Spatial environmental heterogeneity reduces variance in the rate of spread: counterintuitive results from a highly replicated experiment

Background/Question/Methods

Deterministic models predict that spatial heterogeneity in the environment will reduce rates of spread when the spatial scale of dispersal is less than the spatial scale of heterogeneity. However, little is known about how spatial heterogeneity influences the variability or predictability of spatial spread. Using the flour beetle, Tribolium castaneum, as a model organism, we measured rates of spread in laboratory microcosms consisting of discrete patches arranged in linear landscapes under controlled conditions. We contrasted three types of landscapes differing in heterogeneity of food resources: 1) homogeneous, 2) small-scale heterogeneity, 3) large-scale heterogeneity. Each type of landscape was replicated thirty times and populations in each patch were completely censused each generation.

Results/Conclusions

Variance in spread rates between multiple experimental realizations of the spread process was remarkably high, consistent with previous experimental results. Contrary to theoretical models, mean rates of spread were not reduced substantially by heterogeneity in food resources. However, the variance in the rate of spread between experimental realizations was dramatically reduced in heterogeneous landscapes, and more so in the landscapes with small-scale heterogeneity. This result is counter to both intuition and results from simple stochastic models that heterogeneous landscapes should lead to greater heterogeneity in the rate of spread. We conclude that current theory lacks biological processes that have a large influence on stochastic spatial spread in heterogeneous landscapes.