Untersuchungen zu standortunabhängigen Eigenschaften des Wachstums für verschiedene Baumarten unter Verwendung einer Polynommethode und mit Hilfe des Evolonmodells.

The cultivation of fast growing tree species on arable land is deemed to be one of the most promising alternatives amongst the approaches to the cultivation of renewable resources currently available. As the site factors influencing the growth of poplar and willow have not yet been sufficiently identified and quantified, it is not possible to provide reliable national yield expectations as a function of the prevailing soil and climate conditions at a particular site on the basis of the data currently available. The main objective of this study, therefore, was to develop a supraregional site-based yield model for the assessment of yield expectations for poplar and willow grown on arable land over short rotations. In order to achieve this goal, a total of 38 research sites were established across as many regions of Germany as possible. The clones selected for the experiment were the poplar clones Max 1, Hybride 275 (H 275) and AF 2, and the willow clones Inger and Tordis. Each site was also characterised, on the basis of climate and soil data. From the site parameters recorded, variables were defined for modelling purposes. The climate variables comprised temperature (T), precipitation (N) and aridity index (TI) derived from total or average values from selected periods during the year and during the vegetation period. Within the variables the months were indexed as numbers (January = 1), periods were separated by comma. Parameters of the German soil appraisal such as the condition grade (ZS), the soil quality index value (BZ) and the arable land quality index value (AZ) were used as soil variables. The selected texture parameters were the proportions of sand (S[%]), silt (U[%]) and clay (T[%]). Variables of the pore space parameters were the available field moisture capacity (nFK), the air capacity (LK) and the dry bulk density (TRD). Both the texture and the pore space variables referred to a soil depth of 0 to 60 cm. To increase the precision of the results, site clusters were derived, differentiated by main soil type, and the variables were aggregated. The dGZ, measured in odt ha-1 a-1,at the end of the first three-year rotation was chosen as the dependent variable. The main results of the study are outlined in the following: Negative correlations between dGZ and temperatures in the vegetation period, especially in the months July to September, were observed on the sandy and loamy soils. Precipitation variables always had a positive effect on the growth of poplar and willow across all clones and site clusters. Taking into consideration all experimental sites, precipitation in the period from May to July was most important. The parameters of the German soil appraisal revealed a moderate correlation with growth for all sites. From the appraisal of all sites and of the individual site clusters it became apparent that U[%] represents the most important texture variable for poplar and willow growth. Considered for all sites simultaneously, the nFK had the greatest significance for growth across all of the parameters examined. The site-based yield models were all univariate and often comprised aggregated variables. Under consideration of all sites, the model predictors for Max 1, Inger and Tordis were (nFK * TI5.7) and (nFK * N5.7) for AF 2. Each model was calculated with the inverse or sigmoid approach and revealed an R²korr between 0.45 and 0.64 with a RMSE of an average of 2.0 odt ha-1 a-1. The division into site clusters improved the accuracy of the models considerably. In the sandy site cluster, the models exhibited an R²korr of 0.77 to 0.97 and an RMSE of 0.95 to 1.36 odt ha-1 a-1. These comprised (U[%]* N6.7) for Max 1 and F 2, (S[%]/ TI6.7) for H 275 and (S[%]* T7.8) for Inger and Tordis. For the silty site cluster, significant models could only be determined for the clones Max 1 and Tordis. The model predictor for Max 1 was (BZ * TI4.5) and for Tordis solely (BZ). The calculated R²korr values were 0.84 and 0.95 with a corresponding RMSE of 0.22 and 0.62 odt ha-1 a-1, respectively. For the loamy soils the models for Max 1 and Inger comprised the variable nFK, for AF 2 the variables (nFK * N5.6). R²korr varied between 0.86 and 0.98 with RMSE between 0.56 and 1.21 odt ha-1 a-1.