J. D. Sauerländer's Verlag, Bad Orb

The change from the traditional yield tables to individual tree growth models for beech has been met with enthusiasm as well as scepticism. Yield tables are easy to use, and they are „robust“, which means that predictions are usually found to be sufficiently reliable by practitioners, provided the yield table predictions are calibrated with the measured inventory data. Growth estimates for individual trees on the other hand, have many sources of error and simulated graphics may create a false impression that all the predictions are accurate. Comparisons of the actual growth with the predictions are rare, especially at the extremes of density, age and growing site; the few existing ones give rise to scepticism (WINDHAGER, 1999; GADOW and HEYDECKE, 2000). In addition to the greater uncertainty, individual tree growth models have a practical disadvantage when compared with a stand model. Harvest events can be defined by selecting individual trees for removal on a computer screen. But simulating alternative management regimes for a particular stand requires specification of numerous harvest events which should be more or less consistent with the language that foresters use. Thus, there is a need for simpler models of forest dynamics. Such models should be able to a) simulate harvest events, i.e. estimate removals using existing silvicultural terminology and b) predict the growth response following a particular harvest event.

This paper presents an attempt to achieve these two objectives, based on longterm growth studies of European Beech. The European Beech (Fagus sylvatica L.) is the dominant tree species in central Europe. It prefers a maritime climate with moderate fluctuations of temperature and precipitation and is rarely found in areas with extended very cold and dry periods. The European Beech is shade tolerant and prefers moist sites. The geographic range where beech is dominant and the range where it occurs, have been described in numerous research papers1). According to BOLTE et al. (2007) more than 20 maps of its phytogeographical range have been published.

Possibly the first approach to beech growth and yield modeling was the yield table published by PAULSEN (1795). These and subsequent generations of yield tables portray the development of pure beech stands in regular time intervals for a fixed silvicultural treatment and site quality. The empirical data base of beech growth observations was considerably improved and extended since the end of the 19th and during the 20th century. The most widely used German yield tables published by SCHOBER (1972) and DITTMAR et al. (1986) are based on the works of SCHWAPPACH (1911) and WIEDEMANN (1931) who had been the first custodians of the extensive database of the Prussian Forest Research Institute.

During the second half of the 20th century, many yield tables were supplemented or even replaced by flexible stand growth models. Due to improved computer technology and increasing information needs, several innovative single tree growth simulators were developed during the past two decades to evaluate alternative silvicultural strategies (HASENAUER, 1994; STERBA and MONSERUD, 1997; PRETZSCH and KAHN, 1998; PRETZSCH et. al., 2002; NAGEL et al., 2002). It is generally assumed that single tree models are especially useful in uneven-aged forests with varying densities and species combinations. However, the accuracy of their projections, which cannot be estimated for the range of potential applications, remains to be uncertain. One of the reasons may be that the empirical database is heavily biased towards normal growing conditions and normal silviculture (WINDHAGER, 1999; GADOW and HEY - DECKE, 2000). Thus, the dramatic change from the traditional yield tables to individual tree growth models for beech has been met, not only with enthusiasm, but also with some scepticism. Based on our extensive experience with individual tree models and their uncertainties, we anticipate that it is preferable to first develop a sound stand level model and to use that model to constrain the predictions for individual trees.

The objective of this paper is to complement the range of existing beech growth models with a stand-level approach. Specifically, it should be possible to use the dynamic growth model to simulate alternative silvicultural options for prediction intervals of varying lengths. An important requirement of the model, besides accurate and unbiased growth predictions, is ease of specifying silvicultural treatments. Thus the model will be designed to facilitate silvicultural planning and economic analysis of alternative management options.

Download