Community composition is shaped by abiotic conditions and is sensitive to environmental changes and anthropogenic pressures on ecosystems. Existing ecological diagnostic tools already make use of integrative community features such as taxonomy- or trait-based diversity metrics. However, information about taxa co-occurrence is more rarely considered. Yet, co-occurrence networks could offer a complete picture of community organization and have been shown to be sensitive to environmental changes. Applying them to freshwater diatoms could hold great promises for the bioassessment of aquatic systems. As part of the Water Framework Directive (WFD) implementation, biotic and abiotic data are routinely collected to monitor the environmental quality of French river systems and are openly available. Datasets resulting from this monitoring programme offer a unique opportunity to investigate whether co-occurrence networks of freshwater benthic diatoms can be used to predict the ecological quality of rivers. Using data from close to 60,000 distinct diatom sampling operations (and associated measurements of environmental parameters) collected from 2007 to 2023 over 10,000 sites, we propose to: I) analyze co-occurrence networks of benthic diatom abundances in rivers across metropolitan France. II) link co-occurrence network features to the intensities of various categories of anthropogenic pressures. III) compare co-occurrence network features to existing diatom-based biotic indices as predictors of river ecological health. In this work, we evaluate whether current statistical frameworks to analyze highly dimensional co-occurrence networks can be successfully applied to diatom assemblages described quantitatively by microscopic counts. We are also using existing databases on benthic diatom traits to move beyond taxonomic networks towards functional interactions, which are potentially more relevant for diagnosing ecosystem processes and therefore potentially more effective in bioassessing the ecological quality of streams and identifying the probable causes of degradation. Our results illustrate how diatom co-occurrence information can be used to complete existing freshwater bioindication tools.