Dados do Trabalho

TÍTULO

THRESHOLD EFFECTS OF ROADS ON GIANT ANTEATERS IN THE BRAZILIAN CERRADO

Resumo

The Brazilian Cerrado is a biodiversity hotspot. Due to agricultural expansion, it has been severely degraded over the past decades, with natural habitats being converted into pasture and crops. Furthermore, the large road network cutting through the Cerrado enhances fragmentation effects and is responsible for a high number of casualties due to animal-vehicle collisions. After habitat loss, roadkill is one of the biggest threats to terrestrial vertebrates in the Cerrado, particularly to medium and large-sized mammals. The giant anteater (Myrmecophaga tridactyla) is one of the species most affected by roadkill in the Cerrado of Mato Grosso do Sul (MS), Brazil. If this trend continues, it could drive entire local populations of this vulnerable species to extinction. To address this threat, the Anteaters & Highways Project is evaluating the impacts of roads on giant anteater populations in the Cerrado. We used occupancy modelling to evaluate the effects of road and landscape variables on giant anteaters. We sampled 60 landscapes within a 10km buffer zone of MS-040, a recently paved road, with low traffic, located in southeast of MS. We sampled each landscape with three camera-traps for 30 days, during the dry station of 2018, with a total effort of c.5132 days of sampling. We used single-season/single-species occupancy modelling, with one sampling covariate (effort) and four landscape covariates (distance from road, distance from water and percentages of natural forest and pasture within 200ha) as predictors of detection (p) and occupancy (ψ) of the giant anteater. Model selection was based on Akaike Information Criterion corrected for small samples (AICc). Giant anteaters were present in 44 of the 60 landscapes sampled (ψnaïve = 0.73). After modelling for imperfect detection, occupancy increased 14% (ψ = 0.83 ± 0.06; CI 0.66 to 0.92). We found that all four landscape covariates were good predictors for detecting the species, with well estimated effects. Interpreting detection as frequency of use, the giant anteaters used more frequently landscapes with more natural forest and less pasture, that are closer to watercourses and more distant from the road. Further modeling indicated a threshold distance for the road effect: the frequency of use is down to its minimum up to 2750m from the road, from where it starts to increase linearly. We found no relationship between the covariates and the occupancy of the giant anteater, with the null model for ψ being the best-ranked model (ΔAICc < 2.0). Our results are consistent with literature. Despite our models indicating that distance from the road is a strong predictor for detecting the giant anteater, the same effect was not found on occupancy. Giant anteaters may be using the landscape without avoiding the road, which may increase roadkill probability. In fact, results from our telemetry study show that MS-040 does not act as a barrier for the species. It is possible that the giant anteater occupancy is better explained by predictors not tested in this study (e.g. food resources). Nevertheless, our analyses give us insights that can help develop mitigation strategies for preventing road negative impacts.

Palavras-chave

road ecology; occupancy modeling; mammals; xenarthra; Myrmecophaga tridactyla.

Financiamento

CAPES - Coordenação de Aperfeicoamento de Pessoal de Nível Superior

EDGE of Existence programme - Zoological Society of London

Programa Bolsas Funbio - Conservando o Futuro