Active or passive acoustic monitoring? Assessing methods to track anuran communities in tropical savanna wetlands

Abstract The development of new acoustic technologies has promoted remarkable advances in biodiversity monitoring. Passive acoustics is increasingly being used in ecological research to remotely assess animal communities. This recent revolution opens new questions about the precision and efficiency of these emerging methods in tracking biodiversity. However, most of the studies evaluating passive monitoring have been applied on temperate regions and in the short term, limiting the understanding of the potential of these techniques and the development of a general framework for survey design. Based on a year-round monitoring of anuran communities at Brazilian savanna wetlands, we examined efficiency of traditional (active searches, AS) and emerging monitoring methods (passive acoustics, PAM) in detecting species at an increasing scale (hours, days, weeks, and months). We determined species richness on 42 active searches and more than 34,000 audio recordings and defined six comparative scenarios (with unequal and equal sampling efforts) in order to explore the combined effect of sampling effort and sampling duration on the species richness detected by these two methods. Using PAM, we estimated higher regional and local richness (14 species per site on average) than AS (8.3 species per site) and detected all anuran species found by AS. Except when sampling effort was highly biased towards AS (single-hour PAM), passive acoustics was able to find more species than AS in all cases. In scenarios with similar sampling effort, differences in observed species richness depended on the temporal distribution of samples. As PAM covered a wider daily and seasonal period, the number of detected species increased, particularly at the most intensive and prolonged sampling (thirty-full-day PAM). Contrary to expectations, we did not record additional species when using AS. Our results provide evidence that passive acoustics is an efficient monitoring technique, offering a higher capacity to detect vocal species in biodiversity surveys than traditional methods, especially when applied in the long term. Sampling scenarios revealed that species detection on monitoring programs is associated with both sampling effort and sampling duration. Thus, due to the potential of PAM to increase these two parameters, this technique can reach better outcomes and be particularly advised for ecological research. These findings can assist researchers in survey design under tropical environments and in determining suitable sampling efforts when applying acoustic monitoring programs.