Hans-Heinrich Schmidt

Modelling water quality, bioindication and population dynamics in lotic ecosystems using neural networks

The assessment of properties and processes of running waters is a major issue in aquatic environmental management. Because system analysis and prediction with deterministic and stochastic models is often limited by the complexity and dynamic nature of these ecosystems, supplementary or alternative methods have to be developed. We tested the suitability of various types of artificial neural networks for system analysis and impact assessment in different fields: (1) temporal dynamics of water quality based on weather, urban storm-water run-off and waste-water effluents; (2) bioindication of chemical and hydromorphological properties using benthic macroinvertebrates; and (3) long-term population dynamics of aquatic insects. Specific pre-processing methods and neural models were developed to assess relations among complex variables with high levels of significance. For example, the diurnal variation of oxygen concentration (modelled from precipitation and oxygen of the preceding day; R 2 0.79), population dynamics of emerging aquatic insects (modelled from discharge, water temperature and abundance of the parental generation; R 2 0.93), and water quality and habitat characteristics as indicated by selected sensitive benthic organisms (e.g. R 2 0.83 for pH and R 2 0.82 for diversity of substrate, using five out of 248 species). Our results demonstrate that neural networks and modelling techniques can conveniently be applied to the above mentioned fields because of their specific features compared with classical methods. Particularly, they can be used to reduce the complexity of data sets by identifying important (functional) inter-relationships and key variables. Thus, complex systems can be reasonably simplified in clear models with low measuring and computing effort. This allows new insights about functional relationships of ecosystems with the potential to improve the assessment of complex impact factors and ecological predictions.

Modelling population dynamics of aquatic insects with artificial neural networks

We modelled the total number of individuals of selected water insects based on a 30-year data set of population dynamics and environmental variables (discharge, temperature, precipitation, abundance of parental generation) in a small stream in central Germany. For data exploration, visualisation of data, outlier detection, hypothesis generation, and to detect basic patterns in the data, we used Kohonens self organizing maps (SOM). They are comparable to statistical cluster analysis by ordinating data into groups. Based on annual abundance patterns of Ephemeroptera, Plecoptera and Trichoptera (EPT), species groups with similar ecological requirements were distinguished. Furthermore, we applied linear neural networks, general regression neural networks, modified multi-layer perceptrons, and radial basis function networks combined with a SOM (RBFSOM) and successfully predicted the annual abundance of selected species from environmental variables. Results were visualised in three-dimensional plots. Relevance detection methods were sensitivity analysis, stepwise method and Genetic Algorithms. Instead of a sliding windows approach we computed the in- and output data of fixed periods for two caddis flies. In order to assess the quality of the models we applied several reliability measures and compared the generalisation error with the long-term mean of the target variable. RBFSOMs were used to denominate and visualise local and general model accuracy. Results were interpreted on the basis of known species traits. We conclude that it is possible to predict the abundance of aquatic insects based on relevant environmental factors using artificial neural networks.

The influence of environmental variables on the abundance of aquatic insects: a comparison of ordination and artificial neural networks

Two methods to predict the abundance of the mayflies Baetis rhodani and Baetis vernus (Insecta, Ephemeroptera) in the Breitenbach (Central Germany), based on a long-term data set of species and environmental variables were compared. Statistic methods and canonical correspondence analysis (CCA) attributed abundance of emerged insects to a specific discharge pattern during their larval development. However, prediction (specimens per year) is limited to magnitudes of thousands of specimens (which is outside 25% of the mean). The application of artificial neural networks (ANN) with various methods of variable pre-selection increased the precision of the prediction. Although more than one appropriate pre-processing method or artificial neural networks was found, R2 for the best abundance prediction was 0.62 for B. rhodaniand 0.71 for B. vernus.

Extracellular phosphatase activity in sediments of the Breitenbach, a Central European mountain stream

Activity of extracellular phosphatases (phosphomonoesterases) was measured in sandy streambed sediments of the Breitenbach, a small unpolluted upland stream in Central Germany. Fluorigenic 4-methylumbelliferyl phosphate served as a model substrate. Experiments were conducted using sediment cores in a laboratory simulation of diffuse groundwater discharge through the stream bed, a natural process occurring in the Breitenbach as well as many other streams.

Organic Matter Dynamics in the Breitenbach, Germany

1969, when the Breitenbach became the focus of investigations at the Limnological River Station of the Max Planck Institute of Limnology (Limnologische Flupstation des Max-Planck-Instituts fir Limnologie) in Schlitz. The stream fauna has been studied intensively since this time (e.g., I1lies 1971, Meijering 1971, Zwick 1984, Wagner 1986, Becker 1990), and only more recently have aspects such as chemistry, hydrology, bacteria, algae, and POM and DOM dynamics been considered (e.g., Brehm and Meijering 1982, Marxsen 1980, 1988, 1996, Cox 1990, Koch 1990, Marxsen and Witzel 1991, Fiebig and Marxsen 1992, Fiebig 1992). The stream catchment is almost completely forested, chiefly by Fagus sylvatica and Pinus sylvestris. The main channel of the Breitenbach is