Blackwell Science, LtdOxford, LtdOxford, UK ZOJZoological ZOJZoological Journal of the Linnean Society0024-4082The Society0024-4082The nean Society of London, 2004? 2004 140?? 140 487521 Original Article
Lin-
C. ARGOTEVOLUTION OF BORHYAENOIDEA
140,, 487–521. With 11 figures Zoological Journal of the Linnean Society, 2004, 140
Evolution of South American mammalian predators (Borhyaenoidea): anatomical and palaeobiological implications CHRISTINE ARGOT* Laboratoire de Paléontologie UMR 8569 du CNRS, Muséum national d’Histoire naturelle, Paris, France Received September 2002; accepted for publication October 2003
The evolution of South American carnivorous marsupials, the borhyaenoids, borhyaenoids, has been investigated through the functional analysis of postcranial adaptive traits and palaeobiological data. There is evidence that the evolutionary history of Borhyaenoidea proceeded from a noncursorial ancestor. The locomotion and habits of the early Palaeocene Mayulestes ferox probably approached a generalized plesiomorphic pattern for marsupial locomotion, i.e. primarily terrestrial with secondary arboreal adaptations. An exceptionally rich early Miocene Patagonian fauna has yielded various morphological predator types, from scansorial ambusher to terrestrial, incipiently incipiently cursorial, taxa. The most specialized borhyaenoid was the powerful sabretooth Thylacosmilus atrox that survived until the late Pliocene. The evolution from a scansorial pattern towards a cursorial trend, illustrated by Borhyaena tuberata, tuberata , and towards the probable development of postlactational parental care in Thylacosmilus Thylacosmilus,, required by its dental specialization and killing strategy, strategy, suggests a modification of the selective pressures and predatory activities of the group over evolutionary Zoological Journal of the Linnean Society, Society, 2004, 140, 487–521. time. tim e. © 2004 The The Linne Linnean an Socie Society ty of Londo London, n, Zoological
ADDITIONAL KEYWORDS: Cenozoic – functional anatomy – Metatheria – postcranial skeleton.
INTRODUCTION The superfamily Borhyaenoidea includes the marsupial predators that inhabited South America during the Tertiary while the continent was isolated. This superfamily has been revised by Marshall (1976, 1977a, 1978a, 1979, 1981) on the basis of dental remains, and unites ‘dog-like’ taxa (Hathlyacynidae, Prothylacinidae, Borhyaenidae, Proborhyaenidae) and ‘sabretooth’ taxa (Thylacosmilidae). More recently, the discovery of the earliest representative (so far as is known) of the superfamily, superfamily, Mayulestes ferox, ferox, from the early Palaeocene of Tiupampa (Bolivia), added the new family Mayulestidae (Muizon, 1998). The most common remains of borhyaenoids are teeth and jaws, and less than one third of the 35 known genera are represented by both skulls and postcranial remains.
*Current address: Institut für Spezielle Zoologie und Evolutionsbiologie Evolution sbiologie,, Erbertstraße 1, 07743 Jena, Germany. Germany. E-mail:
[email protected]
Dental morphology has therefore been used to determine the various taxa in most cases. Borhyaenoids maintained a position at the top of the South American food pyramid throughout the Cenozoic. They possessed a dental functional complex related to a hypercarnivorous diet, a complex that appeared several times independently in many other groups of mammals, especially thylacinids (see details in Muizon & Lange-Badré, 1997; Muizon, 1999). However, despite such superficial resemblances shared by some borhyaenoids and Australian thylacinids, due to a predatory mode of life, the phylogenetic analyses performed all reached the conclusion (irrespective of the methodology used) that borhyaenoids are more closely related to South American didelphoids than to thylacinids, this latter family being more closely related to dasyuroids (Marshall, 1977b; Archer, 1982; Szalay,, 1982). Szalay 1982) . The borhyaenoid specimens examined were as follows: Mayulestes lows: Mayulestes ferox MHNC 1249 (Tiupampan, early Palaeocene, Bolivia); Sipalocyon gracilis PU 015154, MACN 691–703, MACN 5938–49; Cladosictis pata-
140,, 487–521 © 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140
487
488
C. ARGOT
gonica PU 015046, PU 015170, PU 015702; Prothylacinus patagonicus PU 015700, MACN 706–720; Borhyaena tuberata PU 015701, PU 015120, MACN 2074–78, MACN 6203–6265 (these four taxa: Santacrucian, early Miocene, Argentina); Lycopsis longirostris UCMP 38061 (Laventan, middle Miocene, Colombia); Thylacosmilus atrox FMNH P14531, FMNH P14344 (Huayquerian, late Miocene, Argentina). Intraspecific variation and variation related to sex or age were impossible to assess for these taxa. Sipalocyon,, where Three taxa are less well known: Sipalocyon complete long bones, girdles, and the major part of the axial skeleton are unknown; Borhyaena Borhyaena,, in which most of the thoracolumbar and caudal vertebrae, the pelvic girdle and the hindlimb (except femur) are Thylacosmilus,, where scapula, ulna, unknown; and Thylacosmilus innominate, and most of the thoracolumbar vertebrae and all caudals are unknown. This paper has developed from studies of the postcranial adaptations of various borhyaenoids (Argot, 2001, 2002, 2003a, b, c; in press), and is an attempt to define the structural pattern characterizing these metatherians. Moreover, because the structuralfunctional solution of a lineage appears to be highly dependent on its ancestry, the evolutionary history of Borhyaenoidea is traced from its oldest known representative, Mayulestes ferox. This paper also reviews the existing literature dealing with habitats and faunas associated with borhyaenoids, in order to place these carnivorous marsupials within their palaeoecological framework. Institutional abbreviations: FMNH, Field Museum of Natural History, Chicago, USA; MACN, Museo Argentino de Ciencias Naturales ‘Bernardino Rivadavia’, Buenos Aires, Argentina; MHNC, Museo de Historia Natural de Cochabamba, Cochabamba, Bolivia; PU, Peabody Museum of Yale University, New Haven, USA; UCMP, University of California, Museum of Palaeontology, Berkeley, USA.
ROOTS The oldest borhyaenoid known is Mayulestes ferox from Tiupampa (Santa Lucía Formation, south-central Bolivia), a locality that has yielded a marsupial-rich and taxonomically diverse vertebrate fauna (Pascual & Ortiz Jaureguizar, 1991; Muizon, 1992; Marshall et al. al. 1995; Muizon, 1998; Muizon & Cifelli, 2000). The Tiupampan Land Mammal Age (Pascual & Ortiz Jaureguizar, 1990) is now commonly accepted to be early Palaeocene, about 63–64.5 Mya (Flynn & Swisher, S wisher, al ., 1995; Muizon, 1998). 1995; Marshall et al., The Tiupampa fauna includes only marsupials and placentals, with no indication of nontribosphenic taxa (Table 1). By contrast, the slightly younger Punta Peligro fauna, which represents the oldest Tertiary fauna
Table 1. List of Tiupampan mammals (Santa Lucía Formation, Bolivia) from Muizon (1992, 1998) and Muizon & Cifelli (2000, 2001) GONDWANADELPHIA MICROBIOTHERIA
Microbiotheriidae Khasia cordillierensi cordillierensiss DIDELPHIDA DIDELPHIMORPHIA
Pucadelphydae Pucadelphys andinus Andinodelphys cochabambensis ?Didelphidae Incadelphys antiquus Mizquedelphys pilpinensis Tiulordia floresi Jaskhadelphydae Jaskhadelphys minutus Mayulestidae Mayulestes ferox Allqokirus australis Family Famil y incertae sedis Szalinia gracilis
LEPTICTIDA Palaeoryctidae? cf. Cimolestes sp. PANTODONTA Alcidedorbignyidae Alcidedorbignya inopinata NOTOUNGULATA Henricosborniidae Henricosbornii dae or Oldfielthomasiidae Undetermined taxon PANAMERIUNGULATA Mioclaenidae Kollpaniinae Molinodus suarezi Tiuclaenus minutus, T. cotasi, T. robustus Andinodus boliviensis Pucanodus ga gagnieri Simoclaenus sylvaticus
ARCHIMETATHERIA
Peradectidae Peradectes cf. austrinum SUDAMERIDELPHIA
Caroloameghiniidae Roberthoffstetteria nationalgeographica
from Patagonia (the Peligran is about 61–62.5 Mya: Flynn & Swisher, 1995) includes a monotreme, suggesting biogeographical relationships with Australia and presumably Antarctica, and a multituberculate belonging to an endemic late Mesozoic -early Cenozoic South American or Gondwanan radiation (Bonaparte et al. al . 1993; Flynn & Swisher, 1995). The mammal fauna of Laguna Umayo (Peru), close to the Cretaceous–Palaeocene boundary (although the age of this fauna is still a matter of debate), contains some teeth and jaw fragments of marsupials and condylarths (Flynn & Swisher, 1995). The Chulpas fauna, also from the Umayo Formation but slightly younger than the Laguna Umayo fauna, contains marsupials and notoungulates (Flynn & Swisher, 1995). These faunas are much less diverse than that of Tiupampa and their taxa remain poorly known. Therefore, in the South American faunas from the Cretaceous -Cenozoic transition, the mammalian adaptive types are marsupials
140,, 487–521 © 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140
EVOLUTION OF BORHYAENOIDEA and placentals, the latter being represented mainly by various native ‘ungulates’ (notoungulates, condylarths), whose low-crowned cheek teeth indicate browsing herbivores more adapted to forested than to open habitats (Pascual & Ortiz Jaureguizar, 1990). This period corresponds to the earliest phase of isolation of the continent, and here begins the most autochthonous part of the history of mammals in South America (Pascual & Ortiz Jaureguizar, 1990). Placentals and marsupials of the early Tertiary of South America are commonly presumed to have dispersed southward from North America, various taxa suggesting biogeographical relationships with North America, and possibly other parts of Laurasia, during the Cretaceous -Cenozoic transition (Muizon, 1992; al ., Pascual & Ortiz Jaureguizar, 1992; Bonaparte et al., 1993; Muizon et al., al ., 1997; Flynn & Wyss, 1998). For example, the Tiupampa ‘condylarths’ (Kollpaniinae) and the litopterns are related to the North American Mioclaenidae, but they do not show close relationships with the other South American ‘ungulates’ (Astrapotheria, Pyrotheria, Notoungulata) (Muizon & Cifelli, 2000). By contrast, the marsupials of Tiupampa more closely resemble those known from the late Palaeocene Itaboraí fissures of Brazil, suggesting that they were part of an endemic radiation on the South American continent and therefore that they arrived before the ‘condylarths’ and pantodonts (Pascual & Ortiz Jaureguizar, 1991, 1992; Muizon, 1992; Muizon & Cifelli, 2001). Within the Tiupampa fauna, the placentals were herbivorous and partly omnivorous, whereas the marsupials were insectivorous-frugivorous, omnivorous, and carnivorous, an ecological distribution unique to South America (Pascual & Ortiz Jaureguizar, 1991). The metatherian fauna includes the oldest skulls and skeletons of undisputed marsupials of the American continent, Mayulestes continent, Mayulestes ferox, ferox, Pucadelphys andinus and Andinodelphys cochabambensis, whose adaptations help to shed light on the evolutionary history of various metatherian groups (Muizon & Argot, 2003). Several specimens of Pucadelphys of Pucadelphys and Andinodel phys were found with the majority of bones articulated, suggesting little postmortem dismemberment. The abundance of frogs throughout the deposits where the mammals have been found suggests that they lived near the bank of a river, and were trapped and died as a result of a catastrophic flood (Marshall et al., al ., 1995). A sudden flood is likely to have affected terrestrial species more so than arboreal ones, and it is significant that Pucadelphys andinus is both the least arboreally adapted taxon among the Tiupampan metatherians known from postcranial remains (Muizon & Argot, 2003), and the most abundant mammal species of this fauna (Marshall & Sigogneau-Russell, 1995), with 20 specimens found to date (C. de Muizon, pers.
489
comm.). By contrast, Mayulestes contrast, Mayulestes ferox is known from a single specimen and is the most arboreal of the three Tiupampan metatherians. The locomotion and habits of these three taxa were probably more similar to those of living dasyurids than to those of living didelphids, the former being more predaceous and faster animals. It has been hypothesized that the Tiupampan metatherians represent a generalized plesiomorphic pattern for marsupial locomotion, i.e. primarily terrestrial, with secondary arboreal adaptations more or less emphasized, which is especially true for Pucadelphys and Andinodelphys (Muizon & Argot, 2003). Mayulestes is characterized by more specialized arboreally adapted traits (Muizon, 1998; Argot, 2001, 2002, 2003a). The climbing ability facilitates escape into an arboreal refuge and also makes accessible sources of food and cover that are denied to strictly terrestrial mammals (Wemmer, 1977). Hence, these activities were probably selected by the marsupial taxa in order to avoid attack by terrestrial nonmammalian predators (mainly crocodiles), and to access the potentialities provided by a forested milieu. Mayulestes is separated by more than 45 Myr from the other borhyaenoids examined, and is much smaller. It weighed less than 1 kg (probably less than 500 g), a size at which an animal may be arboreal or scansorial but not cursorial, may be both predator and prey,, and always has to deal with many obstacles, prey obstac les, even when moving on the ground. By contrast, the Miocene taxa were larger, ranging from the size of a cat ( Sipalocyon)) to that of a jaguar ( Thylacosmilus ocyon Thylacosmilus)) (Argot, 2003b, c; in press). Therefore, it is likely that Mayulestidae are totally distinct from the later lineages that gave birth to the Miocene taxa; unfortunately, very few borhyaenoid taxa are known from the period in between. An unusual family of Borhyaenoidea, the Proborhyaenidae, is known from a few specimens that date from the Eocene and Oligocene, at a time when the representatives of other borhyaenoid families are rare (Marshall, 1978a). During the Casamayoran (early Eocene, about abou t 10 Myr after Mayulestes ), a probo Mayulestes), rhyaenid known from postcranial elements had already reached the size of the Santacrucian and Prothylacinus Borhyaena and Prothylacinus (J. Babot, pers. comm.). At the same time, the family Borhyaenidae is represented by one taxon, Angelocabrerus daptes. daptes. Similar in size to Borhyaena to Borhyaena tuberata according to dental measurements, it may represent an early specialized offshoot within the family, the massive P3 suggesting hyaena-like habits (Savage, 1977; Marshall, 1978a). The origins of Proborhyaenidae are unknown and the representatives of this family reached the size of very large bears and became extinct at the end of the Oligocene (Deseadan), simultaneously with very large
140,, 487–521 © 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140
490
C. ARGOT
herbivorous mammals. The differentiation and extinction of these gigantic predators and herbivores appear to be a conspicuous South American example of coevolution, their extinction being interpreted as the effect of decisive climatic-environmental changes (Bond & Pascual, 1983). Considering the fact that during the early Eocene two large proborhyaenid taxa already coexisted, and that the late Palaeocene Itaboraí fissures of Brazil have yielded Patene simpsoni and Nemolestes sp., representatives of two distinct families (Hathlyacynidae and Borhyaenidae, respectively: Marshall, 1978a, 1981), it may be hypothesized that early radiations, as yet unresolved, occurred within Borhyaenoidea.
DETERMINATION OF FUNCTIONAL ADAPTIVE CHARACTERS Although it has been shown that some regions of the postcranial skeleton can help to shed light on the systematics of various groups (see Szalay, 1994; for an analysis of metatherians based on tarsus evidence), it is extremely difficult to clarify the affinities between borhyaenoid families because of the poor sample of specimens known from postcranial elements. However, it is possible to determine whether a common structural pattern can be detected from studying the various responses by borhyaenoids to adapting to different modes of locomotion and hunting strategies under common phylogenetic constraints. Analysis of Mayulestes and of another Tiupampan metatherian, Pucadelphys (Muizon, 1998; Argot, 2001, 2002, 2003; Muizon & Argot, 2003) permits appraisal of possible ancestral retentions as well as of early specializations that occurred in the oldest borhyaenoid known. Two features, a long axial neural process and strong sagittal ventral crests on the axis, C3, and C4, are common to all borhyaenoids observed, and are probably related to their predatory habits. These two features are present in Mayulestes but not in Pucadelphys, and therefore could be qualified as ancestral retentions for borhyaenoids. Axial neural process A long and robust neural process of the axis (Fig. 1) is present in Mayulestes, Cladosictis, Prothylacinus , Borhyaena and Lycopsis. This feature is unknown in Sipalocyon and its presence is uncertain in Thylacosmilus, since the posterior part of the process is broken. However, its thinness and the orientation of the posterior margin (vertical above the postzygapophyses) in Thylacosmilus suggest a relatively reduced posterior extension of the process compared with the Miocene borhyaenoids. A long neural process is also present in marsupials like Dasyurus and Thylacinus,
and in all placental carnivores that feed upon prey smaller than themselves, and shake it when biting the neck, e.g. in felids (Pellis & Officer, 1987) and mustelids (Poole, 1974). By contrast, this posterior extension is not found in carnivores that prey upon animals bigger than themselves or scavenge like wolves and hyaenas. Although the M. obliquus capitis caudalis originates on the axial neural process and inserts on the transverse processes of the atlas, providing rotation and shaking movements of the head, it seems that the morphology of the atlas and axis is not directly related. The transverse processes of the atlas are indeed ovoid in shape in all borhyaenoids except Thylacosmilus, in which they are extended posteriorly (Argot, in press). This condition is similar to that observed in both Hyaena and Smilodon, two taxa where the axial neural process is different. According to anatomical data, the posterior extension of the atlantal transverse processes is probably related to the pull of M. scalenus, a powerful flexor of the neck.
Sagittal ventral crests on the axis, third and fourth cervicals These crests (Fig. 1) form strong triangular processes, a feature that characterizes borhyaenoids. In the Palaeocene Mayulestes the ventral process is incipient on the axis (a condition possibly related to its small size); the other cervicals are unknown. Although incipient processes can also be observed in modern taxa like Thylacinus cynocephalus, Canis lupus and predatory viverrids like Poiana richardsoni, their distribution among taxa and development does not reach that observed in the Miocene borhyaenoids where they are present in all members. The development of these processes was either related to the pull exerted by the long flexors of the neck (Mm. longus capitis and longus colli), or to the presence of strong ventral ligaments not observed in modern taxa, equivalent to the dorsal ligamentum nuchae. In Thylacosmilus the attachment of the M. longus capitis forms rugose scars on the basicranium, in front of the foramen magnum. These flexors probably acted as strong depressors of the head, useful in increasing bite strength, and their development is therefore probably related to carnivorous habits.
Two features, an elongate deltopectoral crest and an asymmetrical metacarpophalangeal joint on the pollex, are present in most borhyaenoids observed and related to manipulation of the forelimb. Deltopectoral crest length In all Miocene borhyaenoids, the deltopectoral crest (Fig. 2) is extremely long (i.e. >60% of the humerus
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
EVOLUTION OF BORHYAENOIDEA
491
axial neural process C7
C7
C5
B
A
C7
D
C
C7
F
E prominent ventral processes
incipient ventral process
Figure 1. Skull and cervical vertebrae in lateral left view, showing the anteroposterior extension of the axial neural process, and the development of strong ventral triangular processes on the axis , C3, and C4 (arrowed). A, Borhyaena tuberata PU 015701 (the fourth cervical is unknown) modified from Sinclair (1906). B, Lycopsis longirostris UCMP 38061 modified from Marshall (1977a). C, Prothylacinus patagonicus PU 015700 (the sixth and seventh cervicals are unknown) modified from Sinclair (1906). D, Cladosictis patagonica PU 015046 (skull) and PU 015170 (cervicals) modified from Sinclair (1906). E, Thylacosmilus atrox FMNH P 14531 (skull, modified from Riggs, 1934), and FMNH P 14344 (cervicals). F, axis in Mayulestes ferox MHNC 1249. The natural curvature of the cervical area is preserved only in Lycopsis (B) and Cladosictis (D). Scale bars: 50 mm in A -E, 5 mm in F.
length). It is particularly strong in two medium to large-sized borhyaenoids that exhibit arboreal capabilities ( Prothylacinus ) or manipulate heavy prey (Thylacosmilus). In these two taxa the crest is prominent anteriorly and ends distally with a strong tuber (Argot, 2003b; in press). This condition has no equivalent in living marsupials. The proximal anterior pro jection of the crest is obviously related to the development of the greater tubercle (that protrudes anteriorly in all borhyaenoids in which it is known: Fig. 2, top), but not the shape of the distal end. In Lycopsis and Cladosictis the crest is extremely long but not prominent distally. Similarly, in the Australian Thylacinus the insertion of the pectoralis is extremely
long but the crest is not prominent. In Sipalocyon and Borhyaena the humerus is fragmentary. In Mayulestes (Fig. 2A) the crest is relatively shorter than in the other borhyaenoids (55% of the humerus length) but an allometric effect cannot be discarded. In living marsupials, the M. deltoideus pars acromialis inserts on the proximal third of the crest (anterolateral deltopectoral area), the pectoralis along the whole crest (anteromedial side). Although the crest does not appear to be particularly strong in living Carnivora, the insertion of the pectoralis is extremely long in taxa where the forelimb retains skilful manipulative capabilities, e.g. felids and ursids (Argot, 2003b). This suggests that borhyaenoids also had good manipulative capabilities.
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
492
C. ARGOT greater tubercle
lesser tubercle
bicipital groove
humeral head
delto pectoral crest length
lateral epicondylar crest
A
B
C
D
E
Figure 2. Humerus in proximal (top) and lateral (bottom) views, showing the variable development of the greater tubercle and deltopectoral crest. A, Mayulestes ferox MHNC 1249. B, Prothylacinus patagonicus PU 015700. C, Thylacosmilus atrox FMNH P 14531. D, Lycopsis longirostris UCMP 38061. E, borhyaenoid UCMP 39250. Not to scale.
Metacarpophalangeal joint of the pollex The very asymmetrical, trochlear-shaped metacarpophalangeal joint of the pollex (Fig. 3) of borhyaenoid taxa in which it is preserved seems to have no equivalent. On the proximal phalanx, the outer (radial) articular facet is transversely narrow but is dorsoventrally long, and exhibits a longer arc of curvature than the inner (ulnar) condyle, which is wider and shorter. This asymmetry is reflected on the distal epiphysis of Mt I and the articulation indicates that flexion -extension of the pollex did not take place in a strict parasagittal plane. This condition is found in Sipalocyon, Cladosictis (Argot, 2003c), and Lycopsis (three taxa very different in size, age and adaptations). In Thylacosmilus, the distal epiphysis of the first metacarpal is not trochlear-shaped as in these three genera, and the proximal phalanx is unknown. However, the distal articular facet of Mc I is much more asymmetrical than in the other metacarpals and also suggests that the pollex was pseudo-opposable (Argot, in press). This condition is unknown in the other borhyaenoids examined because of lack of elements, and no
borhyaenoid has yet been found with a metacarpophalangeal joint similar to that of the other digits. Hence, the acquisition of hand dexterity may have had an important influence on the evolution of feeding and foraging strategy of Borhyaenoidea. The use of the forepaws in grasping prey and pinning it to the ground seems to be a feature that has evolved several times within various orders of mammals (Eisenberg & Leyhausen, 1972; Iwaniuk et al., 1998). According to Eisenberg & Leyhausen (1972), it is present in both the Dasyuridae and those eutherian species, like felids, which are arboreally adapted and retain some grasping ability in the forepaw. Where the forepaw is still employed in grasping, some ability to pronate and supinate it is retained. By contrast, digitigrade viverrids that have little ability to pronate or supinate the forelimbs (e.g. Civettictis, Viverra, Ichneumia and Fossa) never employ them in manipulating prey during feeding. The combination of these two groups of characters (cervical morphology and forelimb dexterity) is not found in any other group of mammals. Other postcranial features vary within the Borhyaenoidea according
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
EVOLUTION OF BORHYAENOIDEA
493
Mc I proximal phalanx ungual phalanx I II V
I
III
IV
III
V IV outer articular facet
II
inner articular facet
proximal phalanx
Mc I unknown
Mc I B
A
V
I
I
IV IV
III
II
II III
proximal phalanx unknown
transverse axis of proximal epiphysis
C
D
transverse axis of distal epiphysis
Figure 3. General morphology of the manus in various borhyaenoids, focusing on the first metacarpal and associated proximal phalanx, showing the asymmetrical metacarpophalangeal joint. A, Cladosictis patagonica PU 015046 (manus as found in matrix); proximal phalanx in dorsal and proximal views, Mc I is in dorsal and ventral views. B, Sipalocyon gracilis PU 015154 (manus as found in matrix); proximal phalanx in proximal view. C, Lycopsis longirostris UCMP 38061 (association of phalanges is conjectural); proximal phalanx in dorsal, ventral and proximal views, Mc I in dorsal, ventral and distal views. D, Thylacosmilus atrox FMNH P 14531 (association of phalanges is conjectural); M c I in dorsal, ventral and distal views. Not to scale. © 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
494
C. ARGOT
to the mode of locomotion and substrate used. Some of these characters, like the curvature of the posterior border of the ulna, the development of the medial epicondyle and of the lateral epicondylar crest of the humerus, and the orientation of the ectal facet on the calcaneum indicate the primary locomotor category (arboreal or terrestrial) that characterizes a taxon (Table 2). Humeroulnar joint In extant taxa, the morphology of the humeroulnar region (Fig. 4) reflects the stability of the elbow joint. In Prothylacinus the humeroulnar joint is poorly stabilized compared with Borhyaena, in which the humeral trochlea is narrower and more concave posteriorly, with a medial margin that is more prominent anteroposteriorly as seen in distal view (Fig. 4, top).
This condition suggests arm movements more restricted to a parasagittal plane in Borhyaena. Seen in distal view, the humeral trochlea is also better developed anteroposteriorly in Cladosictis than in Prothylacinus , which suggests the need for better stabilized movements in the former, smaller taxon. The distal extremity of the humerus of Lycopsis is poorly preserved, but the humeral trochlea is shallower than in Borhyaena, and on the ulna the anconeal process is less prominent anteriorly. In the Palaeocene Mayule stes, the trochlear notch on the ulna is extremely open, which precludes well-stabilized movements. Medial epicondyle and lateral epicondylar crest of the humerus The medial epicondyle is prominent (i.e. the distance between the medial lip of the trochlea and the apex of
entepicondylar foramen
medial epicondyle
A
B
capitulum
C
D
trochlea
E
Figure 4. Distal extremity of the humerus in anterior (top) and distal (middle) views, and ulna (bottom) in lateral view, showing: (1) the presence or absence of the entepicondylar foramen; (2) the protrusion of the medial epicondyle of the humerus; (3) the curvature of the posterior border of the ulna. A, Mayulestes ferox MHNC 1249. B, Cladosictis patagonica PU 015702. C, Prothylacinus patagonicus PU 015700. D, Borhyaena tuberata MACN 2074–78 (humerus) and PU 015701 (ulna). E, Lycopsis longirostris UCMP 38061. Not to scale. © 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
EVOLUTION OF BORHYAENOIDEA the medial epicondyle represents more than 30% of the transverse width of the distal humeral extremity) in Mayulestes, Cladosictis and Prothylacinus (Fig. 4). This distance is reduced in the three borhyaenoids that are primarily terrestrial ( Borhyaena, Lycopsis and Thylacosmilus). In living taxa, the medial epicondyle, where the deep flexors of the digits originate, is not prominent in the most terrestrial forms that exhibit reduced manipulative capabilities of the hand, whatever the group considered (Argot, 2001, 2003b, c). The entepicondylar foramen is absent in two taxa, Borhyaena and Thylacosmilus. Landry (1958) suggested that cursorial and ungulate mammals might have lost the entepicondylar foramen in relation to reduced abduction of the humerus. This could fit well with its absence in Borhyaena, the most terrestrial borhyaenoid examined (Argot, 2003b) but not in Thylacosmilus which apparently had powerful adductors. In all borhyaenoids the lateral epicondylar crest is well-developed. Its length relative to that of the humerus is unknown for Sipalocyon and Borhyaena, for which there is no complete humerus known. However, the preserved part of the crest of Borhyaena suggests that it was relatively reduced compared to Prothylacinus . This is consistent with a reduced crest in the most terrestrial living marsupials examined (e.g. Metachirus, Thylacinus). However, a correlation between the reduction of the crest and of the flexor mass originating from the medial epicondyle is not established in all taxa. For example, the crest is extremely well-developed in Thylacosmilus and Proth ylacinus, whereas the medial epicondyle is much less prominent in the sabretooth form than in Prothylacinus. This could relate to the numerous and antagonist muscular demands placed upon the crest. Posterior border of the ulna A convex posterior border of the ulna (Fig. 4), characteristic of Mayulestes, Sipalocyon, Cladosictis and Prothylacinus, suggests arboreal capabilities. As explained in the study of the forelimb of Mayulestes (Argot, 2001), this condition indicates the combined action of the extensors and flexors of the forearm (triceps brachii caput longum, biceps brachii and brachialis), while they exert powerful tractions both to flex the arm to bring the body of the animal closer to the vertical support, and to pull the body up against gravity (see also Szalay & Sargis, 2001: 183–184). A straight or concave posterior border of the ulna is present in Lycopsis and Borhyaena within borhyaenoids as well as in Thylacinus, i.e. in predatory mammals that are primarily terrestrial and may exhibit an incipient cursoriality. The last step, i.e. a posteriorly concave ulna, is achieved in modern canids and hyaenids (Argot, 2003b).
495
Orientation of the ectal facet of the calcaneum This facet (Fig. 5) is orientated medially in Mayulestes (Argot, 2002) and Lycopsis, a condition which suggests that in a resting posture the plantar side of the foot faced medially in these two taxa, as when appressed against a curved support. In contrast, it is orientated dorsally in Borhyaena, which suggests that in a resting posture the plantar side of the foot lay on a horizontal support. The inclination of this facet is intermediate in Sipalocyon and Thylacosmilus.
The orientation of the femoral head, the width of the femoral condyles and the shape of the tibia denote varied mechanical adaptive solutions related to distinct postures and/or hunting strategies: Orientation of the femoral head The femoral head is much more prominent proximally in Thylacosmilus (Fig. 6I) than in the other borhyaenoids. A similar orientation is also found in bears and suggests the possibility of erect, semibipedal postures while attacking particularly large prey, in order to allow the predator to encircle the prey’s neck with its forelimbs (Argot, in press). However, all sabretooth forms that are expected to perform similar throat attacks on large prey do not exhibit a similar orientation of the femoral head (e.g. in Smilodon the orientation of the femoral head and the height of the greater trochanter are much more similar to a felid pattern: Argot, in press). It seems, therefore, that the peculiar nature of loading exerted on the hindlimb musculoskeletal system clearly results in adaptive modifications within inherited constraints, which are distinct between Thylacosmilus and Smilodon. Femoral condyles Both femoral condyles are approximately equivalent in size in most borhyaenoids, except in Borhyaena in which the medial condyle is wider than the lateral one. Although the medial condyle of Mayulestes is poorly preserved in both femora known, in distal view the lateral condyle is wider than the medial one, but the difference is not so emphasized as in didelphids, microbiotheriids and phalangeriformes. In these groups, the lateral condyle is approximately twice as wide as the medial one whereas in dasyuromorphs, the condyles are equivalent in width. The relative width of the femoral condyles emphasizes the nature of motions that are possible between the thigh and the crus, and the nature of loading in the knee joint (see in particular Szalay & Sargis, 2001). Unequal-sized femoral condyles reflect the locomotor repertoire of highly arboreal marsupials, or of marsupials that likely have an arboreal ancestor, like Metachirus within Didelphidae (Argot, 2002). It is likely that the loading at the knee joint was more similar between Mayulestes and
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
496
C. ARGOT l a i d e m l a i b i t o l a g a r t s a , m i T A ; t e c a f l a r e t a l l a i b i t o l a g a r t s a , l i T A : s
n o i t a i v e r b b A
. a e d i o n e a y h r o B n a i r d b e n t u r o e v f c e i t c a a t s r o r h e t t , c T a r ; a a r h c b e l t a r i n e v a r c r a t s b o m p u l l a , c L i g ; o a l o r h b e t p r r o e m v f l a o i c t v s i r L e c . , 2 C e ; l t b e a c a T f
n o s s t r n t e e c m a m r a o h C c
d e d n a d k o y o h e e i e e c e t t t t v l s e i h f a i t d a n o l l d b u t i n e l l e r e a a t b s t h l r a n s e t a u e a o o o r u s i o l s f f l i r t i t p e g p i t l i o n r p u r e r r p o l n a f s r i r d n i s o o o - x i e r o t r e s r p s w n t o a b i e o o e o a t n r l e t l s a u a w p x i c m p d l t a e r e o e o t r x o t o g o a u s e p t c e o r p e r o r t fl n s p g o p p l p
s u l i m s o c a x l y o r h t T a
g n o l y l r o i r e t s o p o r e t n a
s i r t s s i s o r p i o g c y n o L l
y l r o i r e t s o p o g r e n o t l n a
t d n e n i a n g m n o r o r p t s
a n e a t a y a r h e r o b B u t
y l r o i r e t s o p o r g e n o t n l a
t d n e n i a n g m n o r g o r p n t o s l
s u s n u i c c i a n l y o g h t a o r t a P p
y l r o i r e t s o p o r g e n o t n l a
t d n e n i a n g m n o o r p r t s
s a i c t c i i n s o o d g t a a l a C p
y l r o i r e t s o p o r g e n o t n l a
t d n e n i a n g m n o r g o r p n t o s l
r l a e l a c h i r c o t r d e t e m d p a m n h % y s a s 2 a 3
n o y c s o i l l a i p c a i S r g
?
t d n e n i a n g m o n r o p r t s ?
l a c r i r a t l e e d h e m d c p a m n o r h y a t s s a ?
s e t s e l u y x a o r e M f
y l r o i r e t s o p o r g e n o t n l a
e h t n o t n e s i p i x i c a n i
s s e c o r p l a r u e n l a i x A
, s i t x s l a e r a e c r h l t t a n e n r o v o 4 t l s c e t C a t s , p t o i e 3 t r g c C l e a S D
t n e n i m o r p d n a g n o r t s
t s n n o a e o d i h m x n a s a t e t u g l r o h d n h t t n a i l t a p n i o u s i t r r d f p e r s y a d s t l t e r r i o e l c e r a l a w x g i i o e o b u r d e e t e r p fl t a d e m r
t n e n i m o r p y d l n l a a t s g i d n o l
d e p a t h u s b r l a e a l c h i r t c o e r t m t m o y s n a
g n o l
r l a e a c l h i c r t o r d e t e m d p a m n h % y s a s 8 a 2
t n e y n l d i l n m a a o t s i g r n p d o l
?
?
t o r n a e l s y h e t l d o a n ) d h o t ( c r c s e i t a t e r p a o e r l e r c h s ? l a e x e g l n l a l o a p h e p h o t f p r o a c t n a i t o e j M
% 8 1
% 5 2
% 3 3
% 5 . 3 3
% 2 3
?
% 0 4
% 1 3
?
r n t a o t t s a e c h f i o x l s g e e o c b i u r m a r m e e r e e h n t t t e l e i d o l w n t u l h i t h o o d a r a s l i e i t i r t n e c o r t p t n s r o d f u t h e g j o n l d i n t o t o s r b a i r s t l a n e e o r t s i u s e a c t a t t e f m t e x a e o l a c s c d e fi x l n n r e i l l w r e a e n i e o a l o v x e c e a e n e h o r s t l k n g r r o p fl b o r p fl a p t o i e a c p p r s
e t a i d e m r e t n i
y l t l n a m e i n x i o m r o r p p
h t d i w n i t n e l a v i u q e
l a i d e m
t n y e l n l a i m m i o r x p o t r o p n
] d e g a m a d [
l a s r o d
t n y e l n l a i m m i o r x p o r t o p n
e l y d n o c l r e a d i d i e w m ?
x e v n o c
?
t n y e l n l a i m m i o r x p o t r o p n
n i t n e h l a t d v i i u w q e
t h g i a r t s
x e v n o c
e t a i d e m r e t n i
t n y e l n l a i m m i o r x p o t r o p n
n i t n e h l a t d v i i u w q e
t h g i a r t s
x e v n o c
e t a i d e m r e t n i
?
n i t n e h l a t v d i i u w q e ?
l a i d e m
t n y e l n l a i m m i o r x p o t r o p n
e l y d n o c l r i e d a o d i r m e t w i g a l s
t n e s b a ?
t n e s e r p
t h g i a r t s
t t h g n i e s a r b t a s
t n e s e r p
t n e s e r p
t n e s e r p
t n e s e r p
x e v n o c
% 3 3
% 3 3
s e a h d e e t f s l y s o u d s e r e e n r g o a m p t c i x u p e n h e h e c l h r a t l t a g e t d i s i n p i d e a d w e l M
r f a o l y s u d r r r e n e l a d o r y c o m i u d b a p n n e H n r l e o h o u l / c m i t t i a r g s a p r e n e r e e e r e t o t h t a c l n f s o t L E P
e l e h a h t t t d f c f a o e o e n l n h o a o l i i e t a t n a a r t a o t c e t n l c n m e e a a e i r c f i r f O O
s e l y d n o c l a r o m e F
d i o m g i s
t h g i a r t s
e p a h s a i b i T
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
498
C. ARGOT
CaFi
CaA Su
CaFi CaA
Su
Su
CaA
peroneal process
orientation of the ectal facet
CaA
CaFi
Su peroneal process A
B
probable attachment of a calcaneocuboid ligament
CaCu C
CaCu D
E
Figure 5. Calcaneum in dorsal (top) and anterior (bottom) views, showing the orientation of the ectal facet (arrowed). Abbreviations: CaA, calcaneoastragalar facet; CaCu, calcaneocuboid facet; CaFi, calcaneofibular facet; Su, sustentacular facet. A, Mayulestes ferox MHNC 1249. B, Lycopsis longirostris UCMP 38061. C, Sipalocyon gracilis PU 015154. D, Thylacosmilus atrox FMNH P 14344. E, Borhyaena tuberata MACN 2074–78. Not to scale.
living arboreal forms than between Mayulestes and younger borhyaenoids, especially Borhyaena in which the wide medial condyle suggests a medial displacement of the load line and more parasagittal movements of the crus.
Shape of the tibia In borhyaenoids, only Mayulestes and Thylacosmilus have a sigmoid tibia (Fig. 7A, D). A peculiar sigmoid curvature characterizes the tibia of didelphids, microbiotheriids and caenolestids. This condition probably represents a primitive therian or pretherian feature and this curvature is not likely to be reliably indicative of functional attributes that can be related to substrate preference or locomotor mode (Szalay & Sargis, 2001: 166, 206–209). Moreover, a relation between a sigmoid tibia and asymmetrical femoral condyles cannot be ascertained. For example, Thylacosmilus and Thylacinus exhibit a sigmoid tibia with femoral condyles subequal in width, whereas in the small Mayulestes the sigmoid tibia is associated with a slightly wider lateral condyle.
The position of the anticlinal vertebra, the presence/ absence of a femorofibular contact, the tibioastragalar joint, the development of the calcaneofibular facet, of the peroneal process and of the first metatarsal appear to be related both to the locomotion and to the evolutionary history of Borhyaenoidea. For these characters, the case of Mayulestes and other Tiupampan metatherians is therefore of particular importance, in order to determine the ancestral state of these features.
Anticlinal vertebra The anticlinal vertebra is L3 or L4 in Mayulestes. It is likely to be an ancestral retention, as it is L2 in the two other Tiupampan metatherians, Pucadelphys and Andinodelphys (Muizon & Argot, 2003), whereas it is a thoracic vertebra in Miocene borhyaenoids, T?11 in Cladosictis and Prothylacinus , and T11 in Lycopsis. The anticlinal is unknown in Sipalocyon. In Borhyaena, this condition is unclear because of the poorly preserved axial skeleton, but the orientation of lumbar neural processes suggests that the anticlinal was a lumbar vertebra (Argot, 2003b). The two last lumbars are known for Thylacosmilus, and their vertical neural processes suggests that there was no anticlinal vertebra in this taxon, as in Smilodon (Argot, in press). The flexibility of the lower back of the two sabretooth taxa appears to be noticeably less than in their closer relatives. In the lumbar and posterior thoracic regions, the M. longissimus dorsi (the most powerful extensor of the back) originates by powerful tendons either from the apex of neural processes (which are then usually inclined anteriorly), or more laterally, from the mammillary processes. In this latter case, the neural processes usually stand vertical (Argot, 2003a). The forces generated by these symmetrical muscular attachments primarily act to stretch or stabilize the vertebral column (Shapiro & Jungers, 1994), but also contribute substantially to body propulsion by additive sagittal spine movements allowed by thin and strongly inclined neural processes (Schilling & Fischer, 1999). Therefore, the position of the anticlinal vertebra and the inclination of the posterior neural processes is related to the flexibility of the back.
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
EVOLUTION OF BORHYAENOIDEA
499
ilium prominent dorsally
greater trochanter
A E
F
G
H
L6
B
C
gluteal fossa short iliac neck iliac blade
I
J
iliacus fossa acetabulum D
antero-inferior iliac spine (m. rectus femoris origin)
Figure 6. Innominate in lateral right view and proximal extremity of the femur in anterior view in various borhyaenoids, showing: (1) on the innominate, the general shape and orientation of the ilium, and the development of the anterior inferior iliac spine; (2) on the femur, the height of the greater trochanter and the orientation of the femoral head. A, E, Mayulestes ferox MHNC 1249. B, F, Cladosictis patagonica PU 015702. C, J, Lycopsis longirostris UCMP 38061. D, G, Prothylacinus patagonicus PU 015700. H, Borhyaena tuberata PU 015701. I, Thylacosmilus atrox FMNH P 14531. Not to scale.
Femorofibular articulation The fibula exhibits a broad and distinct articular facet with the femur in didelphids, phalangeriformes and microbiotheriids. By comparison, this facet is reduced in dasyuromorphs. In Mayulestes, the lateral expansion of the lateral femoral condyle suggests a femoro fibular contact in contrast to Sipalocyon, although in the latter taxon the fibular head is prominent proximally. In Cladosictis and Lycopis, a potential contact between the two bones cannot be ascertained because of their poor state of preservation. However, the fibular head of Cladosictis, Sipalocyon and Lycopsis is
more prominent proximally than in Prothylacinus and Thylacosmilus. The absence of femorofibular articulation as observed in Prothylacinus and Thylacosmilus (Fig. 7B, D) is likely to be a derived feature among metatherians (Haines, 1942; Barnett & Napier, 1953; Szalay, 1994).
Tibioastragalar joint A sharp angle between the lateral and medial astragalotibial facets is observed in all borhyaenoids, in Pucadelphys and in various Palaeogene Itaboraian taxa
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
500
C. ARGOT no femoro-fibular contact
convex medial side
A
fibula
B
C
D
deep tibial malleolus
Figure 7. Tibia and fibula, underlining the general shape of the tibia. A, Mayulestes ferox MHNC 1249. B, Prothylacinus patagonicus PU 015700. C, Lycopsis longirostris UCMP 38061. D, Thylacosmilus atrox FMNH P 14344. Not to scale.
(Szalay, 1994). As emphasized by Szalay (1994: 329) this pattern is primarily suited for flexion -extension rather than abduction and inversion and therefore, the tibioastragalar joint of borhyaenoids is anatomically more severely constrained than in many living marsupials like didelphids, phalangeriformes and microbiotheriids. A perpendicular angle between the medial and lateral astragalotibial facets, present in all the borhyaenoids examined whatever the adaptive features of the other parts of the skeleton, may represent a minor modification of an ancestral metatherian condition (Szalay, 1994). More recently, Szalay & Sargis (2001: 203) suggested that a narrow astragalotibial lateral facet bordered by a sharply angled astragalotibial medial facet (which reflects the exceptionally deep tibial medial malleolus), a narrow astragalofibular facet, a huge astragalar medial plantar tuberosity (ampt), and a proximodistally long sustentacular facet extending dorsal to this ampt together represent the primitive sudameridelphian pattern. This pattern can be observed in all borhyaenoids despite the alteration of surrounding elements, especially modifications to the calcaneoastragalar articulation (particularly to the orientation of the ectal facet), or to the orientation of the tibial malleolus, which in Mayulestes exhibits a torsion relative to the proximal epiphysis (Muizon, 1998; Argot, 2002). This fea-
ture is unknown in the other borhyaenoids for which a complete tibia is known ( Cladosictis, Proth ylacinus, Lycopsis, Thylacosmilus). This condition might be related to the capability to invert and evert the foot in Mayulestes, which is consistent with the orientation of the ectal facet of the calcaneum and suggests the ability to walk on irregular supports.
Calcaneofibular facet In borhyaenoids, the calcaneofibular contact is extensive in Sipalocyon and Thylacosmilus (Fig. 5C, D) and considered an ancestral therian condition (Szalay, 1994: 209). However, this may represent a secondarily extended contact, since in the oldest borhyaenoid known, Mayulestes, the calcaneofibular facet is smaller than in Sipalocyon. Despite the absence of femoro-fibular contact in Thylacosmilus, the wide calcaneofibular facet suggests that the fibula in this taxon is still distally a weight-bearer of importance. In Prothylacinus , the calcaneum is unknown. The preservation of the calcaneoastragalar facet in Borhyaena and Cladosictis does not allow us to know if there is also a calcaneofibular facet. In Lycopsis, although slightly damaged, the general shape of the calcaneoastragalar facet does not suggest the presence of a calcaneofibular facet.
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
EVOLUTION OF BORHYAENOIDEA Peroneal process In many Palaeogene taxa, the peroneal process (Fig. 5) is well-developed (Szalay, 1994: 199), and so it is in Pucadelphys and Mayulestes. In the other borhyaenoids this process is reduced compared with Mayulestes. In Lycopsis and Borhyaena no process is observable, in contrast to Sipalocyon in which it is reduced but present. In Prothylacinus , the calcaneum is unknown. The reduction of the peroneal process is a common event which occurred at least two or three times in the Metatheria (Szalay, 1994: 345). Functionally, a laterally extended peroneal process potentially increases the leverage of the M. peroneus longus, an abductor of the hallux. Hallux A powerful grasping hind foot with a widely divergent hallux (Fig. 8) characterizes both the didelphid-microbiotheriid and the primitive syndactylan patterns (Szalay, 1994). The reduction of the hallux appears in both dasyuromorphs and borhyaenoids. This is a derived pattern in both cases, the taxa characterized by a vestigial (or absent) hallux being closely related to those that exhibit a normally developed Mt I. This is the case in Dasyuridae (all the Dasyurus species do not have a reduced hallux), and in borhyaenoids: the Mt I of Sipalocyon and Lycopsis is fully developed, in contrast to that of Prothylacinus and Thylacosmilus, where it is vestigial. However, the development of the first metatarsal does not appear to be correlated with that of the peroneal process, which is prominent in Sipalocyon but not in Lycopsis.
The postcranial skeleton of Mayulestes appears to be different from that of the Miocene taxa especially at the level of the pectoral and pelvic girdles (Figs 6, 9): Development of the supraspinous fossa In three Santacrucian borhyaenoids ( Cladosictis, Borhyaena and Prothylacinus ) and in Lycopsis the supraspinous fossa (Fig. 9) is roughly rectangular in outline, the anterior border joining the scapular neck at a right angle to the scapular spine. In contrast, in Mayulestes this fossa is subtriangular and the anterior border does not join the scapular neck at such an extreme angle. The functional significance of the supraspinous fossa outline is unclear and, according to anatomical data on modern didelphids, cannot be directly related to the development of the M. supraspinatus. Angle between the scapular spine and the vertebral border and posterior extension of the caudal angle The angle between the scapular spine and the vertebral border (Fig. 9), which represents the relative orientation of the lever arms of Mm. trapezius and
501
serratus anterior, affects the efficiency of the scapula in bearing rotatory and tensile forces (Oxnard, 1963, 1968). It is therefore related to substrate preference. In Mayulestes (Fig. 9C), this angle is obtuse (100– 110 ∞), i.e. when the spine is orientated vertically, the highest point of the scapula is located at the posterodorsal angle of the infraspinous fossa. This condition is usually related to a posteriorly extended caudal angle. It clearly differs from that of the Miocene borhyaenoids, in which the scapular spine and the vertebral border are approximately perpendicular. In these taxa, as well as in the modern taxa Thylacinus and Metachirus, the caudal angle is not extended posteriorly. An obtuse angle between the scapular spine and the vertebral border and a posteriorly extended caudal angle are found in arboreal and fossorial taxa (Maynard Smith & Savage, 1956; Roberts, 1974), whereas a more quadrangular scapula is found in terrestrial to cursorial taxa (Jolly, 1967; Taylor, 1974). The particular shape of the scapula of Mayulestes is not found in the other borhyaenoids, or in the two other Tiupampan metatherians known from postcranial skeletons, Pucadelphys and Andinodelphys (Muizon & Argot, 2003). It probably represents an early specialization that reflects arboreal habits. It is more similar in outline to the scapula of arboreal primates (such as lemurs) than to that of Miocene borhyaenoids, and is consistent with the posteriorly convex ulna and wide open trochlear notch that also indicates clear arboreal capabilities (Muizon, 1998; Argot, 2001). Orientation of the ilium relative to the ischium In all the borhyaenoids in which the pelvis is known (i.e. Mayulestes, Cladosictis, Prothylacinus and Lycop sis), the ilium is quadrangular in outline. The gluteal fossa is broad and orientated laterally, whereas the iliac fossa is reduced to a narrow ventral strip, in contrast to what is observed in most living marsupials. However, the ilium of Thylacinus and Metachirus is also quadrangular, quite similar to that observed in borhyaenoids. Within Borhyaenoidea, whereas the ilium and ischium are aligned in Cladosictis, Prothylacinus and Lycopsis, the ilium protrudes dorsally above the dorsal ramus of the ischium in Mayulestes (Fig. 6A), and the functional significance of this feature is unclear. Unfortunately, the left part of the pelvis and the sacrum are unknown in Mayulestes, which precludes the accurate orientation of the pelvic girdle. Iliac neck length The iliac neck is variable within Borhyaenoidea. It is much longer in Mayulestes (16.5% of the total innominate length) than in Prothylacinus (4%); it is intermediate in Cladosictis and Lycopsis (11 – 13%) (Fig. 6). A short iliac neck may be related to the stability of the pelvic girdle and to the reduction of
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
502
C. ARGOT As
Ca
Ca Na
As
Mc En Cu Ec
Cu
Na
Na En
Mc En
Mc
Ec
Cu Ec
vestigial
Mt I
Mt I Mt III
A
C
B
Mt I
vestigial
En
Mt I
articulation with Mt I E D
I
V
II III
IV
Figure 8. General morphology of the pes in various borhyaenoids, showing the variable development of the first metatarsal. Abbreviations: As, astragalus; Ca, calcaneum; Cu, cuboid; Ec, ectocuneiform; En, entocuneiform; Mc, mesocuneiform; Mt I, first metatarsal; Mt III, third metatarsal; Na, navicular. A, Cladosictis patagonica PU 015046. B, Sipalocyon gracilis PU 015154. C, Prothylacinus patagonicus PU 015700. D, Lycopsis longirostris UCMP 38061. E, Thylacosmilus atrox FMNH P 14344. Not to scale.
shearing forces exerted on the ilium (Davis, 1964). Its potential relation to the orientation of the ilium is unclear.
Anterior inferior iliac spine This spine is extremely well-developed in all Miocene borhyaenoids (Fig. 6). The prominent and rugose scar,
where the M. rectus femoris (the biarticular head of the quadriceps) originates, is approximately as long as the iliac neck in the genera where it can be observed (i.e. the four Santacrucian taxa). The innominate of Thylacosmilus is unknown, and it is very damaged in Lycopsis. By contrast, this spine is very weak in the Palaeocene Mayulestes. Functionally, the development of a prominent scar in living macroscelidids that
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
EVOLUTION OF BORHYAENOIDEA
503
angle between the scapular spine and the vertebral border
triangular supraspinous fossa
scapular notch
A
C
B
D
E
Figure 9. Scapula in lateral view showing variations in the shape of the supraspinous fossa and in t he angle between the scapular spine and the vertebral border. A, Borhyaena tuberata PU 015701. B, Lycopsis longirostris UCMP 38061. C, Mayulestes ferox MHNC 1249. D, Prothylacinus patagonicus PU 015700. E, Cladosictis patagonica PU 015170. The scapular spine is broken and the acromion in unknown in all specimens except Mayulestes. Not to scale.
exhibit jumping ability suggests a relationship between this tuberosity and the role of the rectus femoris in the type of locomotion performed. However, the potentially different role of this muscle in borhyaenoids compared to living placental Carnivora (where the spine is reduced compared with the fossils) is unclear.
Acetabulum In Mayulestes the acetabulum is oval in outline and the articular facet is strongly constricted at the ilioischial suture. The dorsal margin is concave in dorsal view and the anterior part of the articular facet protrudes laterally. This condition is more similar to that observed in the living Australian marsupials than in
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
504
C. ARGOT
the Miocene borhyaenoids. This morphology also clearly indicates that Mayulestes was not as adapted to leaping as various modern marsupials [e.g. Metachirus, Antechinomys, Perameles, or the living caenolestids (Szalay & Sargis, 2001)]. The morphology observed in Mayulestes, as well as in Pucadelphys , may reflect an ancestral condition. In the Miocene taxa, the articular facet is more rounded in outline and is not constricted at the ilio-ischial suture.
Torsion of the femur In Mayulestes and Lycopsis the femur is twisted mediolaterally along its proximodistal axis, so that the head projects anteromedially rather than medially relative to the distal femur. The transverse axes of both epiphyses form an angle of 30–40 ∞, whereas they are parallel in the other borhyaenoids. Among living marsupials, a torsion is also found in Sarcophilus and Thylacinus. The functional significance of this torsion is unclear. It might be related to the orientation of the acetabulum, but the acetabulum of Lycopsis is too poorly preserved to be compared with that of Mayule stes (in which the position of the acetabulum relative to the vertebral column is unknown).
Greater trochanter height The greater trochanter (Fig. 6) is clearly higher than the femoral head in Mayulestes, whereas it is particularly low in Thylacosmilus (Argot, in press). This character intergrades in the other borhyaenoids, since it is slightly higher in Borhyaena, as high as the femoral head in Lycopsis, as high or slightly lower in Prothylacinus and Cladosictis. A low greater trochanter is also present in ursids and hyaenids and might be related to the shape and orientation of the gluteal fossa. However, the absence of innominate in Thylaco smilus, as well as the uncertain orientation of the innominate in Mayulestes relative to the vertebral column do not allow a test of this hypothesis. Therefore, it seems that a peculiar association of characters - a long axial neural process; strong ventral processes on the axis, C3 and C4; a long and strong deltopectoral crest; a well-developed lateral epicondylar crest; a pseudo-opposable pollex; a tibioastragalar joint medially constrained; a long tail with a muscular base; an intermembral index (humerus + radius + McIII length/femur + tibia + MtIII length) between 75 and 78% - represents a pattern common to all Borhyaenoidea (Table 2). However, each taxon examined evolved characteristic specializations. Mayulestes and Cladosictis, characterized by arboreal adaptive traits, probably occupied a civet-like ecomorph. Prothylacinus and Thylacosmilus were likely ambush predators, the former also exhibiting distinct
arboreal adaptive traits. Lycopsis and Borhyaena were more terrestrial predators. However, some features characterizing Lycopsis (the pseudo-opposable pollex, the nonreduced hallux, the short metatarsals and the poorly stabilized calcaneoastragalar joint) indicate that the animal was not a fast runner, suggesting that it inhabited a densely forested environment. By contrast, Borhyaena reveals a more specialized change towards a cursorial pattern. The morphology of the girdles appears to be variable within Borhyaenoidea, especially between Mayulestes and the later borhyaenoids. The specializations of Mayulestes are consistent with the hypothesis that the Tiupampan metatherians had already experienced some endemic evolution on the South American continent (Pascual & Ortiz Jaureguizar, 1991, 1992; Muizon, 1992; Muizon & Cifelli, 2001).
PALAEOBIOLOGY AND EVOLUTION OF LOCOMOTION Borhyaenoids with postcranial remains are known from four places: (1) Tiupampa, early Palaeocene, south-central Bolivia ( Mayulestes ferox); (2) Santa Cruz Formation, end of early Miocene, Patagonia, Argentina (Sipalocyon gracilis, Cladosictis patagonica, Prothylacinus patagonicus and Borhyaena tuberata); (3) La Venta, middle Miocene, Colombia ( Lycopsis longirostris ); (4) Andalgalá and Corral Quemado Formations, late Miocene, north-western Argentina ( Thylacosmilus atrox). Habitat characteristics are summarized for each place and the diet and hunting behaviour of the borhyaenoid taxa are discussed.
H ABITAT The Tiupampa beds were deposited in channels of a large meandering river on a flat alluvial flood plain (Marshall et al., 1995). In addition to mammals, they have yielded abundant remains of frogs, turtles, snakes and crocodiles: Amphibia (Anura, Gymnophiona), Reptilia [Chelonia, Squamata (Lacertilia and Ophidia) and Crocodilia] (Muizon, 1998). The presence of several taxa of crocodiles attests to a warm and moist climate, probably subtropical and relatively equable (Pascual & Ortiz Jaureguizar, 1990). The late Cretaceous -early Cenozoic interval was dominated by warm, humid tropical-temperate forested environments throughout South America (Flynn & Wyss, 1998). Sedimentological, palaeopedological and vertebrate palaeontological studies undertaken at Monte Observación indicate that the Santa Cruz sediments were deposited under warm, humid conditions, reflecting a temperate to subtropical coastal alluvial plain (Bown
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
EVOLUTION OF BORHYAENOIDEA & Fleagle, 1993; Genize & Bown, 1994). Most of the mammals recorded (porcupines, echimyids, dasyproctids, anteaters and primates: see Table 3) also suggest a mild climate and a forested milieu (Patterson & Pascual, 1968). However, invertebrate fossil remains and traces have been found in the Santa Cruz Formation (termite nests, scarabeid brood masses and cells produced by digging bees) and indicate relatively open areas (Genize & Bown, 1994). A preliminary palaeoecological interpretation of the Santa Cruz Formation, based on the analysis of 22 fossiliferous horizons located between the Río Gallegos and the Río Coyle, notes six main trends between the lower and the upper horizons (Tauber, 1997): (1) a general decrease in taxonomic diversity; (2) a decrease in diversity within small-sized taxa ( <500 g); (3) a decrease in the number of large taxa like Homalodotherium and Theosodon; (4) a reduction of the megatheriid and megalonychid average body size; (5) an increase in taxonomic diversity of the glyptodonts and toxodonts (which were supposedly grazers), and (6) an increase in the number of taxa with euhypsodont teeth. These elements suggest a shift in the climate as a warm and moist climate and a forested milieu transitioned to a drier climate with more open conditions (Tauber, 1997). To a very large extent, the Santa Cruz beds are made up of volcanic and detrital material derived from the rising Andes to the west, superimposed on marine Patagonian deposits (Scott, 1932; Marshall et al. 1983). It appears likely that stochastic catastrophic reductions in population size, with local and regional extinctions resulting from rapid pyroclastic deposition, profoundly influenced the course of mammalian evolution in this area (Bown & Fleagle, 1993). In the middle Miocene, the area of northern Ecuador, central and western Colombia, and western Venezuela formed a peninsula, without land connection with Central America, and bordered on the south by a seaway extending into the upper Amazon basin (Kay & Madden, 1997: fig. 30.1). The La Venta area was situated on the eastern lowland region of that peninsula, several hundred kilometres from the upper Amazon seaway. Like Tiupampa, this area was characterized by a meandering system of flood plains. There is evidence for periodic flooding and for extensive evergreen rain forest cover, in relation to the high percentage of arboreal species found (presence of forest-dwelling marsupials, bats, lizards, and birds like gruiforms, pelecaniforms, and coraciiforms, whose modern analogues occur in fluvial wetlands, wooded swamps and freshwater marshes: Rasmussen, 1997). The diverse freshwater fish fauna discovered indicates a heterogeneous aquatic environment, including large and open rivers, marginal shallow waters, and even anoxic, temporary waters, i.e. a mosaic of biotopes typical of mod-
505
ern lowland meandering stream systems (Kay & Madden, 1997). The annual rainfall may have been at least 2000 mm per year, and the environment was more probably a forest mosaic than an interrupted multistratal evergreen forest (Kay & Madden, 1997). Unfortunately, the composition of the flora of Miocene tropical South America is still poorly understood. In Africa, open clearings are maintained within forests by the destructive habits of tusked megaherbivores. Four large mammals of the La Venta community Huilatherium ( , Pericotoxodon, Xenastrapotherium and Granastrapotherium ) had tusks, and may have helped in creating and maintaining edge habitats within the forests (Kay & Madden, 1997). Overall species diversity in the Monkey beds (the lowest part of the Villavieja Formation) is comparable to that of modern Neotropical South American communities despite profound differences in taxonomic and palaeoecological composition. For example, the La Venta fauna includes endemic ungulates, sloths, and caviomorph rodents vs. hypsodont artiodactyls, sciurid and murid rodents today (Flynn & Wyss, 1998). The two specimens of Thylacosmilus atrox known from postcranial remains have been found in the late Tertiary beds in and around the Santa María Valley of the Province of Catamarca, north-western Argentina (Riggs, 1934). They come from two different localities, the Corral Quemado and Andalgalá Formations, which are considered to be Huayquerian in age (i.e. c. 6.8–9 Myr old) (Flynn & Swisher, 1995). This age forms a part of the ‘Panaraucanian Faunistic Cycle’ of Argentina divided into the Protoaraucanian subcycle (Friasian + Chasicoan) and the Araucanian subcycle (Huayquerian + Montehermosan) (Pascual, 1989; Pascual & Ortiz Jaureguizar, 1990). During the Protoaraucanian subcycle, most of the preceding mammalian taxa suggesting subtropical woodlands (e.g. platyrrhines, stegotheriine armadillos, palaeothentid and abderitid caenolestoid marsupials, echimyid and erethizontid rodents, and most of the tree-dwelling edentates) became rare or absent in Patagonia (Pascual & Ortiz Jaureguizar, 1990). The Araucanian subcycle is predominantly characterized by cursorial and/or grazing mammals. There is a progressive increase of the high-crowned ungulates, the armoured xenarthrans (armadillos, pampatheres and glyptodonts) experience the greatest diversification ever recorded, and specialized lineages within some rodent families (Dinomyidae, Hydrochoeridae) develop the largest and most cursorial rodents ever known (Pascual & Ortiz Jaureguizar, 1990). Also, the presence of Argyrolagidae, a peculiar lineage of ricochetal marsupials, provides one more indication of open habitats (Pascual & Ortiz Jaureguizar, 1990). The fauna therefore suggests predominant savannah-
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
506
C. ARGOT
like environments. These widespread and varied plains have been distinguished as ‘la edad de las planicies australes’, the age of southern plains (Pascual & Ortiz Jaureguizar, 1990; Ortiz Jaureguizar, 1998).
DIET AND HUNTING BEHAVIOUR Four questions may therefore be asked (paraphrasing Van Valkenburgh, 1985): (1) How many species of predators coexisted within each community? (2) What was the array of prey available? (3) What is the diversity of locomotor morphologies that evolved within predators (i.e. how many arboreal and scansorial species, ambush and pursuit hunters, etc.)? (4) How similar in morphology are the predators to each other (i.e. what is the potential for niche overlap)? Contemporaneous borhyaenoids are expected to differ in behaviour, body size, and/or structural morphology. However, the two last questions will be examined only in the case of the Santa Cruz fauna that includes four taxa known from postcranial remains, whereas a single taxon is known from the three other localities (Tiupampa, La Venta and Catamarca province). Mayulestes ferox According to the taphonomy, only marsupials could have played the role of mammalian predators in Tiupampa (Table 1), and the two largest metatherians were the borhyaenoids Mayulestes and Allqokirus (the latter being probably as large as Mayulestes and carnivorous too, but this taxon is poorly known). The hindlimb of Mayulestes exhibits numerous differences compared with that of arboreal living marsupials (didelphids and phalangeriforms), such as a greater trochanter higher than the femoral head, a femoral trochlea well-defined between sharper ridges, and a higher femorometatarsal index, all features suggesting a faster locomotion than that practised by extant highly arboreal marsupials, which fits well with active hunting strategies (Argot, 2002). Mayulestes, as a predator, could have fed upon the abundant fauna of small insectivorous marsupials listed in Table 1, except Andinodelphys (relatively larger than the other forms although slightly smaller than Mayulestes). Muizon (1998) also suggested that omnivorous small mioclaenids like Tiuclaenus, Molinodus or Pucanodus could have represented occasional prey, as well as some Amphibia, like the leptodactylid frog Estesius. It is also likely that Mayulestes could have fed upon eggs of crocodiles, turtles or birds. It was probably hunting the small condylarths and marsupials on the ground, but was clearly able to climb
tree for resting, foraging for eggs or nesting birds, pursuing some prey, or escaping its own predators such as crocodiles.
Sipalocyon gracilis, Cladosictis patagonica, Prothylacinus patagonicus, Borhyaena tuberata In sympatry, closely related species often differ morphologically in regard to features that reflect diet, foraging strategy, or life history. The extent of such differences is usually ascribed to the intensity of competition and predation and/or to the levels of food resources and environmental stability (Van Valkenburgh, 1985; Wayne et al., 1989). The Santacrucian mammalian fauna, one of the most diverse and rich vertebrate fossil assemblages in South America (Table 3), is characterized by various elements: (1) the predominance and diversity of rodents (exclusively Hystricognatha) and edentates (especially Tardigrada); (2) the quantity of small rodent- and rabbit-like notoungulates, Hegetotheriidae and Interatheriidae; (3) the first and last record of the enigmatic burrowing Necrolestes ; (4) the last record of platyrrhine monkeys in Patagonia ( Homunculus patagonicus in coastal deposits); (5) the last record of the poorly known litoptern family Adianthidae, and (6) the last occurrence of the notoungulate family Notohippidae (Marshall et al., 1983). Cladosictis was a short-legged predatory form. The shoulder height was about 20–25 cm, and the body mass is estimated at 4–8 kg (Argot, 2003c; see also Fig. 10B). The proportions of the limbs recall those of the living South American tayra ( Eira barbara), although Cladosictis had shorter legs than this mustelid. The pseudo-opposable pollex, also found in Sipalocyon, suggests skilful manipulative behaviour. Cladosictis was probably able to climb and certainly specialized more in taking smaller prey, arboreal or terrestrial (small mammals, reptiles, amphibians, invertebrates, eggs and birds) than Prothylacinus . Although Cladosictis exhibits the dental complex related to a carnivorous or hypercarnivorous diet (details in Muizon, 1999: 502), hathlyacynines are considered to have a more generalized dentition than prothylacinids or borhyaenids (Marshall, 1981), which may ensure flexibility of food habits, as in extant omnivorous-carnivorous small canids. Prothylacinus (Fig. 10D) was the size of a wolverine, weighing about 30 kg (Argot, 2003b). The forelimb emphasizes arboreal adaptations, while the hindquarters are powerfully built (especially the thighs), and the axial skeleton is flexible. Prothylacinus was not designed for speed, relying instead on its ability to perform powerful jumps from a crouched position (Argot, 2003b). It appears to have been a much more typical ambush predator than the other Santacrucian
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
EVOLUTION OF BORHYAENOIDEA
507
Table 3. Generic list of Santacrucian mammals (coastal beds) from Patagonia, Argentina (from Scott, 1903a, 1903b, 1904, 1905; Patterson & Pascual, 1968; Marshall, 1978a, 1979, 1981; Marshall et al., 1983; Bown & Fleagle, 1993; Rae et al., 1996; Fleagle et al., 1997; McKenna & Bell, 1997) TRIBOSPHENIDA Necrolestidae Necrolestes
GONDWANADELPHIA MICROBIOTHERIA
Microbiotheriidae Microbiotherium DIDELPHIDA SUDAMERIDELPHIA
Polydolopimorphia Parabderites Sparassodonta Borhyaenoidea Pseudonotictis Perathereutes Sipalocyon Cladosictis Anatherium Prothylacinus Lycopsis Arctodictis Borhyaena Acrocyon GLIRIMETATHERIA
Paucituberculata Caenolestoidea Stilotherium Phonocdromus Pichipilus Abderites Palaeothentes Titanothentes Acdestis Propalaeothentes
XENARTHRA Dasypodidae Anantiosodon Peltecoelus Peltephilus Stegotherium Proeutatus Paraeutatus Prozaedyus Stenotatus Vetelia Glyptodontidae Asterostemma Cochlops Eucinepeltus Metopotoxus Propalaehoplophorus Family incertae sedis Hapalops Eucholoeops Planops Megatheriidae Prepotherium Analcimorphus Hyperleptus Pelecyodon Schismotherium Nothrotheriidae Xyophorus Megalonychidae Megalonychotherium Mylodontoidea Analcitherium Nematherium Myrmecophagidae Protamandua
borhyaenoids. This hunting strategy (the predator lying still watching its prey move past, and running a short distance towards it) involves taking advantage of a covered habitat (Kruuk & Turner, 1967). The combination of stout pelvis, long femur and short metatarsals probably improves the mechanical advantage of muscles stabilizing the trunk and hindlimbs while the forelimbs are working against some resistance, such as struggling prey. It is also known that pouncers on rodent-sized prey have large thigh muscles, which do most of the work in stride acceleration and take-off (Alexander, 1993). The manipulative ability of its forearms also suggests ability to execute skilful attacks on small mam-
LITOPTERNA Proterotheriidae Diadiaphorus Licaphrium Prothoatherium Proterotherium Thoatherium Macraucheniidae Theosodon Adianthidae Adianthus ASTRAPOTHERIA Astrapotheriidae Astrapotherium NOTOUNGULATA Toxodontidae Adinotherium Nesodon Hyperotoxodon Homalodotheriidae Homalodotherium Notohippidae Notohippus Interatheriidae Epipatriarchus Interatherium Protypotherium Hegetotheriidae Hegetotherium Pachyrukhos
RODENTIA Octodontidae Acaremys Sciamys Echimyidae Adelphomys Stichomys Spaniomys Chinchillidae Pliolagostomus Prolagostomus Neoepiblemidae Perimys Dasyproctidae Neoreomys Dinomyidae Olenopsis Scleromys Eocardiidae Eocardia Luantus Phanomys Schistomys Erethizontidae Steiromys PRIMATES Platyrrhini Homunculus
mals and birds. Moreover, the powerful musculature of the limbs and neck of Prothylacinus suggests that it might have been able to drag larger prey to a safe place, far from the large ground-dwelling borhyaenids like Borhyaena or Arctodictis . The curved, dorsoventrally deep claws probably functioned for grasping prey as well as for climbing. A study of extant jaguars in Belize revealed that Dasypus (armadillo), Agouti (paca), and Mazama (brocket deer) account for 94% of the available terrestrial prey, with just over half (54%) of the identified prey consisting of armadillos (Rabinowitz & Nottingham, 1986). According to these authors, the armadillo is a particularly vulnerable prey species, due to its limited mobility, lack of effi-
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
508
C. ARGOT
B
A
C
D
E
Figure 10. Skeletal reconstructions of various borhyaenoids. A, Mayulestes ferox, modified from Muizon (1998). B, Cladosictis patagonica. C, Lycopsis longirostris, modified from Marshall (1977a). D, Prothylacinus patagonicus . E, Borhyaena tuberata. Scale bars: 5 cm in A, 10 cm in B -E. The darkened areas represent the elements preserved. Lycopsis is drawn as found in matrix.
cient defences, and muffled grunting that accompanies its nocturnal foraging. Therefore, it seems clear that apart from rodents, caenolestoids and some notoungulates, armadillos and semiarboreal sloths like Hapalops could have represented prey species for Prothylacinus . In the Santacrucian borhyaenoid radiation, Borhyaena (Fig. 10E) differs in having forelimb movements that were more restricted to a parasagittal plane, relatively reduced distal muscular mass of the limb, restricted pronation -supination capabilities, and a digitigrade manus with relatively elongate metacarpals, short phalanges, and blunt claws (Argot, 2003b). Moreover, Stein & Casinos (1997) developed the idea of an asynchrony in the evolution of the mam-
malian girdles: the mammalian pelvis would have evolved cursorial features earlier than the therian scapula, cursoriality being related primarily to the adduction of the hindlimb. In this context, it is noteworthy that the hip joint of Prothylacinus retains more abductive capability than that of Borhyaena, a condition which is consistent with climbing behaviour that requires a greater abductive range at the hip joint than in runners. Moreover, the medial femoral condyle of Borhyaena, which is larger than the lateral one, suggests a medial displacement of the load line, perhaps reflecting a less abducted position of the femur in an obligate terrestrial animal. Considering these structural modifications, Borhyaena may be qualified as the most cursorial
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
EVOLUTION OF BORHYAENOIDEA taxon known amongst Borhyaenoidea, although it must have been a much slower runner than any modern form. Finding a cursorial form was not unexpected within a group of highly carnivorous mammals. According to Stein & Casinos (1997), there is no highly polytypic order of terrestrial mammals that does not contain at least some taxa that would be considered cursorial. Biomechanical evidence suggests that cursorial mammals could have arisen naturally beyond a certain body weight to maximize locomotor efficiency (Janis & Wilhelm, 1993; Stein & Casinos, 1997). Based on its forelimb proportions, Borhyaena was relatively shorter-legged and with bones relatively more robust and heavier than present-day pursuit predators (Argot, 2003b). The forelimbs of Borhyaena were probably able to perform movements not restricted to a strict parasagittal plane as in all borhyaenoids known, a condition which would have been useful in hunting and overpowering small prey, and overturning rocks and logs. The searching behaviour of striped hyaenas, adapted towards obtaining small food (insects, small vertebrates, fruits and carrion) from the denser part of the vegetation without cooperation between individuals (Kruuk, 1976), is likely to resemble the foraging behaviour of Another similarity between the Borhyaena. borhyaenoid and striped hyaenas is that the strong and powerful neck and anterior part of the trunk of Borhyaena would be advantageous in lifting and carrying heavy prey to denning areas, as striped hyaenas do (Kruuk, 1976). Borhyaena appears to have had the best stereoscopic vision among Santacrucian borhyaenoids, the estimated overlap of left and right visual fields being 50–70∞ (Savage, 1977), vs. 30–50 ∞ for Cladosictis and Prothylacinus . The primitive nocturnal carnivores that cannot visually localize their prey seem to be more dependent upon an immediate rush and/or ambush strategy rather than employing a circuitous stalking approach (Eisenberg & Leyhausen, 1972); this fits well with the hunting strategy hypothesized for Prothylacinus . Concerning the apparently poor running ability of Borhyaena compared with living canids, it is noteworthy that among the Santa Cruz herbivores, only two litoptern families might be considered as potentially cursorials, the Proterotheriidae and Macraucheniidae. The slenderly built proterotheriids were of small to moderate size, and constitute the major Santacrucian stock of equid-like animals characterized by long, slender limbs, and feet that have a very horse-like appearance (because of reduced lateral digits), although the metapodials are not particularly elongate (Scott, 1910). The macraucheniids are represented by a single Santacrucian taxon, Theosodon, one of the most common elements of the Santa Cruz fauna, characterized by a long and
509
graceful neck recalling the guanaco and llama, and by the presence of a short proboscis (Patterson & Pascual, 1968). The feet are functionally tridactyl and almost isodactyl (Scott, 1910). Theosodon’s shoulder height was approximately 100 cm (vs. 45 cm for the antelopelike proteroteriid Thoatherium), i.e. it was taller and especially stouter than the modern llama. It is unlikely that Borhyaena could have preyed upon adults, but it was certainly able to attack juveniles. During the Santacrucian, Sipalocyon, Cladosictis, Prothylacinus and Borhyaena coexisted with six other borhyaenoids, all less well-known (Table 3): (1) two hathlyacynids smaller than Sipalocyon gracilis: Pseudonotictis pusillus and Perathereutes pungens; (2) the largest known hathlyacynid, Anatherium defos sus, slightly larger than Cladosictis patagonica, but known only from fragmentary mandibular rami; (3) the medium-sized prothylacynid Lycopsis torresi; (4) the medium-sized borhyaenid Acrocyon sectorius , smaller than Borhyaena tuberata and much rarer; (5) the very large and quite rare borhyaenid Arctodictis munizi. Adults of Arctodictis were the giants of their time, the largest post-Deseadan borhyaenoids known (Marshall, 1976). It has been suggested that this taxon filled the adaptive zone for a large terrestrial carnivore, available in South America since the disappearance of the very large proborhyaenids at the end of Deseadan (Marshall, 1976; Bond & Pascual, 1983). However, Arctodictis did not reach the size of the gigantic proborhyaenids, and evolved concurrently with new climatic and environmental conditions (Bond & Pascual, 1983). Therefore, the diversity in body size among the borhyaenoids that coexisted during the Santacrucian is considerable. Moreover, the diversity in morphological adaptations is also important between Cladosictis, Prothylacinus and Borhyaena. Sipalocyon is too poorly known to infer conclusions about its mode of life, although the pseudo-opposable pollex suggests, like the deep and sharp ungual phalanges, potential grasping ability (Argot, 2003c). Distinct adaptations are expected to make different prey species available to each hunter, by virtue of substrate preference (habitat selection) and/or various hunting strategies (Rosenzweig, 1966). However, other factors, although important to explain variations in diet of sympatric predators, remain unavailable for palaeoguilds, such as their speed and perseverance during running, or the way in which the different reactions of prey interact with the hunting strategy (Kruuk & Turner, 1967). Response to prey is actually a function of experience and of the stimuli emanating from the prey object itself (Eisenberg & Leyhausen, 1972): ungulates can simply flee, whereas ground sloths can fight and exhibit defensive threat postures. Moreover, the occupation of the same habitat
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
510
C. ARGOT
at different times of the day is a factor impossible to assess for fossil forms. How could these contemporaneous borhyaenoids have shared prey? It is usually thought that large carnivores tend to eat large prey. With increase in body mass, predators change the composition of their food from small to larger prey (Carbone et al., 1999), the latter being nevertheless far more scattered, more active, and thus less productive than the former (Goszczynski, 1986). Hence, numerous exceptions exist (Rosenzweig, 1968). The sharp claws of felids enable them to overpower prey larger than themselves. In contrast, because the limbs of living canids are modified for cursorial locomotion, they use only the jaws as a prehensile killing organ, which limits the size of prey which can be captured and killed by a solitary hunter (Kleiman & Eisenberg, 1973). Hence, solitary canids usually exhibit greater opportunism and more omnivorous tendencies than felids. This seems to have permitted the development of traits that are rare in felids, such as peaceful communal feeding (a step towards pack hunting), or extensive scavenging on the kills of larger predators (Kleiman & Eisenberg, 1973). The general pattern exhibited by the postcranium of Prothylacinus and Borhyaena suggests quite similar differences, since the forelimb of Prothylacinus indicates ability to manipulate while that of Borhyaena lacks great abductive ability, pronation -supination capabilities, and sharp claws, all features that restrict manipulation. Rosenzweig (1966) found that active predators tend to specialize in a certain size of prey, although other sized prey may be taken occasionally as well. For example, wolves that hunt in packs primarily eat ungulates, coyotes eat rabbits and large rodents (and also birds, insects, fruit and carrion), and foxes eat rodents and occasionally vegetable matter (Ewer, 1998). Small predators are usually solitary hunters, must sustain predation pressure from larger sympatric carnivores, tend to hunt small prey, and to be partly omnivorous (Kleiman & Eisenberg, 1973), as hathlyacynids were supposed to be compared to prothylacinids and borhyaenids (Marshall, 1981). Rosenzweig (1966) also argued that the coexistence of closely related predatory species is maintained in large part by differences in body size, or that predators of similar size occupy different habitats or exhibit different hunting habits. Although the adaptive niches of fossil forms cannot be defined accurately, the difference in size between Cladosictis and Prothylacinus limited the competition, whereas the distinct adaptations of Prothylacinus and Borhyaena certainly also limited interactions. The powerful neck and massive skull of Borhyaena, characterized by robust and prominent zygomatic arches, suggest that this taxon could have occasionally scavenged, although its dentition does not suggest
that it was a specialist bone-crusher. Most carnivores tend to scavenge when an opportunity arises (Estes, 1967; Viranta, 1996). A primarily scavenging mode of life has two requirements: the presence of efficient hunters (or some other factor inducing high prey mortality), and that the prey be sufficiently large, because small prey are usually consumed entirely by the predators themselves (Viranta, 1996). Moreover, the occurrence of bone-eating species depends on feeding opportunities (Van Valkenburgh, 1989). Santacrucian environmental conditions were probably not optimum for this, since carcasses are more difficult to find in densely covered habitats and when climates are warm and moist, because of rapid decomposition. What sort of efficient hunters could have killed large prey? During the Santacrucian, the nonmammalian occupants of the carnivore adaptive zone were large, flightless running birds, the phororhacoids, that ranged from Deseadan to Montehermosan (i.e. lasting c. 20 Myr: Marshall, 1978b). Three families are known: the medium and large-sized Psilopteridae and Phorusrhacidae, rather lightly built and probably swift runners, and the gigantic Brontornithidae, including ponderous forms with massive beaks (Patterson & Pascual, 1968; Marshall, 1978b). The biology of these birds is still poorly known. Much of the recorded material is fragmentary, and the association of the elements is often questionable, which increases the difficulty of performing a comprehensive overview. Amongst the best known taxa, Pelecyornis (Fig. 11) was a medium-sized bird, about 80–90 cm tall, with a skull about 18 cm long, and a massive, deep beak. According to Ameghino (1895), the orientation of the articular facets of the cervical vertebrae in Phororhacos (structurally similar to, but larger than, those of Pelecyornis) suggests an important sigmoid curvature of the neck: the head was probably placed in a resting position in the same vertical axis as the scapula, which would have helped to support its weight. The mandible of Phororhacos longissimus suggests a skull about 65 cm long. The tail of Pelecyornis was long and allowed lateral movement. The sternum is that of a typical carinate bird, supporting a prominent ventral keel. The length of the wing equals that of the living South American Cariama, a long-legged, long-necked bird about 70 cm tall (Marshall, 1988). Cariama is credited with the ability to run as fast as a trotting horse (25 miles per hour); it spends most of its time on the ground and is only capable of flight over short distances (Sinclair & Farr, 1932; Marshall, 1978b, 1988). The most striking difference between Pelecyornis and Cariama concerns the development of the beak, which is much greater in the fossil, although Cariama also exhibits carnivorous habits, feeding on small mammals, birds, reptiles and insects (Sinclair & Farr, 1932; Marshall, 1978b). Cariama usually seizes its
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
EVOLUTION OF BORHYAENOIDEA
A
511
B
Figure 11. Pelecyornis australis , from the Santa Cruz Formation (end of early Miocene) of Patagonia, Argentina. A, skeletal reconstruction modified from Sinclair & Farr (1932). Scale bar = 10 cm. B, life reconstruction modified from Scott (1932).
victim in its beak and hurls it to the ground with great force. Moreover, the inner claw of each foot is long, sharp and deeply curved, and is probably used to pin the prey to the ground. This suggests ferocious habits in the much larger phororhacoids. Patterson & Pascual (1968) even suggested that they could have preyed on armadillos and glyptodonts, the Santacrucian taxa being less heavily armoured than the Pleistocene ones. Competition between phororhacoids and large borhyaenoids cannot be excluded, but remains speculative. It is usually considered that these large birds replaced borhyaenoids in the later Tertiary in the savanna grasslands of Argentina (Marshall, 1978a), although some authors consider that the 20 Myr of coexistence between borhyaenoids and phororhacoids resulted in resource partitioning rather than extinction (Patterson & Pascual, 1968; Bond & Pascual, 1983). Although the axe-like beak of these big birds suggests that they probably had predatory habits, there is little reason to think that borhyaenoids were unable to kill a phororacoid, especially the relatively small psilopterids. Ecological theory predicts that if competition for a shared
resource occurs, guild members will either diverge in their resource utilization or become extinct locally (Van Valkenburgh, 1985). However, it is still impossible to prove that borhyaenoids and phororacoids exhibited microhabitat separation or differences in activity pattern. Because the partition of the specimens of Borhyaena within the various fossiliferous horizons of the Santa Cruz Formations is poorly documented, it is difficult to establish faunistic correlations precisely. Perhaps Borhyaena was more abundant in forested areas and the flightless birds in grasslands. Alternatively, Borhyaena might have been present in both environments, hunting under cover or in the open country at the very edge of forests and scavenging on grasslands (the presence of carnivorous birds improving carcass predictability). It may have sought out the nests of grassland birds as hyaenas do in African plains, waiting for the time when the ostrich eggs are left momentarily unguarded (Estes, 1967). Such a partition could also explain the relatively important variation in size found within the different specimens of Borhyaena (Marshall, 1978a) as it is well known that jaguars from the grasslands of the Brazilian Pantanal can be
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
512
C. ARGOT
twice as heavy as those dwelling in the Amazonian forest (Emmons, 1992). Moreover, although these ground-dwelling birds were certainly able to run fast, the way that they killed prey with their strong beak acting as a weapon is still unclear. Therefore, it may also be suggested that borhyaenids killed prey and phororacoids scavenged them. In fact, at typical kills, a definite rank order among African open plains scavengers is maintained, hyaenas coming first, then jackals, then vultures (Estes, 1967). Hyaenas and jackals predominate during the night, and vultures during the day, and some phororacoids might have played the role of modern vultures. Moreover, while hyaenas and jackals may live largely as scavengers on the kills of pure predators (big cats), where such predators are scarce they may be very active hunters (Estes, 1967). Assessing whether such opportunism was practised by fossil species remains speculative.
Lycopsis longirostris The La Venta fauna contains a diverse assemblage of endemic ungulates and edentates that filled a variety of niches as herbivores, as well as the earliest representatives of many clades of living South American mammals, including modern subfamilies of armadillos, anteaters, bats, opossums and New World monkeys (Flynn & Wyss, 1998). There was a high proportion of browsing species, a large number of frugivorous taxa, a large number of arboreal species (especially platyrrhines and some edentates), and the presence of forest-dwelling marsupials, bats, lizards and birds suggests that moist evergreen forests were extensive (Kay & Madden, 1997). Remains of birds found at La Venta reveal the presence of Hoazinoides , a cuculiform whose modern relatives (turacos and hoatzins) underwent evolutionary development in foot and wing structure in order to clamber about in trees (Rasmussen, 1997). Phororhacoids are unknown from this locality, perhaps reflecting the presence of extensive forests. Lycopsis weighed about 15 kg, and the shoulder height was about 35 cm (C. Argot, pers. observ.; see also Fig. 10C). It was relatively longer-legged than the other borhyaenoids known from postcranial remains, but the intermembral index is similar to that of the other borhyaenoids (approx. 78%), in the same range as most felids and mustelids. Morphological analysis of the postcranial skeleton reveals that it was a primarily terrestrial form. The general shape of the scapula and ulna indicates in particular that Lycopsis was clearly less adapted towards an arboreal life than Cladosictis or Prothylacinus . However, it does not exhibit features suggesting an incipient cursorial specialization, in contrast to Borhyaena. The development of the
musculature of the arm (especially Mm. spinati, pectoralis and biceps brachii), as well as the pseudoopposable pollex, nevertheless suggest that Lycopsis might have used its forelimb for a manipulative purpose, consistent with the capture of small prey. On the calcaneum, the orientation of the calcaneoastragalar and sustentacular facets reduces the stability of the calcaneoastragalar joint, a condition which probably precluded fast running. Lycopsis is traditionally compared in the literature with Thylacinus (Marshall, 1978a), although functional analysis of the postcranial skeleton reveals clear differences, especially in the structural pattern of the forelimb, the thylacine’s being more specialized towards a cursorial trend. Despite this trend, Thylacinus is thought to have been more at home in dense forests than open plains, although this preference may have resulted from competition with dingos, Canis lupus dingos, which are better adapted to open country (Smith, 1982). However, the niche overlap between the two species may have been overstated because of differences in dentition and thus probably in diet (Johnson & Wroe, 2003). It has been suggested that the diet of the thylacine in the wild was composed of small macropodids, echidnas, rats and birds (e.g. the Tasmanian native hen Gallinula mortierii) - prey considerably smaller than itself, and better stalked than pursued (Smith, 1982). However, such selection could well have resulted from human disturbance, as the two largest potential Tasmanian prey species of the thylacine, the emu and eastern grey kangaroo, were being decimated or eliminated by the 19th century, while the thylacine was restricted to marginal habitats (Johnson & Wroe, 2003). Living in forested environments and without any cursorial specialized features, Lycopsis probably stalked small mammals, which were abundant at La Venta during the Miocene. Direct evidence that it fed upon rodents exists, because in the body cavity of the holotype UCMP 38061, between the ribs and right tibia, several broken rodent bones and an upper molar of Scleromys colombianus have been found. Additional rodent bones and an incisor were found immediately posterior to the pelvis and below the base of the tail (Marshall, 1977a). Scleromys colombianus weighed about 2.5–3.5 kg (Walton, 1997). Rodents are welldiversified in the La Venta fauna (Table 4) and may have represented a diverse array of prey. Their closest living relatives are generally forest dwellers (Walton, 1997), except for caviids whose ever-growing cheekteeth suggest grazing habits. Their presence suggests that the palaeoenvironment of La Venta was a sort of ecological mosaic, although the fossil caviids may also not have had exactly the same ecological requirements as their modern relatives. Alternatively, the absence of modern caviids in the tropical regions may be ascribed
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
EVOLUTION OF BORHYAENOIDEA to various factors such as competitive exclusion (Walton, 1997). Lycopsis coexisted with a few larger borhyaenoids. These included Dukecynus magnus, poorly known from a distorted partial skull, and an undetermined large-sized borhyaenoid, known from very fragmentary postcranial and cranial fragments that do not allow a generic identification. Labelled ? Arctodictis by Marshall (1978a), the size of the fragments corresponds to that expected for D. magnus (Goin, 1997), but the determination is still in doubt. The last undoubted borhyaenoid is Anachlysictis gracilis , which represents the first appearance of thylacosmilids (Goin, 1997). Its flat skull roof, and the absence of postorbital bar and of dorsal extension of the upper canines suggest a predation strategy distinct from that of Thylacosmilus atrox. The systematic position and borhyaenoid affinities of the enigmatic Hondadel phys are still unclear (Goin, 1997). This taxon was smaller than Lycopsis and probably omnivorous (Goin, 1997). Hathlyacynids are unknown at La Venta, although they are known in later faunas. It may be suggested that Hondadelphys , intermediate in size between Sipalocyon and Cladosictis, and sharing with them similar dental specializations, filled a similar
513
ecological niche, which suggests a possible competitive exclusion between the groups (Marshall, 1978a). The nonmammalian predators of that fauna were crocodiles, particularly well-diversified in terms of taxonomic diversity, body size range, and presumed feeding ecology (Langston & Gasparini, 1997). They included probable active terrestrial carnivores like the very large and deep-skulled Sebecus and other sebecosuchians (Langston & Gasparini, 1997). Other taxa have been recorded, like Charactosuchus (Crocodylidae), Mourasuchus (Nettosuchidae, endemic to South America), Balanerodus (Alligatoridae), and the abundant Gryposuchus (Gavialidae) (Langston & Gasparini, 1997). The variety of crocodiles in the Honda Group is astonishing, as it is unusual in modern faunas that more than three taxa frequent the same stretch of river (Langston & Gasparini, 1997). The ability of Lycopsis to invert its hind foot and grasp branches might have allowed it to occasionally escape these predators. Similarly, some Australian carnivore niches may have been occupied by reptiles (crocodiles, snakes and lizards) from the middle Tertiary to the Pleistocene, although such reptiles were probably much more uncommon and geographically restricted than carnivorous marsupials (Wroe, 2002).
Table 4. Generic list of Laventan mammals from the Honda Group (Monkey beds) of La Venta, Colombia (from Hirschfeld, 1985; Bown & Fleagle, 1993; Carlini et al., 1997; Dumont & Bown, 1997; Edmund & Theodor, 1997; Fleagle et al., 1997; Goin, 1997; Johnson & Madden, 1997; Kay & Madden, 1997; McDonald, 1997; McKenna & Bell , 1997; Walton, 1997; White, 1997) DIDELPHIDA DIDELPHIMORPHIA
Didelphidae Micoureus Thylamys Undetermined taxa SUDAMERIDELPHIA
Sparassodonta Hondadelphidae Hondadelphys Borhyaenoidea Lycopsis Dukecynus ?Arctodictis Anachlysictis GLIRIMETATHERIA
Paucituberculata Caenolestoidea Hondathentes GONDWANADELPHIA MICROBIOTHERIA
XENARTHRA Dasypodidae Anadasypus Pedrolypeutes Nanoastegotherium Pampatheriidae Scirrotherium Glyptodontidae Neoglyptatelus Asterostemma Mylodontoidea Pseudoprepotherium Glossotheriopsis Neonematherium Megalonychidae Undetermined taxon Nothrotheriidae Undetermined taxa Megatheriidae Undetermined taxon Myrmecophagidae Neotamandua
LITOPTERNA Proterotheriidae Prolicaphrium Prothoatheri Megadolodus Macraucheniidae Theosodon ASTRAPOTHERIA Astrapotheriidae Xenastrapotherium Granastrapotherium NOTOUNGULATA Toxodontidae Pericotoxodon Leontiniidae Huilatherium Interatheriidae Miocochilius
Microbiotheriidae Pachybiotherium
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
RODENTIA Dinomyidae ‘Scleromys’ ‘Olenopsis’ cf. Simplimus sp. Dasyproctidae Neoreomys Microscleromys Erethizontidae ?Steiromys sp. Microsteiromys Echimyidae Acarechimys Caviidae Prodolichotis Undetermined taxa PRIMATES Atelidae Neosaimiri Cebupithecia Mohanamico Stirtonia Undet. callitrichine
514
C. ARGOT
Thylacosmilus atrox The highly specialized Thylacosmilus is the only large strictly carnivorous borhyaenoid living in the Huayquerian fauna of northern Argentina. Although a detailed study of the evolution of amphicyonids suggests that increased specialization of predators may occur in a carnivore community as a result of decreased herbivore diversity (Viranta, 1996), the fauna associated with thylacosmilids does not clarify the appearance of such specialized predators (Table 5). During the Huayquerian, notoungulates are restricted to three families, Hegetotheriidae and Mesotheriidae (small to medium-sized rodent-like forms) and Toxodontidae (larger in size, similar to rhinoceros and hippopotamus). It is possible that glyptodonts and giant rodents took advantage of the decline of notoungulates to exploit new adaptive niches (Ortiz Jaureguizar, 1998). Thylacosmilids may have evolved in coevolution with huge ground-sloths, notoungulates (toxodontids), perhaps some litopterns (macraucheni-
ids), and giant rodents (Pascual & Ortiz Jaureguizar, 1990). However, it is unlikely that thylacosmilids attacked the two former groups. Their most probable prey were thus mesotheriids (Notoungulata), macraucheniids (Litopterna), and dinomyids and hydrochoerids (Rodentia). The sabre-tooth adaptation is viewed as an extension of the throat-bite, a learned behaviour in felids (Martin, 1980, 1989). A stab into the side of the neck, with the curvature exhibited by the canines, would bring the edge of the canine back under the carotid and jugular, their tearing resulting in almost instantaneous death (Kurtén, 1952; Martin, 1980, 1989). A stab into the neck requires forelimbs powerful enough to manipulate the prey and position it for the killing slash (Radinsky & Emerson, 1982). The dirk-toothed cats and nimravids whose postcranial skeletons are known (Smilodon, Barbourofelis ) exhibit like Thylacosmilus huge muscle attachments, massive proximal limb bones, and shortened distal segments of the leg
Table 5. Generic list of Huayquerian mammals from Argentina, slightly modified from Marshall et al. (1983) DIDELPHIDA DIDELPHIMORPHIA
Didelphidae Hyperdidelphis Lutreolina Thylatheridium Sparassocynidae Sparassocynus GLIRIMETATHERIA
Simpsonitheria Argyrolagidae Microtragulus SUDAMERIDELPHIA
Sparassodonta Borhyaenoidea Notictis Borhyaenidium Stylocinus Eutemnodus Thylacosmilus
XENARTHRA Dasypodidae Chorobates Doellotatus Paleuphractus Paraeuphractus Pampatheriidae Kraglievichia Vassallia Glyptodontidae Aspidocalyptus Coscinocercus Cranithlastus ?Eleutherocercus Eosclerocalyptus Glyptodontidium Hoplophractus Peiranoa Phlyctaenopyga Stromaphorus ?Urotherium Nothrotheriidae Pronothrotherium Megatheriidae Plesiomegatherium Pyramiodontherium Mylodontoidea Elassotherium Sphenotherus Myrmecophagidae Neotamandua Palaeomyrmidon
LITOPTERNA Proterotheriidae Brachytherium ?Diadiaphorus ?Eoauchenia Epecuenia Macraucheniidae Macrauchenidia Promacrauchenia NOTOUNGULATA Toxodontidae Pisanodon Xotodon Mesotheriidae Pseudotypotherium ?Typotheriopsis Hegetotheriidae Hemihegetotherium Paedotherium Raulringueletia Tremacyllus CARNIVORA Procyonidae Cyonasua
RODENTIA Octodontidae ? Neophanomys Phtoramys Pseudoplateomys Abrocomidae Protabrocoma Echimyidae Carterodon Proechimys Trichomys Chinchillidae Lagostomopsis Dinomyidae Diaphoromys ? Potamarchus Telicomys Tetrastylopsis Tetrastylus Caviidae Cardiomys ?Caviodon Orthomyctera Palaeocavia Erethizontidae Neosteiromys Hydrochoeridae ?Cardiatherium Kiyutherium ? Procardiatherium
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
EVOLUTION OF BORHYAENOIDEA (arm, crus, and metapodials) (Merriam & Stock, 1932; Martin, 1980; Anyonge, 1996; Turner & Antón, 1997). In acquiring the strength to overpower large prey, it seems that sabretooth forms sacrificed speed and endurance. Post-lactational parental care may have been a factor of considerable importance in the evolution of specialized techniques for killing large prey, for it permits the development of killing methods which are beyond the capabilities of young animals, as well as of those which require a considerable degree of experience before they become fully effective (Ewer, 1969). Unfortunately, no fossil clues remain which suggest anything of that sort for Thylacosmilus, although it is hard to see how such complex predatory behaviour could have developed in the young while they were still unable to become specialized killers of prey larger than themselves. One of the major differences between dasyurids and felids in the development of the killing technique is that cats learn how to use skilfully the innate repertoire of paw movements to position prey for the killing bite (Pellis & Officer, 1987), something Thylacosmilus was also expected to learn. By contrast, living dasyurids are not endowed with the same variety of paw movements, although with experience the forepaws can be used during frontal attacks by Dasyuroides in order to grasp and pin prey (Pellis & Officer, 1987). It has been suggested for Smilodon that large kills were provided by a mature animal, allowing young individuals whose canines had not yet reached their full development to feed upon these kills (Hough, 1949). Similar behaviour in Thylacosmilus would suggest cooperation unknown in modern marsupials. An important aspect of hunting strategy concerns the acuteness of senses. Savage (1977) determined the degree of stereoscopic vision for various predators, and found that the estimated overlap of left and right visual fields was 30–50 ∞ in Thylacosmilus, a low value compared with modern predators. It is noteworthy that in the lion’s hunting methods, surprise is of overriding importance for initial capture and also for reducing the risk of self-injury. On the open plains, there is seldom enough cover for concealed stalking, so lions hunting by day must usually wait in high vegetation, often by a watering place, for game to come to them (Estes, 1967). The Huayquerian savannah-like environments suggest quite similar hunting methods for thylacosmilids, which are not equipped for pursuit. Moreover, in the Ngorongoro crater, probably 90% of kills are made at night (Estes, 1967). Although fossils do not carry traces of nocturnal hunting, this might account for the low stereoscopic vision of thylacosmilids. It may be significant that Thylacosmilus had evolved a specialized, unique (compared to other borhyaenoids) ear region, in
515
which the enlarged hypotympanic sinus is contained within a fully ossified, expanded auditory bulla. The large and complex hypotympanic sinus increases middle ear volume which prevents excessive damping of sound energy transmission (Turnbull & Segall, 1984). This condition is consistent with life in a quite arid environment (Hunt, 1974; Savage, 1977). Because sound absorption in air becomes greater with a decrease in humidity and an increase in temperature, selective pressures in warm arid environments tend to favour mammals equipped with a highly inflated bulla, providing a high degree of auditory sensitivity (Hunt, 1974). The inflated bulla of Thylacosmilus may have therefore favoured acute hearing, which could compensate for the poor stereoscopic vision related to small eyes facing laterally. Because the auditory bullae of Thylacosmilus and Smilodon exhibit similar modelling, Turnbull & Segall (1984) suggested that they may reflect a similar role related to hearing, one especially matched to the peculiar demands of the sabretooth way of life. However, this role is not yet understood. During the Huayquerian, Thylacosmilus coexisted with different dog-like borhyaenoids: (1) the smallsized hathlyacynids Notictis ortizi and Borhyaenidium musteloides (Marshall, 1981), which occur in faunas of the same age but from different Argentinian localities (Marshall, 1978a). Notictis was quite similar in dental structure to the contemporaneous didelphid whereas B. musteloides Lutreolina, appears to be more carnivorous than the contemporaneous Didelphis pattersoni (Marshall, 1981); (2) the poorly known medium-sized Eutemnodus spp. ( E. acutidens, E. americanus, E. propampinus), quite similar in size and dental structure to the Santacrucian borhyaenid Acrocyon sectorius (Marshall, 1978a); (3) the large-sized prothylacynid Stylocynus paranensis. This taxon is a relatively large borhyaenoid, probably the most bear-like according to its dental specialization that suggests an omnivorous diet (Marshall, 1979). Therefore, during the Late Miocene and Pliocene, Thylacosmilus was without doubt the most specialized carnivorous mammalian predator, exhibiting a combination of characters that represents the culmination of a long evolutionary history. No member of the superfamily Borhyaenoidea, as far as is known, could seriously compete with it. Also of relevance is the fact that predatory phororhacoids became increasingly significant in the late Tertiary of South America (Marshall, 1994). Because Thylacosmilus was singularly inadequate as a processor of bones (Goin & Pascual, 1987) it would have probably left significant parts of large kill untouched, which suggests that the giant birds were at least partly scavengers, in the absence of ‘dog-like’ marsupial taxa.
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
516
C. ARGOT
CONCLUSIONS Borhyaenoids do not exhibit a constant and regular increase in size and diversity during the 60 Myr of their development. While the oldest specimen known is small, the largest borhyaenoids evolved during the Eocene and Oligocene (i.e. between 54 and 24 Mya), before the main radiation of the superfamily. This radiation occurred during the Santacrucian (approx. 17 Mya), together with a radiation of rodents and small notoungulates, which were potential prey for Hathlyacynidae, Prothylacinidae and Borhyaenidae. As far as is known, borhyaenoids survived until the Chapadmalalan (late Pliocene, 3.5–4 Mya), with the most specialized family of the group, the Thylacosmilidae, having developed during the last 10 Myr. Their evolution is thus characterized by various periods of radiation and extinction, and by the development of a broad range of body plans, the marten-like Cladosictis contrasting with the digitigrade Borhyaena and powerfully built Thylacosmilus. All the borhyaenoids known from postcranial elements appear to have possessed considerable strength in the neck and forequarters, prehensile forepaws and a pseudo-opposable pollex, probably required to capture struggling prey and/or to climb efficiently. Mayulestes is the oldest and only Palaeocene borhyaenoid known from postcranial elements. It is distinct from the other taxa examined by its age, size and particular morphology. Some of these features, particularly the morphology of the pectoral and pelvic girdles, are already specialized, compared with those of the other Tiupampan metatherians (Muizon & Argot, 2003). Because of these specializations, the notion of primitive pattern does not make sense for Mayulestes. Mayulestidae appear totally distinct from the late Palaeocene lineages that gave birth to the Miocene taxa, and in view of the still unresolved appearance of large Eocene taxa, it is suggested that early radiations occurred within the group. The end of the Oligocene and late Miocene represent two critical periods in the evolution of the Borhyaenoidea, probably in relation to global environmental changes. The post-Santacrucian climatic changes in particular led to the rarefaction of subtropical woodlands, which were replaced by savannah-like environments (Pascual & Ortiz Jaureguizar, 1990). This occurred together with the simultaneous predominance of the sabretooth thylacosmilids and growing scarcity of the ‘dog-like’ taxa. The appearance of sabretooth taxa offers an interesting parallel with similar evolutionary developments that occurred within various other groups (Nimravidae, Felidae, Creodonta), demonstrating that this adaptation was not an unusual evolutionary track. The lifestyle and hunting and killing strategies of thylacosmilids are totally dis-
tinct from those of the ‘dog-like’ borhyaenoids, although the associated fauna does not explain the reasons for their appearance. Some dog-like taxa survived in the northern parts of the continent, e.g. in Colombia, probably because the environmental conditions in the equatorial tropics were less affected by the gradual global climatic zonation that occurred in southern South America (Patterson & Pascual, 1968). One question which has been posed concerns why only thylacosmilids survived and why the dog-like taxa did not evolve greater cursoriality in the southern plains. It has been suggested that borhyaenoids were gradually eliminated by placental carnivorans and didelphids that began to develop from the Huayquerian onward. At that time, the radiation of didelphids was characterized by a tendency towards the development of carnivorous types, some of them very specialized, such as Hyperdidelphis or Sparassocynus (Simpson, 1974). These taxa might have occupied the adaptive zone of hathlyacynids, thus representing an ‘endo-replacement’ (Goin, 1989). Concerning the ‘exoreplacement’ of marsupials by placentals, it has been suggested that procyonids might have competed with the omnivorous Lycopsis / Pseudolycopsis lineage, or replaced it, as suggested by the successive occurrence of Pseudolycopsis during the Chasicoan and Cyonasua during the Huayquerian (Marshall, 1978a). However, it is not clear why so few borhyaenoids survived during late Miocene -Pliocene, considering the array of potential prey, especially rodents and some notoungulates. Although much has been said about the extinction of Borhyaenoidea due to competition with large grounddwelling birds or other mammals (see Marshall, 1978a), it is worth stressing the change in environmental parameters that occurred. During the Santacrucian, borhyaenoids were abundant, reaching a climax in variety of size and adaptations, apparently without strong constraints on the rate and direction of morphological evolution, since all the mammalian predatory ecological niches were vacated. The fact that the Santacrucian area was at first largely wooded, with a subtropical climate, may have influenced the evolution of the Borhyaenoidea, which were not confined to the ground with the probable exception of Borhyaena (as far as we know from postcranial elements). They were probably able to exploit all strata of tropical forests, a strategy which provides opportunities for greater diversification in niche occupancy (Kleiman & Eisenberg, 1973). While Palaeocene southcentral Bolivia, early Miocene Patagonia, and middle Miocene Colombia were characterized primarily by a subtropical climate and forested environment, Thylacosmilus is the only borhyaenoid examined that did not live in a densely covered milieu, and the savannah-like environments of the late Miocene -Pliocene
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
EVOLUTION OF BORHYAENOIDEA seem to have been less favourable to the dog-like taxa. With less than one third of the taxa recorded known from postcranial skeletons, improved understanding of the locomotor evolution of Borhyaenoidea is critically dependent on the discovery of further remains. More material is required in order to fully document the evolution of the ecomorphs that occurred within each family.
ACKNOWLEDGEMENTS For access to borhyaenoid specimens, I am grateful to the following people and institutions: José Bonaparte (Museo Argentino de Ciencias Naturales ‘Bernardino Rivadavia’, Buenos Aires), John J. Flynn (Field Museum of Natural History, Chicago), Patricia Holroyd (Museum of Palaeontology, University of California, Berkeley) and Mary Ann Turner (Peabody Museum, Yale University, New Haven). I thank Christian Muizon for critically reviewing the first version of this manuscript, as well as Jean-Pierre Gasc, Pascal Tassy, and two anonymous reviewers for comments that helped improve the text. Financial support was provided by the Muséum national d’Histoire naturelle, Paris.
REFERENCES Alexander R, McN. 1993. Legs and locomotion of Carnivora. Symposia of the Zoological Society of London 65: 1–13. Ameghino F. 1895. Sur les oiseaux fossiles de Patagonie. Boletín del Instituto Geográphico Argentino 15: 1–104. Anyonge W. 1996. Locomotor behaviour in Plio-Pleistocene sabre-tooth cats: a biomechanical analysis. Journal of Zoology, London 238: 395–413. Archer M. 1982. A review of Miocene thylacinids (Thylacinidae, Marsupialia), the phylogenetic position of the Thylacinidae and the problem of apriorisms in character analysis. In: Archer M, ed. Carnivorous marsupials. Sydney: The Royal Society of New South Wales, 445–476. Argot C. 2001. Functional-adaptive anatomy of the forelimb in the Didelphidae, and the palaeobiology of the Palaeocene marsupials Mayulestes ferox and Pucadelphys andinus . Journal of Morphology 247: 51–79. Argot C. 2002. Functional-adaptive analysis of the hindlimb anatomy of extant marsupials, and the palaeobiology of the Palaeocene marsupials Mayulestes ferox and Pucadelphys andinus. Journal of Morphology 253: 76–108. Argot C. 2003a. Functional-adaptive anatomy of the axial skeleton of some extant marsupials, and the palaeobiology of the Palaeocene marsupials Mayulestes ferox and Pucadelphys andinus. Journal of Morphology 255: 279– 300. Argot C. 2003b. Functional adaptations of the postcranial skeleton of two Miocene borhyaenoids (Mammalia, Metathe-
517
ria), Borhyaena and Prothylacinus, from South America. Palaeontology 46: 1213–1267. Argot C. 2003c. Postcranial functional adaptations in the South American Miocene borhyaenoids (Mammalia, Metatheria) Cladosictis, Pseudonotictis, and Sipalocyon. Alcheringa 27: 303–356. Argot C. in press. Functional-adaptive features of the postcranial skeleton of a sabertooth borhyaenoid, Thylacosmilus atrox (Metatheria), and palaeobiologic implications. Alcheringa in press. Barnett CH, Napier JR. 1953. The form and mobility of the fibula in metatherian mammals. Journal of Anatomy 87: 207–213. Bonaparte JF, Van Valen LM, Kramartz A. 1993. La fauna local de Punto Peligro, Palaeocene inferior, de la provincia del Chubut, Patagonia, Argentina. Evolutionary Mono graphs 14: 1–61. Bond M, Pascual R. 1983. Nuevos y elocuentes restos craneanos de Proborhyaena gigantea Ameghino, 1897 (Marsupialia, Borhyaenidae, Proborhyaeninae) de la Edad Deseadense. Un ejemplo de coevolucion. Ameghiniana 20: 47–60. Bown TM, Fleagle JG. 1993. Systematics, biostratigraphy, and dental evolution of the Palaeothentidae, Later Oligocene to Early-Middle Miocene (Deseadan-Santacrucian) caenolestoid marsupials of South America. Journal of Paleontology, Memoir 29: 1–76. Carbone C, Mace GM, Roberts SC, MacDonald DW. 1999. Energetic constraints on the diet of terrestrial carnivores. Nature 402: 286–288. Carlini AA, Vizcaíno SF, Scillato -Yané GJ. 1997. Armored Xenarthrans: a unique taxonomic and ecologic assemblage. In: Kay RF, Madden RH, Cifelli RL, Flynn JJ, eds. Vertebrate paleontology in the Neotropics. The Miocene fauna of La Venta, Colombia. Washington, DC: Smithsonian Institution Press, 213–226. Davis DD. 1964. The giant panda. A morphological study of evolutionary mechanisms. Fieldiana: Zoology Memoirs 3: 1– 327. Dumont ER, Bown TM. 1997. New caenolestoid marsupials. In: Kay RF, Madden RH, Cifelli RL, Flynn JJ, eds. Vertebrate paleontology in the Neotropics. The Miocene fauna of La Venta, Colombia. Washington, DC: Smithsonian Institution Press, 207–212. Edmund G, Theodor J. 1997. A new giant pampatheriid armadillo. In: Kay RF, Madden RH, Cifelli RL, Flynn JJ, eds. Vertebrate paleontology in the Neotropics. The Miocene fauna of La Venta, Colombia. Washington, DC: Smithsonian Institution Press, 227–232. Eisenberg JF, Leyhausen P. 1972. The phylogenesis of predatory behaviour in mammals. Zeitschrift für Tierpsychologie 30: 59–93. Emmons LH. 1992. Les jaguars. In: Seidensticker J, Lumpkin S, eds. Les félins. Paris: Bordas, 116–123. Estes RD. 1967. Predators and scavengers. Natural History 76: 20–29, 38–47. Ewer RF. 1969. Some observations on the killing and eating of prey by two dasyurid marsupials: the mulgara, Dasycercus
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
518
C. ARGOT
cristicauda, and the Tasmanian devil, Sarcophilus harrisii. Zeitschrift für Tierpsychologie 26: 23–38. Ewer RF. 1998. The carnivores. Ithaca, NY: Cornell University Press. Fleagle JG, Kay RF, Anthony MRL. 1997. Fossil New World monkeys. In: Kay RF, Madden RH, Cifelli RL, Flynn JJ, eds. Vertebrate paleontology in the Neotropics. The Miocene fauna of La Venta, Colombia. Washington, DC: Smithsonian Institution Press, 473–495. Flynn JJ, Swisher CC III. 1995. Cenozoic South American land mammal ages: correlation to global geochronologies. Geochronology Time Scales and Global Stratigraphic Correlation, SEPM Special Publication 54: 317–333. Flynn JJ, Wyss AR. 1998. Recent advances in South American mammalian palaeontology. Trends in Ecology and Evolution 13: 449–454. Genise JF, Bown TM. 1994. New Miocene scarabeid and hymenopterous nests and Early Miocene (Santacrucian) paleoenvironments, Patagonian Argentina. Ichnos 3: 107– 117. Goin FJ. 1989. Late Cenozoic South American marsupial and placental carnivores: changes in predator-prey evolution. Fifth International Theriological Congress (Rome 22–29 August 1989, abstracts of papers and posters) 271–272. Goin FJ. 1997. New clues for understanding Neogene marsupial radiations. In: Kay RF, Madden RH, Cifelli RL, Flynn JJ, eds. Vertebrate paleontology in the Neotropics. The Miocene fauna of La Venta, Colombia. Washington, DC: Smithsonian Institution Press, 187–206. Goin FJ, Pascual R. 1987. News on the biology and taxonomy of the marsupials Thylacosmilidae (late Tertiary of Argentina). Anales de la Academia Nacional de Ciencias Exactas, Físicas y Naturales 39: 219–246. Goszczynski J. 1986. Locomotor activity of terrestrial predators and its consequences. Acta Theriologica 31: 79–95. Haines RW. 1942. The tetrapod knee joint. Journal of Anatomy 76: 270–301. Hirschfeld SE. 1985. Ground sloths from the Friasian La Venta fauna, with additions to the pre-Friasian Coyaima fauna of Colombia, South America. University Of California Publications in Geological Sciences 128: 1–90. Hough J. 1949. The habits and adaptation of the Oligocene saber tooth c arnivore, Hoplophoneus. Geological Survey Pro fessional Paper 221-H: 125–137. Hunt RM Jr. 1974. The auditory bulla in Carnivora: an anatomical basis for reappraisal of carnivore evolution. Journal of Morphology 143: 21–76. Iwaniuk AN, Nelson JE, Ivanco TL, Pellis SM, Whishaw IQ. 1998. Reaching, grasping and manipulation of food objects by two tree kangaroo species, Dendrolagus lumholtzi and Dendrolagus matschiei. Australian Journal of Zoology 46: 235–248. Janis CM, Wilhelm PB. 1993. Were there mammalian pursuit predators in the Tertiary? Dances with wolf avatars. Journal of Mammalian Evolution 1: 103–125. Johnson SC, Madden RH. 1997. Uruguaytheriine astrapotheres of tropical South America. In: Kay RF, Madden RH, Cifelli RL, Flynn JJ, eds. Vertebrate paleontology in the Neo-
tropics. The Miocene fauna of La Venta, Colombia. Washington, DC: Smithsonian Institution Press, 355–383. Johnson CN, Wroe S. 2003. Causes of extinction of vertebrates during the Holocene of mainland Australia: arrival of the dingo, or human impact? Holocene 13: 1009–1016. Jolly CJ. 1967. The evolution of the baboons. In: Vagtborg H, ed. The baboon in medical research, Vol. II. Austin, TX: University of Texas Press, 23–50. Kay RF, Madden RH. 1997. Paleogeography and paleoecology. In: Kay RF, Madden RH, Cifelli RL, Flynn JJ, eds. Vertebrate paleontology in the Neotropics. The Miocene fauna of La Venta, Colombia. Washington, DC: Smithsonian Institution Press, 520–550. Kleiman DG, Eisenberg JF. 1973. Comparisons of canid and felid social systems from an evolutionary perspective. Animal Behaviour 21: 67–659. Kruuk H. 1976. Feeding and social behaviour of the striped hyaena ( Hyaena vulgaris Desmarest). East African Wildlife Journal 14: 91–111. Kruuk H, Turner M. 1967. Comparative notes on predation by lion, leopard, cheetah and wild dog in the Serengeti area, East Africa. Mammalia 31: 1–27. Kurtén B. 1952. The Chinese hipparion fauna. Societas Sci entiarum Fennica, Commentationes Biologicae 13: 1–82. Landry SO. 1958. The function of the entepicondylar foramen in mammals. American Midland Naturalist 60: 100–112. Langston W Jr, Gasparini Z. 1997. Crocodilians, Gryposuchus, and the South American gavials. In: Kay RF, Madden RH, Cifelli RL, Flynn JJ, eds. Vertebrate paleontology in the Neotropics. The Miocene fauna of La Venta, Colombia. Washington, DC: Smithsonian Institution Press, 113– 154. Marshall LG. 1976. Evolution of the Thylacosmilidae, extinct saber-tooth marsupials of South America. Paleobios 23: 1– 30. Marshall LG. 1977a. A new species of Lycopsis (Borhyaenidae: Marsupialia) from the La Venta fauna (Late Miocene) of Colombia, South America. Journal of Paleontology 51: 633–642. Marshall LG. 1977b. Evolution of the carnivorous adaptive zone in South America. In: Hecht MK, Goody PC, Hecht BM, eds. Major patterns in vertebrate evolution. New York, London: Plenum Press, 709–721. Marshall LG. 1978a. Evolution of the Borhyaenidae, extinct South American predaceous marsupials. University Of California Publications in Geological Sciences 117: 1–89. Marshall LG. 1978b. The terror bird. Field Museum of Natural History Bulletin 49: 6–15. Marshall LG. 1979. Review of the Prothylacyninae, an extinct subfamily of South American ‘dog-like’ marsupials. Fieldiana Geology N. S. 3: 1–50. Marshall LG. 1981. Review of the Hathlyacyninae, an extinct subfamily of South American ‘dog-like’ marsupials. Fieldiana: Geology N. S. 7: 1–120. Marshall LG. 1988. Land mammals and the Great American Interchange. American Scientist 76: 380–388. Marshall LG. 1994. The terror birds of South America. Sci entific American 270: 64–69.
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
EVOLUTION OF BORHYAENOIDEA Marshall LG, Muizon C, Sigogneau-Russell D. 1995. The locality of Tiupampa: age, taphonomy and mammal fauna. In: de Muizon C, ed. Pucadelphys andinus (Marsu pialia, Mammalia) from the Early Palaeocene of Bolivia. Paris: Mémoires du Muséum national d’Histoire naturelle, 11–20. Marshall LG, Hoffstetter R, Pascual R. 1983. Mammals and stratigraphy: Geochronology of the continental mammal-bearing Tertiary of South America. Palaeovertebrata, Mémoire extraordinaire 1–93. Marshall LG, Sigogneau-Russell D. 1995. Postcranial skeleton. In: Muizon C, ed. Pucadelphys andinus (Marsupialia, Mammalia) from the Early Palaeocene of Bolivia. Paris: Mémoires du Muséum national d’Histoire naturelle, 91–164. Martin LD. 1980. Functional morphology and the evolution of cats. Transactions of the Nebraska Academy of Sciences 8: 141–154. Martin LD. 1989. Fossil history of the terrestrial Carnivora. In: Gittleman JL, ed. Carnivore behaviour, ecology and evolution. London: Chapman & Hall, 382–409. Maynard Smith J, Savage RJG. 1956. Some locomotory adaptations in mammals. Zoological Journal of the Linnean Society 42: 603–622. McDonald HG. 1997. Xenarthrans: Pilosans. In: Kay RF, Madden RH, Cifelli RL, Flynn JJ, eds. Vertebrate paleontology in the Neotropics. The Miocene fauna of La Venta, Colombia. Washington, DC: Smithsonian Institution Press, 233– 245. McKenna MC, Bell SK. 1997. Classification of mammals above the species level. New York: Columbia University Press. Merriam JC, Stock C. 1932. The Felidae of Rancho La Brea. Carnegie Institution of Washington, DC Publications 422: 1– 231. Muizon C de. 1992. La fauna de mamiferos de Tiupampa (Palaeoceno Inferior, Formación Santa Lucía), Bolivia. In: Suarez-Soruco R, ed. Fosiles y facies de Bolivia, Vol. I: Vertebrados. Santa Cruz (Bolivia): Revista Técnica de YPFB, 575–624. Muizon C de. 1998. Mayulestes ferox, a borhyaenoid (Metatheria, Mammalia) from the early Palaeocene of Bolivia. Phylogenetic and palaeobiologic implications. Geodiversitas 20: 19–142. Muizon C de. 1999. Marsupial skulls from the Deseadan (late Oligocene) of Bolivia and phylogenetic analysis of the Borhyaenoidea (Marsupialia, Mammalia). Geobios 32: 483– 509. Muizon C de, Argot C. 2003. Comparative anatomy of the didelphimorph marsupials from the early Palaeocene of Bolivia ( Pucadelphys, Andinodelphys, and Mayulestes). Palaeobiologic implications. In: Jones M, Dickman C, Archer M, eds. Predators with pouches: the biology of carnivorous marsupials. Collingwood: CSIRO Publishing, 43– 62. Muizon C de, Cifelli RL. 2000. The ‘condylarths’ (archaic Ungulata, Mammalia) from the early Palaeocene of Tiupampa (Bolivia): implications on the origin of the South American ungulates. Geodiversitas 22: 47–150.
519
Muizon C de, Cifelli RL. 2001. A new basal ‘didelphoid’ (Marsupialia, Mammalia) from the early Palaeocene of Tiupampa (Bolivia). Journal of Vertebrate Palaeontology 21: 87– 97. Muizon C de, Cifelli RL, Céspedez Paz R. 1997. The origin of the dog-like borhyaenoid marsupials of South America. Nature 389: 486–489. Muizon C de, Lange-Badré B. 1997. Carnivorous dental adaptation in tribosphenid mammals and phylogenetical reconstruction. Lethaia 30: 353–366. Ortiz Jaureguizar E. 1998. Palaeoecología y evolución de la fauna de mamiferos de América del Sur durante la ‘edad de las planicies australes’ (Mioceno superior–Plioceno superior). Estudios Geologicos 54: 161–169. Oxnard CE. 1963. Locomotor adaptations in the primate forelimb. Symposium of the Zoological Society of London 10: 165–182. Oxnard CE. 1968. The architecture of the shoulder in some mammals. Journal of Morphology 126: 249–290. Pascual R. 1989. Late Cenozoic mammal dispersal between the Americas: an overview of the South American evidences. Fifth International Theriological Congress (Rome 22–29 August 1989, abstracts of papers and posters), 279– 280. Pascual R, Ortiz Jaureguizar E. 1990. Evolving climates and mammal faunas in Cenozoic South America. Journal of Human Evolution 19: 23–60. Pascual R, Ortiz Jaureguizar E. 1991. El ciclo faunistico cochabambiano (Palaeoceno temprano). Su incidencia en la historia biogeografica de los mamiferos sudamericanos. In: Suarez-Soruco R, ed. Fosiles y facies de Bolivia, Vol. I: Vertebrados. Santa Cruz (Bolivia): Revista Técnica de YPFB, 559–574. Pascual R, Ortiz Jaureguizar E. 1992. Evolutionary pattern of land mammal faunas during the late Cretaceous and Palaeocene in South America: a comparison with the North American pattern. Annales Zoologici Fennici 28: 245–252. Patterson B, Pascual R. 1968. Evolution of mammals on southern continents. V. The fossil mammal fauna of South America. Quarterly Review of Biology 43: 409– 451. Pellis SM, Officer RCE. 1987. An analysis of some predatory behaviour patterns in four species of carnivorous marsupials (Dasyuridae) with comparative notes on the eutherian cat Felis catus. Ethology 75: 177–196. Poole TB. 1974. Detailed analysis of fighting in polecats (Mustelidae) using ciné film. Journal of Zoology, London 173: 369–393. Rabinowitz AR, Nottingham BG Jr. 1986. Ecology and behaviour of the jaguar ( Panthera onca) in Belize, Central America. Journal of Zoology, London 210: 149–159. Radinsky L, Emerson S. 1982. The late, great sabertooths. Natural History 91: 50–57. Rae TC, Bown TM, Fleagle JG. 1996. New palaeothentid marsupials (Caenolestoidea) from the early Miocene of Patagonian Argentina. American Museum Novitates 3165: 1–10.
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521
520
C. ARGOT
Rasmussen DT. 1997. Birds. In: In: Kay RF, Madden RH, Cifelli RL, Flynn JJ, eds. Vertebrate paleontology in the Neotropics. The Miocene fauna of La Venta, Colombia. Washington, DC: Smithsonian Institution Press, 171–183. Riggs ES. 1934. A new marsupial saber-tooth from the Pliocene of Argentina and its relationships to other South American predacious marsupials. Transactions of the American Philosophical Society, N. S. 24: 1–31. Roberts D. 1974. Structure and function of the primate scapula. In: Jenkins FA Jr, ed. Primate locomotion. New York: Academic Press, 171–200. Rosenzweig ML. 1966. Community structure in sympatric Carnivora. Journal of Mammalogy 47: 602–611. Rosenzweig ML. 1968. The strategy of body size in mammalian carnivores. American Midland Naturalist 80: 299– 315. Savage RJG. 1977. Evolution in carnivorous mammals. Palaeontology 20: 237–271. Schilling N, Fischer MS. 1999. Kinematic analysis of treadmill locomotion of tree shrews, Tupaia glis (Scandentia: Tupaiidae). Zeitschrift für Säugetierkunde 64: 129–153. Scott WB. 1903a. Mammalia of the Santa Cruz beds: Edentata, Dasypoda. In: Scott WB, ed. Reports of the Princeton University Expedition to Patagonia, 1896–1899, Vol. V, Part I, 1 . Princeton University, 1–106, plates I XVI. Scott WB. 1903b. Mammalia of the Santa Cruz beds: Edentata, Glyptodontia and Gravigrada. In: Scott WB, ed. Reports of the Princeton University Expedition to Patagonia, 1896– 1899, Vol. V, Part I, 2. Princeton University, 107–227, plates XVII- XXXV. Scott WB. 1904. Mammalia of the Santa Cruz beds: Edentata, Gravigrada. In: Scott, WB, ed. Reports of the Princeton University Expedition to Patagonia, 1896–1899, Vol. V, Part I, 3. Princeton University, 227–364, plates XXXVI LXIII. Scott WB. 1905. Insectivora and Glires. In: Scott, WB, ed. Reports of the Princeton University Expedition to Patagonia, 1896–1899, Vol. V, Parts 2, 3. Princeton University, 365–499, plates LXIV-LXXI. Scott WB. 1910. Litopterna of the Santa Cruz beds. In: Scott, WB, ed. Reports of the Princeton University Expedition to Patagonia, 1896–1899, Vol. VII, Part I . Princeton University, 1–156, plates I- XX. Scott WB. 1932. Nature and origin of the Santa Cruz fauna. In: Scott, WB ed. Reports of the Princeton University Expedition to Patagonia, 1896–1899, Vol. VII, Part 3. Princeton University: 193–238, plates A-K. Shapiro LJ, Jungers WL. 1994. Electromyography of back muscles during quadrupedal and bipedal walking in primates. American Journal of Physical Anthropology 93: 491– 504. Simpson GG. 1974. Notes on Didelphidae (Mammalia, Marsupialia) from the Huayquerian (Pliocene) of Argentina. American Museum Novitates 2559: 1–15. Sinclair WJ. 1906. Marsupialia of the Santa Cruz beds. In:
Scott WB, ed. Reports of the Princeton University Expedition to Patagonia, 1896–1899 , Vol. IV, Part 3 . Princeton University, 333–460, plates XL-LXV. Sinclair WJ, Farr MS. 1932. Aves of the Santa Cruz beds. In: Scott WB, ed. Reports of the Princeton University Expedition to Patagonia, 1896–1899 , Vol. VII, Part 2. Princeton University, 157–191, plates XXI- XXXV. Smith M. 1982. Review of the thylacine (Marsupialia, Thylacinidae). In: Archer M, ed. Carnivorous marsupials. Sydney: The Royal Zoological Society of New South Wales, 237–253. Stein BR, Casinos A. 1997. What is a cursorial mammal? Journal of Zoology, London 242: 185–192. Szalay FS. 1982. A new appraisal of marsupial phylogeny and classification. In: Archer M, ed. Carnivorous marsupials. Sydney: The Royal Society of New South Wales. 621– 640. Szalay FS. 1994. Evolutionary history of the marsupials and an analysis of osteological characters. New York: Cambridge University Press. Szalay FS, Sargis EJ. 2001. Model-based analysis of postcranial osteology of marsupials from the Paleocene of Itaboraí, Brazil, and the phylogenetics and biogeography of Metatheria. Geodiversitas 23: 139–302. Tauber AA. 1997. Paleoecología de la Formación Santa Cruz (Mioceno inferior) en el extremo sudeste de la Patagonia. Ameghiniana 34: 517–529. Taylor ME. 1974. The functional anatomy of the forelimb of some African Viverridae (Carnivora). Journal of Morphology 143: 307–336. Turnbull WD, Segall W. 1984. The ear region of the marsupial sabertooth, Thylacosmilus: influence of the sabertooth lifestyle upon it, and convergence with placental sabertooths. Journal of Morphology 181: 239– 270. Turner A, Antón M. 1997. The big cats and their fossil relatives. New York: Columbia University Press. Van Valkenburgh B. 1985. Locomotor diversity within past and present guilds of large predatory mammals. Paleobiology 11: 406–428. Van Valkenburgh B. 1989. Carnivore dental adaptations and diet: a study of trophic diversity within guilds. In: Gittleman JL, ed. Carnivore behaviour, ecology and evolution. London: Chapman & Hall, 410–436. Viranta S. 1996. European Miocene Amphicyonidae – taxonomy, systematics and ecology. Acta Zoologica Fennica 204: 1–61. Walton AH. 1997. Rodents. In: Kay RF, Madden RH, Cifelli RL, Flynn JJ, eds. Vertebrate paleontology in the Neotropics. The Miocene fauna of La Venta, Colombia . Washington, DC: Smithsonian Institution Press, 392–409. Wayne RK, Van Valkenburgh B, Kat PW, Fuller TK, Johnson WE, O’Brien SJ. 1989. Genetic and morphological divergence among sympatric canids. Journal of Heredity 80: 447–454. Wemmer CM. 1977. Comparative ethology of the large spotted genet (Genetta tigrina) and some related viverrids. Smithsonian Contributions to Zoology 239: 1–93.
© 2004 The Linnean Society of London, Zoological Journal of the Linnean Society, 2004, 140, 487–521