A modeling approach for identifying recolonisation source sites in river restoration planning.

Context: The colonization of restored river reaches by benthic macroinvertebrates and fish depends strongly on the proximity of source sites. Central European river networks have been fragmented over decades and populations of sensitive species have been eradicated from large parts of the catchments. Objectives: Identification of remaining source sites (i.e., near-natural river stretches with populations of sensitive organisms) allows to protect them and reconnect them to degraded or restored stretches. We developed an approach to identify source sites of fish and benthic invertebrates and applied it to large parts of Germany. Methods: The approach is based on identifying source sites from sampling data (5919 benthic invertebrate and 2584 fish monitoring sites) depending on the occurring number of sensitive species. For river stretches that have not been sampled we conducted statistical modeling with environmental data (e.g. land use, river habitat data) using boosted regression trees to identify source sites characterized by similar environmental conditions. Results: The results are presented as maps on the level of the federal states. Statistical modeling allowed identification of stream type-specific environmental parameters and their thresholds. The maps allow a visual estimation of the recolonisation potential for river sections considered for restoration. Conclusions: The results provide valuable insight into the perspective of restoration in different regions. For restoration planning we suggest application on a catchment level using environmental data with higher resolution and consideration of additional parameters (e.g. fine sediment input) in lowland regions. [ABSTRACT FROM AUTHOR]