Diameter growth of European beech on carbonate: a regional versus forest type perspective

Abstract

We studied the impacts of site, stand and tree variables on the diameter growth of beech trees (Fagussylvatica L.) on carbonate bedrock and examined to what extent the regional diameter growth model can be used at the forest type level. Based on 12,193 permanent sampling plots (500 m2 each) with 94,770 beech trees, we first developed a linear mixed-effect model of the periodic diameter increment at the regional level (Dinaric region, Slovenia, 1.7 thousand km2). Subsequently, we parametrized the model for five forest types within the region (submontane, thermophilous, montane, upper montane and subalpine) and used a homogeneity-of-slopes model to test whether the covariates have different effects in the five forest types. The regional model suggested the positive impact of tree diameter (nonlinear), tree diameter diversity, mean diurnal temperature range and mean annual temperature and the negative impact of basal area, proportion of beech, inclination, rockiness and annual amount of precipitation. Stand basal area and the proportion of beech contributed > 50% of the total explained diameter increment variability, followed by tree diameter (44%), topographic (3%) and climatic variables (< 2%). The regional model was well transferable to forest types; the only variable with a significantly different effect in forest types was tree diameter. However, models at the forest type level differed with respect to the slopes and significance of several predictors, wherein coefficients for some predictors were even of opposite sign. Not all predictors from the regional model were included in the forest type models if predictor selection and model parameterization were performed independently for each forest type. Our study suggests that some growth characteristics of beech can be detected at the regional level only, while analyses at the forest type level can reveal significant differences in beech growth response to tree, stand and environmental variables.