Population-size estimators and unequal catchability in painted turtles

Most methods of estimating population size from mark-recapture data assume equal catchability. Failure to meet this assumption may have profound effects on population-size estimates. We used 3 sampling methods to compare population-size estimates derived from Petersen, Schumacher and Eschmeyer, and Jolly-Seber models with the true size of a closed population of painted turtles (Chrysemys picta) in Algonquin Park, Ontario. We found significant variation in capture probabilities, and almost all population-size estimates were far below the true population size. To try to improve the accuracy of the estimates, we applied 4 techniques commonly recommended for reducing bias when catchability is unequal: (i) changing sampling methods, (ii) using several sampling methods simultaneously, (iii) dividing the population by sex, and (iv) calculating population sizes using the computer program CAPTURE. None of the 4 methods reduced the error that resulted from unequal catchability in any of the estimates sufficiently for these methods to be suitable for management of populations or for ecological research.