Estudo do crescimento diamétrico de cedro (Cedrela fissilis) por regressão quantílica não linear

• Autor(a) principal: Frühauf, Ariana Campos
• Data de Publicação: 2022
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Repositório Institucional da UFLA
• Texto Completo: http://repositorio.ufla.br/jspui/handle/1/50116
• topic: Cedro - Anéis de crescimento; Cedro - Curva de crescimento; Dendrocronologia; Modelo Brody; Regressão não linear; Cedar - Growth rings; Cedar - Growth curve; Dendrochronology; Brody model; Non-linear regression; Estatística

Forests play a key role in sustaining life. However, they are decreasing, and so many tree species, mainly from native forests, as is the case of cedar (Cedrela fissilis), are going into extinction due to their intense exploitation. Therefore, it is necessary to study its growth as an aid in obtaining better management plans for the forests that contain it. Tree growth, in general, is well adjusted by nonlinear regression models. However, it can commonly present problems caused by heteroscedasticity or possible asymmetry in the distribution of residues. Quantile regression can overcome these problems by allowing estimates in different quantiles and thus generating a more complete mapping of the development of the forest under study. The objective of this work was to compare the fit of the nonlinear Logistic, Gompertz, von Bertalanffy, Brody, Chapman-Richards, and Weibull models by the least-squares method and by quantile regression, for the data of the diameter of breast height (DBH) accumulated at the overtime for 56 trees sampled in a native forest using a non-destructive technique and classify them as small, medium and large. The coefficient of determination, the mean absolute deviation, and the Akaike information criterion was used to assess the quality of the adjustments and the adequacy of the models was verified through residual analysis and non-linearity measures. All computational analysis was performed using the R statistical software and the Brody model was the one that best adhered to the data.