Development of a benthic macroinvertebrate predictive model based on the physical and chemical variables of rivers in the Republic of Korea

Abstract Predictive models for the benthic macroinvertebrate community based on environmental variables facilitate the identification of the organisms expected to inhabit an area according to the target environmental conditions when restoring rivers. In this investigation, a biotic community predictive model was developed using benthic macroinvertebrate and environmental variable data collected from 1,210 sites in the Republic of Korea from 2010 to 2020. The sites were classified into six groups according to Two Way Indicator Species Analysis (TWINSPAN) and based on their individual abundance/m2 of benthic macroinvertebrates. The TWINSPAN groups were related to 14 variables by stepwise multi-discriminant analysis. The relative importance of the environmental variables that classified each TWINSPAN group was in the order of mean diameter of particle size, catchment area, altitude, velocity, total phosphorus, latitude, pH, longitude, conductivity, water depth, suspended solids, biochemical oxygen demand, stream order, and total nitrogen. Discriminant functions 1–4 showed statistically significant and a predictive model was developed using functions 1 and 2 based on Wilks’ lambda values. The fit of the derived model was confirmed using Sørensen similarity (number of taxa) and Bray–Curtis dissimilarity (individual abundance/m2) analyses between the predicted organisms and those observed at the sites. The distributions of similarity and dissimilarity that were confirmed by stream type ranged from 0.60 to 0.72 and 0.46–0.56, respectively, based on the mean. Based on the predicted and observed values, the ratio of shredders and scrapers to collectors showed similar results overall for each stream type. The predictive model derived using nationally managed available data is expected to be applicable to stream and river restorations in the future, as it provides a statistical assessment of the biotic communities that are expected to inhabit a given environment. Key highlights points A biotic community predictive model is presented. The model was developed using benthic macroinvertebrate and environmental variable data collected from 1,210 sites across the Republic of Korea from 2010 to 2020. The purpose of the model is to identify communities that should be present in river environments after restoration under modified environmental conditions. The model can function on a larger scale to address the increasing need for river restoration from a broader perspective. Model usage will provide successful and sustainable results and meet the needs of policy makers to restore riverine environments.