Factors affecting the detection probability of a critically endangered flying-fox: consequences for monitoring and conservation

Context: Monitoring is crucial for understanding population trends of threatened species and for assessing the effectiveness of conservation efforts. However, population monitoring is subject to detection probabilities that can vary across factors such as time, type of vegetation cover, weather conditions and observer. Aims: In this study, we investigated the impact of environmental factors (e.g. wind and rain), spatiotemporal factors (e.g. time of night and geographical location) and observer variability, on the detection probability of Pteropus natalis (Christmas Island flying-fox), a critically endangered species that has been monitored across its single (135 km²) island range since 2006, by using active aural and visual detection of foraging individuals. Methods: Surveys were conducted at four visits to 133 sites across Christmas Island, representing the environmental variation of the island, over a 2-month survey period. The survey was conducted in 9 years between 2006 and 2022. Key results: Variable importance analysis showed that distance from the coast, year, and time of night were key predictors of P. natalis detection probability. Detection probability was higher on calmer nights, suggesting higher flying-fox activity or better sound transmission. Detection probability was also higher near roosts earlier and later in the night, indicating that P. natalis gradually moves away from and returns to roosts over the night. Detection probabilities varied between 2012 and 2022 across vegetation types, potentially reflecting changes in diet or phenology. Experienced observers were more likely to detect P. natalis, likely due to familiarity with their vocalisations or visual cues. Analyses excluding environmental and spatiotemporal factors suggested a slight increase in detections since 2012∼ however, once these factors were included, a significant decrease in detection probability between 2019 and 2022 emerged. Conclusions: Our findings highlighted how environmental and spatiotemporal factors can affect detection probability and, consequently, survey results of a mobile, threatened small-island endemic. Implications: This study demonstrated the importance of considering environmental and spatiotemporal factors when designing a monitoring program and, in subsequent analysis, to maximise the accuracy and precision of estimates derived from monitoring programs.