Estimating the diversity of tropical anurans in fragmented landscapes with acoustic monitoring : lessons from a sampling sufficiency perspective

Ribeiro D Anunciação, Paula and Sugai, Larissa Sayuri Moreira and Martello, Felipe and Tavares de Carvalho, Luis Marcelo and Ribeiro, Milton Cezar (2022) Estimating the diversity of tropical anurans in fragmented landscapes with acoustic monitoring : lessons from a sampling sufficiency perspective. Biodiversity and Conservation, 31. pp. 3055-3074. ISSN 0960-3115

Abstract

Determining the distribution and abundance of populations is the first step toward assessing biodiversity conservation status. This step is based on field observations that are largely influenced by the sampling method employed. Autonomous Recording Units (ARUs) are tools developed to improve species monitoring that employ acoustic communication. Although largely employed, the efforts required to achieve good diversity estimates with this technique are still unknown. We investigate the use of ARUs in estimating species richness of anuran assemblages in a tropical region, aiming to determine the sampling sufficiency of species richness at local and regional levels, analyze whether the asymptote point is related to forest cover, and investigate the influence of subsampling type over time on species richness estimates. We monitored amphibians in 14 streams embedded in landscapes representing a gradient from 20 to 70% native forest coverage. We detected a total of 14 species, with the regional sampling sufficiency of total species richness reached in 3448 min and influenced mainly by the terrestrial species’ presence. Forest coverage had no influence on the minimum audio processing time required to achieve local asymptote. The subsampling schemes (temporally stratified and randomly assigned) had similar efficiency when using 5 min/h or more sample efforts. Our findings indicate that passive acoustic monitoring can adequately represent local anuran richness, focusing especially on the arboreal guild. Sampling effort can be optimized, with a 5 min/h duty cycle being sufficient to recover detection of most species, saving up to 75% of the effort devoted to auditing the acoustic dataset.