Individual tree crown width models for Norway spruce and European beech in Czech Republic

Both spatially explicit and spatially non-explicit individual tree crown width models were developed for Norway spruce (Picea abies (L) Karst.) and European beech (Fagus sylvatica L.) using a large dataset from fully stem-mapped permanent research plots (PRPs) located in various parts of the Czech Republic. A number of tree and stand characteristics were evaluated for their potential contributions to the description of the crown width variations. In addition to diameter at breast height (DBH), other significant predictor variables identified for crown width models are dominant height (HDOM), height-diameter ratio (tree slenderness coefficient), height to crown base, DBH sum of all tree species per PRP and proportion of DBH sum for a species of the interest (spatially non-explicit competition measures), and Hegyi's index (spatially explicit competition measure). Among various base functions evaluated, a simple power function was chosen to expand through the integration of tree and stand variables. The PRP-level random effects were also included using mixed effect modeling approach. Both spatially explicit and spatially non-explicit models and their mixed effect versions described large parts of the crown width variations [R-adj(2) = 0.76-0.78 (Norway spruce), 0.70-0.73 (European beech)] without significant residual trends. For both species, spatially explicit mixed effect model described larger part of the crown width variations than its spatially non-explicit mixed effect counterpart. The models showed that after DBH, height diameter ratio for Norway spruce and HDOM for European beech showed the largest contribution to the models. The crown width increased with increasing dominant height, but decreased with increasing height-diameter ratio, height to crown base, and competition among the trees within a stand. For both species, spatially explicit competition exhibited significantly larger effect on crown width than spatially non-explicit ones. This suggests that spatially explicit models can be more appropriate for description of the individual tree growth dynamics than spatially non-explicit ones. However, because of a little difference between the fit statistics of spatially explicit and spatially non-explicit models, the later models can be applied for precise predictions of crown width as they do not require spatially explicit competition measures, which are computationally complex and difficult. (C) 2016 Elsevier B.V. All rights reserved.