MODELOS PARA LA ESTIMACIÓN DEL ÍNDICE DE SITIO PARA Pinus durangensis Martínez EN SAN DIMAS, DURANGO

Abstract

Site index is the most commonly used method in Mexico and in several countries to estimate the productivity of forest lands, since it is a numerical expression easy to understand and density-free. The objective of this work was to adjust model groups to predict the site index of Pinus durangensis from data of a the remeasurement of 12 permanent plots settled in a natural forest in the Ejido Vencedores at San Dimas, Durango. Thirteen mathematical models were adjusted through the mean square procedure, ten of which are linear and three non-linear. Were adjusted three groups of models of linear simple regression, logarithmic,quadratic and a mixed model. In regard to the three non-linear, the Chapman-Richards and Schumacher functions as well as the Weibull model were tested. The best models were selected according to the degree of adjustment or Pseudo R2, the F test, the mean square error (MSE), the test of hypothesis of the estimators and the sign of the parameter. Results show that provided the best site index estimation was the logarithmic 6 (Ln[E/H] = β0 + β1[1/E]) since it revealed the highest values for the residual R2 test, which was 0.7419; MSE was 0.01608 and 221. 0 in the F test.