Predicting endangered species recovery objectives using biological traits and patterns of decline

Background/Question/Methods

Species listed under the U.S. Endangered Species Act are required to have quantitative recovery objectives stated in their recovery plans for determining when they can be delisted. Using recursive partitioning methods (classification and regression trees and random forests) on a comprehensive dataset of 642 endangered and threatened plant species, we tested whether recovery objectives for numbers of individuals or populations can be predicted by biological traits, previous abundances, or a combination of traits and previous abundances. We also examined relationships with listing status and year of recovery plan writing.

Results/Conclusions

Previous abundances alone were relatively good predictors of the population-based recovery objectives: species with fewer previous populations required fewer populations but a greater proportion of historical populations for delisting. Previous abundances also predicted individual-based delisting objectives when the models included both abundances and biological traits, and physiographic division was also an important predictor. Overall, our results suggest that managers are relying on previous abundances and patterns of decline as guidelines for setting recovery objectives. This may be justifiable in that previous abundances inform managers of the effects of both intrinsic biological traits and extrinsic threats that interact and determine extinction risk.