Allosaurus Was Predominantly a Scavenger

Published In

Ecological Modelling

Document Type


Publication Date



Recently, we proposed the hypothesis that Allosaurus were not apex predators, but rather filled the ecological niche of scavengers, because they were able to survive on sauropod carrion generated by natural attrition (Pahl and Ruedas 2021). While we respect the thoughtful rebuttal comments provided by Kane, Healy and Ruxton (2023), and appreciate their consideration of our research, we disagree with their arguments against the paper for a number of reasons. In particular, Kane et al. (2023) begin their commentary by stating that we made an “iconoclastic claim”. We did no such thing. We presented evidence that strongly supported an alternative hypothesis of the foraging ecology in Allosaurus, specifically that it was primarily a scavenger rather than a predator. We presented testable evidence from a variety of data streams, including our modelling efforts, the source code for which was openly available. Notwithstanding our evidence, Kane et al. (2023) disagreed with our hypothesis based on four points: 1. We omitted consideration of empirical evidence from the fossil record. 2. Our model assumes unrealistic disadvantages to predation. 3. Our model assumes some unrealistic aspects of the availability of carrion that favour scavenging. 4. Our discussion mis-represents previous work. Below, we address these points seriatim. It is true that we omitted some reports of injuries in fossil material from our research; in particular Carpenter et al. (2005) and Bakker et al. (2014). These either were not relevant to our research, or in our view, did not constitute evidence of predation. For example, the bite mark on a Stegosaurus plate reported by Carpenter (2005) was likely not caused by predatory behavior because it was too shallow and did not heal. Stegosaurus plates themselves most likely did not constitute high value predation targets; although they were vascularized in life, plate-inflicted wounds such as that reported by Carpenter et al. (2005) would not have caused life threatening damage. The shallow depth of the bite (fewer than four anterior teeth; Figure 17.4 of Carpenter et al., 2005) more likely implicates attempted removal of skin tissue rather than a wound-inflicting bite on the part of an Allosaurus engaged in predation, which to be effective as such would have had to engage a greater proportion of the maxilla and jaw. Furthermore, stegosaur plates were probably meatless. It has been demonstrated that modern carnivores target more profitable parts of a carcass first, whether or not they are scavengers (Selva et al. 2019). As Carpenter et al. (2005) noted, the plate would have been covered in skin over bone, without intervening muscle or other soft tissues, and it may have included a keratinous sheath. Because the plate did not heal and had little caloric value, it is more reasonable that the bite mark, as preserved, was inflicted on the stegosaur postmortem, and after higher value elements of its carcass had been consumed. However, no concrete evidence exists to indicate whether an allosaur killed the Stegosaurus or if it died by other means. Regarding the thagomizer injury to an Allosaurus pubis reported by Bakker et al. (2014), this injury was regarded as fatal to the Allosaurus. While it is reasonable to conclude that this encounter was initiated by a failed predatory attempt on the part of the Allosaurus, this fact strengthens an important suggestion we made in our paper: that although large carnosaurs may certainly have been able to hunt, as even modern avian scavengers do, armored dinosaurs such as Stegosaurus were very dangerous and probably not targeted by theropods except in seasons of desperation. As a modern analogue, predation by coyotes on humans, an extremely rare event, was suggested by Gehrt et al., 2023 to have occurred only after coyotes transitioned to routine predation upon moose, a somewhat less rare occurrence, because climate change is causing small prey to become too rare to support them. Gehrt et al., 2023 suggested that the lack of natural prey would have facilitated the transition to larger and potentially more dangerous prey, a phenomenon they denominated the Limited Resource Hypothesis. We hypothesize that the Limited Resource Hypothesis is a plausible explanation for the rare occurrence of Allosaurus wounds from Stegosaurus, in that these interactions probably only took place when their natural resource pool, sauropod carrion, was unusually low. In any case, we should note that Bakker et al. (2014) was a conference abstract for a paper that was never published. Conference abstracts do not typically constitute defensible publications, because their data cannot be evaluated. Indeed, the specimen number and collection details of the individual are not even mentioned, so it is impossible to replicate the study, which by definition makes Bakker et al., 2014 unsubstantiable. Finally, our model was written specifically to test whether or not sauropod carrion itself, as a primary resource, could have been calorically profitable for Allosaurus, and whether it could have supported long term populations of these dinosaurs. We therefore did not consider it relevant to include ornithischians in the model, which of course could have been prey, but were not important to the main question of our project. However, it should be noted that ornithischians were less common and less diverse than sauropods (Leach et al., 2021). If we had estimated ornithischian population densities in a different model of the ecosystem, scavenger agents would have been able to survive on sauropod carrion just the same, even if survival odds of predator agents were to increase nominally. The argument here is not that Allosaurus was unable to hunt, but that it evolved as a species to obtain most of its calories from scavenged sauropod carcasses, which made it an obligate scavenger both functionally and ecologically. In life, this was probably true if they occasionally hunted small and/or sick herbivores as well. Kane et al. (2023) suggested that our model assumed unrealistic disadvantages of predation. Indeed, while we based predator success odds on those of modern predators such as lions and hyenas, predator agents had a 10% chance of death during predatory encounters in some iterations of our model. While the report in Bakker et al. (2014) may imply that Allosaurus in particular was especially vulnerable to the weapons of large adversaries, a future and similar research project we have just completed may satisfy this criticism, as it did not incorporate such a survival cost into predatory collisions. In any case, if predatory agents had not been included in our model at all, it would not change the fact that sauropod carrion abundance was sufficient to sustain mature populations of Allosaurus, so predation was not necessary for them to survive. We definitely agree with Kane et al. (2023) that selection for predator adaptations usually favors much more robust characteristics, along with binocular vision, that are not observed in Allosaurus. This almost certainly made it a poor predator of larger, powerful dinosaurs such as diplodocid species, to say nothing of the brachiosaurids and camarasaurs that also were present alongside them. To respond to a specific rebuttal point: “...the model assumes a higher metabolic maintenance cost to being a predator than an obligate scavenger.” This is not accurate. Predator and scavenger agents were assigned the same field metabolic rates, based purely on their mass specific metabolic engines. The higher metabolic costs we described for predatory agents were not because predators required more food to survive than obligate scavenging agents, and we think it is possible that the authors are referencing a scripting error or another issue with the project code version as hosted on GitHub. The metabolic costs of predation we described in our paper existed because predatory agents had a 55% chance of failure in predation attempts, which itself amounts to a metabolic and time opportunity cost that obligate scavengers typically do not experience. This fact has been known for decades, and was articulated directly in the work of Houston (1979) on scavenging dynamics, “...there is no advantage in killing prey if the animal could easily obtain good-quality meat by scavenging.” This has been substantiated by recent authors and empirical observations in diverse taxa such as wolves, Polar Bears, and Great White Sharks (Fallows et al 2013; Laidre et al., 2018; Klauder et al., 2021). In all of these examples, carnivores prefer scavenging opportunities even when prey is abundant for them. We hypothesize that this was exemplified in dinosaur ecosystems where gigantic, whale-sized herbivores were common and their carrion from natural attrition would have been present at high levels. Our research supports this principle because there was no advantage for our agents to kill prey when the opportunity to consume large carcasses was naturally high and convenient. To put it another way, if an endothermic brachiosaur died of exhaustion during the dry season, its liver alone may have weighed up to 1.2 tonnes (Delgado-Coello, 2021), with nearly the same caloric value as an entire subadult Stegosaurus (Brassey et al., 2015), but without the risk of death by thagomizer. It is not difficult to recognize that annual sauropod-falls created a low-risk, high-reward resource pool that made predation much less favorable. There also of course is no reason to assume that Allosaurus, or any other large vertebrate, would need to rely on aerial scavengers to locate carcasses, as Kane et al. (2023) suggested. Modern Alaskan wolves do not coexist with vultures and do not need to follow birds to carcasses, but instead become apex scavengers themselves during the winter months, and outcompete smaller, more plentiful carnivores for carrion resources (Klauder et al., 2021) just as we hypothesized would have been true for large theropods. The same is true in Polar Bears, which often rely on beached whale carcasses to sustain them for weeks or months, using olfaction and their characteristically nomadic lifestyle to access these opportunities (Laidre et al., 2018). In fact, the smell of a sauropod-sized dead whale is strong enough to be perceived even by humans from at least 6 kilometers away (Landis-Hanley, 2021). Modern terrestrial animals do this already without help from aerial species, so there is no reason to add unnecessary assumptions to explain theropod foraging behavior. Large theropods probably did not need to follow flocks of birds or pterosaurs, and could smell rotting flesh from multi-tonne dead sauropods even from many kilometers away, and also likely took social cues from one another (Selva et al., 2019). We are confident that these facts are sufficient, both logically and empirically, to overturn conclusions in other work, namely, “...arguments of Ruxton & Houston (2004) still stand: there is no obvious trade-off for non-flying vertebrates that would select for a carnivore to give up the ability to predate entirely and become an obligate scavenger” (Kane et al., 2023). Carnivores already give up predation in favor of scavenging opportunities; there is no reason to think theropods, with consistent access to orders of magnitude more carrion, would have done differently. Kane et al. (2023) did not challenge the sauropod carcass production rate we used, nor did they take issue with carcass distribution or temporal persistence on the landscape; we appreciate this acknowledgement. We also fully recognize that many factors influence consumptive value of carcasses in modern biomes, including ambient temperatures, arthropod abundance, decay state, even stochastic weather events. We attempted to make our carcass decay rate match empirical observations by deriving a calculation that could be applied to carcass objects in a programmatic sense, but models are meant to be simplified abstractions, and loss of realism unavoidable in almost all ecological modelling. Our research was not meant to account for all potential details, and it was incomplete with respect to many other variables as well. It had no ornithischians, no geographic features, and no weather. Its sauropod population density was probably too low as well, and could have been 15 – 33 times higher in life than we estimated (Farlow et al., 2010; Wilkinson et al 2012). If we had modeled a system with 10 coexisting species of Morrison Formation sauropods, to perhaps approach the densities described in Wilkinson et al. (2012), our model would have been more realistic, and certainly more favorable to our hypothesis. But we wanted to remain as scientifically cautious as possible, so we modeled only 1 species at the lowest population density estimated by Farlow et al. (2010). We do not believe changes to these or other details would have significantly altered the results of our research. Having said that, it is important to consider two points with respect to rebuttal arguments about microbial decay agents and the palatability of sauropod carcasses after long periods of time. First, modern obligate scavengers typically are not threatened by microbial infections because vulture stomachs and their immune systems are able to overcome anthrax and other deadly pathogens. It is reasonable to assume that theropod dinosaurs had similar immunological adaptations, because these features evolved more than once in unrelated vultures independently (Zou et al 2021). Direct evidence of similarly advanced, localized immunoresponse to infections for Allosaurus exists in the literature as well: at least one paleopathological infection, on an individual's foot, was surmised to have been caused or exacerbated from contact with a decaying carcass (Laws 1996). Authors have remarked that although many of its injuries healed, the individual described in that paper was too crippled to hunt (Laws 1996; Breithaupt 2001). Another specimen survived long after a traumatic injury deformed the anterior portion of its left dentary, which would have severely compromised its ability to use its mouth in combat (Foth et al., 2015), especially since Allosaurus had such weak jaws to begin with. The same individual also had many other healed pathologies, including multiple infections and tumors, that would have made it difficult or impossible for it to hunt. It is possible that large carnosaurs lived in socially advantageous family groups that supported disabled individuals, thereby allowing these and other injured allosaurs heal, but there exists scant evidence for behavior of this nature. It is more likely that they were able to feed themselves despite being physically unable to hunt, because sauropod carrion was plentiful, and these animals likely evolved to subsist on it in the first place. Second, Allosaurus were very common, and we think it is unlikely that exposed sauropod carcasses in their environments remained undetected for more than a few days at most. As much as 90% of all vertebrate carrion is consumed by vertebrate scavengers rather than by arthropods or microbial decomposers (Houston 1986; Turner et al., 2017). It therefore seems unlikely that sauropod carcasses, particularly those exposed on the landscape, would have remained undiscovered for long periods of time. Large vertebrates typically die in predictable locations, and at predictable times of year (Moleon et al 2019). Modern scavengers capitalize on this fact by learning where and when animals are most likely to die of complications related to environmental conditions, starvation, or other common causes of mortality. There are no logical reasons it would have been impossible for Allosaurus to do this. Carcasses of large, conspicuous animals such as sauropods probably were not difficult to locate for hungry theropods of any kind. As such, if straggler Allosaurus individuals approached an undisturbed 20-day-old sauropod carcass, even if putrefaction had begun, we doubt they would have ignored the food opportunity, because the meat would have sustained them for weeks, and they probably had adapted to resist pathogens just as modern vultures have (Laws 1996; Zou et al. 2021). We think they certainly would have been able to locate fresher carcasses using olfaction, if it were the case that a decay-induced carcass with perhaps 25 tonnes of mass in meat, were not edible. For these reasons, arguments about the palatability of late-stage carcass objects in our model are likely not strong enough to significantly limit the amount of sauropod carrion that was available to these dinosaurs either in our model or in life. While we appreciate this criticism, we also should point out that the title of Ruxton and Houston's 2004 paper is “Obligate vertebrate scavengers must be large soaring fliers”, which unambiguously contradicts portions of its own Discussion section. To quote the paper directly: “Although obligate scavenging is predicted to be less attractive to the terrestrial body-forms than the avian ones, it still seems feasible. Fig. 2b indicates that a 1 tonne mammal or reptile, in an ecosystem yielding carrion at densities similar to the current Serengeti, could have met its energy requirements if it could detect carrion over a distance of the order of 400–500 m.” Carrion detection distance in today's mammals often is much greater than 500 m (Lai et al 2015) and the brain of Allosaurus was largely devoted to odor detection (Rogers, 1998, 2005). It therefore is eminently reasonable to assume that they met this criterion easily, and our model reflected this as well. Fossil evidence strongly indicates that herbivore biomass in the Morrison Formation was orders of magnitude higher than it is in the modern Serengeti, so it should be expected that large theropods such as Allosaurus were able to meet their energy requirements from carrion produced therein, particularly so by means of Ruxton and Houston's own calculations. More importantly, if it were true that only flying vertebrates were efficient enough to survive on carrion alone, even in environments dominated by whale-sized sauropods beyond 20 tonnes in mass, we would expect three things: First, agents in our model would not have been able to survive sustainably on sauropod carrion at the level of a population. Second, empirically, we would expect to find aerial soaring specialists or avian taxa abundantly in the Morrison Formation, since the carrion resource pool created by the natural mortality of giant dinosaurs was so large. Third, we also should expect aerial scavenging species to be common in other places where sauropod carrion could be expected, because sauropods often were the most diverse and common animals of their environments, and the primary causes of mortality for them almost certainly were unrelated to predation. In fact, none of these criteria have been met. First, our model's carnosaur agents were able to survive profitably with no predation at all purely on account of the existence of sauropod carrion, in what we consider a carrion abundance level much lower than existed in natural conditions. Regardless of whether predator agents were modeled according to other authors’ preferences, this fact alone would be enough to cast serious doubt on the hypothesis that only aerial foragers can evolve as obligate scavengers. To argue otherwise is perhaps even irrational, given that our agents also were shackled by a significant locomotor handicap. They only were able to travel 1 km per day, which is so restrictive that we expected that factor to invite scrutiny, rather than anything else. Most large vertebrates travel much further than this on a daily basis, whether they are predators or not, at the elevated functional metabolic rates we used no less. Even animals as small as deer mice, Peromyscus maniculatus (ca. 15–20 g) have been documented moving distances of at least 1320 m in a single night (Rehmeier et al. 2004). Chappell et al. (2004) reported that captive deer mice (average mass, 22.2 g) using a treadmill wheel recorded mean daily distances travelled of 3.005 km and a maximum distance of 16.5 km. Bank voles, Myodes glareolus (ca. 25 g) have been documented traveling almost 1 km per night (Kozakiewicz et al. 2007). The proposition that theropods orders of magnitude larger in mass would have been unable to efficiently search for and locate enough sauropod carrion to survive, simply because they were not aerial, is difficult to justify even in the absence of modelling data. Secondly, birds and soaring specialists—indeed, flying vertebrates of any kind—are extremely rare in the Morrison Formation. Research is clear that this is not due to sampling bias, fossil record incompleteness, or preservational distortion (Leach et al. 2021). A logical explanation for this serious anomaly would be appreciated from Kane et al. (2023) if it were true that only aerial foragers were able to survive as obligate scavengers. It would be interesting to gain an understanding as to why sauropod carrion, which was a highly abundant and highly available resource pool for any carnivore species, was unexploited by either pterosaurs or birds if soaring-specialist taxa are uniquely able to survive as obligate scavengers. We find it difficult to imagine such a scenario, given that even a medium-sized sauropod carcass, perhaps resulting from an animal having died of exhaustion during migration, would have had enough calories to sustain multiple large-bodied theropods for weeks or months, and of course would have been much more valuable for species of lightly built soarers. But, as it stands, there was no competition from aerial vertebrates in this time period. It therefore is reasonable to assume that sauropod carcasses were targeted by large theropods, and probably were their primary food source, because they were the only carnivores present in the ecosystem capable of appreciably consuming them. Finally, Kane et al.’s assertion that vultures are unable to outcompete mammals is demonstrably false. Condors consistently outcompete much larger cougars in South American landscapes at carcass sites; to such an extent that cougars must hunt 50% more often than their North American counterparts to meet their energy budgets (Elbroch and Wittmer 2013). African vulture species remain on carcasses unchallenged by mammalian carnivores up to 84% of the time (Houston 1979). They are able to arrive at carcasses much faster than terrestrial animals, so vultures often do not need to defend themselves from this kind of competition. Having said that, it is worth mentioning again that these issues probably are irrelevant to Morrison Formation ecology because aerial scavengers were absent, and theropods did not need to compete with them for carrion at all.


© 2022 Elsevier B.V.



Persistent Identifier