Assessing Detection Probability and Survey Frequency for the Threatened Magnificent Broodfrog, <i>Pseudophryne covacevichae</i>

ABSTRACT Difficulty in detecting species' presence is a common issue when surveying threatened species. This is particularly relevant when target species occur in remote regions, have small populations, are difficult to detect, or sampling effort is limited. This can lead to underestimation of a species' true occurrence, which can be an issue where developments are proposed that could impact populations through habitat loss or fragmentation. We aimed to identify the environmental variables influencing the probability of detecting the magnificent broodfrog ( Pseudophryne covacevichae ), determine environmental triggers for survey initiation and estimate the number of surveys required to provide confidence in the species' true absence at a location. We analysed repeat site survey data from 13 locations where the species was known to occur. Single‐season occupancy models identified volumetric soil moisture to be the most influential environmental variable in detection, followed by a combination of volumetric soil moisture and accumulated rainfall in the 5 days prior to a survey. These two variables were used to classify survey conditions into poor, average and excellent, defined by their 5th, 50th and 95th percentiles, to estimate the relationship between survey conditions and survey effort. Cumulative detection probability under ‘poor’ environmental conditions remained low, with less than 40% cumulative detection probability following six surveys and high uncertainty in posterior distributions. In contrast, under ‘average’ conditions, detection probability increased to 96% following three surveys, and in ‘excellent’ conditions, a single survey resulted in 98% probability of detection, and certainty in the posterior distributions increased in both instances. These results demonstrate that targeting surveys under good to optimal environmental conditions can improve detection probability, maximise the efficiency of surveys and reduce the likelihood of false absences.