Whereas studies in comparative cognition normally invoke ecology and social organisation to account for differences in social behaviour and cognition across species, dog–wolf differences have so far been explained mostly as a result of direct human selection for desirable traits (e.g., tameness, attention to humans, sociability). Yet, as will be reviewed in this paper, dogs and wolves also differ considerably in both their feeding niche and social organisation (which together we refer to as ‘social ecology’).
We suggest that observed wolf–dog differences especially in their interaction with the environment (e.g., neophobia, persistence, risk taking) and conspecifics (e.g., tolerance, cooperation, communication) need to be considered also in regard to their social ecology. We propose that social ecology alongside human selection should be recognised as a potentially important factor affecting dogs’ behaviour, and suggest a number of potential avenues for future research, which can more directly test the relative importance of these selection forces.
Genetic analyses indisputably conclude that dogs’ closest living relatives are grey wolves suggesting that the common ancestor prior to domestication likely resembled modern day wolves . Pioneering studies by Frank et al. in the 1980s  and more recent research in Hungary  and Austria  have highlighted a host of behavioural differences between wolves and dogs. To account for these differences, a series of ‘dog domestication hypotheses’ have been proposed suggesting that human selection for specific traits such as increased attention to humans ; acceptance of humans as social partners ; reduced fear and aggression of humans ; a tamer temperament , human-directed social competence  lead to increased cooperativeness and related cognitive skills in dogs.
Although differing in the specifics, all the hypotheses above (see Ref. [10•] for a review) attempt to account for observed differences in wolf–dog behaviour & cognitive skills based solely on direct selection by humans for specific dog traits, which relate to face-to-face interactions with people (we collectively refer to these as the ‘human-oriented’ hypotheses). However, the direct interaction with humans is not the only change that occurred during the domestication process, and as we hope to clarify, wolves and dogs differ in a number of other important aspects.
The location of dog domestication is still controversial , but it is widely accepted that dogs were well established across Eurasia by the late Pleistocene, that is, before the advent of cultivation. Two scenarios have been proposed for the early stages of dog domestication: the human-selection and the self-domestication scenario.
The ‘human-selection’ scenario (or ‘directed pathway’ [12•]) postulates that people actively adopted wolf pups and selected only the most docile animals for reproduction . Circumstantial support for this scenario comes from archaeological and anthropological records showing that adoption and taming of various Canid species by indigenous populations occurred across South America before the arrival of domestic dogs , suggesting this may have occurred also with wolves.
Differently, the ‘self-domestication’ or ‘commensal pathway’ hypothesis proposes a two-stage process. In the first stage, dogs’ ancestors started exploiting a new feeding niche provided by humans reducing their reliance on conspecific group hunting and increasing their dependence on human generated food/refuse [12•, 15]. Already at this stage, selective pressures may have acted on proto-dog’s characteristics with less fearful individuals being more likely to approach human settlements, thus maintaining an advantage in the exploitation of the new niche compared to their more timid conspecifics . Proof of concept that proto-dogs were able to acquire and maintain distinct characteristics despite continued significant gene flow with wolves comes from recent studies of two wolf populations in North America [16, 17].
Despite living sympatrically and with regular gene flow between them, these populations have specialised in different prey and habitat use; one adopting a migratory lifestyle (following caribou), the other a more territorial organisation dependant on local prey. They also show a suite of genetic and morphological differentiations such as a lighter coat colouring in the migratory population. The second stage of the ‘self-domestication’ hypothesis, proposes that humans then started a direct selection by removing unwanted animals (e.g., too aggressive) and favouring animals with desired characteristics (e.g., more attentive, responsive, cooperative).
Only a time-machine would allow us to determine which scenario occurred, and quite possibly both processes played a role. However, independently of which pathway dogs took during domestication, the feeding niche of today’s wolves and dogs is remarkably different from each other and likely has been since the advent of cultivation.
Although the common perception of the ‘dog’ is that of a ‘pet dog’, free-ranging dogs actually represent between 70 and 80% of the world dog population [18, 19] and in many areas free-ranging dogs appear to form a quite distinct population deriving from ancestors native to the region rather than a recent admixture of modern breeds (e.g. Africa ; South and Central Asia, ; Eurasia,  but ).
These populations are not, as is often thought, made up of ex-pet dogs (which represent the minority), but rather they are self-sustaining populations of ‘free-breeding’ animals that independently choose their mates [24, 25] and whose existence is affected by dispersal patterns, food abundance, mate choice as well as human activities [15, 26].
Hence in the remaining part of this review we aim to (1) highlight the differences in the feeding ecology of wolves and free-ranging dogs and, likely as a consequence, (2) in their social organisations; and (3) suggest that to fully explain the behavioural differences between dogs and wolves, their respective social ecologies need to be taken into account.
Most wolves predominantly rely on group hunting of ungulates. Both the location and abundance of prey change seasonally, hence predictability is low. Furthermore, hunting is risky, considering most ungulates have the capacity to seriously injure their attackers. Finally, hunting requires extraordinary persistence considering success rates are only between 10% and 49% .
Although the relationship between hunting success and pack size is non-linear but dependent on prey type  and defence from scavengers , it appears to play a role both in hunting success and territorial defence .
Free-ranging dogs typically live in proximity to human settlements  and their survival rates, at least in some areas, are significantly affected by direct human intervention . Their diet consists predominantly of human waste (between 50 and 88%) with grains and human faeces (20%) as major components [33, 34•]. While hunting (including group hunting) wild animals is part of dogs’ feeding ecology [34•, 35], in most populations it plays a minor role and occurs mostly if human-provided food sources are not sufficient [36, 37] or when dogs live at greater distances from humans [34•, 38]. The importance of starch, and clear evidence of a shift in feeding ecology from hunting to scavenging, emerges from dogs showing genetic adaptations for starch digestion, which are absent in wolves  but already present in the ancient Newgrange dog (4800 BP-11). Scavenging on human refuse appears to be mostly a solitary or dyadic activity  likely due to the fact that the food is itemised and dispersed and does not require multiple individuals to obtain it. In fact if more animals are moving together and find the same small resource, it would likely be counter productive allowing potential for conflict to emerge. Indeed, Ethiopian wolves (also part of the Canis genus) show a social structure like that of grey wolves (a bonded pair with offspring), but due to the dispersed, itemised and reduced dimension of their prey, which consists mostly of burrowing rodents, they are mostly solitary foragers, only occasionally showing group hunting of small ungulates .
Wolf packs are predominantly composed of a breeding pair (which form a long-term bond), their adult and sub-adult offspring as well as pups from the most recent litter . However, unrelated individuals may also join . All members participate in pup-rearing by regurgitating food for the pups (and the lactating female) as well as defence from predators. Pups start actively participating in hunts only at 7–8 months of age; until then they are almost totally reliant on other pack members [44•]. Both litter size and pup survival is linked to pack size (and food abundance) . Overall, it appears that wolves are highly reliant on a cohesive and functional pack structure allowing them to successfully forage, defend their territory and raise pups [44•].
Free-ranging dogs appear to be ‘facultatively social’, potentially depending on food abundance [26, 46], breeding status of females and season . Most studies report that free-ranging dogs form groups of 2–8 individuals although stable packs of up to 27 individuals have also been reported [47•]. Packs are multi-male, multi-female, with some females choosing to breed with just one male, but the majority being promiscuous . Pups are raised mostly by their mothers (including regurgitation [47•]), although a few cases of paternal and ‘grandmotherly’ care have been observed [48, 49]. By the age of 10–11 weeks, mothers no longer provide for them . Pups may remain with their native pack (and having reached sexual maturity breed themselves) or disperse.
Interestingly, in both wolves and dogs, there are examples of different feeding strategies being adopted: some wolves have been observed scavenging on human refuse  and some dogs cooperatively hunting [34•, 35]. Similarly, mating strategies may also be flexible to a certain extent, with one example reported in dogs where only a single female was observed breeding in a family pack of feral dogs in Alaska  and occasional reports of multiple breeding females in wolf packs [44•]. In a number of Canid species, mating strategies appear to vary in accordance with resource abundance/distribution [46, 53], highlighting the strong link between feeding ecology and social organisation (hence the term ‘social ecology’). Likely this is the case also for grey wolves and dogs; nevertheless, comparing their more typical feeding ecologies (Table 1) and social organisations (Table 2), potentially important differences emerge.
In particular, humans’ long-term food storage capabilities provide waste uniformly over the course of a year and in predictable locations. Free-ranging dogs have adapted to this niche, and reliance on this fairly constant, predictable and safe food source has likely affected all aspects of their social life. Pair-bonding is no longer required since lactating females can den close to the food source  and can leave pups alone for only brief periods to forage.
This diminishes the need for alloparental investment as pups are rapidly able to provide for themselves [15, 18]. As a consequence, dogs’ reliance on pack members for both foraging and pup-rearing is much reduced compared to wolves. We suggest that such differences in social ecology may help explain behavioural differences observed in comparative studies of wolves’ and dogs’ interactions with their social partners and asocial environment.
Although the major focus of wolf–dog comparative research has been on their interactions with humans [10•], there is a growing number of studies also comparing their behaviour with conspecifics and the environment. The actual results appear to be difficult to account for based only on a ‘human-oriented’ view of domestication. Indeed, whereas a direct selection for tameness would predict higher tolerance and therefore higher cooperative abilities in dogs compared to wolves , a number of studies have not supported these predictions.
As regards tolerance, compared to wolves, pack-living dogs are less tolerant of proximity during feeding on a monopolizable food source tending to avoid conflict by maintaining distance rather than using communication to negotiate access as wolves do [54, 55]. In the same populations during daily observations of the animal’s social interactions, we found that low intensity aggressive encounters (not involving physical contact) were more frequent in wolves while higher intensity conflicts (involving physical contact) were more frequent in dogs, with reconciliation occurring in the former but not the latter (Cafazzo et al., submitted); indeed, after conflicts, dog opponents avoided each other by limiting their interactions.
From these results it emerges that dogs tend to adopt an avoidance of conflict strategy, which does not however equate to tolerance, since when it comes to food, dominant dogs monopolize the resource more successfully than dominant wolves .
In line with the above, a series of older studies comparing group living wolves and dogs showed that regardless of breed, dogs showed more escalations into serious fighting than wolves . Moreover, in a developmental study, the frequency of agonistic behaviours in dogs was more similar to that of solitary living and scavenging jackals than of wolves . Based on these observations, Feddersen-Petersen  suggested that dogs cooperate less than wolves, even at the most basic level of synchronising their behaviours with conspecifics.
Indeed, when comparing equally raised and kept pack-living wolves and dogs, the latter showed a reduced capacity for imitative learning from a con specific , and were less successful in intra specific cooperative tasks (Marshall-Pescini et al., submitted).
Taken together, such results are in line with the social ecology of the two species since dogs’ reduced reliance on pack members (for both hunting and pup-rearing) and greater dependence on small dispersed food sources which are sought mostly solitary, may have relaxed the need for intra specific communication and tolerance around food sources, as well as the need for social learning abilities and cooperative skills
Differences have also been found between wolves’ and dogs’ independent problem solving abilities and interactions with their environment. Predictions based on the selection for tamer temperament would be that a reduction in fear would allow them a greater ease/propensity to explore and manipulate novel objects/environments .
However, results seem to be more in line with the different ecologies of the two species with wolves being more persistent (and successful) in manipulative problem solving tasks [61, 62, 63] and more risk-prone when given a choice between a safe low quality reward vs. a high quality but less certain one . Furthermore, in line with their more dangerous and complex foraging strategy [65•], compared to dogs, wolves are more neophobic, in that they take longer to approach a novel object, but then they spend more time exploring such objects as well as novel environments [66, 67] (Figure 1).
Finally, as regards interaction with humans, domestication has without doubt eased the propensity of dogs to accept humans as social partners with very little exposure to humans necessary for dogs to become socialised  in contrast to the intensive human contact needed for wolf pup socialisation . However, it is interesting to note, that, when acceptance of humans as social partners is achieved, differences between wolves and dogs in understanding social cues and cooperating with them are smaller than initially thought. Indeed, at the Wolf Science Centre, where both are similarly extensively socialised to humans, wolves show a similar capacity to socially learn from a human partner , are equal to, if not better, at following human gazing cues into distant space and around barriers [71, 72] and are as likely to use gaze alternation to gain help from a human in obtaining an out-of-reach reward .
In conclusion, we suggest that, although human selection for favourable traits (e.g., tameness, human sociality, attention) played a major role in dog domestication, changes in their social ecology (i.e., both their feeding niche and social organisation) may also have played a significant part in moulding dogs’ behavioural and cognitive profile and should therefore be integrated in dog domestication hypotheses and given more attention in future research (Box 1).
Future research which may help better understand the effect of wolves’ and dogs’ social ecology on their observed behavioural differences
Future research which may help better understand the effect of wolves’ and dogs’ social ecology on their observed behavioural differences.
Disentangling the relative role of social ecology and human-selection on dogs’ social behaviour and cognition is difficult as there is no dog population where no direct selection by humans has occurred. Nevertheless some areas of research may help to address this question:
By: Sarah Marshall, Pescini, Simona Cafazzo, Zsófia Virányi and Friederike Range
Papers of particular interest, published within the period of review, have been highlighted as:Acknowledgements
We thank Rachel Dale, Kurt Kotrschal, Anindhita Bhadra and Ákos Pogány for comments on earlier versions of the manuscript. We further thank Rooobert Bayer and Roberta Massimei for the photos of the wolves and dogs respectively (used in Figure 1). Sarah Marshall-Pescini, Simona Cafazzo and Friederike Range were supported by funding from the European Research Council under the European Union’s Seventh Framework Programme(FP/2007-2013)/ERC Grant Agreement no. . Sarah Marshall-Pescini was further supported by the Vienna Science and Technology Fund (WWTF CS15-018) as was Zsófia Virányi (WWTF project CS11-026).