Estuaries are highly variable environments where fish are subjected to a diverse suite of habitat features (e.g., water quality gradients, physical structure) that filter local assemblages from a broader, regional species pool. Tidal, climatological, and oceanographic phenomena drive water quality gradients and, ultimately, expose individuals to other habitat features (e.g., stationary physical or biological elements, such as bathymetry or vegetation). Relationships between fish abundances, water quality gradients, and other habitat features in the Sacramento-San Joaquin Delta were examined as a case example to learn how habitat features serve as filters to structure local assemblages in large river-dominated estuaries. Fish communities were sampled in four tidal lakes along the estuarine gradient during summer-fall 2010 and 2011 and relationships with habitat features explored using ordination and generalized linear mixed models (GLMMs). Based on ordination results, landscape-level gradients in salinity, turbidity, and elevation were associated with distinct fish assemblages among tidal lakes. Native fishes were associated with increased salinity and turbidity, and decreased elevation. Within tidal lakes, GLMM results demonstrated that submersed aquatic vegetation density was the dominant driver of individual fish species densities. Both native and non-native species were associated with submersed aquatic vegetation, although native and non-native fish populations only minimally overlapped. These results help to provide a framework for predicting fish species assemblages in novel or changing habitats as they indicate that species assemblages are driven by a combination of location within the estuarine gradient and site-specific habitat features.