In pratikunterwegs/klepto-move-evol: R code for the Kleptomove Simulation

Effect of Local Dispersal

In order to focus on adaptive movement strategies, we chose to implement large dispersal distances in our default simulation setup, which we refer to as 'global' natal dispersal. Under global dispersal, offspring are homogeneously distributed over the entire landscape (dispersal radius = 512). Our results are not changed in any way when dispersal is much more strongly localised, which we refer to as simply 'local' natal dispersal. In this implementation, the natal dispersal distance is comparable in magnitude as the distance between resource peaks. If offspring dispersal is more local, the spatial population dynamics may become more intricate, and kin competition or local adaptation may become influential. We therefore ran the simulations presented in the main text also under local dispersal (dispersal radius = 2).

In summary, scenarios 1 and 3 yield similar results under local as under global dispersal, while scenario 2 shows some interesting dynamics typical of reaction-diffusion systems. In scenario 1 (see Fig. S13), the resource landscape plots A, the activity budget and intake plots B and C, as well as the evolved movement strategies E exactly match the simulation results shown in Figure 1 of the main text. Only the correlations between number of foragers and cell productivity are higher under local dispersal than under global dispersal (panel D). This is a straightforward consequence of local dispersal, where individuals occurring on more productive cells have a higher intake rate and therefore produce more offspring than individuals on less productive cells. Thus, under local dispersal many agents already start out on more productive cells. This does not seem to impact movement strategies. The same is true for scenario 3 (Fig. S14): After the initial depletion of the landscape, kleptoparasitic behavior spreads, and the landscape is somewhat replenished again. Also here, the landscape snapshots, the activity budget, as well as the intake plot and the evolved movement strategies match the global dispersal case. The difference in competition strategy (panel F) corresponds to the observed bistability (compare Main Text Fig. 6). Again, the correlation between number of foragers and cell productivity is higher under local dispersal than under global dispersal, in the latter averaging in late generations around 0.1, and in the former around 0.2.

Scenario 2 is the only one where we observed a marked difference between local and global dispersal (see Fig. S15). As soon as kleptoparasites occur, they spread and become locally abundant, driving foragers to local extinction. The kleptoparasites themselves then wither away due to a lack of foragers to steal from, after which foragers may colonize the area once again. This spatial instability repeats itself over wide parts of the landscape, driven by the extinction, recolonization and diffusion of foragers and kleptoparasites. Kleptoparasites and foragers here effectively form a reaction-diffusion system. Snapshots of this dynamic pattern can be seen in Fig. S15A. As a consequence, the proportions of kleptoparasites and foragers, as well as the total per capita intake of the population fluctuate widely (panels B and C). The correlations between individual densities and cell quality lie around zero and are therefore not much different from the results observed under global dispersal (Main Text Fig. 2D). An interesting contrast with global dispersal is to be found in the movement strategies. While kleptoparasites have similar preferences under global and local dispersal, foragers have much stronger item preferences under local dispersal. Due to the pattern of extinction and recolonization under local dispersal, there are parts of the landscape not only rich in food items, but also free from kleptoparasites, and thus a strong preference for items becomes beneficial.