Many animals rely on facial traits to recognize their kin; however, whether these traits have been selected specifically for this function remains unknown. Using deep learning for face recognition, we present the first evidence that interindividual facial resemblance has been selected to signal paternal kinship. Mandrills (Mandrillus sphinx) live in matrilineal societies, in which females spend their entire lives not only with maternal half-sisters (MHS) but also with paternal half-sisters (PHS). We show that PHS have more differentiated social relationships compared to nonkin, suggesting the existence of kin recognition mechanisms. We further demonstrate that facial resemblance increases with genetic relatedness. However, PHS resemble each other visually more than MHS do, despite both kin categories sharing similar degrees of genetic relatedness. This paternally derived facial resemblance among PHS indicates selection to facilitate kin recognition. This study also highlights the potential of artificial intelligence to study phenotypic evolution.

Phenotype matching, a learning mechanism that evolved based on phenotypic cues shared among relatives, may provide animals with the ability to recognize unfamiliar kin. The generalization of this mechanism across animal species is debated, however, because appropriate tests are difficult to design due to possible confounding effects of familiarity. Hence, only a few studies have examined evidence for the existence of such a mechanism in natural populations. Here, we tested the phenotype matching hypothesis based on visual cues in a semi‐free‐ranging population of mandrills (Mandrillus sphinx) that contains individuals related to different degrees and where familiarity is controlled for. Using an experimental design based on the presentation of photographs, we show that mandrills discriminate unfamiliar relatives using facial cues alone. Our results build on earlier studies, showing that primates use phenotype matching to recognize and subsequently discriminate unfamiliar kin. We suggest that facial features along with other visual and non‐visual cues provide a proximate mechanism for kin selection to operate.

No abstract

Animal faces convey important information such as individual health status1 or identity2,3. Human and nonhuman primates rely on highly heritable facial traits4,5 to recognize their kin6–8. However, whether these facial traits have evolved for this specific function of kin recognition remains unknown. We present the first unambiguous evidence that inter-individual facial similarity has been selected to signal kinship using a state-of-the-art artificial intelligence approach based on deep neural networks and long-term data on a natural population of nonhuman primates. The typical matrilineal society of mandrills, is characterized by an extreme male’s reproductive skew with one male generally siring the large majority of offspring born into the different matrilines each year9. Philopatric females are raised and live throughout their lives with familiar maternal half-sisters (MHS) but because of male’s reproductive monopolization, they also live with unfamiliar paternal half-sisters (PHS). Because kin selection predicts differentiated interactions with kin rather than nonkin10 and that PHS largely outnumber MHS in a mandrills’ social group, natural selection should favour mechanisms to recognize PHS. Here, we first show that PHS socially interact with each other as much as MHS do, both more than nonkin. Second, using artificial intelligence trained to recognize individual mandrills from a database of 16k portrait pictures, we demonstrate that facial similarity increases with genetic relatedness. However, PHS resemble more to each other than MHS do, despite both kin categories sharing similar degrees of genetic relatedness. We propose genomic imprinting as a plausible genetic mechanism to explain paternally-derived facial similarity among PHS selected to improve kin recognition. This study further highlights the potential of artificial intelligence to study evolutionary mechanisms driving variation between phenotypes.