Using logistic regression to analyze the sensitivity of PVA models: a comparison of methods based on African wild dog models

We used logistic regression as a method of sensitivity analysis for a stoch astic population viability analysis model of African wild dogs (Lycaon pict us) and compared these results with conventional sensitivity analyses of st ochastic and deterministic models. Standardized coefficients from the logis tic regression analyses indicated that pup survival explained the most vari ability in the probability of extinction, regardless of whether or not the model incorporated density dependence. Adult survival and the standard devi ation of pup survival were the next most important parameters in density-de pendent simulations, whereas the severity and probability of catastrophe we re more important during density-independent simulations. The inclusion of density dependence decreased the probability of extinction, but neither the abruptness nor the inclusion of density dependence were important model pa rameters. Results of both relative sensitivity analyses that altered each p arameter by 10% of its range and life-stage-simulation analyses of determin istic matrix models supported the logistic regression results, indicating t hat pup survival and its variation were more important than other parameter s. But both conventional sensitivity analysis of the stochastic model which changed each parameter by 10% of its mean value and elasticity analyses in dicated that adult survival was more important than pup survival. We evalua ted the advantages and disadvantages of using logistic regression to analyz e the sensitivity of stochastic population viability models and conclude th at it is a powerful method because it can address interactions among input parameters and can incorporate the range of parameter variability, although the standardized regression coefficients are not comparable between studie s. Model structure, method of analysis, and parameter uncertainty affect th e conclusions of sensitivity analyses. Therefore, rigorous model exploratio n and analysis should be conducted to understand model behavior and managem ent implications.