Emergent properties of individual-tree growth models

Several individual-tree growth models have been developed within the last 15 years. All these models comprise a set of mathematical equations (sub-models) describing the establishment, increment, and mortality of individual trees. The status of a whole forest stand (basal area, volume, mean diameter, etc.) can be obtained by summing up or averaging the attributes of all individual trees. Hence, a specific dynamic of the whole stand is expected to emerge from the interactions of all sub-models. From previous research (e.g. field experiments) several basic principles of forest stand dynamics are known: (1) Due to early and heavy thinnings individual trees change their stem taper, become more stable, and therefore less susceptible to breakage by snow and wind. (2) The relationship between increment and stand density can be expressed by an optimum curve, i.e. maximum increment does not directly correspond to maximum stand density but can be found at a slightly lower level of stand density. Thus, any deviation from this so-called "optimal stand density" results in a decreased increment. (3) Site productivity is not solely described by dominant height, but is further determined by potential density which depends on site factors others than those which determine dominant height. (4) At a given site a specific succession of the tree species composition can be expected depending on their different preferences for particular soil properties, their different shade tolerances, and therefore their different abilities to respond to stand treatments like thinning or harvesting. (5) In unmanaged, natural forests a typical succession of development phases can be observed, for example, regeneration phase, phase of optimal growth, and finally, old growth phase with increased mortality of the largest trees due to senescence effects. The objective of this research project is therefore to test whether these basic principles do emerge in an individual- tree growth model simply from the interactions of the respective sub-models.

The project addressed the individual tree growth simulators BWINPro, MOSES, PrognAus and SILVA, which are widely used in Germany and Austria. The focus was to investigate if the simulated forest growth corresponds to commonly accepted principles of forest growth in stands. The following principles were investigated: The ratio of tree height:tree diameter (h/d-ratio) is an important measure for stand stability. Trees with high height:diameter ratios are prone to snow damage, but also to wind damage. Open grown trees have very low h/d-ratios, whereas trees in dense stands have very high h/d-ratios. All four growth simulators could reasonably well predict the relationships between height:diameter ratio and stand density. Trees need water, light and nutrients to grow. Since these are limited resources, there are only a certain number of trees that can grow on a certain site at a certain age. This maximum number of trees for a site is know from growth and yield experiments and forest inventory data. With respect to the maximum number of trees that could be grown on a certain area, the results of the simulators differed significantly. This is probably due to one of the sub-models of the simulator, the model for the mortality of trees. This model should therefore be revised. Concerning the steady state species composition - simulating relatively long periods without management influences - the estimations by PrognAus corresponded well with the expert`s expectations for sites where Norway spruce and European beech are expected to be the natural tree species. As most of the Austrian forests belong to this group of sites, PrognAus can be considered to be a reliable tool to support Austrian forest management in predicting the long-term development of species composition when refraining from human influence. The results obtained with BWINPro showed that the simulated steady state species composition is independent of the site conditions; thus, developing a site dependent model for estimating forest regeneration for BWINPro represents a worthwhile task for future research. From observations in natural forests it is known that in the temporal forest development certain stages can be recognised. The results of this project show that PrognAus and BWINPro realistically simulate these stages. The project provided important findings concerning the realism of temporal forest development simulation using forest growth simulators, and emphasised the practical value of forest growth simulators as important tools to support planning and decision making in sustainable forest management.