The oldest known species of the genus Didelphis (Mammalia, Marsupialia, Didelphidae) is described here. The holotype and only known specimen comes from the late Miocene deposits of the Solimões Formation at Patos, Acre River, Brazil. The new species is one of only a few marsupials known from this assemblage. The new species differs from other known species by its smaller size, low mandibular ramus, smaller molars relative to premolars, and p3 with inflated crown. The last feature suggests more frugivorous habits than in other species of the genus.
Living species of Didelphis Linnaeus, 1758 (Mammalia, Marsupialia, Didelphidae), the largest and most conspicuous marsupials of the Americas, range from Canada to Patagonia. D. virginiana Kerr, 1792, is distributed from British Columbia to Costa Rica. D. marsupialis Linnaeus, 1758, is known from Mexico to tropical South America. The remaining 2 species are strictly South American. D. albiventris Lund, 1841, occurs through a mostly peri–Amazonian belt from Colombia and Venezuela to Argentina, whereas D. aurita Wied–Neuwied, 1826, is from southeastem Brazil (Brown 2004). Two Pampean (Argentina), extinct species of this genus also have been reported, the late Pliocene (Chapadmalalan South American Land Mammal Age) D. crucialis (Ameghino, 1904) and the late Pliocene–early Pleistocene (Chapadmalalan–Ensenadan South American Land Mammal Ages) D. reigi Simpson, 1972 (Coin 1991; Coin and Pardinas 1996).
Goin (1991) postulated that the evolutionary history of Didelphis could be traced back in time to the beginning of the Neogene. Because of the reduced exposures and consequently relatively poor fossil record for tropical regions of the Americas, most of the Neogene history of this (and many other didelphid genera) remains obscure. Remains of didelphid marsupials were reported for the Acre region of Brazil (Czaplewski 1996), but no identification at generic or lower levels was possible. Here we report the oldest record of a species of Didelphis, coming from the late Miocene deposits of the Amazonian region in Brazil.
Materials and Methods
Locality, geology, associated fauna, and age.—Specimen UFAC 5180 was collected from the Solimões Formation at Patos, at the border between Pern and Brazil along the upper Acre River (10°55′55″S, 69°55′20″W) in a riverbank cut that was approximately 38 km upriver from the towns of Assis Brasil (on the Brazilian side) and Iñapari (on the Pemvian side; Fig. 1). The outcrop is visible only during the dry season and consists of an intraformational conglomerate with clasts of clay and silt. It is part of a large paleochannel (Fig. 2). The base of the unit is not visible and its upper part is covered by vegetation.
The Solimões Formation, of Tertiary age, has a large regional extent in southwestem Amazonia. From sediment cores, the formation has a maximum thickness of 1,800 m and is about 800 m thick in the Acre region. Surface outcrops occur along road cuts and banks of the main rivers in this region, reaching a maximum thickness of 70–100 m. Latmbesse et al. (1997) interpret the Solimões deposits as being part of a megafan system with its headwaters in the Pemvian Andes. The system was a sedimentary response to tectonic movements in the Andean foreland basin. The relevant tectonic phase was initiated in the Bolivian and Pemvian Andes approximately 10 ± 2 million years ago (Jordan et al. 1983; Marshall and Sempere 1993).
The temporal interval of deposition for the formation has been reported as spanning from Paleocene to Pleistocene but examination of recent data indicates that its range may be from late early Miocene to late Miocene (Cozzuol, in press; Hoom 1993, 1994a, 1994b, 1995; Hoorn et al. 1995). Outcrops in the Acre region, including the ones containing the specimen studied here and other vertebrates, are considered to be of late Miocene age (time equivalent to Huayquerian South American Land Mammal Age) based on the amniote taxa (Cozzuol, in press). Recent radiometric dates from ashes in sediments correlative to the Solimões Formation at 2 Peruvian localities close to the Brazilian border gave dates of 9.01 and 3.12 million years ago (Campbell et al. 2001). The dates are consistent with the age indicated by the fossil vertebrates, but the precise stratigraphic position of the lower ash is not clear (Campbell et al. 2000:77).
Other specimens found at Patos (mostly under study), include po– tamotrygonid rays; catfishes of the families Callichthyidae, Doradidae, and Pimelodidae; characiform fishes of the family Characidae; crocodiles of the families Alligatoridae and Gavialidae; aquatic turtles; and mammals. Among the latter are primates of the families Cebidae and Atelidae (Kay and Cozzuol 2006), xenarthrans of the families Dasypodidae and Glyptodontidae, as well as indeterminant sloths. Rodents are represented mainly by teeth but some more– complete specimens also were recovered. They belong to the families Caviidae, Erethizontidae, Dinomyidae, Dasyproctidae, and Echimyidae. Fragmentary remains of notungulates also were found. Frailey (1986) provides a faunal list from the same locality that is generally consistent with ours.
The geological interpretation is consistent with and age-equivalent to the Huayquerian South American Land Mammal Age (9–6.8 million years ago) for the mammalian faunas (Flynn and Swisher 1995; Marshall et al. 1983). According to this chronology, the temporal constraints placed on the age of the specimen described here make it 3–6 million years older than previous records of Didelphis.
Abbreviations and terminology.—We used the following abbreviations: UFAC, Laboratório de Pesquisas Paleontológicas, Universidade Federal do Acre, Rio Branco, Acre, Brazil; MLP, División Paleontología Vertebrados, Museo de La Plata. Dental nomenclature and terminology for length (L) and width (W) of teeth follows Goin et al. (2003). Measurements with an asterisk are estimated. All measurements are in millimeters.
Superorder Marsupialia Illiger, 1811 Order Didelphimorphia Gill, 1872 Family Didelphidae Gray, 1821 Genus Didelphis Linnaeus, 1758 Didelphis solimoensis, new species Figs. 3a–c
Type.—UFAC 5180, fragment of right dentary with alveoli of pi, roots and posterior one–third of p2, and complete p3–m2. Hipodigm.—The type only.
Etymology.—solimoensis, after Solimões Formation, which yielded the holotype and only known specimen.
Locality and age.—The type comes from Patos, along the upper Acre River (10°55′55″S, 69°55′20″W), in a riverbank cut approximately 38 km upriver from the towns of Assis Brasil (on the Brazilian side) and lñapari (on the Pemvian side). Solimões Formation, late Miocene (equivalent to Huayquerian South American Land Mammal Age).
Measurements (in mm).—Lpl–p3: 12.5*; Lpl–m2: 19.5*; Lp2: 5.3; Wp2: 1.7; Lp3: 4.2; Wp3: 2.1; Lml: 3.9; Wml: 1.8; Lm2: 4.0; Wm2: 2.1.
Diagnosis.—Differs from all other species of Didelphis in the following features: smaller size, low horizontal ramus, p3 with inflated crown, smaller molars relative to the premolar series.
Description.—Even though incomplete, the completely empted p3 suggests that the holotype of D. solimoensis belonged to an aduh or subadult individual. Thus, the gracility of the dentary bone distinguishes this species among the others in the genus. The lingual face of the dentary is relatively flat. At its posterior end, the symphysis reaches a point located below the anterior root of p2. The labial face of the dentary is convex and relatively wide, even below the premolar teeth. The anterior mental foramen is proportionally larger than that of other species of Didelphis, and is located below the anterior root of p2. Judging from the preserved roots it can be appreciated that, even though absent, p1 was not very much reduced in relation to p2. In turn, and judging from the length and the size of its root, p2 (which is almost completely broken) was larger than p3, as in all modem didelphids. The most obvious feature of the crown of p3 is its bulbous, swollen aspect. This single feature clearly distinguishes D. solimoensis from all other species of the genus. The p3 is single–cusped and has its posterior crest longer and slightly more subvertical than the anterior one. Another feature that easily distinguishes D. solimoensis from other species of the genus is the proportionally smaller size of the lower molars, both in absolute measures as well as relative to the premolar teeth. The 2 preserved molars are only slightly worn; on ml the paraconid is somewhat labiolingually compressed, more so than in living species of Didelphis. The talonid of this molar is proportionally larger than that of m2. The entoconids of both molars are well developed, and their hypoconids are only slightly salient laterally. Both molars have a small, labial, single–cusped cingulid at the crown base between the hypoconid and the entoconid. The anterobasal cingulid is well developed on both molars, being narrower than that of other species of the genus.
Comparisons.—The new species can only be assigned to the genus Didelphis. Species of Lutreolina have the p2 much larger (higher and longer) than the p3, all premolars are laterally compressed, and molars have shorter talonids and more reduced entoconids. The new species differs from Philander because in the latter the p2 and p3 are laterally compressed and have more vertical anterior and posterior crests. Molars also are more laterally compressed and have comparatively longer paracristids. When compared with extinct didelphids, the only similarity between Hyperdidelphys and D. solimoensis is a short diastema between p1 and p2 (because of the former's camivorous adaptations—see a review in Goin and Pardinas ). Ii contrast, the premolars of Hyperdidelphis are more laterally compressed than in Didelphis; the molars have more robust trigonids with trenchant paracristids, and short, reduced talonids. Finally, the extinct Thylophorops is very large, its p3 has a well–developed “talonid” posterior to its posterior crest, and its molars are wider and more robust than those of any Didelphis (see also Goin and Pardinas 1996).
Didelphis solimoensis is the smallest known species of Didelphis; its p3 is comparatively wider and more bulbous, whereas ml and m2 are proportionally smaller and have narrower trigonids than in other members of the genus. Even comparing the holotype of D. solimoensis with juvenile specimens of other species of the genus, the diastema between p1 and p2 is shorter in the former. Other than these most obvious features, the fragmentary nature of the holotype of D. solimoensis makes very difficult other comparisons with living or extinct species of Didelphis. Compared with the living ones, D. solimoensis shows closer resemblances with D. marsupialis: the length of the diastema between p1 and p2 is smaller in D. marsupialis than in other living species; the crown of its p3 is proportionally more swollen (although not as much as in D. solimoensis), and its posterior talon is less developed; and the anterobasal cingulum in ml and m2 is well developed. The extinct Pampean Neogene species of the genus Didelphis, D. reigi, is much larger, its p1 is poorly reduced, the p3 is much more trenchant, the diastema between p1 and p2 is proportionally larger, and the lower molars are stronger, compared to D. solimoensis. In tum, D. crucialis is more similar in size to D. solimoensis; however, its molars are stronger, p2 is proportionally larger, and p3 has a better developed, more trenchant anterior crest.
In short, D. solimoensis is the smallest known species of Didelphis. Its known dental and mandibular features suggest closer affinities with D. marsupialis than to any other species of this genus.
Living species of Didelphis have omnivorous, opportunistic feeding habits that frequently show seasonal variations. Inneotropical forests, the diet of D. marsupialis consists of a wide variety of items, including insects, frogs, birds, small mammals, earthworms, fruits, and other types of plant material (Cordero and Nicolas 1987). Even gum exudates from several tree species have been reported as a feeding item of D. albiventris from the Atlantic forests of southeastem Brazil (Aléssio et al. 2005). Interestingly, Cáceres (2002) noted that D. albiventris from southem Brazil is an effective seed disperser. Analyzing scat contents produced by this species, he found fruit remains in 76% of the scat sample (compared to 100% occurrence of invertebrates and 58% occurrence of small vertebrates). Seeds of several plants remained viable (as later tested through germination experiments) after passing through the opossum's guts. The inflated p3 of D. solimoensis, as well as the weak development of crests in the lower molars, may be interpreted as an indication of more strictly frugivorous feeding habits in this extinct, Amazonian species. Such habits already have been inferred for several taxa of extinct (Paleocene) “opossum–like” marsupials bearing an enlarged p3 (Marshall 1987).
Because of their generalized postcranial skeleton, species of Didelphis, as well as many other didelphids, are capable of a wide range of locomotory pattems, including arboreal, terrestrial, and aquatic (McManus 1974). If more strictly frugivorous, it could be the case that D. solimoensis was adapted for a more arboreal locomotion, such as that characteristic of living caluromyids (Bucher and Hoffmann 1980). This hypothesis needs to be tested with (still unknown) postcranial materials of D. solimoensis. Judging from comparable parts, D. solimoensis probably resembled in size and body weight the living Caluromys and Philander, whose mean body masses have been estimated between 275 g (Caluromys derbianus) and 531 g (Philander opossum—Gordon, 2003).
Summing up, D. solimoensis, from the late Miocene of Amazonia, is the smallest known species of the genus. Its dental specializations suggest more frugivorous feeding habits than its living counterparts. It shares more features in common with D. marsupialis than with any other known species of Didelphis, living or extinct. D. solimoensis may represent an early offshoot of the Didelphis clade, with no ecological counterparts among the living species of the genus.
We thank the Laboratório de Pesquisas Paleontológicas of the Universidade Federal de Acre (UFAC) for allowing us to study the material and for providing other information related to the locality. The specimen here described was collected with funding from a grant from Conselho Nacional de Pesquisas Científicas e Tecnológicas, the Brazilian Science Council, to MAC. M. Tomeo prepared the photograph of the holotype.