Projecting population cycles from individual life histories: The role of extrinsic and intrinsic factors in rodent population dynamics explored by individual based models

Background/Question/Methods

Rodent population cycles have fascinated ecologists for almost a century; however there is still no unambiguous explanation about what drives these cycles. Currently the literature is dominated by mathematical models stating that one extrinsic factor – namely predation by specialist predators – is sufficient to explain rodent cycles. However, an interaction of an extrinsic factor (predation) with intrinsic factors (e.g. sociality and dispersal) was suggested to lead to the generation of the cycles, but was not formally tested. Here, we tested this hypothesis with an individual-based model fully parameterized with an exceptionally rich empirical database on vole life histories that allowed for a realistic presentation of density-dependence and social mechanisms operating on all demographic processes. We contrasted rival hypotheses about the factors supposedly driving the population cycles using the full factorial design that included models with the following factors: predation only, predation and sociality, predation and dispersal, and predation and both sociality and dispersal. To compare the results of these four models with the cyclic natural vole populations subjected to long term monitoring we used a comprehensive set of metrics: period and amplitude of the cycles, autumn population size and yearly population growth rate.

Results/Conclusions

Only the full model which included both intrinsic factors and predation resulted in a distribution of cycle periods, amplitudes and autumn population sizes closest to those observed in nature. Additionally, all models except for the one including predation only produced multimodal distributions of the yearly population growth rates, a pattern that was observed in half of the inspected long-term vole population time-series. Surprisingly, such an aggregated and commonly used population metric as population growth rate has so far been understudied in the research of vole population dynamics; it merits closer inspection in cyclic populations which may lead to new insights. Our approach allows to model, as emergent properties of individual life-histories, the sort of non-linear density- and phase-dependence in aggregated population parameters that is expected to destabilize population dynamics. We propose that individual-based approach, pending on more detailed data on rodents, will be useful for addressing the effects of other mechanisms operating on finer temporal and spatial scales than have been explored with models so far.