Understanding the mechanisms that contribute to local and regional extinction of species is an essential goal of conservation biology and ecology. In particular, a critical question is whether extinction processes are primarily selective and based on species traits, or are primarily stochastic and therefore based strictly on species abundance. Previous studies have found that invasive predators, such as fish, cause selective local and regional extinctions, leading to homogenization of aquatic diversity. These past studies, however, have not taken into account the potentially critical role that variation in species abundance may play in the local extinction process, therefore underestimating the importance of random probability in the extinction process. To determine the importance of abundance and stochastic processes in species’ extinction risk, we used a null model approach to analyze local macroinvertebrate extinctions from a mesocosm experiment in which predatory green sunfish (Lepomis cyanellus) were added. Mescosms were surveyed for macroinvertebrate richness and abundance prior to, three weeks after, and six weeks after the addition of sunfish. Our null model follows a rarefaction approach that assumes species loss is determined by the random removal of individuals following a disturbance.
Our preliminary results provide evidence for both mechanisms of extinction. We compared the expected number of extinctions based on our rarefaction simulation to the observed number of extinctions after both time periods. Plots of observed extinctions vs. expected extinctions demonstrate that in general, more extinctions occurred in treatment tanks than expected by the null model. However, linear regression analysis indicates that null-expected stochastic extinction is a significant predictor of observed extinctions in the first post-treatment survey (p = 0.039, r2 = 0.27). This pattern was weaker and non-significant in the regression analysis of the second post-treatment survey (p = 0.09). These results suggest that abundance is a primary determinant of species’ extinction risk and that local extinctions following a disturbance may be driven by stochastic processes that are masked by additional selective extinctions that would not be predicted based on abundance alone. Although the importance of abundance to extinction risk has long been recognized, we suggest that it is essential to explicitly account for stochastic extinction processes prior to inferring selective or deterministic extinction processes.