Reprinted with permission, from C. Busse and W. J. Hamilton III, “Infant Carrying by Male Chacma Baboons,” SCIENCE 212:1281-83 (1981) Copyright © 1981 AAAS

A frequently observed but poorly understood feature of savanna baboon (Papio spp.) societies is that an adult male sometimes carries an infant during interactions with another male (1–3). Explanations of this behavior, originally termed agonistic buffering (4), are that by carrying an infant a male reduces his probability of being attacked (1–3) or increases his probability of gaining access to an estrous female (2) or food (2, 3). The assumption is that the carrying male gains an advantage by exploiting an infant who is probably the opponent's offspring (2). In this report we present the opposite interpretation: that the infant is probably the carrying male's offspring and that carrying protects the infant from potentially infanticidal immigrants. Thus, we identify infant carrying as a form of paternal care. This interpretation provides a new link in the understanding of paternal investment (5) and infanticide (6) in polygamous primate groups.

During a 3-year study of chacma baboons (Papio ursinus) in the Moremi Wildlife Reserve, Botswana, we saw adult males kill only two infants and wound two others (7). Three of the attackers were immigrant males; the fourth was unidentified. [Male baboons have also attacked infants at other study sites (3, 6, 8).] Infanticide may be infrequent because resident troop members protect infants from immigrant males (6, 9).

Our analysis of infant carrying by males focuses on the infant's paternity. Males compete to mate with females, and dominant males sire most of the infants in a troop (10). This conclusion is based on field observations of mating patterns 3 days before the female's estrous swelling begins deflating (“D – 3”). Studies of reproduction in captive baboons suggest that matings on D – 3 have the highest probability of resulting in conception (11). Another factor influencing paternity is that males leave one troop and join another at least once in their lifetimes(9): a male joining a troop after an infant's conception cannot be the father.

The effect of infant death on a mother's reproductive behavior (12) is another important aspect in this analysis. Baboon infants are weaned at about 12 months of age, and mothers conceive 18 months (532 ± 23 days, N = 13) after the birth of an infant who survives to weaning. However, if an infant dies before weaning, the mother becomes sexually receptive after a few weeks and conceives 4 months (134 ± 19 days, N = 8) after the death. Thus, by killing an unweaned infant, a male enhances his opportunity to impregnate the mother earlier than he could otherwise (6, 13).

Top-ranking, or alpha, immigrant males may benefit most from killing infants because they have the highest probability of subsequently impregnating the mother. The median tenure of males at alpha rank is 5 months (range, 1 to 12 months, N = 9). By killing an infant, an alpha male substantially increases the probability that when the mother becomes sexually receptive again he will still be alpha.

We observed two chacma baboon troops in Moremi between September 1979 and June 1980 and recorded 112 interactions in which infants were carried. The troops were designated C and W and averaged 70 and 72 members, including nine and six adult males, respectively. The average number of interactions per observation hour per troop was 0.4. The median duration of these interactions was 10 seconds (range, 1 to > 360 seconds). One interaction escalated into a fight, with canine fencing between the males, but there were no injuries during any interaction (3).

Evidence that males carried their own offspring is as follows: (i) nine of ten carrying males were troop members when the carried infant was conceived, and these males carried infants in all but one of the 112 interactions (Table 1) (3); (ii) the nine males had had significantly higher than average social rank when the infants they carried were conceived (P < .05, sign test, N = 9); and (iii) we identified the probable father (based on D – 3 observations) of ten of the 25 infants carried. All of these males subsequently carried their probable offspring during interactions with other males.

The opposite pattern prevailed for the opponents of carrying males. Nine of the 12 opponents observed immigrated after the carried infant's conception (Table 2) (1, 3). Of the 112 interactions, three involved opponents who were troop members at the time of the infant's conception. Two of these opponents, WSM and CJT, were low-ranking subadults at the time of conception and unlikely to be fathers. Thus, all the opponents, whether present or absent at conception of the carried infants, were probably not the fathers.

In 28 of 29 carrying male-opponent combinations, the opponent was dominant to the carrying male (P < .001, sign test) (14). Carrying males held high rank at the time of an infant's conception, but had dropped significantly in rank by the time they carried the infant (P < .05, Wilcoxon signed-ranks test, N = 9). Opponents were high-ranking nonfathers (Table 2), that is, were most likely to benefit from killing the infants. These results support the interpretation that carrying males are protecting their own offspring rather than exploiting unrelated infants.

Why should carrying an infant be an effective means of protecting it against a dominant male? We suggest that infant carrying communicates the willingness of fathers to protect their progeny regardless of the opponent's rank. In these confrontations the payoff to the carrying male, the continued survival of an offspring, is greater than the payoff to the opponent, a higher probability of siring an infant in the future.

Other features support our interpretation. (i) Unweaned infants less than a year old were carried in 101 of 112 interactions (1, 3). Unweaned infants may be at the greatest risk of attack by immigrant males (13). (ii) In 13 interactions the infants elicited protective carrying by approaching and threatening immigrant males (3). In 18 interactions the infants initiated contact with protective males and in 12 interactions the latter initiated contact. (iii) Baboon mothers rarely interfere with males who carry their infants, and are constant associates of these males (8). In only one interaction did a mother attempt to retrieve her infant before the interaction ended. Females observe the interactions from a distance, usually 5 m or more, and are available to retrieve infants when the interactions are completed. (iv) Mothers carrying unweaned infants show alarm by raising their tails when in close proximity to immigrant males (15). Sometimes they scream when closely approached by immigrants (9), which may attract protective males. This antagonism by mothers and infants toward immigrants suggests that these males are a threat to the lives of the infants.

Contests over food resources or estrous females were not a general feature of infant carrying in our population. Opponents were in consort with estrous females during nine interactions, and carrying males never gained access to these females. Also, carrying males sometimes spatially displaced their opponents (16), but only once gained access to food. These results further support rejection of the infant exploitation hypothesis.

In a recent study of infant carrying by male baboons (P. anubis) at Gombe, Tanzania, Packer (3) concluded that alternative explanations of this behavior are needed. Observations at Gombe (1, 3) are consistent with our interpretation, which thus may also apply to other savanna baboon populations.