Primate social organization evolved from a flexible pair-living ancestor

Abstract

Explaining the evolution of primate social organization has been fundamental to understand human sociality and social evolution more broadly. It has often been suggested that the ancestor of all primates was solitary and that other forms of social organization evolved later, with transitions being driven by various life history traits and ecological factors. However, recent research showed that many understudied primate species previously assumed to be solitary actually live in pairs, and intraspecific variation in social organization is common. We built a detailed database from primary field studies quantifying the number of social units expressing different social organizations in each population. We used Bayesian phylogenetic models to infer the probability of each social organization, conditional on several socioecological and life history predictors. Here, we show that when intraspecific variation is accounted for, the ancestral social organization of primates was inferred to be variable, with the most common social organization being pair-living but with approximately 10 to 20% of social units of the ancestral population deviating from this pattern by being solitary living. Body size and activity patterns had large effects on transitions between types of social organizations. As in other mammalian clades, pair-living is closely linked to small body size and likely more common in ancestral species. Our results challenge the assumption that ancestral primates were solitary and that pair-living evolved afterward emphasizing the importance of focusing on field data and accounting for intraspecific variation, providing a flexible statistical framework for doing so.

Keywords: monogamy; social structure; social system.

Fig. 1.

The distribution of social organizations across extant primate populations. The Top panel demonstrates how we coded social organization per population as solitary, male-female (MF) or pair living, single male multifemale (MFF), single female multimale (FMM), or multimale and multifemale (MMFF). Three examples taken from field research on the slender loris (Loris lydekkerianus), common marmoset (Callithrix jacchus), and lion-tailed macaque (Macaca silenus) are shown. Large circles (Middle) around pictures represent different populations of the species. Smaller circles within each large circle represent a single social unit within a population, with color corresponding to the social organization observed. The phylogeny reflects a simple contour mapping of overall IVSO (# units deviating from primary social organization/total # units) across taxa in our database. Note that the branch lengths have been arbitrarily modified for visual clarity and should not be directly interpreted. The Lower panel shows the total number of populations in our dataset exhibiting each form of primary social organization, as well as overall IVSO (binwidth = 0.05). Left, primary SO: Dark gray bars represent uncertainty in the primary social organization for populations exhibiting two social organization with equally high frequency. Right, overall IVSO: Light gray hashed bars represent uncertainty in the level of IVSO due to studies including only a single social unit.

Fig. 2.

Variation in social organization among extant primates. (A) The proportion of variation in each social organization and overall IVSO accounted for by phylogenetic history, ecological and life-history factors (“ecological predictors”: habitat type and heterogeneity, diet, foraging style, substrate, activity pattern, and body size), as well as remaining residual (unexplained) variation among populations, species, and superfamilies. Dots indicate posterior medians, and lines indicate 90% Bayesian CIs. See SI Appendix, Table S4 for direct and total effect estimates for each predictor. (B) Total effects of the ecological and life history factors (activity level, substrate, and body size) used to predict ancestral social organization and IVSO. Thick lines indicate posterior medians, and ribbons indicate 90% Bayesian CIs.

Fig. 3.

Ecologically informed predictions of social organization in ancestral primate populations. Predicted probabilities of a social unit exhibiting each social organization and some form of IVSO within ancestral populations, assuming that the ancestral was primarily nocturnal, arboreal, and small-bodied (~50 g/−2 SD from the mean body size of extant species), as well as average within-superfamily sampling effort. See SI Appendix, Table S6 for predictions at different levels of ancestral body size, from −0.5 SD to −2 SD. Scaled posterior densities are shown, with posterior medians indicated by the dotted line.