Optimizing sampling effort and timing to improve inference of small mammal diversity in southeastern US forests

Abstract

Wildlife managers and researchers frequently utilize the number of species present in a given area as a metric to evaluate ecological function as well as progress of and impact of conservation actions. Small mammals serve vital roles as seed and fungal dispersers, regulating plant and insect populations, and the primary prey item of many species. Their ecological importance, coupled with their ubiquity across terrestrial ecosystems and the rapid response of small mammal populations to ecological changes makes live capture box trap surveys of small mammal communities a common tool to evaluate ecological health. Despite the extensive use of small mammal diversity surveys in research and management, few empirical guidelines exist on fundamental aspects of box trap surveys like duration and time of year, particularly in southeastern US forests. To address this, we evaluated bi-monthly box trap surveys of small mammals that were conducted between March 1997–September 2006 at the Savannah River Site, South Carolina for the impact of season and survey length on the number of species detected. Using a 10-year dataset from 12 long-term trapping grids in upland loblolly and longleaf pine forest, we first quantified how sampling duration and season influence species richness estimation by modeling species accumulation and time to achieve near-complete community characterization. The probability of detecting new species declined rapidly after approximately 7–8 days of sampling but achieving ≥90% of estimated richness required substantially longer effort (17–26 days), particularly in winter. We then expanded this analysis to evaluate species-specific detection probabilities in relation to seasonal conditions and short-term weather variation (precipitation and temperature) to identify mechanisms driving observed patterns and determine how survey efforts could be improved. Detection probability varied consistently across seasons and in response to weather, demonstrating that variation in capture rates reflects not only differences in community composition but also dynamic changes in species’ activity and availability for detection. These results indicate that commonly used short-duration surveys likely underestimate community composition and that sampling efficiency is strongly mediated by seasonal and environmental conditions. By linking species accumulation dynamics with species-specific detection processes, this study provides empirically grounded recommendations for optimizing trapping design in southeastern forests and highlights the importance of aligning sampling effort with study objectives when using small mammals as indicators of ecosystem condition.