Baboon travel progressions as a "social spandrel" in collective animal behaviour

Abstract

How individuals in a group move relative to one another can influence both their survival and fitness. Spatial positioning has been well studied in baboons (Papio spp.), which travel collectively in line formations or "progressions." Early studies of baboon progressions presented contradictory findings on the progressions' order - some reporting random positioning of individuals, while others reporting non-random positioning, thought to protect more vulnerable group-members. Here, we revisit this topic and use high-resolution GPS tracking data to study travel progressions in a group of chacma baboon (Papio ursinus) on Cape Peninsula, South Africa. We identify 78 progressions over 36 d and find that progression orders are not random. We test four non-exclusive hypotheses to explain progression orders: vulnerable individuals position themselves in the middle (risk hypothesis), subordinate individuals position themselves at the front to gain better access to resources (competition hypothesis), dominant individuals assume leading positions (group decision-making hypothesis), or progression order is an emergent outcome of underlying social bonds (social spandrel hypothesis). We find no evidence that progression orders are adaptive responses to minimize an individuals' risk, maximize their resource acquisition, or are the result of decision-makers leading the group. Instead, we find that individuals' positions are predicted by pairwise affiliations, resulting in consistency in order, with more dominant individuals occupying central positions in progressions. This non-random structuring of individuals during progressions can be considered a side-effect or outcome of underlying social forces acting among individuals, providing an example of a "social spandrel" in collective animal behaviour.

Keywords: chacma baboons; collective behaviour; group progressions; social dominance; spatial positioning.

Fig. 1.

A) Visualization of the timing of group progressions (red rectangles) through each day, over the polarization of the group at each second. B) Individual trajectories during one example day (8th August 2018), with colours indicating the identified group progressions. C) Frequency map of the relative position of all individuals to the group centroid (0,0) across all group progressions. The group’s direction of movement is indicated by the positive values of the x axis. D) Distribution of the times during which a progression was identified across all days (red histogram). The coloured lines (and shading) are fitted moving averages and confidence intervals of speed, polarization, and elongation. All metrics are higher during the evening group progressions, corresponding to when the group travels back to the resting site (Figure S5).

Fig. 2.

A) Distribution of individual spatial position across all progressions. Each distribution describes one of the 13 individuals, coloured by dominance rank where 1 = highest ranking. B) Correlation between individuals’ daytime and evening progression rank of the spatial positions. For example, “1st” corresponds to an individual being, on average, the most frontal individual. Diagonal line corresponds to individuals not changing their rank.

Fig. 3.

A) Baboon spatial associations define order during group progressions. Every data point shows the average interindividual distances for all dyads of individuals during group progressions and the rest of the day. B) Correlation of dominance and individual eigenvector centrality during group progressions (both during evening and daytime). Higher centrality corresponds to individuals that are on average closer to, and closer to more, other individuals. C) Correlation between individual dominance rank and standard deviation of the individual spatial positions. D) Variation of nearest neighbours during group progressions (both during evening and daytime). Consistency is measured as the per second probability that an individual keeps a specific individual as its nearest neighbour. Standard errors are shown as vertical lines. Despite the effect size appears low, the probabilities are measured on a time scale of one second, resulting in a big effect considering the duration of an entire progression.