Optimising sampling methods for small mammal communities in Neotropical rainforests

Abstract

Quantifying mammalian biodiversity is a critical yet daunting challenge, particularly in species‐rich ecosystems. Non‐volant small mammals account for >60% of the mammalian diversity and often require several survey methods to estimate their species richness and abundance, because of differences in their size and behaviour. Using 117 studies at 278 sites in a species‐rich biome, the Brazilian Atlantic Forest, we determined the influence of trap configuration, trap type, and sampling effort on measures of species richness and abundance. We used generalised linear mixed‐effects models to determine which methodological approaches influenced estimates of species richness and abundance at the sampling sites. We used estimates of beta (β) to determine which methods improved species richness and abundance estimates, and generated predicted values for the overall species richness as a function of trap configuration (line transect, grid, or both), trap type (pitfall traps, live‐traps or both), and sampling effort (number of trap‐nights). Our results indicated that sites in which pitfall traps alone were used generated higher estimates of the overall small mammal species richness and abundance, and rodent abundance, than sites in which only live‐traps were used. Sites in which pitfall traps alone were used also produced higher estimates of species richness and abundance, and rodent species richness and abundance, than sites in which both trap types were used. Increased sampling effort led to increased estimates of species richness, but sites in which pitfall traps were used alone or together with live‐traps had higher estimates of species richness with less sampling effort than sites in which live‐traps only were used. Using pitfall traps greatly reduced the number of trap‐nights necessary to obtain a good estimate of small mammal species richness in a community. We found no influence of trap configuration (line transects or grids) on estimates of species richness, but the abundance of rodents was estimated to be higher in sites where both line transects and grids were used, than in sites where only one trap configuration was used. Our review shows the importance of using pitfall traps in research, monitoring, and environmental impact studies on species‐rich small mammal communities.

Defaunation and fragmentation erode small mammal diversity dimensions in tropical forests

“…Path models were implemented with generalized linear mixed-effects models framework (GLMM), using trap type (Pitfall, Sherman and Tomahawk traps) and region (northeast, southeast and south) as random effects (Lefcheck 2016). We did not use sampling effort (trap-nights) as a random effect in our analysis because it was strongly correlated with trap type (Bovendorp et al 2017a). We obtained path coefficients by using the standardized β coefficient.…”

A comparison of eDNA to camera trapping for assessment of terrestrial mammal diversity

“…In addition, post-processing remains laborious as manual 50 tagging of images is required, although promising developments with machine learning have 51 been made recently (Tabak et al, 2018). For small mammals, live-trapping is more common but 52 the type of trap, bait and sampling design can strongly affect the probability of detection of 53 particular species (Bovendorp, McCleery, & Galetti, 2017;Harkins, Keinath, & Ben-David, 2019). 54 Moreover, trapping is highly invasive, labor-intensive and generally requires onerous permitting.…”

Disentangling drivers of small mammal diversity in a highly fragmented forest system

“…We compiled the largest dataset on the distribution of small mammals in the AF to date, including published literature, to create a binary dataset containing presence–absence information for small, non‐volant mammals distributed across 340 sites throughout the AF (Bovendorp, McCleery, et al, ; de la Sancha, Higgins, Presley, & Strauss, ; Table S1). Of the 340 sites about a third were small (100 sites <100 ha) and ranged in size (165 sites < 500 ha; 189 < 1,000 ha; 240 < 5,000 ha; 269 < 10,000 ha) with 71 sites larger than 10,000 ha.…”

“…Non‐volant small mammals in the AF are an ideal group to study the effects of fragmentation on patterns of diversity because of their smaller home range and inability to disperse in a short amount of time. They are comprised primarily of didelphid marsupials and sigmodontine rodents (Bovendorp, McCleery, & Galetti, ; Bovendorp, Villar, et al, ; Figueiredo et al, ). These subgroups represent evolutionary lineages that entered SA at different times.…”