Modelling the influence of weed competition on growth of young Pinus radiata. Development and parameterization of a hybrid model across an environmental gradient

Abstract

A hybrid tree growth model sensitive to competition from a diverse range of weed species was developed and validated using data from a nationwide series of 24 plots covering an extensive environmental gradient. Tree growth in plots with weeds over the first 3 years following establishment was predicted by reducing potential growth from an empirically determined optimum rate for the site (weed free) using a competition modifier, which accounts for the degree of weed competition for both light and water availability. Diameter growth of trees in plots with weeds was initially predicted by including a light competition modifier into the model developed for weed-free plots. This model accounted for 87% of the variance in diameter of trees growing with weeds. Although inclusion of this modifier provided unbiased predictions of tree diameter growth on wet sites, model predictions overestimated diameter growth on dryland sites by on average 11%. Addition of a competition index for water based on treatment differences in average fractional available root-zone volumetric water content significantly (p < 0.001) improved the precision (R2 = 0.96) and reduced the bias of the overall model. A cross validation of this final model indicated it was unbiased and relatively accurate (R2 = 0.95).