Analysis of infant handling and the effects of female rank among Tana River adult female yellow baboons (Papio cynocephalus cynocephalus) using permutation/randomization tests.

Abstract

Infant handling has been documented in numerous species. Among cercopithecines, interaction motivations are reported to range from aunting to kidnapping; these interactions are often distressful for both mother and infant. Here we examine handling by adult female yellow baboons (Papio cynocephalus cynocephalus) at the Tana River National Primate Reserve, Kenya, using a relatively new, computer-intensive statistical approach of permutation/randomization tests to deal with repeated measures effects and a skewed sample. We hypothesized 1) a tendency for handlers to handle the infants of females ranked similarly or lower than themselves, and 2) more successful infant handling by higher-ranked females, particularly with very young infants. We collected focal data on 23 females (11 mother-infant pairs) over an 11-mo period, with a total of 303 attempted and/or successful "handles" utilized in the permutation analyses. The general patterns apparent in the data seemed to support our hypotheses. However, the permutation tests showed that while females are somewhat more likely to attempt to handle the infants of females ranked "same or lower" than themselves, lower-ranked females are able to prevent more than three-fourths of the attempted interactions, and there is no statistically significant trend for females to successfully handle these infants. Further refinement of the analyses showed no significant tendencies for females to handle those infants ranked "lower" or "immediately lower" than themselves, casting doubt on the significant finding for "same or lower" attempts. Further, there was no significant effect for higher-ranked females to successfully handle an infant during its first month. Thus, rank does not seem to offer any privileges in terms of handling an infant in this population. We believe the permutation tests are an effective way to analyze repeated measures data and offer a more sensitive analysis tool for determining true significance.