Using model scenarios to predict and evaluate forest-management impacts on soil base saturation at landscape level

Abstract

Silviculture, forest conversion and technical tools of ecosystem management, such as forest liming, display their effects at the landscape level. Therefore their planning and control should take place at the same scale. The primary objective of this work was to assess soil chemical properties and their changes in relationship to ecosystem management, especially forest conversion and forest liming. We calculated scenario models, based on regression analysis, which allow such an examination in the context of understanding landscape processes which can be expected to operate in the sampling region. Stepwise multiple linear regression analysis was used to predict soil chemical attributes (base saturation, pH, C/N content and stock, exchangeable stocks of Ca and Mg) as indicators of site stability or “off-site effects” of forest ecosystems such as effects on clean drinking water from forested watersheds. Because of space limitations, in this paper only the modelling results of base saturation are presented. Base saturation was used as an integrative example for other soil chemical properties. The transformation of measurements to the regional scale, i.e., the regionalization, was calculated for the forested parts of two test regions in the Black Forest on the basis of measured chemical properties of 90–150 soil profiles per test region. The models have a spatial resolution of 50×50 m, which is a spatial scale relevant for forestry practice and forest management. Topographic variables (e.g., elevation, aspect, slope gradient, and slope length), the stratigraphic classification of the geologic substrate, stand characteristics from forest inventory data, and finally technical information about forest liming were the auxiliary variables (secondary site properties) that provided indirect information about base saturation and were available for the whole forested area of the test regions. Base saturation could be predicted with an accuracy of ~50–70% (in terms of the multiple R²) by using these properties as predictor variables in multiple linear regression analyses. The explained proportion of variance was unexpectedly high considering the high geomorphological heterogeneity of the two test regions. Based upon the regionalization models, it was possible to establish scenarios showing the landscape-related effects on base saturation that may be achieved by forest conversion towards a higher proportion of forests with broad-leaved mixed stands and by forest liming. These scenarios allow the interactions between several influencing factors and management strategies and the impacts on the target variable to be synoptically judged. Thus the presented regionalization models achieve the role of decision support tools for the planning of forest management at the landscape level. They allow an assessment of the environmental effects of forest management strategies in terms of site sustainability or preservation of water resources in forested catchments.