Research Impact Statement: New framework developed for modeling the upper limits of biodiversity as a function of streamflow using principles of the river continuum concept and widely available hydrologic and ecological data. ABSTRACT: Effective water resource management requires practical, data-driven determination of instream flow needs. Newly developed, high-resolution flow models and aquatic species databases provide enormous opportunity, but the volume of data can prove challenging to manage without automated tools. The objective of this study was to develop a framework of analytical methods and best practices to reduce costs of entry into flow-ecology analysis by integrating widely available hydrologic and ecological datasets. Ecological limit functions (ELFs) describing the relation between maximum species richness and stream size characteristics (streamflow or drainage area) were developed. Species richness is expected to increase with streamflow through a watershed up to a point where it either plateaus or transitions to a decreasing trend in larger streams. Our results show that identifying the location of this "breakpoint" is critical for producing optimal ELF model fit. We found that richness breakpoints can be estimated using automated low-supervision methods, with high-supervision providing negligible improvement in detection accuracy. Model fit (and predictive capability) was found to be superior in smaller hydrologic units. The ELF model ("elfgen" R package available on GitHub: https://github.c om/HARPgroup/elfgen) can be used to generate ELFs using built-in datasets for the conterminous United States, or applied anywhere else streamflow and biodiversity data inputs are available.

Water-resources managers are challenged with maintaining a balance among beneficial uses throughout river networks and need robust means of assessing potential risks to aquatic life resulting from flow alterations. This study generated ecological limit functions from species-streamflow relations to quantify potential fish richness response to flow alteration and compared results to currently accepted streamflow management guidelines. Modeled responses of absolute richness change were watershed specific and varied among sample sets derived from hydrologic unit classifications of different sizes (large HUC 6 basins to regional scale HUC 8). With a 20% flow reduction, 10% of HUC 8 predicted a richness decrease in one or more taxa. While absolute richness change was consistent across streams within a HUC, percent richness change was stream size dependent. Comparisons with Instream Flow Incremental Methodology habitat models predicted habitat loss greater than percent richness change; however, predictions for habitat and richness decreased similarly as stream size decreased. Watershed-specific responses from flow reductions could allow water-resources management decisions to be made locally based on the predicted richness change for certain sized streams. Quantitative results highlight the utility of a richness-based framework for generating watershed-specific risk assessments that validate and inform currently employed water-resources management practices.