In this study, we conducted a first exploratory comparison of captive wolves and admixed wolves. Largely in line with our hypotheses (Table 1), we consistently found behavioural differences in how wolves and admixed wolves reacted to unfamiliar humans and novel objects, and in the cohesiveness of their social networks (Table 1). As predicted (Prediction 1), when exposed to unfamiliar humans in the experimental task, wolves were more vigilant, fearful, and aggressive than admixed wolves, and less likely to interact with humans (i.e., approaching within 5 meters and sniffing them). Unexpectedly, however, wolves were also more likely than admixed wolves to spend time in human proximity. When exposed to novel objects, as predicted (Prediction 2), wolves were more aggressive than admixed wolves within the experimental context, less likely to spend time in object proximity, and more likely to interact with objects. Unexpectedly, however, wolves were less vigilant and as fearful as admixed wolves. Finally, in line with our predictions (Prediction 3), we revealed more cohesive social networks in wolves than admixed wolves, although weighted densities for proximity networks were similar in wolves and one group of admixed wolves (AW1).
When exposed to unfamiliar humans, admixed individuals were less vigilant, less fearful and less aggressive than wolves, and they were also more likely to approach humans, without showing fearful or aggressive behaviour. These results are in line with the hypothesis that the introgression of dog genes might trigger behavioural changes in how wolves relate to humans. Comparative studies have indeed provided compelling evidence that dogs are less neophobic than wolves toward humans (e.g., Bentosela et al., 2016; Klinghammer & Goodmann, 1987; Range & Marshall-Pescini, 2022), likely as a result of the domestication process selecting individuals with higher tolerance and/or attraction to humans (Hare et al., 2002; Li et al., 2013; vonHoldt et al., 2017; vonHoldt & Driscoll, 2016). Given the genetic basis of these behaviours (e.g., Salomons et al., 2021; vonHoldt et al., 2017), dog admixture may thus lead to a decrease in wolves’ levels of fear and aggression toward humans, and an increase in the likelihood of interactions, as suggested by our results. In contrast to our predictions, however, wolves were more likely than admixed individuals to remain at the same distance or move closer to unfamiliar humans in the first ten seconds of the human task. In this study, however, proximity to humans did not necessarily reflect increased attraction or tolerance, but it could also result from wolves being more territorial and/or defensive in the presence of unfamiliar humans. Therefore, it is possible that wolves, in this task, perceived humans as a potential danger (as confirmed by the higher occurrence of vigilant, fearful and aggressive behaviour within the experimental context) and were also more likely to maintain proximity, without retreating, as an expression of their territorial behaviour (e.g., Mech, 1993; Mech & Boitani, 2003).
When exposed to novel objects, differences between wolves and admixed individuals were less clear-cut. As predicted, admixed wolves showed less aggressive behaviours and were more likely than wolves to be in proximity of novel objects, suggesting a link between dog admixture and lower neophobia. However, admixed individuals were also more vigilant than wolves, and as fearful. Therefore, we only found partial support to the hypothesis that dog admixture is linked to lower object neophobia. These results can be explained in at least two ways. First, it is possible that domestication has selected for individuals that were specifically more tolerant and/or attracted to humans, but not to the human environment (Hare et al., 2002; Li et al., 2013; vonHoldt & Driscoll, 2016). If so, the introgression of dog gene variants would lead to clear changes in wolf reaction to humans, as suggested by our results in the human task, but not to changes in their reaction to human artefacts. Second, it is possible that multiple evolutionary pressures contribute to shaping how species react to novel objects. Wolves, in particular, might be more neophobic than dogs (and perhaps admixed wolves), because dog domestication has selected for individuals that are more attracted to humans and their artifacts (Fritts et al., 2003; Hare & Tomasello, 2005; Kaulfuβ & Mills, 2008), but also because high wolf persecution by humans likely provided selective advantages to more fearful wolves (Fritts et al., 2003; Range & Marshall-Pescini, 2022). However, wolves and dogs also differ in terms of the ecological challenges they face: in contrast to dogs, wolves largely live on the preys they hunt, which implies that high neophobia may be detrimental for their survival (see Moretti et al., 2015; Peterson & Ciucci, 2003). Moreover, wolves are also more territorial than dogs, and they might thus be more explorative and persistent when interacting with novel objects (Marshall-Pescini et al., 2017; Moretti et al., 2015; Rao et al., 2018), as also confirmed by our object task, in which wolves were more explorative than admixed individuals. Therefore, how admixed wolves react to novel objects might be the result of a complex interplay of multiple factors, as also shown in dogs and wolves (see Range & Marshall-Pescini, 2022, for a review). This complex interaction of different factors might thus explain why wolves in our study approached novel objects more cautiously, but also showed less vigilance and more attempts to interact with them (for similar results on dogs and wolves, see Moretti et al., 2015).
Social network analyses suggested higher cohesiveness in wolves than admixed wolves, in terms of grooming and social play, whereas wolf proximity networks were similarly cohesive only to one group of admixed wolves (AW1), the one only including siblings. Grooming, for instance, was only observed in wolves, whereas play and proximity were only observed in wolves and (to a minor extent) in one of the two admixed groups (AW1). These differences between wolves and admixed individuals are in line with studies showing that dogs generally form looser bonds with their conspecifics and are more solitary than wolves (e.g., Beck, 1973, 1975; Berman & Dunbar, 1983; Daniels, 1983; Daniels & Bekoff, 1989; Ortolani et al., 2009; Rubin & Beck, 1982; see Bonanni & Cafazzo, 2014; Cafazzo et al., 2018). Therefore, it is possible that, through the introgression of dog genes, admixed wolves become less likely to engage in affiliative behaviours with other group members, leading to the formation of less cohesive groups. These findings, however, should be taken with caution, as it is not possible to rule out alternative explanations for our results. In our study, for instance, wolves lived in a smaller enclosure than both admixed groups, and this might have favoured a higher rate of encounters and affiliative behaviours in the group. However, this explanation is unlikely, as proximity networks were similar for W and AW1, but social interactions were still more common in admixed dyads. Moreover, as limited space may be a source of stress, affiliative interactions might have been more frequent in wolves than in hybrids as an effective way of reducing stress levels in the group (see Aureli & Smucny, 2000; Bonanni et al., 2010). Furthermore, the lack of social interactions in AW2 could have depended on the fact that group members were not all siblings (in contrast to AW1) but came from different litters (although this was also the case for W), and were moved in the new enclosure only shortly before our observations. Therefore, more wolf and admixed groups with similar size, kinship, and enclosures will have to be included to confirm our preliminary findings.
Overall, our study provides a first direct comparison of behavioural traits in admixed and non-admixed wolves, but the results should be taken with caution, due to several important limitations. First, our study sample was limited to only two admixed groups and one wolf group. This implies that the specific characteristics of the study groups (e.g., group size and composition, kin relationship across individuals, previous life experiences, exposure to visitors, time since group establishment) cannot clearly be disentangled from the test predictor (i.e. admixture). To avoid these confounds, future studies should ideally include more groups with different characteristics, although this will not be an easy endeavour, given that groups of admixed wolves are rarely found in captivity. However, it should be noted that most of these confounds cannot alone explain our results. Group size, for instance, was larger in wolves than admixed groups, but this should have led to lower neophobia and lower densities in the social network (e.g., Balasubramanian et al., 2017; Farine & Whitehead, 2015), which was not the case in our study. Moreover, behavioural differences between wolves and admixed individuals were mostly in different directions depending on the task (e.g., wolves were more vigilant and fearful in the human task, but not in the object task), suggesting that group size, at best, played a marginal role in explaining these differences. Especially important is the fact that our study subjects were not raised for the purpose of the experiment, so that their life experiences were comparable, but not identical. Although all the admixed wolves spent the first weeks of their life with their mothers, they were then raised by humans in captivity from an early age, and this might have made them less fearful and aggressive to humans, as compared to admixed wolves living in the wild. Therefore, it is not possible to exclude that early experiences partially affected the behaviour of our study subjects, and future studies should ideally compare wolves and admixed individuals with strictly controlled life experiences, and/or living in the wild. However, it should also be noted that wolves, like admixed individuals, were born and housed in a zoo, where they were regularly exposed to keepers and visitors from an early age, even more frequently than admixed wolves. A second important limitation of our study is that our results may not be generalizable to all cases of dog-admixture, because different parental dog breeds are partly characterized by different behavioural traits (e.g., Serpell & Duffy, 2014). Third, the facilities in which we conducted our studies did not allow individuals to be separated during the tasks, because experimental manipulations could be stressful for the groups. Therefore, it is possible that individuals would show different responses toward novel stimuli when tested alone, because the presence of other group members might facilitate (e.g., social facilitation) and/or hinder (e.g., presence of higher-ranking conspecifics) individuals’ responses. Finally, while our study provided a first assessment of individuals’ reaction to novel stimuli and the cohesiveness of their social networks, differences in other behavioural and cognitive traits can also be expected (e.g., innovation, risk taking, reconciliation patterns), as these traits have been reported to differ between dogs and wolves (Range & Marshall-Pescini, 2022).
Despite the above caveats, our study provides a first significant exploration of possible behavioural differences between admixed and non-admixed wolves in captivity. Our results suggest that dog admixture may decrease cohesiveness in wolf social groups, and also explain variation in how wolves react to unfamiliar humans, and partially to novel objects. These findings may help increasing our understanding of the impact that anthropogenic hybridization has on phenotypic variation in wild population of wolves and, if supported by future studies, they might also have key implications for conservation. A better understanding of how anthropogenic hybridization, and the extent of admixture, might affect wolf behaviour will contribute to making more informed decisions about the management of admixed individuals in the wild (Donfrancesco et al., 2019). Based on our findings, however, we maintain that by no means it can be assumed that admixed and non-admixed wolves behave similarly. Further studies should investigate the behaviour of admixed wolves, both in captivity and in the wild, possibly adopting genomic tools to ascertain the genetic origin of behavioural variations in admixed individuals.