Efeito de características da paisagem sobre atropelamentos de vertebrados em rodovias do Mato Grosso do Sul
Data
2022-11-29
Autores
Silva, Thais Martins da [UNIFESP]
Orientadores
Vieira, Elisa Hardt Alves [UNIFESP]
Semensatto Junior, Décio Luis [UNIFESP]
Tipo
Dissertação de mestrado
Título da Revista
ISSN da Revista
Título de Volume
Resumo
O atropelamento de fauna constitui uma grave ameaça à biodiversidade e segurança humana, estando também associado a altos custos econômicos para a sociedade. Entender como a paisagem influencia na ocorrência de atropelamentos pode nortear a escolha dos locais onde a aplicação de medidas mitigadoras seja mais eficiente, assim como otimizar os custos decorrentes de suas instalações. O presente estudo teve como objetivo avaliar a influência de características da paisagem sobre a mortalidade de vertebrados nas rodovias BR-158 e BR-262, estado do Mato Grosso do Sul, e propor medidas de mitigação nos locais avaliados como potenciais para a ocorrência de atropelamentos. Para tanto, foram utilizados dados secundários de coletas semanais realizadas nas áreas de estudo durante os anos de 2016 a 2018. A distribuição geral dos atropelamentos foi avaliada pela ocorrência de hotspots utilizando o método 2D Hotspot Identification, que identifica os trechos com maior intensidade de eventos de colisão. Para a criação dos modelos preditores da paisagem, foram consideradas as medidas de área e distância das classes de uso e cobertura da terra no entorno de 1 km das rodovias, como: agricultura, água, área urbana, campo antrópico, cerradão, vegetação ciliar, silvicultura e vegetação natural em regeneração. Utilizando duas diferentes variáreis respostas (ocorrência de atropelamentos e abundância por segmento), os modelos foram avaliados através da aplicação de Modelos Lineares Generalizados, e selecionados a partir do critério de informação Akaike. Tendo como base a presença de hotspots e elementos preditivos da paisagem, foi proposto um critério de avaliação para a determinação dos locais prioritários para a instalação de medidas mitigadoras como passagens de fauna, redutores de velocidade e placas sinalizadoras. Os resultados mostram que os atropelamentos não estão distribuídos de forma aleatória pelas rodovias de estudo, podendo ser encontrados 12 trechos com maior concentração de mortalidade. Dentre as espécies registradas, Cerdocyon thous (cachorro-do-mato) e Myrmecophaga tridactyla (tamanduá-bandeira) foram as mais atropeladas nas rodovias avaliadas, e a ocorrência de espécies ameaçadas ressalta o valor de conservação da região. De modo geral, os atropelamentos estiveram positivamente associados às áreas de vegetação ciliar, campo antrópico e hotspots, e trechos próximos a fragmentos de cerradão e vegetação natural em regeneração. Características ligadas ao uso antrópico da terra, como silvicultura e agricultura, estiveram negativamente associadas às ocorrências. Através da aplicação dos critérios de seleção para a definição dos locais a serem mitigados, foi identificado o total de oito trechos classificados como prioridade 1, seis trechos de prioridade 2 e 16 trechos indicados como prioridade 3, e propostas medidas de custo-benefício equivalentes aos cenários avaliados. É esperado que o presente estudo possa auxiliar no norteamento de tomadas de decisões em escalas mais amplas, além de trazer alternativas mitigadoras aplicáveis a ecossistemas secos como o Cerrado.
Wildlife roadkill seriously threatens biodiversity and human security and is also associated with high economic costs for society. Understanding how the landscape influences roadkill occurrence can guide the choice of places where the application of mitigating measures is more efficient and optimize the costs resulting from their installations. The present study aimed to evaluate the influence of landscape features on vertebrate mortality on highways BR-158 and BR-262, state of Mato Grosso do Sul, and to propose mitigation measures in those places that are considered as potential for the occurrence of roadkills. For accomplishing these measures in this study, I used secondary data from weekly collections carried out in the study areas from 2016 to 2018 were used. For the creation of the predictive models of the landscape, the measurements of area and distance of the classes of road usage and land cover around 1 km of the highways were considered, such as agriculture, water, urban area, anthropic field, cerradão, riparian vegetation, forestry and regenerating natural vegetation. I used in this study two different response variables (occurrences and abundance per segment), and the models used evaluated using Generalized Linear Models and selected based on the Akaike information criteria. Based on the presence of hotspots and predictive elements of the landscape, an evaluation criterion was proposed to determine priority locations for the installation of mitigating measures such as such as safety passages, speed radar, and fauna crossing signs. The results show that roadkills are not randomly distributed along the study roads, and 12 stretches with a higher concentration of mortality can be found. Among the species recorded, Cerdocyon thous (crab-eating fox) and Myrmecophaga tridactyla (giant anteater) were the most roadkill on the evaluated roads, and the occurrence of threatened species highlights the conservation value of the region. Roadkills were generally positively associated with areas of riparian vegetation, anthropic grassland, hotspots, and stretches close to fragments of cerradão and regenerating natural vegetation. Characteristics linked to anthropic land use, such as forestry and agriculture, were negatively associated with the occurrences. Through the application of selection criteria to define the places to be mitigated, a total of eight stretches classified as priority 1, six stretches of priority 2 and 16 stretches indicated as priority 3 were identified, and cost-benefit measures were equivalent to those evaluated scenarios. It xiii is expected that the present study can guide decision-making on broader scales, and bring mitigating alternatives applicable to dry ecosystems such as the Cerrado.
Wildlife roadkill seriously threatens biodiversity and human security and is also associated with high economic costs for society. Understanding how the landscape influences roadkill occurrence can guide the choice of places where the application of mitigating measures is more efficient and optimize the costs resulting from their installations. The present study aimed to evaluate the influence of landscape features on vertebrate mortality on highways BR-158 and BR-262, state of Mato Grosso do Sul, and to propose mitigation measures in those places that are considered as potential for the occurrence of roadkills. For accomplishing these measures in this study, I used secondary data from weekly collections carried out in the study areas from 2016 to 2018 were used. For the creation of the predictive models of the landscape, the measurements of area and distance of the classes of road usage and land cover around 1 km of the highways were considered, such as agriculture, water, urban area, anthropic field, cerradão, riparian vegetation, forestry and regenerating natural vegetation. I used in this study two different response variables (occurrences and abundance per segment), and the models used evaluated using Generalized Linear Models and selected based on the Akaike information criteria. Based on the presence of hotspots and predictive elements of the landscape, an evaluation criterion was proposed to determine priority locations for the installation of mitigating measures such as such as safety passages, speed radar, and fauna crossing signs. The results show that roadkills are not randomly distributed along the study roads, and 12 stretches with a higher concentration of mortality can be found. Among the species recorded, Cerdocyon thous (crab-eating fox) and Myrmecophaga tridactyla (giant anteater) were the most roadkill on the evaluated roads, and the occurrence of threatened species highlights the conservation value of the region. Roadkills were generally positively associated with areas of riparian vegetation, anthropic grassland, hotspots, and stretches close to fragments of cerradão and regenerating natural vegetation. Characteristics linked to anthropic land use, such as forestry and agriculture, were negatively associated with the occurrences. Through the application of selection criteria to define the places to be mitigated, a total of eight stretches classified as priority 1, six stretches of priority 2 and 16 stretches indicated as priority 3 were identified, and cost-benefit measures were equivalent to those evaluated scenarios. It xiii is expected that the present study can guide decision-making on broader scales, and bring mitigating alternatives applicable to dry ecosystems such as the Cerrado.