Assessing detection probabilities for the endangered growling grass frog (Litoria raniformis) in southern Victoria

Abstract

Assessment of the efficacy of survey techniques for determining species occurrence is crucial for the validation of wildlife survey data. We analysed repeated site-survey data for adults and larvae of the growling grass frog (Litoria raniformis) in order to estimate probabilities of detection for the species using alternative survey techniques. The estimated probability of detecting adults of L. raniformis at occupied sites using diurnal searches was much less than 1.0 (0.107; 95% credible interval: 0.045, 0.192). The estimated probability of detecting adults using nocturnal spotlight searches was considerably higher, but still less than 1.0 (0.696; 95% credible interval: 0.585, 0.796). These results indicate that nocturnal searches are a much more efficient and reliable means of detecting the presence of adult L. raniformis than diurnal searches, but detection using either technique is less than certain. The probability of detecting tadpoles of L. raniformis using either funnel-trapping or dip-netting techniques was estimated at 0.350 (95% credible interval: 0.151, 0.567). Together, these results indicate that reliance on single-site visits during surveys for this species is likely to result in severe under-estimation of the proportion of sites that are actually occupied. We urge other workers to use repeated site-survey data and appropriate methods of data analysis to assess and report probabilities of detection when documenting the results of wildlife surveys.