Multinomial logit estimation of a diameter growth matrix model of two Mediterranean pine species in Spain

Abstract

Understanding diameter growth of the Mediterranean pine species is fundamental for evaluating and making appropriate strategic decisions in forest management. A matrix diameter growth model for two Mediterranean pine forest ecosystems in Spain has been developed. Methods Multinomial logistic (MNL) regression was employed to estimate the transition probabilities of a matrix growth model. The model combines individual tree and stand attributes and explores the effects of independent variables and their relationships with tree size and the probabilities of stand diameter growth. The aim was to predict growth of individual trees by diameter class for a 5-year period. Results MNL model results for diameter growth gave better predictions for Maritime pine (Pinus pinaster Aiton) than for Scots pine (Pinus sylvestris L.). Stand simulations showed that diameter growth probabilities depended on productivity and stand density. Stand simulations under fixed conditions showed that the probabilities of diameter growth increased as site productivity increased and decreased with increased stand density index. Conclusions This study demonstrates the usefulness of matrix growth models as tools to predict growth in Mediterranean pine forests. Stand density and site productivity are key factors in explaining Scots pine and Maritime pine forest growth in the study area.