LINEAR MIXED MODEL TO DESCRIBE THE BASAL AREA INCREMENT FOR INDIVUDUAL CEDRO (Cedrela odorata L.)TREES IN OCCIDENTAL AMAZON, BRAZIL

• Autor(a) principal: Cunha, Thiago Augusto da
• Data de Publicação: 2013
• Outros Autores: Finger, César Augusto Guimarães, Schneider, Paulo Renato
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Ciência Florestal (Online)
• Texto Completo: https://periodicos.ufsm.br/cienciaflorestal/article/view/10557
• topic: Mixed model; generalized last-squares; tree-morphometry; competition indices.; modelo misto; mínimos quadrados generalizados; morfometria da árvore; índice de competição

http://dx.doi.org/10.5902/1980509810557Reliable growth data from trees are important to establish a rational forest management. Characteristics from trees, like the size, crown architecture and competition indices have been used to mathematically describe the increment efficiently when associated with them. However, the precise role of these effects in the growth-modeling destined to tropical trees needs to be further studied. Here we reconstructed the basal area increment (BAI) of individual Cedrelaodorata trees, sampled at Amazon forest, to develop a growth-model using potential-predictors like: (1) classical tree size; (2) morphometric data; (3) competition and (4) social position including liana loads. Despite the large variation in tree size and growth, we observed that these kinds of predictor variables described well the BAI in level of individual tree. The fitted mixed model achieve a high efficiency (R2=92.7 %) and predicted 3-years BAI over bark for trees of Cedrelaodorata ranging from 10 to 110 cm at diameter at breast height. Tree height, steam slenderness and crown formal demonstrated high influence in the BAI growth model and explaining most of the growth variance (Partial R2=87.2%). Competition variables had negative influence on the BAI, however, explained about 7% of the total variation. The introduction of a random parameter on the regressions model (mixed modelprocedure) has demonstrated a better significance approach to the data observed and showed more realistic predictions than the fixed model.