Fig. 9: Agent-based modelling of individual movement.

a All simulation trajectories were based on the correlated random walk (CRW) model. A CRW trajectory is constructed from a sequence of vectors \(\overrightarrow{{v}_{t}}\), produced by randomly drawing a step length L and a turn angle θ from an exponential and a wrapped normal distribution, respectively. b The upper map represents the nest grid upon which trajectories are simulated. The nest includes two overlapping ‘spatial modules’ (central and peripheral), characterised by a ‘distance score’ increasing non-linearly with the distance of each site to the nest centre. To model the three movement mechanisms, the CRW is modified as follows. In the focal-point attraction model (point I in map), the CRW vector \(\overrightarrow{{v}_{CRW}}\) is modified by adding a bias vector \(\overrightarrow{{v}_{bias}}\), which points towards the closest point on the border of the individual’s primary module (magenta cell). In the boundary effect model (points II–III), the CRW is modified in the vicinity of the module boundary by increasing (decreasing) the CRW turn angle by σ when approching the boundary from the inside (outside). In the locomotion adjustment model, the L and θ distributions are modified according to location, so individuals make shorter (larger) steps and larger (smaller) turns when inside (outside) their primary module (points IV & V, green & orange distributions, respectively). c Comparison between empirical and simulated trajectories. Grouped columns indicate the five movement metrics described in the main text, and letters below individual columns give species name abbreviations. Formulas provide a brief description of the statistical model conducted for each movement metric, and cell colours encode the sign and value of the coefficient of the predictor in the statistical model (see text for the model definitions). To facilitate comparisons within metrics, the coefficients are normalised by the maximum absolute value of all coefficients for each metric. Source data are provided as a source data file.