Carnivoran Remains from the Malapa Hominin Site, South Africa

Recent discoveries at the new hominin-bearing deposits of Malapa, South Africa, have yielded a rich faunal assemblage associated with the newly described hominin taxon Australopithecus sediba. Dating of this deposit using U-Pb and palaeomagnetic methods has provided an age of 1.977 Ma, being one of the most accurately dated, time constrained deposits in the Plio-Pleistocene of southern Africa. To date, 81 carnivoran specimens have been identified at this site including members of the families Canidae, Viverridae, Herpestidae, Hyaenidae and Felidae. Of note is the presence of the extinct taxon Dinofelis cf. D. barlowi that may represent the last appearance date for this species. Extant large carnivores are represented by specimens of leopard (Panthera pardus) and brown hyaena (Parahyaena brunnea). Smaller carnivores are also represented, and include the genera Atilax and Genetta, as well as Vulpes cf. V. chama. Malapa may also represent the first appearance date for Felis nigripes (Black-footed cat). The geochronological age of Malapa and the associated hominin taxa and carnivoran remains provide a window of research into mammalian evolution during a relatively unknown period in South Africa and elsewhere. In particular, the fauna represented at Malapa has the potential to elucidate aspects of the evolution of Dinofelis and may help resolve competing hypotheses about faunal exchange between East and Southern Africa during the late Pliocene or early Pleistocene.


Introduction
The South African palaeontological record of the Plio-Pleistocene is rich and diverse. Most deposits from this time period are constrained within a relatively small geographical area in the Sterkfontein Valley (Cradle of Humankind). These cave sites range in age from ca. 2.5 Ma to the Holocene. The analyses of faunal remains from these sites have consistently revealed a change in ecology from the older Sterkfontein site (ca. 2.5 Ma), hosting the largest record of the hominin Australopithecus africanus and characterized by relatively heavy tree cover [1], to younger deposits (Kromdraai, Swartkrans, Drimolen, Coopers) which commonly include the hominin taxa Paranthropus and Homo and are characterized by more open habitats [1,2]. These latter sites provide evidence of faunal evolution in the region generally within the period between 1.8 Ma and 1.5 Ma. Historically, the period from approximately 2.0 Ma to 1.8 Ma has remained largely unsampled in southern Africa. This is a critically important time period as it encompasses the ecological shift from faunas typified by assemblages such as that of Sterkfontein, associated with Au. africanus, to the faunas and ecology of the younger Paranthropus/Homo-bearing deposits. The recent discovery of Australopithecus sediba, the youngest known Australopithecus species dating to 1.977 Ma [3], and associated faunal remains provide a unique window to interpreting faunal changes and ecology in southern Africa during this critical time period. This contribution focuses on the analysis of the carnivoran remains recovered from this site to date, paying particular attention to the Dinofelis specimens.
The site of Malapa was discovered in August of 2008 during a survey of cave sites in the region. The geological setting of Malapa is unique when compared to the majority of sites in the region; it is thought to have been a death trap based on the lack of carnivore-damaged bones and the high number of articulated specimens [4]. At present, Malapa appears as a deroofed cave deposit with an exposed area of approximately two metres by one and half metres by two metres deep. Cosmogenic analyses suggest that the site today is the bottom of what would have been a very deep cave system 1.9 million years ago [4]. Malapa has produced over 130 specimens attributed to the new hominin species Australopithecus sediba [5]. In addition to the hominin specimens, the site has produced a diversity of bovid and carnivore taxa that enable us to suggest a biochronological date of between 2.36 and 1.5 Ma [4]. Uranium-lead (U-Pb) dating and palaeomagnetic studies further constrain the likely deposition of the sediments to 1.977 Ma [3].

Results
To date 81 specimens have been attributed to the order Carnivora (Table 1). Of these, 69 have been identified to at least the family level and are described here. The remainder are listed as Carnivora indet.
Comments Specimen UW 88-812 is tentatively identified as Vulpes cf. V. chama based on M 1 morphology and the presence of M 2 and M 3 alveoli. The corpus is low and slender, with the coronoid process rising at about 45u posterior to the M 3 alveolus. The P 4 is tall and compressed. It has no mesial accessory cusp. The main cusp is sharply pointed as is normal in Vulpes. However, unlike all other specimens of Vulpes we have seen, it lacks a distal accessory cusp, there being only a very low crest in its place on the distal face of the main cusp. The M 1 is typically canid, with a short, low paraconid and taller protoconid. The metaconid is nearly the same height as the paraconid and set slightly posterior to the protoconid.
Compared to other small fossil and extant Canidae our material clearly differs from Otocyon in both size and morphology. It further differs from V. pulcher and V. pattisoni in being smaller. It is most similar to the extant V. chama, but features such as the lack of a distal accessory cusp on P 4 make us hesitant to ascribe it to this species.
The other two specimens listed were found in association with the mandible. UW 88-814 matches V. chama in morphology, whereas UW 88-813 is placed here because it represents a small canid and because of the association with the other two specimens. All are likely to belong to a single individual.
Canini indet. (Large species) Material UW 88-838, right distal femur (incomplete). Comments While this specimen is definitely from the family Canidae and judging solely by size belongs in the tribe Canini, it would appear not to be the extant wild dog (Lycaon pictus) although it is of similar size ( Figure 2). Interestingly, it appears similar in size to the Gladysvale specimen (GV 466) that has been attributed to Lycaon sekowei [6], but since there is no overlap in skeletal representation between the specimens, no specific attribution can be made. More material is required before we can make an assignation beyond the family level.
Family VIVERRIDAE, Gray 1821 Subfamily VIVERRINAE, Gray 1821 Genus Genetta Cuvier, 1817 cf. Genetta sp. Material UW 88-834, right mandible fragment with P 4 , M 1 , and tooth roots. Comments The position and morphology of the P 4 , M 1 , alveoli and roots enable us to tentatively place this specimen in the genus Genetta. Overall size limits the assignation to either Herpestidae or Viverridae and as Figure 3 illustrates, the fossil specimen shares overall gross morphologies to extant Genetta. In addition, the high middle cusp of the M 1 as well as the overall narrowness of the M 1 compared to the M 1 of herpestids of similar size justifies the assignation to cf. Genetta sp. The identification of which species of Genetta is represented at Malapa requires considerably more complete material.
Family  (Figure 4).  Comments This specimen is a morphological match to Ditsong Museum specimen KA-86, previously described as Herpestes mesotes [7]. At the time it was indicated that the specimen likely was on the lineage to Atilax and in agreement with other authors [8]  Comparisons of UW 88-694 to morphological characteristics of both fossil and modern material rule out referral to Viverridae and Atilax, but provide a close match in overall size and morphology to specimen TM 32670, an extant yellow mongoose (Cynictis penicillata). However, the gross morphology of upper canines show little variability among herpestids, and therefore we have decided to refer the specimen to cf. Herpestidae pending new discoveries.
Based on morphological comparisons with KA 86, specimen UW 88-822 does not belong to the genus Atilax. Based on the curvature and size of the cranial vault the specimen closely resembles modern Rhynchogale melleri, but does not preserve enough morphological characters to positively refer it to a specific taxon. Therefore we refer it to cf. Herpestidae.
Size comparisons suggest that the mentioned postcranial remains (UW 88-770 and UW 88-771) belong to small herpestids, but lack of enough diagnostic morphological elements forces us to refer them to cf. Herpestidae, though we note a general similarity in size and shape between the fossil sacrum (UW 88-770) and that of Herpestes ichneumon.
Family HYAENIDAE Gray, 1821 Subfamily HYAENINAE Gray, 1821 Genus Parahyaena Hendey, 1974 Parahyaena brunnea Thunberg, 1820 Material    Figure 5). Comments These specimens are all a good match for extant brown hyaena both in morphology ( Figure 5) and metrics ( Figure 6). Figure 6A shows that the Malapa P 2 lies within the confidence ellipses of P. brunnea and C. crocuta but well outside that of H. hyaena, while Fig. 6C shows the M 1 trigonid to be much shorter relative to tooth width than that of C. crocuta but similar in length to P. brunnea and H. hyaena. The Malapa specimens also markedly differ from extinct hyaenids such as Pachycrocuta and Chasmaporthetes. The majority of the specimens are likely to belong to a single individual, but the duplication of left P 3 shows that at least two individuals are involved.
cf Comments Parsimoniously, these are specimens that are all highly likely to be P. brunnea, as no identifiable craniodental material of any other hyaenid species has been recovered from Malapa. However, we do not have adequate comparative material of deciduous dentitions to make a positive determination of said teeth. In addition postcrania of brown hyaena are also much more difficult to distinguish from those of C. crocuta (except by size) than the teeth, especially in the absence of extensive comparative material. Therefore we do not definitively assign these specimens to P. brunnea. The presence of deciduous teeth in the sample alongside teeth with some wear suggests the presence of at least three hyaena individuals at Malapa.     Figure 7). Comments Given the age of Malapa at 1.977 Ma [3], any Dinofelis specimen could potentially belong to either D. barlowi or D. piveteaui. This is the only specimen of Dinofelis from Malapa with the potential to discriminate between these two species. The mesial accessory cusp of this specimen is very small or nearly-absent, in agreement with the specimens of D. barlowi from Bolt's Farm and Sterkfontein [9]. Specimens identified as D. piveteaui from both East and South Africa all have a well developed mesial accessory cusp [9]. The absence of this cusp in UW 88-627 is strongly suggestive of its affinity with D. barlowi. At 1.977 Ma, this would make the Malapa specimen the youngest certain record of D. barlowi.
cf  (Fig. 19B, 19G [10]) identified as D. piveteaui on the basis of association with craniodental remains, in both size and morphology. However, UW 88-527 has a flatter distal end, showing that this specimen does not belong to D. piveteaui, and suggesting more power but less mobility of the first digit in the Malapa individual.
The proximal MC II UW 88-597 is similar in size to the homologous element in both D. aronoki and D. piveteaui from East Africa (Figs. 13C, 19B, 19H [10]) but differs clearly from the latter Crocuta crocuta (spotted hyaena), together with two specimens from Malapa, as labeled. A: length and width of P 2 showing the Malapa specimen UW 88-520 to be similar to P. brunnea and C. crocuta and different from H. hyaena in its greater width. B: length and width of P 3 . One Malapa specimen lies within the ellipses of P. brunnea (the sample correlation for this species is not significant and no regression axis can be defined) and C. crocuta, while the other lies close to all three ellipses. C: trigonid length of M 1 versus tooth width. The Malapa specimens lies near the ellipses of P. brunnea and H. hyaena and far from that of C. crocuta. doi:10.1371/journal.pone.0026940.g006 in the much narrower proximal articular surface. In this UW 88-597 is more like the MC II of D. aronoki, again suggesting that the Malapa specimen cannot be attributed to D. piveteaui.
The remaining specimens are tentatively identified as Dinofelis because they are intermediate in size between Panthera leo and P. pardus, which is the size interval where the two South African Dinofelis species, D. barlowi and D. piveteaui, are found. In addition, the articulated right ankle UW 88-747 formed part of a pilot study using digital techniques to virtually isolate the individual bones of the specimen without having to prepare out the articulated fossil [11].
Subfamily Comments The mandible is broken anterior to the carnassial roughly in the coronal plane, and the remaining section includes part of the anterior margin of the ascending ramus and masseteric fossa, though the angle, coronoid and condylar processes are absent. The M 1 is missing the majority of the crown of the paraconid (from the mesial margin of the tooth to the carnassial notch), though the mesial-most point of the tooth is preserved, along with the majority of its cross-section, except for a small fraction (,0.5 mm) of the mesial outline. The preprotocristid is fairly straight, while the postprotocristid, though mainly vertical, does show a distinct concave curvature. There is a very small remnant of the talonid at the base of the mesial end of the protoconid.
Based on its size and the concave distal profile, we ascribe this specimen to Panthera pardus. Dinofelis spp. and P. leo can be excluded based on their large size, and the specimen is also significantly larger than any modern Felis, or Caracal. This leaves only three possible felid taxa in this size range: Acinonyx sp., Megantereon sp., and P. pardus. Acinonyx can be excluded because that taxon has a relatively straight distal margin and a small but highly distinct talonid that is not found in any pantherine. Also, the M 1 of extant Acinonyx is also on average smaller than this specimen, though this may not apply to fossil Acinonyx from South Africa. On the other hand, leopard and Megantereon spp. lower carnassials are metrically indistinguishable, and the Malapa specimen falls within the range of variation of both ( Figure 9). Instead, we consider it unlikely that the specimen belongs to Megantereon because of the concave distal profile. Although some older specimens of the genus (e.g., some European M. cultridens) show a concave distal margin, none of the African specimens of the genus display anything but a very straight distal border. This can be clearly seen in Fig. 8, which compares the mandibles of fossil P. pardus and M. whitei to the Malapa fragment. This figure clearly shows how the M 1 of the Malapa specimen (Fig. 8B) strongly resembles the M 1 of P. pardus (Fig. 8A) and differs from the more 'upright' M 1 of M. whitei, with its straighter posterior margin (Fig. 8C). Other early Panthera sp., such as the lion-sized Panthera from Laetoli show the concave distal M 1 margin even more clearly [12]. Therefore, we attribute the Malapa specimen UW 88-661 to P. pardus.  The MC IV (UW 88-613) is of the size of that of P. pardus. In addition, fourth metacarpal of leopards is more transversely slender than that of Dinofelis, matching the morphology of this specimen. It is also less oval in lateral view and the proximal articulation is more vertically aligned. Interestingly, it is more similar to the MC IV of Homotherium, but the latter is a much larger animal. Thus, in both size and morphology UW 88-613 is closely similar to leopards and we assign the specimen to P. pardus.

Comments
The two unciforms are pantherine in morphology, being shorter and less robust than those of the machairodonts Dinofelis, Megantereon, and Homotherium, as well as less square in lateral view ( Fig. 13D-F, 19 L-N [10]). The specimens are too small for P. leo but very large for P. pardus and we prefer to leave them in open nomenclature.
Genus Felis Linnaeus, 1758 Felis nigripes Burchell, 1824 Material UW 88-517, fragment of left maxilla with P 3 , L 5.1 W 2.4 ( Figure 10). This very small tooth has a relatively tall main cusp and reduced distal accessory cusp. The mesial accessory cusp and distal cingulum are also small. In metrics the tooth is within the range of variation of a small (N = 5) sample of Felis nigripes and outside the range of variation of similar samples of F. silvestris and F. margarita, the only other African Felidae in this approximate size range ( Figure 11). The reduced distal accessory cusp is a diagnostic autapomorphy of F. nigripes, as is the relatively tall main cusp [13].     Comments All of the above materials could not, either due to preservation or breakage, be identified beyond a classification as Carnivora. It can be noted that UW 88-697 and UW 88-696 were found in conjunction with one another.

Discussion
The following discussion will focus on some specimens of particular interest among the Malapa Carnivoran sample. The smaller taxa, such as Herpestidae and Viverridae, will be discussed in detail at a later date as the fossil record of these groups are much less well known than the larger carnivorans, and therefore require extensive primary research before definitive conclusions about taxonomy, systematics, and environmental adaptations can be drawn. Any conclusions drawn here based on these taxa are necessarily preliminary.

Canidae
Small Canidae are uncommon in the South African Plio-Pleistocene cave sites. The first such material to be described was Vulpes pulcher from Kromdraai [14]. Ewer later tentatively referred a specimen from Swartkrans to the same species [15], which is described as somewhat larger than the extant V. chama (which is confirmed by the published measurement data). In a later publication [16], Broom described a rather larger canid mandible from Taung as Vulpes pattisoni, which is distinguished from V. chama by several features in addition to size. Hendey described material of Otocyon recki (often referred to the genus Prototocyon) from Swartkrans [17], while Turner identified this species as well as V. chama from Member 3 of the same site [18]. Most recently, O'Regan and Menter assign some isolated teeth from Drimolen to V. chama or Canidae cf. V. chama [19]. The Malapa record is thus rare, but in line with the representation of Vulpes at other South African cave sites.
The other specimen identified to Canidae is the large distal femur UW 88-838. While the size of the specimen is on par with L. pictus the morphological characteristics rule it out as being that species. Further material is needed before we can determine if the specimen belongs to L. sekowei or to another, heretofore unknown, form of large Canidae. All of the material attributed to Canidae will be examined in greater detail in a later publication.

Hyaenidae
The majority of the South African cave sites include material of both Parahyaena brunnea and Crocuta sp., often in association with extinct forms such as Pachycrocuta brevirostris  [8]. Although Malapa is an exception at this point, as we can only unequivocally identify material from a single hyaenid species, P. brunnea, we believe that this is a sampling issue and predict that with increasing sampling at Malapa, material belonging to at least one more hyaenid species (likely Crocuta) will be found. However, it should be noted that hyaenids are commonly the most abundant fossils at these sites, both in terms of the number of specimens and the number of species. The latter is thus far not true at Malapa and it is possible that the difference we see is due to taphonomic processes rather than sample size.
Parahyaena brunnea is the only carnivore species identified from Malapa with an MNI greater than one. Since the sample includes both juvenile and adult worn teeth, which include two worn left P 3 , we are confident that we have at least three individuals present. Specimens UW 88-520, 521 and 522 were found in association, as were UW 88-623, 681 and 682. The only other category with an MNI greater than one is cf. Panthera sp., which may or may not represent the same species.

Felidae
The presence of Felis nigripes at Malapa represents the first fossil record of this very small (1.3-2.3 kg) cat. The black-footed cat is endemic to the grasslands of southern Africa and generally displays a relatively small home range [20]. The presence of F. nigripes thus provides an indication of the habitat around Malapa at the time of deposition.
In the Malapa sample, the only preserved Dinofelis tooth is a P 3 , a tooth that presents morphologically diagnostic features. All reported specimens of D. barlowi to date have in common the absence of a prominent anterior cusp on the P 3 , whereas all known specimens of D. piveteaui from East and South Africa show a strongly developed anterior cusp on this tooth [9,10] (Figure 7). The absence of an anterior cusp on the Dinofelis P 3 from Malapa strongly suggests its inclusion within the D. barlowi hypodigm although definite confirmation of this taxonomic attribution must naturally await the recovery of additional specimens of Dinofelis at Malapa. This provisional identification suggests a last appearance datum (LAD) for D. barlowi of possibly 1.977 Ma, younger than previously suggested. This younger age has implications for the evolution of the genus Dinofelis in Africa, more specifically bearing on the origin of D. piveteaui, the last known species of the genus. Dinofelis piveteaui is known from both East and South Africa [10,21,22] but the details of its evolution are contentious [9,10,19]. Two rival scenarios are available. Dinofelis piveteaui may have evolved in South Africa from the earlier South African species D. barlowi, and subsequently migrated to East Africa [19,22]. Alternatively, D. piveteaui evolved in East Africa from the older D. aronoki and subsequently migrated to South Africa [10]. Choosing between these hypotheses involves making progress along several lines of inquiry, including better dated localities, particularly in South Africa, and better and more complete character analyses of the existing material.
At present, in the African continent, at least six Dinofelis species may be recognized, including, from chronologically oldest to most recent: D. cf. D. diastemata, D. petteri, D. aronoki, D. darti, D. barlowi and D. piveteaui, although there is no general consensus in this classification [9,10,15] The genus Dinofelis has sometimes been placed in the subfamily Felinae along with modern cats and sometimes in the subfamily Machairodontinae, sabertooth cats. The main source of this confusion lies in its dental morphology. The upper canines are not highly mediolaterally compressed as they are in typical machairodonts such as Smilodon and Homotherium, yet they lack the longitudinal ('feline') grooves characteristic of most extant cat species, especially the larger members of the genus Panthera, with which Dinofelis is most often compared. A recent study resolves this question by showing that Dinofelis should be placed in Machairodontinae on the basis of derived characters such as the size differential between upper and lower canines, reduction of anterior cheek dentition, and a large pit superomedial to the trochlear notch of the ulna [10]. By this interpretation, for a large part of the genus' existence species Dinofelis played an ecological role similar to that played by species of the genus Panthera today.
Given this ecological context, the evolution of D. piveteaui is of particular interest, as it represents an evolution away from the ''Panthera-like'' paradigm followed by its ancestors, towards a distinctly more machairodont morphology, with more compressed upper canines, further reduced anterior cheek teeth, and a more machairodont skull and postcranium. This change in an evolutionary trend almost certainly represents a response to other changes in the carnivore guild, which makes understanding the evolution of D. piveteaui one of the keys to understanding the turnover in the carnivore guild in Africa between 2.0 and 1.5 million years [24].
Dinofelis barlowi is found at Sterkfontein, being restricted to Members (Mb) 2 (ca. 2.8-2.6 Ma) and Mb. 4, [25,26,27,28]. A previous report of D. barlowi at Sterkfontein Mb. 5 [25] is an annotation error (A. Turner, personal communication to RL, March 29, 2011). This taxon has also been reported from a series of pits at the site of Bolt's Farm [23,29], and although it is generally believed that parts of this site may exceed 2.0 Ma in age, dating remains problematic. Unconfirmed reports have indicated that D. barlowi was also recovered from Gladysvale [30], but this claim remains contentious and in any case, no stratigraphic information is available for this possible identification. D. barlowi remains as a South African endemic species. Its East African counterpart is D. aronoki, which is known from a number of sites in the Turkana Basin dating between ca. 3.0 Ma and ,1.6 Ma [10].
In East Africa, D. piveteaui has been reported from Koobi Fora, and Kanam East in Kenya [23,31], and Konso Gardula in Ethiopia [32]. These sites are temporally constrained to between 1.64 and 1.0 Ma, although the younger date assigned needs further confirmation. In South Africa, D. piveteaui is presently recognized from Kromdraai (which includes the type specimen, KA 61), Motsetse [9,33], Gladysvale [9], and more recently from the site of Drimolen [19]. Other Dinofelis specimens recovered from the caves of Swartkrans, Coopers, and the Kromdraai B deposits [9,10,18] have not been conclusively assigned to species. With the exception of Motsetse, whose age was largely inferred based on the D. piveteaui remains, the ascribed faunal-based ages of Kromdraai A and Drimolen are ca. 2.0-1.5 Ma [34,35,36] The Gladysvale D. piveteaui specimen derives from material without precise stratigraphic context [30].
The picture that emerges from these taxonomic identifications is that South African fossil localities older or around 2.0 Ma record the presence of D. barlowi, while the chronologically younger sites are associated with D. piveteaui remains. The earliest record of D. piveteaui in East Africa, on the other hand, dates to less than1.64 Ma, which is the lower boundary datum of the Okote Mb. of the Koobi Fora Fm. [37]. Thus, either D. piveteaui evolved earlier in East Africa than its present first appearance datum (FAD) implies (i.e., before 1.64 Ma), or the ages of Kromdraai and Drimolen may be best constrained at the younger end of their age range as D. piveteaui has been recovered from both sites. However, if D. piveteaui evolved in South Africa from its putative ancestor D. barlowi [23], then it may not be unreasonable to expect the recovery of intermediate forms between 2.0 and 1.5 Ma [19]. D. piveteaui specimens from Kromdraai include a well preserved skull with teeth (KA 61), a P 3 (KA 62) and an M 1 (KA 63), which together with lower dentition of the Motsetse specimens provide important information in assessing differences between D. barlowi and D. piveteaui [9,10]. Most notably, the expanded P 4 metastyle and increase in length of M 1 contrast with the morphology of D. barlowi characterized by relatively smaller P 4 metastyles and shorter M 1 [9].
The D. barlowi specimen from Malapa at 1.977 Ma therefore bears on the evolution of Dinofelis, increasing the plausibility of a South African origin by lessening constraints on the FAD of D. piveteaui. Thus, with a D. barlowi LAD of over 2 Ma, the FAD of D. piveteaui must be close to the lower age limit of the sites at which it has been found, while a younger LAD for D. barlowi such as that represented by Malapa allows for a D. piveteaui FAD well within the suggested age range for sites such as Drimolen. On the other hand, D. piveteaui shares with D. aronoki morphological traits that it does not share with D. barlowi. Thus, even though Malapa represents a step forward, only additional well-preserved fossils from well dated geological contexts, especially in South Africa, can resolve the issue.
The Carnivora identified from Malapa to date suggest the presence of two or more microhabitats in proximity to the cave. Today, F. nigripes is endemic to southern Africa and associated with the more arid regions that provide some cover in the form of grasses and scrub [38,39]. With home ranges recorded from 9.99 km 2 (females) to 20.69 km 2 (males) [40], the presence of F. nigripes at Malapa indicates that at least some of the region surrounding Malapa consisted of grasses, brush and scrub. The identification of Vulpes cf. V. chama supports the hypothesis of open grasslands as these, along with scrub and desert, are the preferred habitats of V. chama today [39]. We also note that, with few exceptions the family Canidae and the genus Felis today are both associated with open habitats [39]. Parahyaena brunnea historically has been associated with dry open habitats [39,41,42], and has not been documented to inhabit closed, forest-like environments. Panthera pardus today is exceptionally eurytopic and tolerant of a variety of habitats [39] and thus the presence of P. pardus is of little use with regard to interpreting the palaeoenvironment associated with Malapa. The presence of the genus Genetta is problematic as G. genetta today inhabits open regions whereas G. tigrina today is found in more closed environments [39]. The other two taxa, Atilax and Dinofelis, are clear indicators of more closed habitats. In the case of Atilax, the modern species has a strong affinity for wet to very wet habitats along streams or lakes, though we don't know if that applies to the fossil species as well. The brachial index of Dinofelis spp. indicates closed habitat adaptation [9] in line with paleoecological interpretations from other Dinofelis sites. In addition, yellowwood (Podocarpus sp.) along with other forest plant species have also been identified from pollen recovered from a coprolite at Malapa [43]. Thus, plant species indicate a moist, forest environment that corroborates the information from two of the identified carnivore species while open grasslands are indicated by two other identified carnivore species.
It is also worth noting that, if the geological interpretations are correct in that the Malapa assemblage is attributed to a death trap caught at a single moment in time, the social behaviours of the carnivores identified to date may be reflected in the assemblage. Even though the assemblage is small, and it is still early in excavations, the MNI's reflect the social behaviours in that P. pardus, F. nigripes, V. chama, Genetta and Atilax all have an MNI of one and today are all solitary in nature while Parahyaena brunnea (MNI of 3) is a clan oriented social species. The modern analogues of all of the carnivores identified to date are also territorial, thus unlike bovids which can roam far in search of browse or grazing, the carnivores would remain within the territories (unless the specimens we recovered happen to be younger animals looking to establish a territory) suggesting that both the open grasslands and the forested woodlands existed within ca. 20 km 2 , the smallest of the territories (F. nigripes) identified to date.
We have described the carnivore assemblage associated with the Au. sediba remains from Malapa, South Africa [3,4,5,44,45,46,47]. The analysis of this assemblage and its chronological age provide clues to the evolution of several mammalian taxa, with possible LAD for D. barlowi and the FAD for F. nigripes. The Malapa assemblage currently lacks extinct genera such as Megantereon (contrary to initial reports [4]), Chasmaporthetes or Pachycrocuta. Based on the carnivore remains described here, we suggest that the environments of Au. sediba are characterized by presence of relatively closed habitats with nearby open grasslands. This interpretation is in keeping with that described based on plant species [43] and bovid taxa [4] represented at the site.

Methods
All specimens were identified using reference collections housed at the Institute for Human Evolution (IHE) and Bernard Price Institute (BPI), University of the Witwatersrand as well as material housed at the Ditsong Museum of Natural History in Pretoria (formerly known as the Transvaal Museum), with other measurement and photographic data provided by the personal files of LW.
All measurements were taken with digital callipers to the nearest 1/10 millimetre (mm) and follow von den Driesch [48] unless otherwise noted. Tooth lengths (L) are mesial-distal and widths (W) are buccal-lingual. Both BK and LW took measurements. Other abbreviations: MNI: minimum number of individuals; NISP: number of individual specimens.