Long-term vertebrate roadkill monitoring in Central Amazon: Vehicle traffic effect and hotspot inconstancy over time

In this study, we present 21-years of vertebrate roadkill monitoring along a 121 km stretch of highway (BR-174) crossing the Waimiri Atroari Indigenous Territory (TIWA). The TIWA is a well-preserved area of high vertebrate diversity located in the Central Amazon. We conducted regression and spatial analyses on subsets of the 21-year period, depending on the question being addressed. To assess roadkill dynamics, we identified the taxonomic groups most killed and, for each class, we investigated the influence of rainfall and vehicle traffic rates over the number of roadkills by using Generalized Linear Models on data spanning eleven years of monitoring. To identify scales and areas of significant roadkill clusters (hotspots), we used modified K-statistics and the HotSpot 2D-Identification over an eight-year period. From 1997 to 2017, 11,635 vertebrates roadkills were recorded. From these, mammals and reptiles were the most common followed by birds and amphibians. Vehicle traffic explained roadkill variation for all taxa and rainfall was important to explain mortality only for mammals, with a negative effect, and for reptiles with a positive relation with a one-month lag of roadkill relative to rainfall. Roadkill hotspots varied by taxon and over time, though some hotspots persisted or oscillated within each group, likely reflecting local animal abundance or changes in resource availability. Our findings suggest that even on low-traffic highways, vehicle traffic rates should be incorporated into roadkill studies. Furthermore, hotspots monitoring should be conducted over medium to long terms to ensure that mitigation measures cover priority factors and areas over time.