Object and tool manipulation diversity in an urban capuchin monkey (Sapajus libidinosus) group in Brasília National park

Abstract Capuchin monkeys are highly skilled primates recognized for their diverse object manipulation, a crucial evolutionary trait. This study explores the influence of the urban, artificial setting of Brasília National Park (BNP)on capuchin monkeys’ object manipulation behaviors, aiming to offer more systematic data on this activity within urban Cerrado environments. Object manipulation data were systematically gathered using the “all occurrences” method and video analysis, examining the object, action, and target of each manipulation, along with contextual elements such as date, actor, time, function, type, and food acquisition. Tool use involves modifying an object, substrate, food, another individual, or the self; proto-tool use consists of hitting an object or food against a fixed substrate; simple manipulation refers to handling or carrying a single object without contact. Statistical tests conducted in SPSS 26.0 and R console, including GLMM, binary logistic regression, and Spearman correlation, evaluated individual differences in manipulation involvement, predicted acquisition rates, and assessed links between simpler manipulations and tool use, excluding unclear actor data from statistical but not descriptive analysis. The artificial environment of BNP notably influenced the monkeys’ manipulation behaviors, introducing novel anthropogenic materials and altering their techniques relative to natural populations. While overall manipulation frequency varied by sex and age (with males and juveniles manipulating more), specific manipulation types did not show such variation, and tool use correlated with simpler manipulations, supporting a learning-based progression Unexpectedly, tool use yielded lower rates of food acquisition than other manipulation types. Furthermore, food acquisition through tool use was more frequent among males, possibly attributable to the dominant male’s lack of tool use, which did not provide a reliable model for this behavior.