Wirkung von Klimaveränderungen in mitteleuropäischen Wirtschaftswäldern
M. Lindner (July 1998)
The projected change in global climate may have various impacts on forest ecosystems. Changes in the growth of different forest species will also influence the competitive relationships between species, the potential species composition in unmanaged forests and the choice of species in managed forests. While impacts of climate change on physiological processes and the potential natural species composition in forests have been intensively investigated in the last few years, there has been hardly any research on possible consequences in managed forests.
Since the traditional prognostic methods of German forestry, the growth and yield tables, are not suitable for applications under changing environmental conditions, there is a strong need for simulation methods, which are also applicable under changing climatic conditions. For more than 10 years forest gap models have been applied in forest ecology to simulate the impacts of climate change on natural forests. However, the forests of Central Europe have been managed intensively over several hundred years and therefore the results of common forest gap models, which simulate the potential natural species composition (PNV) in unmanaged forests, are of little practical relevance. More realistic analyses of climate change impacts on forests in this region require that forest management activities are included in the simulation experiments. Thus the aim of this thesis was to develop a forest gap model for regional analyses of the impacts of climate change in managed forests of Central Europe.
Application of the forest gap model FORSKA for simulating natural forest dynamics in Central Europe
The forest gap model which was applied in this study, FORSKA2, was originally developed to simulate natural forest dynamics in boreal forests of Scandinavia. Various parameter modifications were necessary to adapt the model to Central Europe and then the model was tested extensively at various spatial scales. The model applications indicated that there were limitations in the approach of the existing model. Especially the lack of representation of interannual climate variability led to problems in simulating realistic species distribution limits for beech in the subcontinental climate of Northeastern Germany. Furthermore, the applicability of the model under variable environmental conditions in different parts of Europe was unsatisfactory. For example, it was not possible to simulate realistic species composition and productivity in boreal and temperate forests with one species parameter set. In large-scale model applications across Central and Eastern Europe, however, the model simulated a plausible distribution of major forest types. In order to simulate regional forest species composition in the state of Brandenburg, a nitrogen response function was implemented and parameterized. The comparison of simulation results with a map of PNV suggested that the model is able to realistically simulate regional patterns of forest distribution as well. In several simulation experiments model comparisons were made with another forest gap model (ForClim) and the global biogeography model BIOME to test the plausibility of the model results. The model comparison with ForClim indicated specific strengths and weaknesses of alternative model approaches in gap models, but in general both models simulate qualitatively comparable responses of forest ecosystems to climate change.
Extensions and modifications to the FORSKA model
As a precondition for the implementation of forest management routines it was necessary to improve the simulation of stand structure in the model. The introduction of a density-dependent height growth function improved the simulation of individual tree dimensions and stand structure considerably. To begin with, five alternative height growth functions were tested against measurements from a long-term thinning experiment in a beech stand at Fabrikschleichach (Bavaria). The modified model, FORSKA-HD, was initialized with data from the first stand inventory in 1870 and simulated stand development over 120 years. The function showing the best correspondence with observations uses the relative radiation intensity in the centre of a tree crown as a competition index. Whereas simulated stand structures improved significantly with the modified height growth function, characteristic differences in simulated size distributions remained in comparison with observations. Another model modification, the increase of the growth efficiency of small and intermediate trees, further improved the simulation results.
The modifications of the growth function introduced three new parameters into FORSKA-HD. Long-term forest observation data were used to parameterize spruce, pine and oak. Parameters of other species were estimated depending on the light response and the crown shape in relation to the 4 main forest species.
In the next step initialization and management routines were included into the model. The initialization routines contains numerous algorithms to generate individual tree data from stand level characteristics in forest inventory data. The management module uses a thinning routine which is based on the weibull distribution. Through parameter modifications of the weibull function it is possible to simulate different management strategies. Thinning intervals depend on the height growth of the dominant trees of the stand and thinning intensity follows the development of basal area in the beech and pine yield tables of the former GDR.
Management techniques in the context of global change
Common forest management strategies for stand regeneration, silvicultural treatments and harvesting differ considerably with respect to the response options which they offer in the context of changing environmental conditions. Whereas the conservative thinning from below does not increase the adaptation potential of a forest stand, modern selective thinning strategies enable the development of rich mosaics of forest structure or the early introduction of a new forest species into the stand. In general, forest management may not only respond to climate change at the stand level, but there are also response options in forest planning at the district level, where the diversity of species and forest types can also be increased as a means of regional risk reduction.
Model application in managed forests
The model performance of the extended FORSKA-HD with management routines was analysed in a forestry district in the Dübener Heide, Northeast Germany. This district is characterized by a relatively large variety of forest types and site classes. Besides the usually prevailing pure stands of Scots pine there are numerous stands with deciduous species such as beech, oak, birch, lime or hornbeam. The model application comprised 65 forest stands with a total area of 330 ha. Two management scenarios (conservative and adaptive forest management) were compared with a forest preservation scenario and the PNV-simulation. The comparison of simulated PNV with the actual forest composition in 1993 underlines the strong impact of forest management on the species composition in this area. If management is suspended under current climate, the management history is still visible after 110 years of forest development in the forest preservation scenario because a considerable share of old pine trees remains in many forest stands. In both management scenarios broadleafed species dominate the forests after harvesting of mature pine stands, because in most stands in this district regeneration of broadleafed species was already present under the pine canopy in the year 1993.
Impact of climate change in managed forest stands
All management scenarios were also simulated with a climate change scenario that was developed by the Department of Climate Research at PIK for a nearby meteorological station. In the scenario, temperature increases by +3K and precipitation remains fairly constant. In all management scenarios climate change led to a shift in simulated species composition accompanied in some cases by fairly strong reductions of stand productivity. There were clear differences between scenarios. While in the conservative management scenario beech still dominated many forest stands, it was of little importance in the adaptive management scenario, in which drought-tolerant species were favoured. Since beech is no longer very productive under this climate scenario, the simulated biomass is strongly reduced with the conservative forest management. In the adaptive management scenario, the selection of climatically better-adapted species was able partially to mitigate the drop in productivity.
With regard to species composition, the smallest changes are simulated for the forest preservation scenario. Only on the poorest site class the dieback of beech caused a distinct shift in species composition. As in the conservative management scenario, productivity is significantly reduced. It is interesting to note that in the study region climate change reduced productivity more strongly in those scenarios where the shift of the species composition was small. Differences in the adaptability between managed and unmanaged forests can be explained by the fact that management frequently changes stand structure and thus enables or speeds up the adaptation of the forest to the changes in environmental conditions.
Evaluation of the extended forest gap model
The validation of forest gap models is rather difficult because suitable data are hardly available. The qualitative performance of the model can be tested with long-term observation data from growth and yield experiments. However, it should not be expected that a general model, which was designed to simulate forest development over several hundred years under very different environmental conditions, may at the same time quantitatively simulate the stand development of an individual forest stand with high precision. Validation of growth dynamics in natural and managed mixed stands is generally difficult because there are very few long-term measurements in mixed stands. While simulated thinning regimes in monocultures can be compared with many growth and yield experiments, suitable data from mixed stands are scarce and thinning strategies are more complex. Besides the size-related thinning probabilities for individual trees, the competition effects between species must also be correctly represented in the thinning routine.
Numerous model comparisons were conducted to test the plausibility of the model. The comparison of regional simulation results with a map of PNV (which is an independent model concept based on the floristic composition of the forest vegetation) showed that the spatial pattern of forest types in the state of Brandenburg can be satisfactorily simulated. Model comparisons of two different forest gap models underlined specific strengths and weaknesses which can be attributed to characteristics of the alternative modelling approaches. In general the model performance for Central Europe is plausible. Moreover, comparisons of different model versions indicate that targeted model modifications may considerably improve model results.
Probably for the first time, empirical data from long-term forest trials were applied in this study to validate and improve a forest gap model. The results suggest that such data are very valuable in this respect. They could be used much more often, for example in the validation of the mortality function or the scaling of growth responses to site conditions and climate.
Development of management strategies
Because of the remaining limitations of available data and methods the interpretation of simulation results must be cautious. Nevertheless, the need for improved decision support in forest management under global change calls for more detailed simulation studies to assess the sensitivity of forest ecosystems to changes in climate. Decision making should incorporate risk assessments and risk reduction strategies which acknowledge the uncertainties of current scientific understanding. The forest gap model with management routines is a suitable tool to accomplish this task.
The simulation results demonstrate the strong influence of management on forest development within the next 110 years under current climate and under a scenario of climate change. Different simulation results include valuable information which may be used in forest planning. The impact of climate change on simulated PNV indicates which species are well adapted to the new climatic conditions. Moreover it is possible to use indices from the environmental response functions as a direct measure of the adaptability of different species to the prevailing quality and climate of a site. These indices may be used to define environmental conditions under which a species may be favoured by forest management. Finally the comparison of management scenarios shows how species productivity responds to different silvicultural treatments. This information offers a good basis for the development and assessment of adaptive management and mitigation strategies.
The development of strategies can be based both on positive and negative criteria. The "negative choice" selects stressed species to be cut as part of the regular stand treatment or in precautionary harvests. The "positive choice" in contrast selects well adapted species to be favoured by forest management. Another important question is the reference time for the determination of species suitability. Alternatively to the current climate conditions it is also possible to use the projected future climate conditions against which species adaptation is tested.
The comparison of different management strategies indicates how much climate change restricts forest management in the future. Socio-economic consequences can be estimated, for example, by comparing the cost of adaptation strategies with the potential gains in wood production or by analysing the impact of a shift in species composition on different types of nontimber forest value.
