Multilayer networks in landscape ecology: a case study to assess changes in aquatic habitat connectivity for flying and non-flying benthic macroinvertebrates in a Danube floodplain

Context In efforts to mitigate anthropogenic impacts on floodplain biodiversity, restoration measures that enhance habitat connectivity have been applied. However, these approaches have either neglected the spatial position of water bodies or the dynamic nature of the floodplain ecosystem. Objectives This study focuses on the novel application of the multilayer network framework to assess changes in the aquatic habitat connectivity in floodplains, showcasing its application in the context of aquatic passive dispersal (drift) of two indicator groups of benthic macroinvertebrates (Oligochaetes and Chironomids) Methods Our case study is located in the Donau-Auen National Park in Austria and follows floodplain restoration measures (side-channel reconnection) applied in the mid-1990s. Multilayer networks were constructed to represent the conditions before, short-term, and long-term after restoration to quantify habitat connectivity across inundation frequencies. Our network analyses involved multilayer correlation, static and dynamic monolayer centralities (centrality profiles), and multilayer centrality assessments. We used a Partial Least Squares Regression analysis as a variable selection tool to identify which centrality measures better explained the variance in α diversity and Local Contributions to Beta Diversity (LCBD) of benthic macroinvertebrates. Results In the short-term, our connectivity analysis indicated an increase in habitat connectivity. However, centrality profiles, multilayer correlation, and multilayer centrality techniques identified a long-term decrease in connectivity. Multilayer centralities had higher Variable Importance in the Projection scores (VIP) than their monolayer counterpart in explaining variations in α diversity and LCBD for strict aquatic dispersers. Meanwhile, for flying dispersers, monolayer centralities had the highest VIP scores for explaining α diversity. Conclusions This study underscores the relevance of integrating dynamic aspects of water-mediated transport beyond traditional pairwise distances. Although in this study we apply this tool by showcasing the aquatic passive dispersal mode, the application of this method can be extended to other dispersal modes and representative abilities for diverse groups of aquatic organisms. The expanding cross-disciplinary applications and open-source tool development for multilayer networks offer practical implications for planning and evaluating management measures.