Abstract

RODRIGUES, Abel et al. Linear Mixed Modelling of Cone Production for Stone Pine in Portugal. Silva Lus. [online]. 2014, vol.22, n.1, pp.1-27. ISSN 0870-6352.

This study aimed to model the cone production weight of stone pine (Pinus pinea L.) trees using two approaches: classic linear mixed modelling and Bayesian mixed modelling. The field data were collected in the Setúbal Peninsula, the main production area in Portugal, where 51 plots for monitoring cone production per tree during three production periods (2004-2005, 2005-2006 and 2006-2007) were set up. Linear mixed models with a random intercept term were fitted to the whole dataset (416 trees) for the three production periods. Clustered longitudinal mixed models, under a Bayesian approach with random terms related to intercept, time slope and nested effects, were fitted to a subset of the global dataset (9 plots and 76 trees), corresponding to the plots where production data for all trees in the three years were available. The selected models included, as independent variables, crown width, basal area per hectare, tree height and accumulated rainfall in the five-year periods, prior to cone collection. Despite the small period considered in this study, the Bayesian analysis proved to be useful for calculations with smaller samples, and it was indicative about the tree and plot biometric complex dynamics, subjacent to cone production.

Keywords : stone pine; cone production; mixed models; Bayesian analysis; clustered longitudinal data.

        · abstract in Portuguese | French     · text in English     · English ( pdf )