The importance of individual variation in the dynamics of animal collective movements

Abstract

Animal collective movements are a key example of a system that links two clearly defined levels of organization: the individual and the group. Most models investigating collective movements have generated coherent collective behaviours without the inclusion of individual variability. However, new individual-based models, together with emerging empirical information, emphasize that within-group heterogeneity may strongly influence collective movement behaviour. Here we (i) review the empirical evidence for individual variation in animal collective movements, (ii) explore how theoretical investigations have represented individual heterogeneity when modelling collective movements and (iii) present a model to show how within-group heterogeneity influences the collective properties of a group. Our review underscores the need to consider variability at the level of the individual to improve our understanding of how individual decision rules lead to emergent movement patterns, and also to yield better quantitative predictions of collective behaviour.This article is part of the theme issue 'Collective movement ecology'.

Keywords: Lagrangian models; aggregation; behavioural rules; collective motion; context-dependent factors.

Figure 1.

Example snapshots of simulations. Locations (grey dots) and velocity vectors (dark arrows) for each individual are presented across four time points (left to right). In the upper panels (a), the parameterization corresponds to a swarming scenario, where the individuals are spatially coherent but oriented in random directions (note, due to toroidal boundary conditions, what appears to be two groups in the top-rightmost panel is actually one group). The top row of panels shows the homogeneous case, while the second row shows what happens when heterogeneity in all parameters is introduced: looser aggregations and more isolated individuals. The lower panels (b) show the aligned parameterization, where (in the homogeneous case, third row) there is not only spatial coherence but also similar alignment across individuals, and in the heterogeneous case (bottom row) the grouping is much more diffuse and no longer aligned.

Figure 2.

Collective movement statistics presented against increasing heterogeneity in sociability (coefficient of variation in parameter α, ranging from 0 to 1). Each point represents an average across 10 snapshots in 20 scenarios against 36 values of parameters r_l and s (7200 simulations per summary point). Darker and lighter lines show result from the swarming and aligned scenarios, respectively. Circles and solid lines represent simulations with 100 individuals, triangles and dashed lines represent simulations with 30 individuals. Vertical bars are the standard errors around the means. The collective movement metrics are related to distance (left panels): nearest neighbour distance (a) mean and (b) standard deviation, (c) number of isolates (i.e. total number of individuals at a distance greater than r_l); related to velocities (middle panels), with (d) mean and (e) standard deviation of speed, and (f) the alignment coefficient which ranges from 0 (total randomness) to 1 (perfect alignment of all individuals); and group size (g) mean and (h) standard deviation.