Tanya L Rogers, Samuel M Bashevkin , Christina E Burdi, Denise D Colombano, Peter N Dudley, Brian Mahardja, Lara Mitchell, Sarah Perry, Parsa Saffarinia

Identification of the key biotic and abiotic drivers within food webs is important for understanding species abundance changes in ecosystems, particularly across ecotones where there may be strong variation in interaction strengths. Using structural equation models and four decades of integrated data from the San Francisco Estuary, we investigated the relative effects of top-down, bottom-up, and environmental drivers on multiple trophic levels of the pelagic food web along an estuarine salinity gradient and at both annual and monthly temporal resolutions. For zooplankton and estuarine fishes, bottom-up effects appeared to be stronger in the freshwater upstream regions, while top-down effects were stronger in the brackish downstream regions. Interestingly, this contrasts with hypotheses that freshwater systems have stronger top-down effects than marine systems due to the more limited spatial range of their predators. The net effect of environmental drivers was similar to or greater than bottom-up and top-down effects for all food web components. Some relationships (e.g., bottom-up effects of phytoplankton on zooplankton) were seen primarily at annual timescales, whereas others (e.g., temperature effects) were only observed at monthly timescales. Overall, our results provide strong evidence that environmental gradients can structure the relative strengths of bottom-up, top-down, and environmental drivers within food webs. This advances our understanding of the influence of environmental drivers on species interactions, as well as the mechanisms behind variability in the relative strengths of bottom-up and top-down drivers among habitats and ecosystems. Furthermore, these findings can help identify which trophic levels or environmental factors could be targeted by management actions to have the greatest impact on estuarine forage fishes. This approach of leveraging long-term datasets to identify trophic interactions is applicable to a wide range of systems, including complex, dynamic systems along environmental gradients.