Research Article: It’s a trap: Optimizing detection of rare small mammals

Date Published: March 5, 2019

Publisher: Public Library of Science

Author(s): Kristina M. Harkins, Doug Keinath, Merav Ben-David, Bi-Song Yue.

http://doi.org/10.1371/journal.pone.0213201

Abstract

Improving detection probabilities for rare species is critical when assessing presence or habitat associations. Our goal was to create a new small mammal trapping protocol that improved detection of rare species, such as the olive-backed pocket mouse (Perognathus fasciatus). We used three trap and bait types and trapped an area 4.4 times larger than the standard grid. We also assessed the effect of captures of non-target species on detection probability of pocket mice. Regardless of species, trap success was higher for Havaharts. We found that bait and trap type selection varied significantly by species, with pocket mice showing strongest selection for Havahart traps baited with bird seed. Increasing grid size, while maintaining a similar trapping effort, resulted in higher detection probability, although our analyses showed that effective grids can be about three-quarters of the size we use to achieve similar results. We were also able to demonstrate that by deploying a combination of different traps and baits it is possible to overcome the potential effect of non-target species (e.g., deer mice, Peromyscus maniculatus) on the detection probability of pocket mice. Our results show that simple changes to standard small-mammal trapping methods can dramatically increase the detectability of rare and elusive small mammals. Increasing detection probability of rare components of a community can improve the results and understanding of future studies.

Partial Text

Rapid and global environmental change increases the need to identify and conserve critical habitats for at-risk and endangered species [1]. Identifying such critical habitats often relies on occupancy modeling, which has become a common tool to monitor wildlife populations when insufficient captures prevent the use of mark-recapture analysis or when individual identification is impossible [2]. Occupancy modeling is a logistic regression-based framework that estimates the probability of a species occupying sampled sites (occupancy) while accounting for the probability of detecting the species using the given sampling methods (detectability). Imperfect detection is a frequent sampling problem when animals are rare, populations are small, or individuals are difficult to observe [3,4]. However, even when species are fairly easy to observe and identify, detection is often imperfect [5,6]. Survey methods and conditions can alter the behavior of animals or affect surveyor performance, further reducing the ability to distinguish between a true negative, where a species is absent from the survey area, and a false negative, when a species is present but undetected during the survey [2,7]. Including estimates of detectability in occupancy analyses is crucial because false negatives can lead to a biased and imprecise estimation of habitat characteristics and their relative influence on the occurrence of the target species [8,9]. Accounting for detection probability increases the performance of models estimating occupancy [4], especially for rare or difficult to detect species [10,11]. As detection probability increases, estimated occupancy becomes more precise (i.e. tighter confidence intervals; [2]). In turn, that precision directly increases our ability to track temporal changes in occupancy in surveyed areas and create refined habitat suitability models. Both these results translate into better ecological understanding of the systems in question and, consequently, an improved ability to make effective management decisions.

Our results suggest that Havahart traps, especially when baited with bird seed, are more effective at increasing capture success, and thus detection probability, of pocket mice compared with the standard trap and bait combination of Sherman traps and peanut butter-oats mix. Grid size further affected pocket mouse detection. These results indicate that using standard small mammal trapping methods (i.e., a 10×10 grid of Shermans spaced 10 m apart and baited with peanut butter and oats) will likely result in negative bias in modeling occupancy of pocket mice and other rare species.

Our results suggest that simple modifications to the standard small mammal trapping-methods could improve their effectiveness, thereby increasing the robustness of conclusions from studies targeting rare and difficult to detect species with differing diet preferences. For pocket mice, this increased level of detection has the potential to improve our knowledge of their ecology, habitat suitability and occupancy trends across their range. In addition, our protocols should be considered by researchers attempting to assess community and ecosystem dynamics. Failure to detect some members of the community may yield incorrect estimates of community composition, species richness and interactions, niche size and overlap, as well as energy flow in ecosystems [65,66]. For example, whole community energy consumption will be relatively constant when the abundance of competing species alternate [66]. The effects of low detection of one of these interacting species, even when they are at high densities (as would be the case for pocket mice trapped with peanut butter and Sherman traps), could result in miscalculation of whole community energy consumption. In turn, failure to properly describe community interactions will reduce the credibility of management decisions for species and ecosystems in the face of a rapidly changing environment.

 

Source:

http://doi.org/10.1371/journal.pone.0213201

 

Leave a Reply

Your email address will not be published.