calcancum shows a partially plantar orientation of the cuboid facet which can be interpreted as foreshadowing the specialisation ol later Didelphidae; the situation then is more advanced than in the “plesiomorphic metatherian morphotype” of Szalay (1982 a. b; 1984); the only characters of the latter persisting in Pucadelphys are the large peroneal process and the “remarkably broad transverse dimensions from peroneal process to the medial margin of the sustentaculum". Collectively these characters support the view, based on the study of the skull and dentition (Marshall & Muizon, 1995), that Pucadelphys represents the plcsiomorphic taxon within the family Didelphidae. Functional considerations of the skeletons suggest that Pucadelphys was essentially terrestrial, quite agile, and possessed limited bounding and digging capabilities. Marshall. L. G. & Sigogneau-Russell, D.. 1995.— Part III: Postcranial skeleton. In: Muizon, C. de <ed.). Pucadelphys andinus (Marsupialia, Mammalia) from the early Paleocene of Bolivia. Mem. Mus. natn. Hist. licit.. 165 : 91-164. Paris ISBN : 2-85653- 223-3. LARRY G. MARSHALL & DENISE SIGOGNEAU-RUSSELL 92 RESUME Troisieme partie : Ie squelette postcranien Les squclettes les plus ancicns et les plus complets de metatheriens fossiles connus a ce jour sont representcs par quatre specimens de Pucadelphys andinus, en provenance de la Formation Santa Lucia (Paleocene inferieur), a Tiupampa. dans le sud de la Boli vie centrale. Deux ensembles de cc que nous avons interprete comme des couples male-femclle ont Cte conserves en trois dimensions, en position de vie. et disposes tete-bechc dans ce qui scmble avoir ete la berge d’un meandre fluviatile. Ces animaux sont apparemment morts a la suite d’une inondation qui les a pieges dans leur terrier en remplissant ce dernier d’eau et de boue. L'etudc detaillee et comparee dcs os postcraniens a revelc que la grande majorile des caracteres de Pucadelphys peuvent etre considers comme etant dans un etat plcsiomorphe pour les Mammiferes, les Tribosphenida et/ou les Metatheriens (e.g. atlas imperfore et gardant une suture entre l’intercentre ossifie et I’arche atlantale ; axis depourvu de canal transverse, mais possedant une cote libre persistante : absence de veritable canal transverse sur CV7 ; fibula robuste ; presence d’un os marsupium ossifie; etc.). L’existence d’une seule vertebre fulcrale (SI) et d'une longue queue non prehensile constituent, eux, des caracteres de polarite incertaine. L’articulation du tarsc est de type «bicontact» comme chez les Didelphidac actuels ; en outre la facette cuboi’de du calcanCum presente une orientation en partie plantaire qui peut etre interprets comme prCfigurant la specialisation des Didelphidae ultericurs ; ce tarse est done plus specialise que celui du «plesiomorphic metatherian morphotype» de Szalay (1982a.b ; 1984). ne le rappclant que par le grand processus peroneen du calcaneum et la largcur remarquable qui sCpare cc processus et le bord medial du sustentaculum. Prise dans son ensemble, cette analyse du squelette soutient Popinion, basee sur l'etudc du crane et de la denture, selon laquellc Pucadelphys represente le taxon plcsiomorphe a Pinterieur des Didelphidac. Les considerations fonctionnelles resultant de Petude de ces squelettes suggerent que Pucadelphys etait un animal essentiellement terrestre. plutot vif et capable, dans certaincs limites. de sauter aussi bien que de creuser. RESUME DEVELOPPE Les restes de squelettes postcraniens des mammiferes paleogenes sont extremcmenl rares et. qui plus est, difficiles a identifier en raison de Pabsence dissociation avec les dents, organes sur Iesquels est essentiellement fondee la taxonomic de ces animaux. Or les squelettes etudies dans cette troisieme partie sont non sculcment associes a dcs cranes munis de leur denture (voir Part. II de ce volume), mais ils sont tr£s complets (95% des os sont represents) et tres bien conserves. Ils sont enfin les plus anciens restes squeletliques connus de marsupiaux, cette position etant jusqu'ici tenue par des os tarsiens isoles du Paleocene superieur d’ltaborai. Bresil. L’etude detaillee de ces squelettes est done du plus haul interet. Les comparisons ont ete faites, d’une part avec les rares squelettes connus de mammiferes mesozoiques: I'eothericn Eozostrodon, du Jurassiquc inferieur d’AIriquedu Sud. lethcricn non tribosphenique Henkelotherium, du Jurassiquc superieur du Portugal, et les placentaires du CretacC inferieur de Mongolie ; avec d'autre part les petits didclphides generalises actuels : Metachirus, Monodelphis, Mannosa et Didelphis, avec aussi Perameles. La colonne cervicalc de Pucadelphys andinus est celle d’un petit didelphide actuel, si ce n’est que l’atlas montre la persistance d'une suture entre i'intercentre et fare neural, et que Taxis est depourvu de canal transverse et garde des cotes axiales, deux caracteres primitifs. La colonne thoracique est non moins gcneralisee dans son ensemble ; par contre la colonne lombaire est interpretee commederivCe. en raison de Tallongement progressif du corps vertebral et des apophyses transverses, et de la hauteur des epines neurales dirigees vers Tavant: une telle morphologie ne se retrouve pas chez les didclphides examines, mais bien chez la forme fouisseuse Perameles. Dans ce contexte. le sacrum est considere comme specialise, avec deux vertebres donl une seulement s’appuie sur T i lion. La longueur de la queue reste imprecise, mais les vertebres caudales conserves ne montrent aucune specialisation prehensile, contrairement a cellcs d’Henkelotherium ou de Didelphis. La ceinture scapulaire est identique a celle dcs didelphides actuels les plus primitifs ; Thumerus montre une vaste surface d’insertion pour les extenseurs, comme celui des petites formes terrestres ; Ic cubitus et le radius sont encore tres robustes. Les os du carpe et de la main n’ont pas ete conserves. La bassin, bien que massif, presente, comme le sacrum, des specialisations de type peramelide, avec une grande expansion dorso-ventrale de Taile antericure de Tilion, un grand foramen obturateur, et des os marsupiaux reduits. Au contraire, les os de la cuisse et de la jambe sont plus primitifs que ceux des petits didclphides actuels. L’astragalc n’est pas tres bien conserve, mais il semblc que sa morphologie etait plus plcsiomorphe que celle de Didelphis. Quant au calcaneum, il est depourvu de facette fibulaire ; il est done plus evoluc que celui defmissant le “morphotype metathcrien” de Szalay (1982a. b) et se rapproche du “morphotype didelphide” ; il conserve pourtant un fort processus peroneen. interprete dans cc contexte particulier comme un caractere plCsiomorphe. Un point interessant concerne Torientation de la facette cuboidienne, consideree comme annonQant la condition des didelphides actuels. Le pied lui-meme parait avoir etc long et relativemcnt rigide. Source MNHN, Paris PUCADELPHYS ANDINUS: POSTCRANIAL SKELETON 93 Dc P ensemble de ces caracteres, les auteurs deduisent pour Pucadelphys andinus un mode de vie terrestre et non arboricole (une adaptation considdrce comme primitive pour les Theria Tribosphenida. Krebs. 1991): I’astragale et le calcancum en particulier nc montrent pas les caracteres relids a unc telle specialisation ; malheureusement la configuration de lajonction tibia-astragale n'a pu etre precisee. nous ignorons done s’il y avait renversement du pied comme observe par Jenkins & Me Learn (1984) chez quelques didelphidcs arboricoles. Dans le detail, la brievetd des apophyses epineuses des vertebres cervicales suggere une bonne mobilite du cou. L’etroitesse de I’espace separant radius et cubitus imposait une rotation limitee de I’avant-bras. Le ddveloppement des metapophyses dorsales, la longueur des apophyses epineuses des vertebres lombaires et la largeur dc Pextremitd distale de I’humdrus constituent autant de potentialites fouisseuses ; de memc, la longueur des epincs lombaires et la conformation du sacrum (possible mobilite de la jonction sacro-iliaque, grand angle ilio-sacre, largeur et orientation de la surface iliaque destinee aux abducteurs et extenseurs de la cuisse) devaient favoriser le fouissement, sans qu’une telle specialisation soit. chez Pucadelphys, poussee aussi loin que chez Perameles. Par ailleurs les proportions relatives des membres superieuret inferieur sont compatibles avec une bonne agilite mais ne correspondent pas a celles d'un animal coureur; enfin cettc espece sembie avoir ete capable de sauter. mais sans etre reellement specialisee dans cette direction. En ce qui concerne le mode de vie. cette forme devait etre nocturne comme la plupart des petits didelphidcs. dormant le jour dans un nid-terrier ct cherchant sa nourriture durant la nuit. La connaissance des ma-urs des petits didelphides actuels (qui sont habituellement solitaires) conduit h penser que ces quatre squelcttes associes deux a deux representent ceux de couples formes durant la periode de reproduction ; les terriers etaient creuses dans la berge d'unc riviere, dont la crue subite a rempli ces nids de sediments ayant favorise la conservation en position naturelle. Cette conclusion est corroborec par l’abondance des fossi les trouves dans le gisement (comportant en particulier des squelcttes presque complcts de grenouilles). Peu de caracteres du squelette permettent de preciser la position phylogenique de Pucadelphys andinus, si ce n'est la configuration du tarse. Celle-ci, jointe aux conclusions concernant le crane et la denture, font considerer Pucadelphys comme le taxon le plus pldsiomorphe a I'interieur des Didelphidae. INTRODUCTION Postcranial bones of mammals are very important in phylogenetic studies; yet these elements are extremely rare in rocks of Cretaceous and Paleocene age, a time when the basic branches of this class became established. Associated skeletons which provide functional and phylogenetic information are even rarer, and those with associated dentitions and skulls are almost non-existent. Because of a dearth of such specimens, little is known of the early postcranial evolution of Tribosphenida (metatherians, eutherians and related forms with tribosphenic dentitions; sensu McKenna, 1975). In fact, the postcranial character states in the direct ancestor(s) of metatherians and eutherians are currently inferred from the study of all too few isolated Cretaceous and Paleocene elements, of Eocene to Pleistocene fossils, and of living taxa. There are, to date, only two non-tribosphenid therians (both eupantotheres) for which associated postcranial material is known. One is Henkelotherium guimarotae from the Late Jurassic of Portugal, described by Krebs (1987; 1991), and the other is Vincelestes neuquenianus from the Early Cretaceous of Argentina, which is not yet described (see Bonaparte & Rougier, 1987 ; Rougier et al., 1992 ; Wibble & Hopson, 1993). Eutherians are the best known of Late Cretaceous tribosphenids, and partial skeletons of Asioryctes, Zalambdalestes and Barunlestes have been described from the ? Late Santonian and/ or Campanian of Asia (Kielan-Jaworowska, 1977, 1978). In contrast, postcranial remains of metatherians from the Late Cretaceous and Paleocene are presently known only from isolated calcanea and astragali (Szalay, 1982a and b; 1984); associated skeletons have not been reported. The earliest nearly complete skeleton of a metatherian 94 LARRY G. MARSHALL & DENISE SIGOGNEAU-RUSSELL Fig. 22. —The “sarigue fossile” from Montmartre, Cuvier collection MNHN 7905 (top) and 7904 (bottom): opposite halves of same specimen, type of Peratherium cuvieri Fischer, 1829. X 1. Pig. 22. — La sarigue fossile de Montmartre, collection Cuvier MNHN 7905 (haul) et 7904 (has): moities opposees du meme specimen, type de Peratherium cuvieri Fischer, 1829. X 1. Source: MNHN. Paris PUCADELPHYS AND IN US: POSTCRANIAL SKELETON 95 SARIGUE fossile . Fig. 23. — The “sarigue fossile” from Montmartre (after Cuvier 1804. PI. 19). type of Percitheriuni cuvieri (Fischer. 1829). Fig. 23. — La sariguefossile de Montmartre (d'apres Cuvier 1804. Pi 19). type de Peratherium cuvieri (Fischer. 1829). Source MNHN. Paris 96 LARRY G. MARSHALL & DENISE SIGOGNEAU-RUSSELL was long represented by the classic “sarigue fossile” collected from the Butte Montmartre in northcentral Paris (Figs 22, 23). The fossil is from the “Gypse de Montmartre”, assigned to the Late Eocene (i.e. Headonian Land Mammal Age, Savage et al., 1994). It was first described by Cuvier (1804), named Didelphis cuvieri by Fischer (1829), and is now classified as Peratherium cuvieri (Crochet, 1980). Hence the major interest of the four nearly complete skeletons, two of which have articulated skulls, of the metatherian Pucadelphys andinus, from the Early Paleocene Santa Lucia Formation at Tiupampa in southcentral Bolivia. These are currently the earliest and most complete articulated skeletons of metatherians yet known, and they provide the first opportunity to securely assess aspects of the postcranial structure of a member of this group at the “Beginning of the Age of Mammals”. Unless otherwise specified, all numbers cited below, in the figure captions and in the Appendix (i.e. 6105, 6106, 6110, 6111) pertain to YPFB. SYSTEMATIC PALEONTOLOGY The skeleton-pairs described in this study were collected with articulated skulls, one of which (6105) was designated the type of Pucadelphys andinus Marshall & Muizon, 1988. A detailed study of the dentitions and skulls of 6105 and 6110 (as we! 1 as numerous other specimens), shows that both skeleton-pairs are referable to this species (Marshall & Muizon, 1995). The systematic position of the skeletons is thus as follows: Legion TRIBOSPHENIDA McKenna, 1975 Infraclass METATHERIA Huxley, 1880 Order DIDELPHIMORPHIA (Gill, 1872) Marshall et al., 1989 Family DIDELPHIDAE Gray, 1821 Genus PUCADELPHYS Marshall & Muizon, 1988 Pucadelphys andinus Marshall & Muizon, 1988 Diagnosis (postcranial skeleton morphology only). — Atlas not perforated by transverse canal, and with a persisting suture between ossified intercentrum and atlantal arch; absence of transverse canal on axis, with possible unfused rib; absence of enclosed transverse canal on CV7; lumbar series specialized compared to that of other didelphids (gradual lengthening of vertebral body and transverse processes, long anteriorly directed neural spines); single fulcral vertebra (SI); specialized pelvis (ilium dorsoventrally expanded anteriorly, large obturator foramen, small ossified os marsupium); possible movable sacro-iliac joint; long (±30 caudals) non-prehensile tail; digging specializations of the humerus (no third distal articular surface, large areas for extensors of forearm and carpus); robust fibula; calcaneum with bicontact upper ankle joint (U AJ) (Szalay, 1982a, b), large peroneal process and remarkably broad transverse dimensions from peroneal process to medial margin of sustentaculum. Source; MNHN, Paris PUCADELPHYS ANDINUS: POSTCRANIAL SKELETON 97 DESCRIPTION Measurements (\n mm) of the individual bones of the specimens of Pucadelphys described below are given in Tables 1 to 20 in the Appendix. General Features. — Skeleton-pairs YPFB Pal 6105 and 6106 (Figs 2, 24A and 25A). The animals are in a snout-rump position; 6105 faces to the right and 6106 to the left (referring to the positions as in the photos). Part of the dorsal surface and the entire ventral surface of both individuals are presently visible, and the latter was presumably on the floor of the burrow when the animals died and were fossilized. Both individuals are adults as evidenced by the facts that the skull of 6105 has a slightly worn adult dentition and the epiphyses, although still distinct, are all firmly attached to the diaphyses. 6106 is slightly larger than 6105 (see Appendix). On 6105, the seven cervical and eleven thoracic vertebrae are in a nearly straight line. The left forelimb is extended posteriorly and parallel along the body. The left hindlimb (as seen on the opposite surface of the block) has the femur on top of the thoracic region of 6106, the tibia and fibula extend posteriorly over the thoracic vertebrae to T11, and the pes is on the right side of that specimen. These features indicate that the pelvic region of 6105 was lying upon the upper thoracic, neck and possibly head region of 6106. On 6106. the body is in an arched position as shown by the arrangement of the articulated T1 to C9 vertebrae. The right forelimb lies along side and nearly parallel to the body and, as shown by the humerus, was extended posterolateral ly. The proximal part of the left forelimb is extended anterolaterally under the posterior lumbar-pelvic region of 6105, with the ulna and radius flexed sharply anteriorly along the posterolateral side of 6105. The right hindlimb is flexed anteriorly under the abdominal area. The left hindlimb is extended laterally, with the tibia and fibula under the thoracic region of 6105. Skeleton-pairs YPFB Pal 6110 and 6111 (Figs 3,4 and 26). The animals are also in a snout-rump position; 6110 faces to the right and 6111 to the left (referring to the positions as in the photos). The dorsal surfaces of both individuals are visible. Both individuals are subadults as demonstrated by the facts that the skull of 6110 has an unworn adult dentition and the epiphyses are unfused and often separated from the diaphyses in both individuals. 6110 is slightly larger than 6111 (see Appendix) and both individuals are notably smaller (average about 20%) than 6105 and 6106. On 6110, the pelvic area is nearly horizontal. The tail bends sharply dorsolaterally to the left and the end of the tail lies upon, and parallels, the thoracic vertebrae of 6111. In the posterior thoracic region the body begins to twist to the left, with the cervicals and head completely on their left side. The proximal end of the right forelimb is extended posterolaterally and the ulna is flexed sharply anteriorly. The left forelimb extends parallel along the posterior side of the proximal part of the right forelimb. The relationship of the forelimbs and cervical vertebrae clearly shows that the animal was lying on its left shoulder. The right hindlimb was extended nearly perpendicular to the body, while the left hindlimb was in a tightly anteriorly flexed position directly under the body. Source; MNHN, Paris 98 LARRY G. MARSHALL & DENISE SIGOGNEAU-RUSSELL Source: MNHN, Paris I'UCADELPHYS AN DIN US: POSTCRANIAL SKELETON 99 6111 was lying on its right shoulder with the right forelimb extending nearly perpendicular on the left side of the body (as seen by the position of the distal end of the humerus and proximal ends of the ulna and radius). The left forelimb was flexed tightly against the thoracic region of the right side of the body and the “elbow" abuts vertebrae T8 and T9. The proximal end of the right hindlimb extends anteriorly under the body such that the femur parallels the lumbar vertebrae and the distal end (tibia and fibula) extends nearly perpendicular on the left side of the body. The left hindlimb is extended anterolaterally on the left side of the body, and the distal parts of the tibia and fibula (and pes) lie upon the lumbar region of 6110. Collectively, the four specimens include about 95% of the complete skeleton. The only missing elements are the manus (see p. 127), the first metatarsal, some tarsals and phalanges of the pes, and an estimated 17 posterior caudal vertebrae. Axial Skeleton. 6105 7 cervicals (left half of atlas, axis, CV3-CV7) 9 thoracics (T1 -T9, fragments of T10-T11) 6106 13 thoracics (T1 -T13) 6 lumbars (L1-L6) 2 sacrals (SI-S2) 9 caudals (C1-C9) 6110 7 cervicals (atlas, axis, CV3-CV7) 8 thoracics (T6-T13) 6 lumbars (L1-L6) 2 sacrals (S1-S2) 9 caudals (C1-C5, C16?-C 17?, C207-C21 ?) 6111 7 thoracics (T7-T13) 6 lumbars (LI-L6) 2 sacrals (S1-S2) 9 caudals (C1-C9) Cervical Vertebrae Allas (Figs 25B, 27 and 28A, B; Table 1).— Two elements: 6105, left half (attached to axis), ventral and lateral views; 6110, isolated left and right halves, nearly complete. Fig. 24. — Pucadelphys cmdinus. Stereophotos. A, specimen-pair 6105-6106, partial view. X 1; B, 6106, right scapula, anterior view; thoracic vertebrae and ribs, ventral view. X 3. Fig. 24. Pucadelphys andinus. Stereophotos. A. coupledesspecimens 6105-6106, vueparlielle. X /; B. 6106, omoplatedroite. vue anterieure; vertebres thorciciques et cotes, vue ventmle. X 3. 100 LARRY G. MARSHALL & DENISE SIGOC.NEAU-RUSSELL Source: MNHN, Paris