Estimativas e padrões temporais e espaciais de fatalidades de sapos-cururu (Rhinella gr. marina) numa ferrovia da Amazônia brasileira

• Ano de defesa: 2019
• Autor(a) principal: Rubem Augusto da Paixão Dornas
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; Brasil; IGC - DEPARTAMENTO DE CARTOGRAFIA; Programa de Pós-Graduação em Análise e Modelagem de Sistemas Ambientais; UFMG
• Palavras-chave em Português: Ecologia de estradas; Atropelamento; Agregações; Hotspots; Hot moments; Anfíbio; Remoção de carcaças; Detectabilidade; Modelo; Regressão geograficamente ponderada; Modelagem de dados – Aspectos ambientais; Ecologia; Estrada de Ferro Carajás (MA e PA)
• Link de acesso: http://hdl.handle.net/1843/33853

Transportation infrastructures are directly responsible for killing billions of animals worldwide annually. Although the understanding about road impacts have recently increased, the impact of railroads on wildlife has received less attention. The current knowledge concerning the impacts of railroads focuses mainly on large mammals although amphibians might be affected. This study aims to unravel temporal and spatial patterns of Rhinella toad fatalities on a Brazilian Amazonian railroad, to comprehend how toads are killed and to estimate the magnitude of toad fatalities. Data collection was carried out on foot on an 871-km stretch of the Estrada de Ferro Carajás from 2013 to 2017. Different potential causes for fatalities were identified: being run over, desiccated or barotrauma. A surprisingly high carcass persistence probability of about 38 days was estimated. After correcting for the bias from carcass detection and removal, it was estimated that approximately 10,000 toads are killed per year (approx. 11 fatalities/km/year). A generalized linear model (GLM) model showed that toads were more likely to be killed in the dry to wet transition and wet seasons. Critical zones of fatalities were identified and prioritized according to their intensity. The highly critical segments encompass more than 10% of all fatalities although they cover only 1.5% of the railroad. A Geographically Weighted Regression (GWR) non-stationary model indicated forest and urban areas as predictors of fatalities, presenting positive relation with proportions of the forest and urban areas, as well as the distance to urban areas and a negative relation with the distance to forest. The predictive model generated by cross-validation presented a moderate prediction capacity, with a low incidence of false positives in highly critical regions of fatalities. This study is the first one to address carcass detection and persistence on railroads and to unravel temporal and spatial patterns of fatalities of an amphibian species in a tropical climate. A better understanding of patterns of animal fatality on railroads is of fundamental importance to plan, manage, and mitigate this impact. The work meets the needs highlighted by CAPES Committee on Environmental Sciences by promoting research and the social insertion of research in regions with a high index of social vulnerability, environmental vulnerability, and geographical isolation.