Refine
Has Fulltext
- no (2)
Document Type
- Article (2)
Language
- English (2)
Is part of the Bibliography
- yes (2)
Keywords
- Boosting (1)
- Mboost (1)
- Non-stationarity (1)
- Predictive vegetation mapping (1)
- Site ecology (1)
- Species distribution modelling (1)
- boosting (1)
- model selection (1)
- nonstationarity (1)
- spatial autocorrelation (1)
Institute
Questions Can forest site characteristics be used to predict Ellenberg indicator values for soil moisture? Which is the best averaged mean value for modelling? Does the distribution of soil moisture depend on spatial information? Location Bavarian Alps, Germany. Methods We used topographic, climatic and edaphic variables to model the mean soil moisture value as found on 1505 forest plots from the database WINALPecobase. All predictor variables were taken from area-wide geodata layers so that the model can be applied to some 250 000 ha of forest in the target region. We adopted methods developed in species distribution modelling to regionalize Ellenberg indicator values. Therefore, we use the additive georegression framework for spatial prediction of Ellenberg values with the R-library mboost, which is a feasible way to consider environmental effects, spatial autocorrelation, predictor interactions and non-stationarity simultaneously in our data. The framework is much more flexible than established statistical and machine-learning models in species distribution modelling. We estimated five different mboost models reflecting different model structures on 50 bootstrap samples in each case. Results Median R2 values calculated on independent test samples ranged from 0.28 to 0.45. Our results show a significant influence of interactions and non-stationarity in addition to environmental covariates. Unweighted mean indicator values can be modelled better than abundance-weighted values, and the consideration of bryophytes did not improve model performance. Partial response curves indicate meaningful dependencies between moisture indicator values and environmental covariates. However, mean indicator values <4.5 and >6.0 could not be modelled correctly, since they were poorly represented in our calibration sample. The final map represents high-resolution information of site hydrological conditions. Conclusions Indicator values offer an effect-oriented alternative to physically-based hydrological models to predict water-related site conditions, even at landscape scale. The presented approach is applicable to all kinds of Ellenberg indicator values. Therefore, it is a significant step towards a new generation of models of forest site types and potential natural vegetation.
Species distribution models are an important tool to predict the impact of global change on species distributional ranges and community assemblages. Although considerable progress has been made in the statistical modeling during the last decade, many approaches still ignore important features of species distributions, such as nonlinearity and interactions between predictors, spatial autocorrelation, and nonstationarity, or at most incorporate only some of these features. Ecologists, however, require a modeling framework that simultaneously addresses all these features flexibly and consistently. Here we describe such an approach that allows the estimation of the global effects of environmental variables in addition to local components dealing with spatiotemporal autocorrelation as well as nonstationary effects. The local components can be used to infer unknown spatiotemporal processes; the global component describes how the species is influenced by the environment and can be used for predictions, allowing the fitting of many well-known regression relationships, ranging from simple linear models to complex decision trees or from additive models to models inspired by machine learning procedures. The reliability of spatiotemporal predictions can be qualitatively predicted by separately evaluating the importance of local and global effects. We demonstrate the potential of the new approach by modeling the breeding distribution of the Red Kite (Milvus milvus), a bird of prey occurring predominantly in Western Europe, based on presence/absence data from two mapping campaigns using grids of 40 km 2 in Bavaria. The global component of the model selected seven environmental variables extracted from the CORINE and WorldClim databases to predict Red Kite breeding. The effect of altitude was found to be nonstationary in space, and in addition, the data were spatially autocorrelated, which suggests that a species distribution model that does not allow for spatially varying effects and spatial autocorrelation would have ignored important processes determining the distribution of Red Kite breeding across Bavaria. Thus, predictions from standard species distribution models that do not allow for real-world complexities may be considerably erroneous. Our analysis of Red Kite breeding exemplifies the potential of the innovative approach for species distribution models. The method is also applicable to modeling count data.