Discussion

In this study, we investigated behavioural contagion in captive wolves and wolves admixed (i.e., introgressed) with dogs. Our results showed that, in both admixed and non-admixed wolves, individuals who had the opportunity to observe a yawn were significantly more likely to yawn within the subsequent two minutes than those who had not observed the event. Crucially, this effect was stronger in non-admixed than admixed wolves. Furthermore, individuals that were more socially integrated in the group were more likely to elicit behavioural contagion in both admixed and non-admixed wolves. In contrast, neither individual rank nor dyadic relationship quality between the yawner and the observer significantly influenced the probability of yawning contagion.

The presence of behavioural contagion in both admixed and non-admixed wolves supports Prediction 1a and aligns with previous findings in wolves (Romero et al. 2014). More broadly, our findings are consistent with evidence of contagious yawning across a range of social mammals and birds (e.g., Gallup et al. 2015; Miller et al. 2012; Norscia et al. 2021; Rossman et al. 2020), supporting the view that behavioural contagion represents a widespread mechanism facilitating behavioural synchrony (Duranton and Gaunet 2016; Massen and Gallup 2017). Our study extends previous work by showing that also admixed wolves, like non-admixed ones, show behavioural contagion when observing conspecifics yawning, suggesting that the perception–action mechanisms underlying this phenomenon (Chartrand and Bargh 1999; Preston and de Waal 2002; Yoon and Tennie 2010) might be preserved in admixed wolves despite introgression from domestic dogs, for which evidence of behavioural contagion in response to conspecifics is still lacking.

At the same time, we found that behavioural contagion was significantly stronger in non-admixed than in admixed wolves, supporting Prediction 1b. This difference is in line with the hypothesis that dog domestication may have favoured the emergence of social responsiveness and attentional biases toward interspecific rather than intraspecific partners, (Call et al. 2003; Miklósi et al. 2003; Hare and Tomasello 2005), possibly decreasing the likelihood of intraspecific behavioural contagion in dogs and, through the introgression of dog genes, in admixed wolves. Indeed, there is no conclusive evidence yet for dog–dog contagious yawning (Harr et al. 2009; O’Hara and Reeve 2011), although dogs have repeatedly been shown to yawn contagiously in response to human stimuli (Joly-Mascheroni et al. 2008; Romero et al. 2013; Madsen and Persson 2013; Silva et al. 2012; Neilands et al. 2020). In the future, it will be important to confirm these findings by systematically testing dog–dog social responsiveness, attentional biases and behavioural contagion in individuals living under different conditions (e.g., dogs raised by humans, dogs in kennels, free-ranging mongrel dogs), to better disentangle evolutionary and ontogenetic factors that might explain variation in these behaviours. Moreover, our findings align with previous work suggesting that social networks might be more cohesive in non-admixed than in admixed wolves (Amici et al. 2024). This indicates that the introgression of dog genes might potentially influence fine-scale mechanisms of behavioural coordination, which, in wolves, may facilitate synchrony and group cohesion (Duranton and Gaunet 2016; Conradt and Roper 2000). In the future, it will be essential to confirm the differences in behavioural coordination and group cohesion that we revealed, by including more wolf and admixed groups, possibly including recent hybrids (sensu Stronen et al. 2025) and ideally in natural settings.

Individuals’ integration in the social network mediated behavioural contagion but, in contrast to Prediction 2d, in a similar way for admixed and non-admixed wolves. In all groups, individuals that were more central in the social network were more likely to elicit yawning contagion, when the yawning event had been observed. This suggests that socially well-integrated individuals might exert greater influence on the behavioural responses of group members. Although the effect size appears modest, it mirrors findings in primates, where more central individuals elicited higher levels of contagion (Zhang et al. 2022). Social centrality may therefore represent a robust proxy for social prominence, capturing an individual’s embeddedness within the group and, consequently, its social salience (see Farine 2017). From an attention-based perspective (Massen et al. 2012; Massen and Gallup 2017), individuals that are more integrated may attract greater visual monitoring from others, thereby increasing opportunities for behavioural contagion.

In contrast, neither dominance rank nor dyadic relationship quality (CSI) significantly mediated behavioural contagion, neither in admixed nor in non-admixed wolves. The absence of a CSI effect is particularly noteworthy, as previous studies have reported higher levels of contagious yawning between closely bonded individuals in chimpanzees and bonobos (Campbell and de Waal 2011, 2014; Demuru and Palagi 2012; Palagi et al. 2014), geladas (Palagi et al. 2009) and wolves (Romero et al. 2014). Our findings therefore do not support Prediction 2a and contrast with studies linking contagion to relationship quality or familiarity. One possible explanation is methodological. Our CSI was calculated exclusively on spatial proximity (Silk et al. 2009), which may not fully capture the multidimensional nature of affiliative bonds. Wolves, in particular, are known to form complex and highly differentiated social relationships within their packs (Packard 2003, 2012). These relationships are expressed not only through spatial association but also through coordinated and cooperative activities such as hunting (MacNulty et al. 2009, 2012), breeding (Mech 1999; Packard et al. 1992) and territorial defence (Harrington and Mech 1979; Packard 2003). By relying solely on proximity as an indicator of bond strength, we may therefore have overlooked important behavioural components of social relationships, potentially limiting our ability to detect bond-related modulation of contagion. However, it is also possible that social bond strength is simply not a relevant factor in contagious yawning, in line with other literature on domestic dogs (Neilands et al. 2020; O’Hara and Reeve 2011; Madsen and Persson 2013), primates (e.g., Massen et al. 2012; Madsen et al. 2013; Pedruzzi et al. 2025; Valdivieso-Cortadella et al. 2023) and parrots (Gallup et al. 2015). Similarly, the absence of rank effects contrasts with our Prediction 2b and suggests that dominance status alone does not significantly influence behavioural contagion in admixed and non-admixed wolves.

Several limitations of our study should be acknowledged. First, the number of study groups and individuals was necessarily limited, as facilities housing groups of wolves admixed with dogs are extremely rare. Therefore, variation in enclosure size, group size and composition, as well as early life experiences, could not be fully controlled for. In particular, most admixed wolves spent few weeks in the wild with their mother and were subsequently raised in captivity, whereas the zoo-born wolves were born and raised entirely in captivity, with regular exposure to keepers and visitors. These differences were mostly graded rather than categorical, making it impossible to fully capture them with simple descriptors. It therefore cannot be excluded that such developmental differences influenced social responsiveness or attentional patterns. Moreover, wolf groups in our study were slightly larger than admixed groups, which raises the possibility that group size may have contributed to the observed difference. In principle, larger groups might promote behavioural contagion because individuals are exposed to a greater number of potential demonstrators and thus to more opportunities for contagion. In the present study, however, we accounted for this by modelling behavioural contagion as the odds that an individual yawned as a function of whether it had observed a trigger event or not. This approach reduced the likelihood that higher rates in larger groups simply reflect the presence of more potential yawners. Nevertheless, it is also possible that behavioural contagion plays a stronger functional role in larger groups, where mechanisms promoting behavioural synchrony may be especially important to maintain group cohesion and coordinated activity. If so, variation in behavioural contagion may reflect the specific socio-ecological conditions experienced by a group, rather than selective pressures working at the evolutionary level, an avenue that warrants further investigation. Second, as in most previous studies of contagious yawning (e.g., Romero et al. 2014; Massen and Gallup 2017), our data were collected in captive settings. Although our modelling approach allowed us to control for differences in the baseline occurrence of trigger events, which might also be linked to specific living conditions, caution is still warranted when extrapolating these findings to wild populations. Third, wolf-dog hybridization represents a continuum, and even if our findings were confirmed with larger samples, caution would remain necessary when generalising to more recent hybrids (sensu Stronen et al. 2025) or introgressed individuals with different degrees of admixture, as deviations from species-typical wolf behaviour may be expected to scale with the proportion of dog ancestry. Fourth, although we conducted genetic analyses to confirm introgression with dogs in our admixed groups, we did not investigate the genomic architecture potentially underlying behavioural variation. Yet recent genomic research suggests that even low levels of dog introgression may affect specific brain function and behaviour in wolves (Pilot et al. 2021), including variants involved in neurotransmission and neurodevelopment. Given the strong genetic basis of behavioural traits in dogs (Morrill et al. 2022; Salomons et al. 2021), and the extensive impact of artificial selection on the dog genome (Bergström et al., 2020; Freedman et al., 2016), it is plausible that introgressed alleles may contribute to specific behavioural variation in wild wolves subject to various extents of introgression from dogs (Leonard et al. 2013). If the observed differences between admixed and non-admixed wolves are replicated, future research integrating behavioural data with finer-scale genomic analyses will be essential to determine whether specific introgressed gene variants are associated with variation in social responsiveness and coordination. Finally, for logistic constraints, the timing of data collection varied across the five groups studied, introducing additional potential confounding factors that we could not fully account for. Temperature, for instance, is known to affect yawning behaviour (Campos and Fedigan 2009; Eldakar et al. 2015; Gallup et al. 2011; Massen et al. 2014), and although our models partially controlled for baseline yawning propensity (e.g., by also modelling individual’s likelihood to yawn when not observing the trigger event), future studies should ideally incorporate environmental temperature more explicitly as a predictor in the models. Similarly, we did not measure circulating hormone levels or account for variation in reproductive or gonadal status across individuals and groups, although this would ideally have been controlled for, given that behavioural sampling occurred both within and outside the breeding season. This may be relevant, as hormonal state might be linked to empathy-related processes and contagious yawning (e.g., Kis et al. 2020). Future work should therefore consider including hormonal status or proxies thereof to better control for its potential effects on behavioural contagion.

In conclusion, our findings support the view that dog introgression may modulate behavioural contagion in admixed wolves, possibly by modifying individual attentional biases. Given that hybridization with dogs is an emerging threat for several European wolf populations (Ciucci et al. 2026), understanding how introgression might affect behavioural processes linked to pack cohesion is of growing importance. While our findings should not be interpreted as evidence of broad behavioural disruption, they indicate that genetically detectable levels of admixture may indeed influence fine-scale social dynamics. Behavioural modifications linked to admixture could, in turn, have cascading ecological consequences: changes in attentional biases and social responsiveness may affect pack size and cohesion, social interactions within and between packs, reproductive success through alloparental care and dispersal dynamics (Newsome et al. 2017; Sparkman et al. 2012). Integrating behavioural analyses with ecological, demographic and genomic data will therefore be essential to determine whether such differences translate into measurable consequences for group coordination, cohesion and eventually the ecological role of wolves.