Estimação de riqueza de espécies de macroinvertebrados bentônicos utilizando o modelo michaelis-menten: uma abordagem bayesiana

• Ano de defesa: 2010
• Autor(a) principal: Machado, Eustáquio José
• Orientador(a): Muniz, Joel Augusto
• Banca de defesa: Louzada, Júlio Neil Cassa, Savian, Taciana Villela, Beijo, Luiz Alberto, Sáfadi, Thelma
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: UNIVERSIDADE FEDERAL DE LAVRAS
• Programa de Pós-Graduação: DEX - Departamento de Ciências Exatas
• País: BRASIL
• Palavras-chave em Português: Inferência bayesiana; Modelo não-linear; Estimação em espécies; Método de monte carlo via cadeias de markov; Macroinvertebrados bentônicos; Bayesian inference; Non-linear model; Estimation of species; Monte carlo method via markov chain; Benthic macroinvertebrates
• Área do conhecimento CNPq: Estatística
• Link de acesso: https://repositorio.ufla.br/handle/1/4303

A SURVEY of number of species is the first step in any study about biological diversity and essential for evaluating the conservation status being the extinction an irreversible process, and the lack of information implies in misdeeds in conservation strategies, increasing consequently, the species loss which in most part, have not been studied yet. In this study, the number of biological species called benthic macroinvertebrates was estimated through the non-linear regression model known as Michalis-Menten equation in the ecological context, from samples collected in clear and clean-water stream in Lima´s study (2008). The methodology used was the Bayesian inference for the model parameters estimation through Monte Carlo methods via Markov chains using the Gibbs´ sampler and Metropolis-Hastings algorithm. The autocorrelation function was used for producing independence to the sample obtained. The convergence analysis of the chains was monitored through criteria: Raftery & Lewis (1992), Gelman & Rubin (1992) and Geweke (1992), so that the implementation of the analysis was carried out through routines and BOA pack executable in the R (2010) software. The results found indicated a considerable increasing of the accuracy in parameter estimation using HPD (highest posterior density) interval, Monte Carlo error and standard deviation as criteria. Michaelis-Menten model fitted well to observational data in three data mass used. Because of the results presented, one can conclude they are quite satisfactory and promising by using the Bayesian approach in ecological studies on estimation of species.