Analysis of holm oak intraspecific competition using Gamma regression

Analysis of plant competition is a major issue in ecology and forestry, as it influences plant growth and plant-environment interactions. Competition is expected to be lower in the sparse tree stratum of open woodlands and agroforestry systems than in closed forests. We have analyzed competition in open woodlands of Quercus ilex in the Iberian Peninsula by studying a 10-year diameter growth increment from collected samples and from consecutive National Forest Inventories. Density was the competition index selected in all models, outperforming more complex distance-dependent indices. The models showed that competition is playing a role in growth but that the covariate most correlated with growth is age or dbh as a surrogate of age. Therefore, below-ground competition is likely to be limiting tree growth, but below-ground competition is lower in open woodlands compared with that in denser forests and thus potential growth (which is strongly linked to age) is almost expressed. Model behavior was improved when data were fitted directly using generalized linear models, which do not require transforming of the dependent variable. Our results showed that modeling growth with the gamma probability distribution resulted in better models compared with Gaussian linear models. Gamma regression offers a great potential for many forestry applications.