Why do species go extinct? An experimental test of the contribution of rarity to extinction risk

Background/Question/Methods

It is generally understood that rare species are at greater risk of extinction than more common species.  Analyses of past extinctions support this understanding, since endemic species and species with low local abundances have been shown to have high rates of extinction in response to disturbances, as compared to more widespread species.  Two possible explanations exist for this pattern.  First, extinctions of rare species may occur because rare species have a higher probability of going extinct owing to probability alone (i.e., stochastic extinction).  Alternatively, rare species extinctions may occur because the species are particularly susceptible to the disturbance in question.  Thus, when extinctions of rare species occur, it can be difficult to disentangle selective (e.g., trait-associated) from nonselective (stochastic) extinction mechanisms in observation or retrospective studies of extinction.  We addressed this particular issue in an outdoor aquatic mesocosm experiment in which we manipulated the occupancy and abundance of rare species by augmenting their populations.  We then introduced a disturbance in the form of an invasive predator (green sunfish).  By comparing the effects of green sunfish in augmented and non-augmented treatments, we attempted to disentangle the contributions of abundance, occupancy, and species identity to the extinction probability of aquatic invertebrates.

Results/Conclusions

We based our analyses on a simulation model of random extinction, which allows us to determine if extinction may have occurred for strictly stochastic reasons related to low abundance or occupancy, versus extinctions which were unlikely to have occurred as a random event.  We interpret the latter group of extinctions to be selective.  Of the 16 focal species of aquatic invertebrate in this study, 50% went extinct after fish were added to the non-augmented treatment.  Our extinction simulation suggested that none of these extinctions would be unexpected based on a random extinction process acting on species occupancy alone.  In contrast, when we augmented populations of our 16 focal species, only 3 species went extinct after fish were introduced.  Based on our simulation model, two of these three extinctions would be unexpected under a random extinction scenario (p-value of 0.07 and 0.01), suggesting that these two invertebrate species were selectively driven extinct and are disproportionately susceptible to fish predation, despite having had their populations augmented.  Collectively, our results suggest that extinctions in this system are overwhelmingly consistent with a random extinction model based only on species occupancy.