Forensic Application of Stable Isotopes to Distinguish between Wild and Captive Turtles.

Wildlife traffickers often claim that confiscated animals were captive-bred rather than wild-caught to launder wild animals and escape prosecution. We used stable isotopes (δ13C and δ15N) derived from the claw tips of wild wood turtles from Maine and captive wood turtles throughout the eastern U.S. to develop a predictive model used to classify confiscated wood turtles as wild or captive. We found that the claw tips of wild and captive wood turtles (Glyptemys insculpta) were isotopically distinct. Captive turtles had significantly higher δ13C and δ15N values than wild turtles. Our model correctly classified all wild turtles as wild (100%) and nearly all captive turtles as captive (94%). All but two of the 71 turtles tested were successfully predicted as wild or captive (97.2% accuracy), yielding a misclassification rate of 2.8%. In addition to our model being useful to law enforcement in Maine, we aim to develop a multi-species model to assist conservation law enforcement efforts to curb illegal turtle trafficking from locations across the eastern United States and Canada.

Evaluating the effect of expert elicitation techniques on population status assessment in the face of large uncertainty.

Population projection models are important tools for conservation and management. They are often used for population status assessments, for threat analyses, and to predict the consequences of conservation actions. Although conservation decisions should be informed by science, critical decisions are often made with very little information to support decision-making. Conversely, postponing decisions until better information is available may reduce the benefit of a conservation decision. When empirical data are limited or lacking, expert elicitation can be used to supplement existing data and inform model parameter estimates. The use of rigorous techniques for expert elicitation that account for uncertainty can improve the quality of the expert elicited values and therefore the accuracy of the projection models. One recurring challenge for summarizing expert elicited values is how to aggregate them. Here, we illustrate a process for population status assessment using a combination of expert elicitation and data from the ecological literature. We discuss the importance of considering various aggregation techniques, and illustrate this process using matrix population models for the wood turtle (Glyptemys insculpta) to assist U.S. Fish and Wildlife Service decision-makers with their Species Status Assessment. We compare estimates of population growth using data from the ecological literature and four alternative aggregation techniques for the expert-elicited values. The estimate of population growth rate based on estimates from the literature (λmean = 0.952, 95% CI: 0.87-1.01) could not be used to unequivocally reject the hypotheses of a rapidly declining population nor the hypothesis of a stable, or even slightly growing population, whereas our results for the expert-elicited estimates supported the hypothesis that the wood turtle population will decline over time. Our results showed that the aggregation techniques used had an impact on model estimates, suggesting that the choice of techniques should be carefully considered. We discuss the benefits and limitations associated with each method and their relevance to the population status assessment. We note a difference in the temporal scope or inference between the literature-based estimates that provided insights about historical changes, whereas the expert-based estimates were forward looking. Therefore, conducting an expert-elicitation in addition to using parameter estimates from the literature improved our understanding of our species of interest.