Genetic effects on mating success and partner choice in a social mammal

Abstract

Mating behavior has profound consequences for two phenomena--individual reproductive success and the maintenance of species boundaries--that contribute to evolutionary processes. Studies of mating behavior in relation to individual reproductive success are common in many species, but studies of mating behavior in relation to genetic variation and species boundaries are less commonly conducted in socially complex species. Here we leveraged extensive observations of a wild yellow baboon (Papio cynocephalus) population that has experienced recent gene flow from a close sister taxon, the anubis baboon (Papio anubis), to examine how admixture-related genetic background affects mating behavior. We identified novel effects of genetic background on mating patterns, including an advantage accruing to anubis-like males and assortative mating among both yellow-like and anubis-like pairs. These genetic effects acted alongside social dominance rank, inbreeding avoidance, and age to produce highly nonrandom mating patterns. Our results suggest that this population may be undergoing admixture-related evolutionary change, driven in part by nonrandom mating. However, the strength of the genetic effect is mediated by behavioral plasticity and social interactions, emphasizing the strong influence of social context on mating behavior in socially complex species.

Figure 1. Male genetic background, genetic diversity, and dominance rank influenced consortship probabilities

Probability density functions for the proportion of consortship opportunities (see Methods) converted into successful consortships by each male (A and B) or each male rank position (C), where the male or rank position is the unit of analysis and summary values are from the raw data without controlling for other variables. Distributions are partitioned to represent two data sets, one including males above the median and one including males below the median for A) male hybrid score; B) male genetic diversity or C) male rank. Males with higher hybrid scores, higher genetic diversity, and higher dominance rank have a higher probability of consorting, given an opportunity, than other males.

Figure 2. Assortative mating tendencies and male genetic background influenced consortship probabilities

The probability of a sexual consortship occuring increased with both male hybrid score and with assortative pairings by hybrid score. The large central figure shows the relationship between female hybrid score, male hybrid score, and the probability of consortship, based on the fitted effects from the consortship model. Note that the fitted effects are a combination of two of the idealized results shown on the left: consortships were more likely to occur if the male was more anubis-like, and also if both the male and the female had a similar genetic background (assortative mating).

Figure 3. The size and significance of effects varied across social groups

The six variables identified as significant predictors of consortship occurrence in the overall analysis were investigated in five social groups (excluding the two groups with the smallest available sample size): A) male hybrid score; B) male dominance rank; C) assortative mating index; D) male-female dominance rank interaction; E) genetic distance; F) male genetic diversity. Colors indicate strength of agreement with the effect identified in the combined analysis across all groups: dark blue represents a significant effect that is concordant with the direction of the effect in the full analysis; light blue presents a nonsignificant effect that is concordant with the direction of the effect in the full analysis; light yellow represents a nonsignificant effect in the opposite direction of the effect in the full analysis; and dark yellow represents a significant effect in the opposite direction of the effect in the full analysis. *p <= 0.01; **p <= 0.05.

Figure 4. Rank-related effects and genetic effects predicted consortship behavior

Genetic variables and rank-related variables both predicted consortship outcomes better than random chance, but rank-related variables performed better. When the data set was permuted, neither set of variables predicted better than random chance, as expected (area under the curve, AUC, = 0.5, shown by the dashed gray line). Boxplots show the distribution of AUC values for 100 iterations of leave-k-out model fitting and prediction, where k = 500 possible consortship pairs.