Modelagem hierárquica Bayesiana de comunidades de anfíbios e de seu fungo patogênico (Batrachochytrium dendrobatidis) em riachos da Mata Atlântica
Carregando...
Data
2018-07-20
Autores
Ribeiro Junior, José Wagner [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
A Mata Atlântica brasileira se encontra altamente fragmentada e conta com menos de 15% da sua cobertura original, todavia, essas florestas ainda abrigam uma das maiores diversidades de anfíbios com alta proporção de espécies endêmicas e raras. As espécies enfrentam diversas ameaças como perda de habitat e doenças infecciosas nesse bioma. Dessa forma, é alto o interesse para entender como os anfíbios estão distribuídos na paisagem e respondem aos gradientes ambientais. Neste trabalho, nós usamos modelagem hierárquica bayesiana para investigar a probabilidade de ocorrência das espécies e comunidades de anfíbios e de seu fungo patogênico (Batrachochytrium dendrobatidis, Bd) em riachos da Mata Atlântica. No primeiro capítulo, nós investigamos como as espécies e comunidades de anfíbios respondem as covariáveis da paisagem (i.e., cobertura florestal nativa, agricultura, área da microbacia, densidade de riachos e inclinação do terreno), enquanto consideramos a detecção imperfeita das espécies. Nossos resultados demonstraram que a probabilidade de ocupação ao nível de comunidade foi positivamente relacionada com a cobertura florestal ao redor dos riachos (buffer 200 m), enquanto a agricultura teve um efeito negativo. Além disso, nossos resultados enfatizaram que a probabilidade de ocorrência ao nível de comunidade é maior em zonas ripárias de riachos menores com topografia mais plana, embora algumas espécies mostraram uma associação com riachos maiores. No segundo capítulo, nós avaliamos a contribuição relativa das variáveis locais do riacho, da bacia hidrográfica e de variáveis espaciais sobre a ocorrência dos anfíbios associados a riachos através de Modelagem Hierárquica de Comunidades de Espécies (MHCE). Nesse capítulo, nós consideramos a estrutura espacial hierárquica do sistema estudado e testamos quatro modelos representando possíveis vias de dispersão (i.e., dispersão terrestre, dispersão através do curso d’água e dispersão pelas duas vias). As comunidades de anfíbios estiveram principalmente relacionadas aos fatores locais do riacho e às características da bacia de drenagem, e essas relações foram mediadas pelos atributos das espécies. Encontramos baixo suporte ao papel da dispersão limitada ou excessiva influenciando a distribuição das espécies em nosso estudo. No último capítulo, nós avaliamos como a probabilidade de ocupação e detecção do fungo B. dendrobatidis, está relacionada com variáveis abiótica e bióticas. Nossas principais descobertas foram que a proporção de riachos da Mata Atlântica abrigando anfíbios infectados por Bd é alta (estimativa média = 75% dos riachos). Não contabilizar a falha de detectar o Bd subestima o número de riachos com animais infectados. A probabilidade de os riachos abrigarem anfíbios infectados por Bd foi positivamente relacionada com a densidade de riachos e inversamente relacionada com diversidade de anfíbios. Espécies reprodutoras terrestres demonstram maior prevalência de Bd. A cobertura florestal é positivamente relacionada com a prevalência de Bd, portanto a probabilidade de detectar o patógeno é maior em áreas mais prístinas.
The Atlantic Forest is highly fragmented with less than 15% of its original cover remaining, however these fragmented forests harbour one of the most diverse amphibian communities with high proportion of endemic and rare species. The amphibians face many threats in this biome, such as habitat loss and infectious diseases. Thus, there is a high interested on understanding how the amphibians respond to the environmental gradients and how they are distributed on the landscape. Here, we used Bayesian hierarchical model to study the occurrence probability of amphibian species and communities and its pathogenic fungus (Batrachochytrium dendrobatidis, Bd) on Atlantic Forest streams. On the first chapter we assessed both species- and community-level responses of amphibians to important landscape-scale characteristics (i.e., forest cover, agriculture, catchment area, stream density, and slope). Our results showed that mean occupancy probability at the community level was positively related with the forest cover around the streams (buffer 200 m), while the agriculture had a negative effect. Furthermore, our results emphasized that many amphibian species have the highest occurrence probabilities in riparian zones with small catchment area within flat landscapes, although some species were associated with wider streams. On the second chapter, we evaluate the relative contribution of both stream and watershed-level characteristics, and the spatial component on the occurrence of stream-dwelling amphibians through Hierarchical Modelling of Species Communities (HMSC). We considered the hierarchical spatial structure of the studied system to test four models representing different dispersion ways (terrestrial, watercourse and both ways). Amphibians communities were related with both local and watershed-scale characteristics that were mediated by species traits. There was low support of relationship between the spatial component (i.e., dispersal limitation and mass effects) and species distribution in our study. On the last chapter, we evaluated how the occupancy and detection probabilities of the fungus B. dendrobatidis is related with biotic and abiotic variables. Our mains findings showed a high proportion of Atlantic Forest streams harbouring amphibians infected by Bd (mean = 75% of streams). Not accounting for imperfect detection underestimated the number of streams with infected animals. Occupancy probability of Bd was positively related with stream density and inversely related with amphibians diversity. Terrestrial-breeding species had a higher prevalence of Bd, thereafter. The forest cover is positively related with Bd prevalence; therefore, the detection probability of pathogen is higher in pristine areas.
The Atlantic Forest is highly fragmented with less than 15% of its original cover remaining, however these fragmented forests harbour one of the most diverse amphibian communities with high proportion of endemic and rare species. The amphibians face many threats in this biome, such as habitat loss and infectious diseases. Thus, there is a high interested on understanding how the amphibians respond to the environmental gradients and how they are distributed on the landscape. Here, we used Bayesian hierarchical model to study the occurrence probability of amphibian species and communities and its pathogenic fungus (Batrachochytrium dendrobatidis, Bd) on Atlantic Forest streams. On the first chapter we assessed both species- and community-level responses of amphibians to important landscape-scale characteristics (i.e., forest cover, agriculture, catchment area, stream density, and slope). Our results showed that mean occupancy probability at the community level was positively related with the forest cover around the streams (buffer 200 m), while the agriculture had a negative effect. Furthermore, our results emphasized that many amphibian species have the highest occurrence probabilities in riparian zones with small catchment area within flat landscapes, although some species were associated with wider streams. On the second chapter, we evaluate the relative contribution of both stream and watershed-level characteristics, and the spatial component on the occurrence of stream-dwelling amphibians through Hierarchical Modelling of Species Communities (HMSC). We considered the hierarchical spatial structure of the studied system to test four models representing different dispersion ways (terrestrial, watercourse and both ways). Amphibians communities were related with both local and watershed-scale characteristics that were mediated by species traits. There was low support of relationship between the spatial component (i.e., dispersal limitation and mass effects) and species distribution in our study. On the last chapter, we evaluated how the occupancy and detection probabilities of the fungus B. dendrobatidis is related with biotic and abiotic variables. Our mains findings showed a high proportion of Atlantic Forest streams harbouring amphibians infected by Bd (mean = 75% of streams). Not accounting for imperfect detection underestimated the number of streams with infected animals. Occupancy probability of Bd was positively related with stream density and inversely related with amphibians diversity. Terrestrial-breeding species had a higher prevalence of Bd, thereafter. The forest cover is positively related with Bd prevalence; therefore, the detection probability of pathogen is higher in pristine areas.
Descrição
Palavras-chave
Anuran, Automated acoustic recorders, Community model, Deforestation, Detection error, Habitat loss, Landscape, Tropical forest, Amphibian disease, Chytridiomycosis, Occupancy model, Tropical streams, Watershed, Spatial component, Stream-dwelling amphibians, Hierarchical modelling of species communities