Spatial bias in sampling small rodents in the Atlantic Forest: A landscape and accessibility perspective

Abstract

Understanding the effects of habitat loss and fragmentation on species spatial distribution is challenging, mainly because knowledge of species occurrences is biased. Survey efforts are unevenly distributed causing spatial sampling biases that are normally neglected. Assessing sampling bias is particularly urgent for threatened ecoregions, such as the Atlantic Forest, a global biodiversity hotspot. Here, we assess spatial sampling biases of small rodents in the Atlantic Forest, using an integrative approach with accessibility and landscape metrics. We built a robust dataset of 11,495 primary records of the Atlantic Forest’s small rodent species, based on information from digitally accessible repositories. We expect that well-sampled sites are spatially aggregated and nearer roads, urban centers, on landscapes with larger forest fragments, and with higher percentage of forest cover. We also expect gaps of small rodents sampling in rare landscape conditions. Our results indicated that only less than 1% of the Atlantic Forest (at 1 km2 cell-size resolution) are well sampled. Following our expectations, the well-sampled sites were spatially aggregated biased toward roads, urban centers, larger forest fragments, and landscapes with higher percentage of forest cover. We also found a survey gap on common landscape conditions. Our findings suggest that the spatial distribution of small rodents at landscape level (1 km2) remains unknown across most of the Atlantic Forest spatial extension. Our findings also point to new priority sites for small mammals sampling on common landscape conditions, in smaller fragments and on remote areas improving spatial distribution knowledge and contributing to conservation policies at landscape level.

Keywords

• Atlantic forest
• Survey completeness
• Macroecology
• Biodiversity shortfalls