Métodos de processos pontuais para análise de interação entre árvores de espécies nativas da amazônia

• Ano de defesa: 2018
• Autor(a) principal: Almeida, Elianara Martins de
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Lavras; Estatística e Experimentação Agropecuária; UFLA; brasil
• Palavras-chave em Português: Configurações pontuais; Manejo florestal; Testes de hipóteses; Método de Monte Carlo; Processos estocásticos; Point pattern; Forest management; Hypothesis testing; Monte Carlo Method; Stochastic processes; Estatística
• Link de acesso: http://repositorio.ufla.br/jspui/handle/1/29072

Understanding the spatial patterns of native trees, with so many species that can coexist in tree communities is a relevant challenge for Biologists and Forest Engineers. Point process methods are the standard tools to analyze the spatial distribution of tree species. The Ripley's K function is one of the most popular techniques in the analysis of spatial point processes since it allows detecting spatial interactions at different scales. Nevertheless, the K-function characterizes only the second-order property of the stochastic process that generated the point pattern. In addition, other functions can be applied to characterize the second-order properties, such as the J-function, which has been less popular in the forestry. The use of the K and J functions, without prior analysis of the first-order effects, can lead into incoherent results. First and second-order analyzes can be performed in both unmarked (single species trees) and marked (trees of different species) configurations. Once the first and second-order properties are characterized, spatial stochastic models can be fitted for a complete characterization of the spatial point pattern. In the same way, the aim of the present work is to use spatial point process methods to characterize the spatial distributions of cumaru-ferro, garapeira and sumaúma species of the Brazilian Amazonian native forest, located in Canary Farm, Bujari, Acre. The results showed that all species present cluster patterns of trees. Poisson cluster stochastic models were successfully fitted for the garapeira and sumaúma species, which allowed a complete characterization of the spatial distribution of the trees of these species in the study region. Bivariate point pattern analyzes have shown that trees of different species tend to repel each other. Finally, the results showed that with the appropriate use of spatial point process methods is possible to characterize the complex spatial interactions that exist among species of native forest trees.