INTRODUCTION

The history of the dog family Canidae has been greatly clarified and illuminated by the work of Richard Tedford and his colleagues, Beryl Taylor and Xiaoming Wang (e.g., Tedford, 1978; Tedford et al., 1997; Wang et al., 1999). We now recognize three successive (but overlapping) radiations over the family's approximately 40 m.y. history, each corresponding to a subfamily (Wang et al., 1999: fig. 143). The initial radiation, the Hesperocyoninae, consisted of small to medium-sized species with diets that varied from mesocarnivory to hypercarnivory. It was succeeded by the Borophaginae, an extensive array of North American species that included hypocarnivores, mesocarnivores, and hypercarnivores, several of which appear to have had hyena-like habits. The borophagines disappeared about 2 Ma, at a time when the current subfamily, the Caninae, was diversifying, primarily in the Old World (Hunt, 1988). Like the Borophaginae, the Caninae include hypo-, meso-, and hypercarnivores, but unlike them, canines have not shown a tendency towards bone cracking.

The Borophaginae was the largest of the three subfamilies of canids, including some 66 species and spanning approximately 34 m.y., all of which were confined to North America. They reached their peak diversity of 16 species in the early Barstovian (16–15 Ma). Relative to extant canids, this is a remarkable level of continental diversity; at present, the maximum number of canid species on any continent is found in South America, where there are 11 (Nowak, 1991). Over the course of their evolutionary history, borophagines tended to become larger and more hypercarnivorous, with the most derived species exhibiting extreme craniodental adaptations for durophagy (Wang et al., 1999). It seems likely that some of the larger, highly carnivorous species hunted in packs as big canids do today, but until recently, there was no way to predict which borophagines were the likely pack hunters. The identification of pack hunting canids in the fossil record could lead to further interesting explorations of predator–prey coevolution.

Van Valkenburgh and Koepfli (1993) argued that canids that regularly hunt prey larger than themselves are very likely to be social. Moreover, among living species, the hunters of large prey can be recognized by a suite of adaptations including deep jaws, broad muzzles, and enlarged incisor and canine teeth, all of which reflect the heavy loads placed on their skulls and teeth by large prey (Van Valkenburgh and Koepfli, 1993). Unlike felids, which retain strong, flexible forelimbs capable of grasping prey, extant canids must capture, hold, and kill with jaws only. Pack hunting canids can work together to down large ungulates, with some individuals holding prey by the throat or muzzle while others wound the prey. Thus, if it can be determined that an extinct canid or canidlike species did not have flexible forelimbs and took prey larger than itself, then it can be assumed that it probably did so in groups.

In this paper, we first make overall comparisons of the craniodental morphology of borophagines, canines, and hyaenids using principal component analysis. We then use two approaches to predicting the probable prey size of extinct borophagines. In the first, we estimate prey size based on the correlation between maximum jaw depth and typical prey size in living canids. In the second, we use a recently established relationship between body mass and diet in carnivorans (Carbone et al., 1999) to predict which borophagines regularly would have taken prey larger than themselves. Carbone et al. (1999) showed that among living carnivores, almost all species larger than about 21 kg take prey as large or larger than themselves, whereas smaller species feed mostly on prey that is 45% or less of their body weight. The exceptions to this pattern were mostly bears, which are mainly omnivorous. Using an energetic model, they argued that it becomes increasingly difficult to subsist on small prey items as predator body mass increases due to limitations on intake rate and foraging time. If their model is correct, then it should be possible to identify hunters of large prey in the fossil record by estimating predator body mass. For the borophagines, we estimated body mass based on previously published regressions of body mass against skull or tooth length in living canids (Van Valkenburgh, 1990).

One complication that was not dealt with in either Van Valkenburgh and Koepfli (1993) or Carbone et al. (1999) is the problem of distinguishing hunters from scavengers. Strong jaws and teeth might be required for both activities, and the foraging energetics might be similar as well. This is especially relevant to the study of borophagines, as they are well known for their similarity to hyenas in craniodental morphology (Werdelin, 1989; Munthe, 1998). Scavenging behavior is not associated with group foraging in living carnivorans. There are no canid species that scavenge frequently at present, but there are two hyena species that do so, the brown hyena (Parahyaena brunnea) and the striped hyena (Hyaena hyaena) and they usually forage alone. Brown hyenas have diets that include fruits as well as small vertebrates that they kill, but most of their prey is scavenged (90–95%; Mills, 1990). The striped hyena is also omnivorous, scavenging most of its food, but catching insects and small vertebrates (Kruuk, 1976; Macdonald, 1988). Spotted hyenas (Crocuta crocuta) are the largest and most carnivorous species of the family, often hunting in groups for prey much larger than themselves (Kruuk, 1972). In an attempt to more confidently infer pack hunting in borophagines, we examine craniodental features associated with hunting as opposed to scavenging within extant Hyaenidae (exclusive of Proteles cristata, the aardwolf, a termite specialist). The observed differences are used to interpret foraging behavior (scavenging vs. hunting) within the group of borophagines predicted to have subsisted on large prey.

MATERIALS AND METHODS

RESULTS

DISCUSSION

The Borophaginae were intermediate in morphology between extant canids and hyaenids. Like the extant subfamily Caninae, the borophagines included hypo-, meso-, and hypercarnivorous species (Wang et al., 1999), but borophagines tend to exhibit deeper jaws, larger teeth, and smaller dental grinding areas than their putative ecological equivalents among canids. In these features, they resemble hyenas. However, although a few species equal extant bone-cracking hyenas in features such as jaw muscle leverage and relative carnassial blade size, they differ markedly from hyenas in retaining postcarnassial molars and not enlarging their anterior premolars (Werdelin, 1989). Unlike hyenas, they must have used their rearmost lower premolar (p4) and upper carnassial (P4) to crack large bones (ibid.). Smaller bones and bone fragments were likely crushed with the carnassials and postcarnassial molars just as in extant canids (Van Valkenburgh, 1996). Because gape diminishes as the jaw joint is approached, the more rearward position of the bone cracking teeth results in limitations on the size of bones that can be broken easily and consumed. Moreover, as Werdelin (1989) pointed out, the proximity of the bone-cracking teeth to the cutting blades of the carnassials results in heavy apical wear on the cutting teeth, compromising their slicing function.

We identified 9 of the 26 borophagine species in our sample as probable consumers of prey larger than themselves based on estimated body mass and the energetic model of Carbone et al. (1999). According to this model, all carnivorous species estimated to have a body mass of at least 21.5 kg are assumed to take large prey because of energetic constraints on foraging time and efficiency. The jaw depth approach identified two additional species, Aelurodon stirtoni (body mass = 20 kg, prey size = 40 kg), and Paratomarctus euthos (body mass = 12 kg, prey size = 22 kg) as probable consumers of large prey. Given that our body mass figures are estimates, it would not be surprising to find that A. stirtoni was somewhat larger, above the 21.5 kg boundary. However, the assignment of P. euthos as a hunter of large prey is notable. It reflects the fact that P. euthos has both a deeper jaw and smaller m1 relative to other borophagines of similar skull length. The deep jaw suggests large prey and the small m1 results in a lowered body mass estimate.

The notion that a 12-kg predator would have regularly hunted prey larger than itself is not inconsistent with the Carbone et al. model. The authors did not argue that species smaller than 21.5 kg cannot take large prey; rather they made the point that species above this weight are “forced” to take large prey to meet their nutritional needs. Because jaw depth is expected to reflect typical loads placed on jaws during killing and feeding, we retain the full list of 11 species as likely to have relied on prey as large or larger than themselves. This should be considered a minimum number for the subfamily because, due to incomplete material, our sample excluded a number of large borophagines that were close relatives of several of the 11 species.

Of the 11 species, which were primarily hunters and which were primarily scavengers? This is a difficult question to answer. Based on the comparison of the three extant hyena species, there is no single morphological feature that clearly distinguishes hunters from scavengers. They do not differ in jaw strength despite the fact that one struggles with large, live prey while the others simply consume already dead prey. Dentally, spotted hyenas have less grinding area and longer cutting blades in their dentition than the other two, reflecting their commitment to a purely carnivorous diet. Notably, the spotted hyena is much heavier than either of the other two species, averaging 50–70 kg as opposed to 35–50 kg for the brown hyena and 25–40 kg for the striped hyena (Richardson and Bearder, 1984). Moreover, it is much more abundant and has a greater geographic range than any other living hyaenid. Undoubtedly, this reflects their ability to acquire food by hunting as well as scavenging. Given this, it is expected that the largest, most common borophagines with the most reduced dental grinding areas should be assumed to have hunted most of their food. Based on their craniodental morphology and abundance in the record, this would certainly include at least 5 of the 11 species predicted to have taken large prey: Epicyon saevus, E. haydeni, Borophagus secundus, Aelurodon ferox, and A. taxoides. It should not be concluded that these species never scavenged; they almost certainly did so opportunistically, just as spotted hyenas and lions do today. However, their size and abundance are more consistent with group hunters than solitary scavengers. Searching for food to scavenge is best done alone, as the returns are not usually large enough to share. In support of this, both brown and striped hyenas forage alone although they may aggregate at clumped resources (e.g., striped hyenas at dumps; Macdonald, 1988) and for raising young (brown hyena dens; Mills, 1990).

The remaining six species should not be assumed to have scavenged most of their food. They range in estimated size from 12 kg (Paratomarctus euthos) to 29 kg (A. mcgrewi), and all have relatively large grinding areas and blade sizes that are comparable to those of extant Caninae that take large prey (e.g., Cuon alpinus, Lycaon pictus, Canis lupus). The fact that they do not appear to have been as abundant as the five borophagine species listed above does not preclude their being hunters. After all, the African wild dog, L. pictus, is a pack hunter that exists today at low densities relative to the two other social hunters, lion and spotted hyena (Creel and Creel, 1998). The wild dog would be expected to have (and does have) a much more limited fossil record than the two more abundant species (Turner, 1990). Thus rarity in the record should not be assumed to indicate habitual scavenging in hypercarnivorous borophagines. On the other hand, a large, hypercarnivorous borophagine that is common in the record probably existed at fairly high densities and therefore is more likely to have sustained itself by hunting than scavenging.

Van Valkenburgh and Koepfli (1993) argued that unlike felids, canids that kill prey much larger than themselves must do so in groups because they do not have the forelimb flexibility to grapple with and hold prey. Instead canids work together, with some individuals holding the prey while others wound the prey with repeated bites. This association between group hunting and relative prey size holds for living canids because their ability to rotate their paws inwards (supinate) is constrained by the morphology of the radioulnar joint. However, many of the Borophaginae displayed more mobility in this joint and were capable of moderate supination of the forepaws (Munthe, 1989). This suggests that they may have had some ability to grapple with their prey. Nevertheless, they did not have sharp, curved, retractile claws as do felids; instead the nails are blunt and relatively straight as in living Caninae (Munthe, 1989). Thus, it may have been difficult for borophagines to hold large prey, but their moderate supinating abilities could have been used to hit prey and knock them down during a chase, as brown bears have been observed to do when hunting ungulates (Cole, 1972; Boertje et al., 1988).

In her study of borophagine postcranial morphology, Munthe (1989) identified Epicyon saevus and E. haydeni as having limbs that come closest in morphology to those of extant large Caninae. They are relatively long, slender, and restricted in flexibility. On that basis she identified these species as probable pursuit hunters, and we would agree, adding that they probably usually hunted in groups. Because of their somewhat more flexible and robust forelimbs, Munthe (1989) felt that most or all Aelurodon species were solitary, ambush hunters of large prey. Although we agree that aelurodonts probably were not the extreme sort of endurance runner that we see today among canids, their running abilities are difficult to determine. After all, spotted hyenas are very capable pursuit hunters and yet their limb proportions and ankle morphology would suggest something much slower (Van Valkenburgh, 1987; Janis and Wilhelm, 1993). It is possible that groups of Aelurodon pursued large prey over limited distances, perhaps swatting them down with their forepaws and then killing them with multiple bites.

The youngest species in our sample, Borophagus secundus, is also the most robust in craniodental morphology and postcranial skeleton. According to Munthe (1989), it had the most flexible forelimbs of the large Borophaginae, as well as heavily blunted teeth, suggesting to her that it was much more of a scavenger than a hunter. Werdelin (1989) also classified Borophagus as a scavenger rather than a hunter, based on tooth wear. Arguing against a scavenger hypothesis is the abundance of this species in the record, and its tendency to occur in sizable numbers at single localities (e.g., Coffee Ranch, Optima; see Wang et al., 1999). Were this species similar to the brown hyena in its habits, it would be expected to have existed at low densities over a fairly limited range. Instead, it ranged from Central America to California and Nebraska and dominates the preserved carnivore fauna at several sites. The heavy apical tooth wear certainly suggests it consumed bones regularly, as do wolves today, but its abundance implies it hunted much of the time. Sharp teeth are probably not essential for killing prey; museum collections include numerous lions, leopards, hyenas, and other carnivores that survived for considerable lengths of time with broken and blunted canine and cheek teeth (Van Valkenburgh, 1988).

In conclusion, we believe that at a minimum, 11 species of Borophaginae regularly ate prey larger than themselves. Moreover, many or most of these probably hunted much of their prey, and did so in groups. In particular, both species of Epicyon, Aelurodon ferox, A. taxoides, and Borophagus secundus are predicted to have been social hunters based on their hypercarnivory and relative abundance in the fossil record. As noted above, previous workers have favored a predominantly scavenging lifestyle for most large borophagines, with the possible exception of Epicyon (Werdelin, 1989; Munthe, 1998). Their conclusions are based largely on the heavily blunted teeth and lack of gracile limb proportions that characterize pursuit-hunting canids today. However, as expressed above, we do not find these reasons sufficient to conclude that most hypercarnivorous borophagines rarely hunted. The only living carnivoran that can be described as a hypercarnivorous scavenger is the brown hyena. This species has a geographic range limited to southern Africa, and is only abundant where spotted hyenas are relatively rare (Mills, 1990). The low diversity and geographic restriction of this ecomorph undoubtedly reflects the difficulty of its lifestyle, and argues against such ecomorphs ever having been common within communities.

One further reason for believing that borophagines could not have been pack hunters was proposed by Radinsky (1969, 1973). According to Radinsky (1969), canid brains have a larger prorean gyrus than those of felids, and pack hunting canids exhibit still further expansion of this region. Because borophagines do not have an expanded prorean gyrus, Radinsky (1973) inferred that they could not have displayed packlike social organization. Although the difference in brain morphology is intriguing, it does not seem to be sufficient to exclude social behavior in the borophagines. The functional relationship between external brain morphology and complex behavior such as sociality is not well established. On the other hand, the association of group hunting with hypercarnivorous dentition, strong jaws, reduced forelimb flexibity, and body mass in excess of 21 kg follows from both energetic and functional considerations.

All 11 species predicted to consume large prey appear in the Barstovian (16 Ma) or later. Although we were unable to include all species within the subfamily in our study, it seems clear that the subfamily shifted in its ecological emphasis around 16 Ma. Prior to that time, the subfamily was dominated by consumers of smaller prey and more omnivorous species (Wang et al., 1999). In the future, it would be useful to examine the 11 borophagine species proposed here to have consumed large prey in an ecological context. Notably, the initial diversification of the highly carnivorous borophagines occurs at a time when the species richness of other North American hypercarnivores is either low (felids, nimravids) or declining (amphicyonids and hemicyonine bears; Van Valkenburgh, 1999). The subsequent decline to extinction of the hypercarnivorous borophagines occurs steadily from about 14 Ma to 2 Ma, alongside an increase in large felid diversity. By examining suites of sympatric borophagines and other carnivorans (i.e., paleoguilds), it might be possible to more confidently assign various taxa to scavenging and hunting roles given that purely scavenging species are likely to have existed at low diversity as they do today. Typical prey sizes for all extinct carnivores can now be more confidently estimated using the jaw depth and body size models, allowing us to better track the evolution of predator–prey relationships as well as ecological separation among predators. Of course, such paleoecological analyses are greatly enhanced by having a solid and well-delineated phylogeny in hand, and few groups have received as thorough a treatment as the Borophaginae. It is to be hoped that more such monographs appear on extinct North American felids, amphicyonids, and ursids in the not-too-distant future.