Dunbar (1992, Behav. Ecol. Sociobiol., 33, 35-49) argued that constraints on social time limited the size to which savannah baboon, Papio cynocephalus, troops in any given population could grow before fissioning. Since this should be reflected in population structure, we have elsewhere (Henzi et al. 1997a, Anim. Behav. 53, 525-535) constructed a model, based on a rising probability of fission, that fits the observed distribution of troop sizes of mountain baboons, P. c. ursinus, in the Drakensberg mountains of South Africa and which predicts that the probability of fission will rapidly increase once a troop has more than 23 members (or 8.7 females). We test this prediction in this paper. Since Dunbar argued that females will drive fission once they cannot engage in the grooming necessary to sustain alliances, we compared the grooming interactions of adult females from four troops in the Drakensberg mountains. The mean female grooming clique size reached an asymptote at 7.4 females, so that females in cohorts of eight or more no longer attempted to groom all other females, and mean grooming bout length declined as the cohort grew to 7.9 females and then increased again. These values are coincident with the female cohort size predicted by our model of troop growth and fission. We argue that females attempt to groom all other females as well as sustain closer relationships with a few females through longer bouts of reciprocated grooming. When the demands of grooming all other females reduce bout length to a point when no reciprocated bouts are possible, female clique size is capped. As a troop continues to grow, the mechanical difficulties involved in gaining access to grooming partners leads to a reduction in the diversity of grooming relationships. This weakening of the total female network, as cliques become more differentiated, is likely to facilitate fission. We conclude that our data provide the first within-population validation of Dunbar's hypothesis concerning the mechanism underpinning fission. In the Drakensberg, where there is no advantage to female coalitions, we propose, as an amendment, that females will leave a troop not to escape local competition, but to follow a male with whom they have a close friendship. (C) 1997 The Association for the Study of Animal Behaviour.
