Inferring Kinship from Behaviour: Maternity Determinations in Yellow Baboons

Kinship is commonly inferred from behaviour in primate field studies, but the validity of such inferences has not yet been documented. A comparison of the relationships of six three-year-old yellow baboon (Papio cynocephalus) females with 14 adult females showed that when a juvenile's mother was living she could easily be identified from behavioural data. The most useful behaviour in this context was Presenting For Grooming. When the mother was not living, however, the juvenile compensated by forming a strong relationship with a less closely related or unrelated adult female. If compensation also occurs in other populations and other species, past attempts to infer kinship from behaviour have probably included a few cases in which a female was incorrectly identified as a juvenile's mother. Multivariate statistical techniques revealed differences between mother-daughter relationships and strong relationships based on compensation among other individuals. These differences involved comparisons of the frequencies of certain behaviours (frequency of Grooming by adult versus frequency of Grooming by juvenile, frequency of Grooming versus frequency of aggression) as well as the magnitudes of frequencies (amount of Grooming, number of Interventions). In field studies of primates, the identity of an individual's mother is often inferred from observed behavioural relationships. This has been done with chimpanzees (Pan troglodytes) (Van Lawick-GoodaU 1968), baboons (Papio spp.) (Cheney 1977, 1978a, 1978b; Lee & Oliver 1979), and Japanese macaques (Macacafuscata) (Kawai 1965; Kurland 1977), among others. For very young animals, persistent suckling from one female identifies that female as the mother. Although females do adopt and suckle infants that are not their own (Marsden & Vessey 1968; Berman, personal communication), this phenomenon is so rare that suckling is a very reliable criterion for identification of maternity (Koyama 1967; but see Kurland 1977). In all of the above studies, however, the identities of mothers of fully weaned juveniles, and in some cases even of adults (Kurland 1977), were inferred. The criteria used in these cases include play, grooming, proximity, interventions in ongoing agonistie interactions, and unrestricted access to infants (Lee & 0liver 1979); proximity, grooming, presenting for grooming, and sleeping in proximity (Cheney, personal communication); grooming and play (Van Lawick-GoodaU 1968); and close association (Kurland 1977). Evidence for the reliability of inferences about older animals consists of correlations between *Present address: Museum of Vertebrate Zoology, 2593 Life Sciences Building, University of California, Berkeley, California 94720. kinship and behaviour among individuals whose kin relationships are known. Such data exist for proximity, grooming, play, and interventions in rhesus macaques (M. mulatta) (Sade 1965, 1972a; Kaplan 1977, 1978; Berman 1978), for grooming in Japanese macaques (Yamada 1963), and for interventions in other macaques (Massey 1977; De Waal 1977). This evidence is inadequate bebecause, except for Yamada's study (1963), it considers only mean relationships, whereas the validity of inferences depends greatly on variance within relationship classes. The means of two groups may differ, but if there is sufficient overlap between the groups it may be impossible to determine to which group a particular case belongs (see Yamada 1963). Primates generally are long-lived and bear young infrequently. Yellow baboon (P. cynocephalus) females, for example, may live at least as long as 15 years, and probably considerably longer. They bear their first infant at about 6.5 years of age, and the interval between births of viable offspring is about 22 months (J. Altmann et al. 1977). Genealogies based on known relationships thus take many years to establish. It is important to establish the validity of kinship inferences, since the information obtained from them is so valuable. Two other factors make this an especially urgent task. First, the effects of kinship on behaviour are currently of much interest. Lacking evidence of the validity of kinship inferences, studies that use such inferences when examining