Validation of diagnostic tests in wildlife: the case of chytridiomycosis in wild amphibians

Abstract

We validated the diagnostic tests for the high-profile disease, chytridiomycosis, in wild amphibian populations. We compared histological samples with a Taqman real time quantitative PCR (qPCR) test in five species of frogs at different times of the year at six locations in the wet tropics of northern Queensland. The sensitivity and specificity of each test were estimated using prior estimates from previous laboratory studies and Bayesian methods. The qPCR test was almost three times as sensitive 72.9% (62.7–82.2%) than histology 26.5% (19.9–33.9%) but was less specific 94.2% (89.3–98.6%) than histology 99.5% (98.4–100%), which was likely caused by contamination. Monitoring of the negative control success rate of the qPCR test is potentially a good indicator of specificity. It is likely that using individual gloves for handling amphibians reduces cross-contamination and, therefore, improves specificity rather than cause inhibition of the qPCR. Classifying indeterminate results as positive will increase the qPCR test sensitivity but will lower specificity to a lesser degree depending on the likelihood of contamination. Although PCR is the preferred test for amphibian populations, histology is useful when wishing to confirm a diagnosis of infection and in situations where observing the severity of infection and pathology in skin is desired. In this study, we show that diagnostic test validation in wild animals is now relatively straight forward using modern computing power and can incorporate prior knowledge generated from laboratory studies using Bayesian approaches.