Efeito da competição no incremento em área transversal de Araucaria angustifolia (Bertol.) Kuntze em floresta ombrófila mista, no Sul do Brasil

Abstract

The competition is considered the most important factor on growing trees in a forest. Hence, we aim to describe asymmetric competition and annual periodic increment in cross-sectional areas of trees of Araucaria angustifolia (Bertol.) Kuntze in a mixed tropical rainforest. Dada was from from the PELD – “ Conservation and Sustainable management of forest ecosystem”- FLONA of Sao Francisco of Paula, RS and FLONA of Irati. For the last one, data was used just to compare annual periodic increment in cross-section area with Johann index. First, data was analyzed, and when it was necessary to correct outlier diameters, the mean was applied, or the linear regression from the ordinary last squared, quantic and robust regression. Crown measurements were made, what gave information about horizontal vertical and the distance between pseudo whorls, percentage of green-leafs, sociologic position and crown dimension. Competing tree was made by the concept of Bella and Basal area factor (1 and 4) and Hegyi, BAL and Johann competition index with simulations of sum, average and maximum value of “A” for inter and intra competition, respectively, between araucarias, broadleafs and total. It was made 3 modifications on the Johann index to include height of the competing tree. The building of the model IPAg used dendrometric variables, Johann index and transformations by STEPWISE. Generalized linear model (MLG) was used, and was necessary to use logarithmic tansformation on the IPAg variable (NORMAL distribution and logarithmic link function). To improve models estimatives, it was selected crown variables by Pearson and Kendall correla-tions, principles component analisys, and grouping. The assessment was made by the coen-ficient of determination (R²aj), standard error (Sxy), AIC and BIC and on the residual charts. Results showed that on FLONA-SFP, RS the intra and interspecies competition had hight cor-relation with growth of A. Angustifolia and explain better increment variations of trees. Compe-tition index tested, except BAL and MOD1 and MOD2 presented same distribution in function of IPAg. As competition increase, IPAg drops. However, trees with same competition value presented IPAg low and height. IPAg model building with the competition Johann did not de-scribe variability found on the forest (R²aj.=13,7%, Sxy=18,4, AIC=1.284,4 e BIC=1.295,6), being necessary to include croen variables (FV, PS and shape, to improve the fit and precision (R²aj.=44,0%, Sxy=8,18, AIC=336,9 and BIC= 348,8). It was concluded that 2 principal com-ponents are enough to explain the variation of data (68.2%) and used on the discrimination of two groups of trees, the first with dominant and codominant trees and the second with domi-nated trees. It was noted that reducing crown dimensions was more severe the bigger and closer was the competitor; and it occur in the sides and up-down direction. This fact explain the heterogeneity of the increments found on trees with same diameter, height and level of competition whe described by the index tested. Thus, IPAg is best when estimated the model of regression have variables to describe crown.