An overall lack of generality still exists in relationships between degree of flow modification and ecological/geomorphological response, important for better understanding and management of freshwater resources. Here, we model direct and indirect effects in flow-habitat-biota relationships in the littoral zone of a 230 km stretch of the lower River Murray (Victoria, Australia), a large and highly-regulated lowland river. We rely on a novel (to lotic ecology) statistical method called the 'd-sep test'. A number of theory-laden (including PHABSIM-based) competitive models were assessed, and model reliability further evaluated against alternative models. The model of choice was one with flow directly affecting all abiotic (substratum composition) and biotic (aquatic vegetation, zooplankton abundance, fish abundance) components. The latter were in turn linked in a bottom-up series, with inclusion of a direct effect of aquatic vegetation on fish abundance. However, an alternative model with a direct effect of substratum composition on fish abundance was also partly supported by the data. We discuss the ecological significance of the proposed model(s), along with their value, relevance, and implications in (fish) habitat assessment studies.