Comparing light interception with stand basal area for predicting tree growth.

Empirical and process-based tree growth models have been developedconcurrently; however, their growth predictions have rarely been compared directly. A major difference between the model types is the explicit quantification of foliage biomass as a key variable in process-based models. The aim of this work was to test if this difference has a significant impact on model behavior, especially when simulating silvicultural practices such as intensive thinning. A method was developed to evaluate leaf area and light interception of the mean tree of an even-aged stand from yield table data for Norway spruce (Picea abies (L.) Karst.) in the northern French Alps. Two scenarios were analyzed: (1) a closed stand where leaf area was limited by a maximum leaf area index-represented by young, dense stands, and (2) an open stand where leaf area was limited by the height of the crown base-represented by old, sparse stands. Light interception was calculated based on interpolation between a closed stand (Beer-Lambert law) and an isolated tree (light interception proportional to leaf area). This approach was then used to build a growth model in which competition was described by the ratio of light intercepted by a mean tree of the stand to light intercepted by an isolated tree of the same size. This process-based model was compared with a simpler empirical model in which competition was described by stand basal area. Both models fit well to yield table diameter increment data, the simpler model being slightly better. Simulation of long-term growth, interspersed with thinning, revealed differences between the models. The empirical model was sensitive to thinning and simulated a discontinuous growth pattern, whereas the model based on light calculation showed a smoother growth response to thinning. Simulations of heavy thinning in a dense stand highlighted these differences. The empirical model simulated heavy thinning in a dense stand unrealistically: after thinning, trees immediately grew as fast as trees of similar diameter in an unthinned stand at the same density. In contrast, leaf area played a regulatory role in the model based on light interception: trees with short crowns, as a result of a previous period of growth at high density, benefited little from an increase in light following thinning. It is concluded that models based on physiological or ecological processes have qualitative behaviors different from those of classical empirical models. This is especially important when models are used to make extrapolations far from reference data, for example, to forecast the long-term effect of a new silvicultural strategy.