Cofeeding tolerance in chimpanzees depends on group composition: a longitudinal study across four communities

Abstract

Social tolerance is generally treated as a stable, species-specific characteristic. Recent research, however, has questioned this position and emphasized the importance of intraspecific variation. We investigate the temporal stability of social tolerance in four groups of sanctuary-housed chimpanzees over eight years using a commonly employed measure: experimental cofeeding tolerance. We then draw on longitudinal data on the demographic composition of each group to identify the factors associated with cofeeding tolerance. We find appreciable levels of variation in cofeeding tolerance across both groups and years that correspond closely to changes in group-level demographic composition. For example, cofeeding tolerance is lower when there are many females with young infants. These results suggest that social tolerance may be a "responding trait" of chimpanzee sociality, reflecting individual-level behavioral responses to social changes. Additional, experimental research is needed to better model the causal drivers of social tolerance within and among species.

Keywords: Animals; Evolutionary Biology; Zoology.

Graphical abstract

Figure 1

Aerial view of the enclosures Aerial view of the four enclosures at Chimfunshi (modified from Google Maps).

Figure 2

Time series of cofeeding tolerance in four groups of chimpanzees Each bar plots the central 90 percent credible interval of maximal cofeeding tolerance (i.e., initial cofeeding tolerance) in each group of chimpanzees in each year. Bars are jittered around year for visual clarity, but all empirical observations were matched in time. We observe substantial variation in cofeeding tolerance, both across groups and within groups across years. See also Figures S1–S6 and S8–S13, Tables S3 and S4, and Video S1.

Figure 3

Density plots of covariates Density plots of the effects of various covariates on maximal cofeeding tolerance, with central 90% credible intervals in blue. The group-specific plots show the within-group effects of a change in a covariate on the maximal cofeeding tolerance in that group (i.e., all parameters are random effects by group). The overall plots (the bottom plot in each panel) show the between-group effects (i.e., when the same parameters are shared across all groups). Note that for the covariates “infants ($<$3 years)” and “females (5-12 years)” very little of the central 90% credible intervals overlap with 0, indicating reliable within- as well as between-group effects. For the parameters “group size” and “maternal kin”, this is only the case for the between-group effects. See also Figure S8.

Figure 4

Estimates from the social relations model Each bar plots the central 90 percent credible interval of the regression coefficients giving the change in log-odds of a cofeeding tie as a function of the indicated variable. Bars with little or no overlap with 0 indicate reliable positive or negative effects of the respective variable. Bars are clustered by age category on the left axis, with color showing unique estimates by sex within each age category and shading indicating sample size. In the top frame, we plot individual-level random effect estimates for the interaction of sex and age category on residence in the food zone. In the bottom frame, we plot random effect estimates for dyads composed of females with or without infants under 3 years old (indicated by color) and males of various age categories (indicated by rows). Adult females with infants under age 3 and young females are less likely to coreside in the food zone with adult or adolescent males, while adult females without infants under age 3 are more likely to coreside with adult males. Females, with or without infants, are more likely to coreside with male ages 5 to 8 than with males of other age classes.