Results

The percentage of yawning events observed by a partner that elicited a yawn was 41% in non-admixed wolves (N = 90/221) and 24% in admixed wolves (N = 48/204). Across individuals, the proportion of yawning events that elicited a yawn after being observed was, on average (mean ± SD), 0.41 ± 0.13 in European wolves, 0.37 ± 0.25 in Hudson wolves, and 0.21 ± 0.12 in admixed wolves. When partners did not observe the yawning event, however, the percentage of yawning dropped to 6% in both non-admixed wolves (N = 10/167) and admixed wolves (N = 11/183). Across individuals, the proportion of non-observed yawning events followed by a yawn was 0.17 ± 0.15 in European wolves, 0.00 ± 0.01 in Hudson wolves, and 0.10 ± 0.13 in admixed wolves.

The full model significantly differed from the null one (GLMM, χ2 = 102.28, df = 9, p < 0.001), showing a significant effect of the 2-way interactions of both introgression and yawner centrality with observation of the event on the log odds of yawning (Table 2). First, post-hoc tests revealed that both admixed and non-admixed wolves were more likely to yawn after observing than not observing a yawning event (both p < 0.001). Moreover, the probability of a partner yawning after observing a previous yawning event was higher for non-admixed than admixed wolves (p = 0.016), although there was no difference between the two, if the partner had not observed the event (p = 0.902; Fig. 2). Second, the log odds of yawning increased with increasing yawner centrality, for both admixed and non-admixed wolves, but only after observing a previous yawning (Fig. 3).

Predictors | Estimate | SD | 95% CIs | LRT | df | P
Intercept | − 2.84 | 0.52 | − 3.86 to − 1.83 | - | - | -
Observation * Introgression | 1.25 | 0.58 | 0.11 to 2.39 | 1 | 4.75 | 0.029*
Observation * Yawner centrality | 0.74 | 0.32 | 0.12 to 1.36 | 1 | 5.38 | 0.020*
Observation * Yawner rank | − 0.13 | 0.33 | − 0.77 to 0.52 | 1 | 0.15 | 0.702
Observation * CSI | − 0.30 | 0.26 | − 0.81 to 0.21 | 1 | 1.26 | 0.262
Observation | 1.55 | 0.38 | 0.79 to 2.30 | - | - | -
Introgression | − 0.36 | 0.52 | − 1.38 to 0.67 | - | - | -
Yawner centrality | -0.64 | 0.28 | − 1.19 to − 0.09 | - | - | -
Yawner rank | 0.11 | 0.30 | − 0.47 to 0.69 | - | - | -
CSI | 0.50 | 0.23 | 0.04 to 0.95 | - | - | -
Yawner sex | − 0.33 | 0.27 | − 0.87 to 0.21 | 1 | 1.40 | 0.237
Partner sex | 0.34 | 0.28 | − 0.22 to 0.89 | 1 | 1.47 | 0.226

Fig. 2: For non-admixed and dog-admixed wolves, predicted probability that other group members would yawn after observing (in blue) or not observing (in grey) another yawning event. The thick horizontal lines represent the estimated mean probabilities as assessed from the model, the horizontal ends of the boxes represent the estimated means ± SD, and the ends of the whiskers represent the 95% confidence intervals. Post-hoc significant differences are marked with an asterisk. Dog-admixed wolves included second- or later-generation backcrosses to wolves, as determined by biparental and uniparental genetic markers (Caniglia et al. 2020)

Fig. 3: For all study animals, predicted probability that other group members would yawn after observing (in blue) or not observing (in grey) another yawning event, as a function of the yawner’s centrality in the social network. Dots represent the mean probability of yawning for each individual and the lines represent the fitted model, as described in the text