Designing screening protocols for amphibian disease that account for imperfect and variable capture rates of individuals. Issue 5 (1st July 2014)

• Record Type: Journal Article
• Authors: Canessa, Stefano; Martel, An; Pasmans, Frank
• Is Part Of: Ecological applications. Volume 24:Issue 5(2014)
• Page Start: 1204
• Page End: 1212
• Publication Date: 2014-07-01
• Subjects: Batrachochytrium dendrobatidis -- Bayesian statistics -- binomial model -- Bombina variegata pachypus -- chytridiomycosis -- detectability -- Jolly-Seber model -- Liguria -- mark–recapture -- monitoring -- predictive models -- surveillance; Ecology -- Periodicals; Environmental protection -- Periodicals; Biology, Economic -- Periodicals; 577.05
• DOI: 10.1890/13-0103.1
• Languages: English
• ISSNs: 1051-0761

Abstract

Abstract : The amphibian chytrid fungus, Batrachochytrium dendrobatidis, is one of the main factors in global amphibian decline. Accurate knowledge of its presence and prevalence in an area is needed to trigger conservation actions. However, imperfect capture rates determine the number of individuals caught and tested during field surveys, and contribute to the uncertainty surrounding estimates of prevalence. Screening programs should be planned with the objective of minimizing such uncertainty. We show how this can be achieved by using predictive models that incorporate information about population size and capture rates. Using as a case study an existing screening program for three populations of the yellow‐bellied toad ( Bombina variegata pachypus ) in northern Italy, we sought to quantify the effect of seasonal variation in individual capture rates on the uncertainty surrounding estimates of chytrid prevalence. We obtained estimates of population size and capture rates from mark–recapture data, and found wide seasonal variation in the individual recapture rates. We then incorporated this information in a binomial model to predict the estimates of prevalence that would be obtained by sampling at different times in the season, assuming no infected individuals were found. Sampling during the period of maximum capture probability was predicted to decrease upper 95% credible intervals by a maximum of 36%, compared with least suitable periods, with greater gains when using uninformative priors. We evaluated model predictions by comparing them with the results of screening surveys in 2012. The observed results closely matched the predicted figures for all populations, suggesting that this method can be reliably used to maximize the sampling size of surveillance programs, thus improving their efficiency.