Optimising tropical forest bird surveys using passive acoustic monitoring and repeated short-duration point counts

Estimation of avian biodiversity is a cornerstone measure of ecosystem condition, with turnover in avian community composition underpinning many studies of land-use change in tropical forests. Surveys conducted using autonomous recorders have been frequently found to be more efficient than traditional point-count surveys. However, there has been limited research into optimal survey duration, despite autonomous recordings allowing for many more repeats of short-duration surveys with relative ease in comparison to traditional survey methods. We use an acoustic dataset collected from a region of very high avian biodiversity - the eastern Brazilian Amazon - to test the effect of using short-duration surveys to increase temporal coverage without increasing total survey duration. We use this dataset to assess whether a survey protocol consisting of 240 15-second surveys at 29 locations, short-duration surveys, has an influence on resulting alpha and gamma diversity, and detection frequency, than standard-duration surveys of four 15-minute surveys per location. We find that repeated short-duration surveys outperform longer duration surveys in every metric considered herein, with short-duration surveys predicted to detect approximately 50% higher alpha diversity, and 10% higher gamma diversity. Short-duration surveys also detect species more often, at more survey locations. Conversely, standard-duration surveys are almost four times more likely to produce false negatives (i.e. to fail to detect species presence). Whilst there is no difference between the proportion of uncommon species detected by the two methods, when considering species detected multiple times at multiple locations, short-duration surveys detected three times more uncommon species than standard-duration surveys. We conclude that short-duration recorded surveys should be considered the primary method for sampling the species richness of bird communities in tropical forests and is likely to be preferable to longer duration or traditional surveys in most environments.