Does the analysis unit affect the consistency of the factors influencing small vertebrate road-kills?

Abstract

Road-kill studies are important to understand the magnitude of the impact on the persistence of the surrounding populations. The size and the decision between the use of buffers or segments as analysis unit to data collection of landscape factors influencing roadkill are arbitrary. It may influence the results and the definition and placement of mitigation measures. In this study we aimed to test this issue by using units with different shapes (buffers and segments) and sizes (daily movement, standard and dispersal) in five small vertebrate species. The road-kill data were collected weekly in 2005 in an extension of 137 km encompassing two Brazilian Federal paved roads. We analyzed records of two reptilians, water snake (Helicops infrataeniatus) and D’Orbigny’s slider (Trachemys dorbigni), and three mammals, white-eared opossum (Didelphi salbiventris), nutria (Myocastor coypus) and skunk (Conepatus chinga). We used those species because they have different life-history traits and are frequently killed on the assessed roads. The percentage of variance explained independently for each land-use class was determined using hierarchical partitioning (HP) of multiple regression models. Our results showed that buffer shape is the best shape to be used on the analyses, and that the role of size of units is complex and may be species-specific. However standard size (1000 m) seems to be a reasonable solution as it worked well for species with both low and high home-range values. Our study highlights the importance of prior analysis with several scales and shapes to identify the appropriate unit for road-kill modeling.