Thursday, August 7, 2008 - 4:20 PM

COS 92-9: Extinction in experimental populations: Effects of habitat quality, size, and metapopulation configuration

John M. Drake and Blaine D Griffen. University of Georgia


Stochastic population theory makes clear predictions about the effects of reproductive potential, carrying capacity, and metapopulation structure on characteristic time scales of extinction. At the same time, the effects of habitat size and habitat quality on reproduction and regulation have been debated. We performed two experiments with replicate Daphnia magna populations to trace the causal relationships among these factors. First, we looked at the effects of habitat size and quality on extinction time under crossed treatments of habitat size (2 levels) and habitat quality (food quantity; 3 levels). Second, we looked at effects of the configuration of source and sink patches in crossed treatments contrasting source-sink structure with homogeneous patch structure (3 levels), overall metapopulation size (2 levels), migration rate among subpopulations (2 levels), and the degree of metapopulation subdivision (3 levels).


For the first experiment, we used statistically derived estimates of key parameters to relate experimental treatments to persistence time through their effects on carrying capacity and the population's intrinsic rate of increase. We found that carrying capacity and the intrinsic rate of increase were each influenced similarly by habitat size and quality and that the estimated effect of the intrinsic rate of increase on population persistence exceeded the estimated effect of carrying capacity by a factor of three, expressed as the relative effect of a doubling of habitat quality or a doubling of habitat size. We expected habitat quality to have a greater influence on extinction. However, because of an unexpected effect of habitat size on reproductive potential, habitat size and quality were similarly important for population persistence. Preliminary results from the second experiment suggest that the source-sink versus homogeneous patch structure is by far the most important predictor of persistence (p=<0.0001; relative importance=0.66). Estimated effects of total metapopulation size and subdivision differ depending on statistical model, though both are detectable. Effects of the migration treatment are negligible. In this system, configuration of habitats and resources is evidently more important than migration rate.