Forest stand-by-environment interaction invalidates the use of space-for-time substitution for site index modeling under climate change

Quantification and prediction of forest site productivity potential under climate changes are core issues of growth and yield modelling and crucial for the effective management of forest stands. Climatic factors typically enter these models as long-term baseline data of mean air temperature and sum of precipitation. Such models frequently are parameterized using spatially distributed data, an approach that can be considered as indirect space-for-time substitution (SFTS), which is based on the assumption that the growth response across spatial environmental gradients is equivalent to the dynamic growth response in time when site factors shift in situ due to environmental changes over time. In site index modelling this assumption has never been thoroughly tested. We applied procedures to test the validity of the SFTS approach using altitudinal transect analysis of data from long-term experimental plots with Norway spruce (Picea abies) in southwest Germany. The following procedures were applied 1. Partitioning of spatial and temporal variance components of site index and air temperature; 2. Test of the equivalence of spatial and temporal variation in models predicting site index from air temperature; 3. Bias analysis of the SFTS-and the altitudinal range-specific individual models. The results show that the response of site index to changes in growing season air temperature as estimated by the SFTS-based approach is signifi-cantly different from the observed temporal dynamics within the plots. The estimated sensitivity of site index to changes in growing season air temperature is significantly larger in the SFTS-model. Bias trends increase with increasing air temperature in all altitudinal ranges. We found significant effects of stand-by-environment interaction on site index responses to changing growing season air temperature. Possible reasons for the found differences are discussed. Furthermore, based on an independent data set from the German National Forestry Inventory (NFI), a SFTS-based climate-sensitive site index model was compared to a state-space modelling approach (SSM) which considers the dynamics of environmental changes over time. The results show that the SSM is statistically superior to the SFTS approach, corroborating the findings of the altitudinal transect analysis. Finally, future site index dynamics of Norway spruce were simulated on the basis of climate change scenarios using the SSM approach. We recommend for all climate sensitive forest growth and yield models based on spatially and temporally distributed data to properly account for within-and between-group effects to avoid erroneous model predictions under climate change.

Automated summary

Quantification and prediction of forest site productivity potential is crucial for effective forest management. Climate change is an important factor that needs to be considered in growth and yield modeling. The assumption that growth response across spatial gradients is equivalent to dynamic growth response over time has never been thoroughly tested in site index modeling. The state-space modeling approach is statistically superior to the space-for-time substitution approach in predicting site index dynamics.