Evolution of dispersal rates in an experimental metapopulation of confused flour beetles (Tribolium confusum)

Background/Question/Methods

Spatial structure has long been considered a key part of population ecology. This is especially true when analyzing metapopulations, where populations are assumed to reside in discrete patches of suitable habitat connected by dispersal. Due to occasional extinction of individual patches, typically summarized as catastrophe, recolonization by dispersal is essential in order to avoid the ultimate extinction of a metapopulation. The American pikas (Ochotona princeps) residing in the ore dumps of Bodie, California are among the most studied metapopulations, resulting in one of the strongest ecological data sets available. However, models of the Bodie site have relied on rough estimates for dispersal parameters, due to the rarity of O. princeps’s natal dispersal. To inform future models, we have developed a laboratory metapopulation of confused flour beetles (Tribolium confusum), designed to behave as an experimental analog for field-metapopulation systems. Our structure comprises of a central chamber connected to four outer chambers by small tubes. Migration tests have assisted in determining dispersal behavior in the beetles and the development a computational model of dispersal habits.

Results/Conclusions

Here we show our experimentally verified model of metapopulation dynamics in T. confusum. Our laboratory studies have shown that dispersal in T. confusum is innate, in that beetles actively seek out new environments as opposed to incidentally blundering into new patches, and that dispersal is linear with respect to density. Given this information, we have developed a generalized ordinary differential equation to describe dispersal behavior. The discretized form of this equation has been combined with Dennis et al’s LPA (larvae, pupae, adult) model for Tribolium spp. population dynamics. Long term surveys of our T. confusum populations within metapopulation systems have shown this model to be accurate in predicting population behavior over several months. The model predictably shows that, given our static laboratory environment, a dispersal rate of zero (no dispersal) should be evolutionarily preferred, despite the high dispersal rates witnessed in our T. confusum populations. However, the addition of stochastically determined extinction events reverses this evolutionary preference.