Wildlife Roadkill Driven by Hydrological Regime in a Subtropical Wetland

Abstract

This study explored the role of the hydrological regime as a trigger factor for wildlife roadkill along a 22 km road crossing the Taim Wetland, a Ramsar site of international importance in South Brazil. The north–south crossing of BR-471, a federal highway, causes fatalities to numerous animals due to collisions with vehicles. An 11-year long-term time series study encompassing monthly roadkill incidents, water level, and rainfall was analyzed by considering three different periods related to a Wildlife Protection System (WPS): (i) 3 initial years before the WPS implementation (BWPS); (ii) 4 intermediate years after the WPS implementation (AWPS), which includes fences, fauna tunnels, cattle guard stocks, bumps, and speed control; (iii) 4 final years during partial destruction of the WPS (PDWPS). A pseudo-2D full hydrodynamic cell model was employed to fill missing water level data. The model had a good to very good performance (NSE: 0.73 to 0.87; R2: 0.79 to 0.90). The relationship between roadkill incidents and the WPS periods (BWPS, AWPS, and PDWPS) was modeled using Generalized Additive Models for Location, Scale, and Shape (GAMLSS), considering rainfall and water level as predictor variables. The analysis revealed a higher incidence of wildlife roadkill in BWPS compared to AWPS and PDWPS, suggesting the effectiveness of the implemented measures. Critical season and interplay between water levels, rainfall, and the roadkill were assessed. Mammals was the most common roadkill class identified (~92%), followed by reptiles (13%) and birds (2%), with no change in these percentual in the BWPS, AWPS, and PDWPS. Among mammals, capybara (Hydrochoerus hydrochaeris) and coypu (Myocastor coypu) were the most frequent victims (~93% of mammals). Winter, followed by autumn, recorded the highest number of roadkill incidents (>60%), and this pattern remained consistent during the three periods. While rainfall did not emerge as a determining variable for roadkill, water levels above certain thresholds (>3.3 m) drastically diminished the effectiveness of the WPS, mainly due to fauna tunnel submersion. These findings offer valuable insights for enhancing wildlife conservation strategies in this protected area by incorporating hydrological information providing a baseline for designing WPS in similar environments.