Application and validation of a new approach for modelling benthic invertebrate dispersal: First colonisation of a former open sewer system

Abstract

Within a heavily modified catchment, formerly polluted streams are now free of untreated wastewater. Additionally, the morphology of streams has been improved by physical habitat restoration. Both water quality and structural improvements offered a unique opportunity to investigate the recolonisation of restored sections by benthic macroinvertebrates. As dispersal is a key mechanism for recolonisation, we developed a method to predict the dispersal of 18 aquatic insect taxa to 35,338 river sections (section length: 2m) within the catchment. Source populations of insect taxa were sampled at 33 sites. In addition, 14 morphologically restored sites were sampled and constituted the validation dataset. We applied a "least-cost" modelling approach within a raster-based GIS model, combining taxon-specific aquatic and terrestrial dispersal capabilities with the "friction" that physical migration barriers impose on dispersal of aquatic and terrestrial stages. This taxon-specific modelling approach was compared to a conservative modelling approach, assuming a Euclidean distance of 5km as the maximum dispersal distance for any source population regardless of dispersal barriers. Least-cost modelling showed a significantly better performance in terms of the correct classification rate (CCR) and true predicted absences (specificity), with on average 37% points higher CCR and 42% points higher specificity. Sensitivity was 18% points lower. At 71% of the validation sites, recolonisation was predicted with at least a modest goodness of fit (CCR>70%). Conversely, the conservative modelling approach achieved a modest goodness of fit for only 14% of the validation sites. For 44% of the taxa, least-cost modelling showed a high CCR (=100%), whereas the conservative approach showed a high CCR for none of the taxa. Our approach can help water managers select appropriate sites for restoration to increase recolonisation and biological recovery.

Keywords: Benthos; Dispersal; Fresh waters; Modelling/statistics; Restoration.