Estimating Ages of Turtles from Growth Data

Abstract

Abstract Age estimation is important for management of turtle populations, but techniques such as growth ring counts or skeletochronology may be unreliable or impossible to perform. An alternative is to estimate age from growth models using Bayesian inference. However, individual variation in growth parameters needs to be incorporated into these models for them to generate realistic prediction intervals. For long-lived ectotherms such as chelonians, it is also important that models allow for changes in growth at sexual maturity, and that the growth models are combined with prior distributions reflecting realistic age structures. We describe how a hierarchical biphasic growth model fitted to a long-term data set of carapace length measurements for North American snapping turtles was combined with prior age distributions generated from survival estimates for the same population. The model was used to generate individual posterior age distributions for turtles captured on 2 or more occasions, and also for hy...