During my nine (all-to-short and very enjoyable!) months at NCEAS, I investigated the predictive accuracy of extinction risk models. This report provides a brief overview of my main research activities.
In collaboration with Hugh Possingham, Drew Tyre and Jeremy Day, I wrote a review of methods for assessing predictive accuracy of PVA (population viability analysis) models, which has been accepted in Conservation Biology. This review encompassed work from a range of disciplines to investigate ways of testing the variation in predicted population sizes, not just the average prediction. In addition to this review, I worked on a number of case studies in which the appropriate tests were applied to actual models. One of these has been accepted for publication (McCarthy, Lindenmayer and Possingham in press).
A simulation study conducted as part of the Extinction Risk Working Group confirmed previous work by others (e.g.,Taylor, Ludwig) that the absolute predictions of PVA models are subject to considerable uncertainty. However, in contrast to assertions by several authors, it appears that the relative accuracy of PVA models (changes in risks of extinction in response to management) are equally uncertain. Despite this uncertainty, the simulations demonstrated that PVA models could still help management decisions by more often than not identifying the best of two possible management strategies. This work is being prepared for journal submission in collaboration with Sandy Andelman and Hugh Possingham.
I also developed a competition for assessing the risk of extinction, in which I built models representing the population dynamics of nine species, and used these to generate survey data. These simulated data are being made available on the NCEAS web site to anyone wishing to enter the competition. The aim of the competition is to use these data to try to predict the risks of population decline of the species. The person making predictions that are closest to those of the original model will be invited to attend an Extinction Risk Working Group meeting at NCEAS to explain their method for assessing risks.
While at NCEAS, I attended the Society for Conservation Biology meeting in Missoula, Montana. At that meeting I presented a talk describing my development of a stochastic density-dependent population model of mountain pygmy possums (Burramys parvus). I obtained analytical solutions of the risks of population decline using this model, and then used Bayesian methods to assess the predictive accuracy. I am finalizing some of the details of this model before preparing it for publication. While at the Missoula meeting, I also helped to coordinate a workshop on methods for assessing conservation status.
In addition to projects related to the Extinction Risk Working Group, I contributed to a study of the effects of toe-clipping on return rate of amphibians (Parris and McCarthy in press), and used stochastic dynamic programming to assess management options for serotinous Banksia species (McCarthy, Possingham and Gill in press). This latter project completed some work I initiated as part of the Managing Uncertainty Working Group that was organized by Shea and Possingham.
Finally, I’d like to thank everyone at NCEAS for providing a stimulating, productive and extremely enjoyable research environment.
NCEAS papers accepted for publication
McCarthy, M.A., Lindenmayer, D.B., and Possingham, H.P. (in press). Assessing spatial PVA models of arboreal marsupials using significance tests and Bayesian statistics. Biological Conservation.
McCarthy, M.A., Possingham, H.P., Day, J.R. and Tyre, A.J. (in press). Testing the accuracy of population viability analyses. Conservation Biology.
McCarthy, M.A., Possingham, H.P., and Gill, A.M. (in press). Determining optimal management of strategies for Banksia ornata using stochastic dynamic programming. Journal of Applied Ecology.
Parris, K.M. and McCarthy, M.A. (in press). Effects of toe-clipping
on anuran return rates. Amphibia reptilia.