Modelling approaches for relating effects of changein river flow to populations of Atlantic salmon andbrown trout

Modelling approaches for relating discharge to the biology of Atlantic salmon, Salmo salar L., and brown trout, Salmo trutta L., growing in rivers are reviewed. Process-based and empirical models are set within a common framework of input of water flow and output of characteristics of fish, such as growth and survival, which relate directly to population dynamics. A continuum is envisaged incorporating various contributions of process and empirical structure as practical and appropriate to specific goals. This framework is compared with, and shown to differ from, approaches whose output is in the form of quantity and form of habitat (or usable area) based on its frequency of use by fish, which then is assumed to have some relationship with fish performance. A simple conceptual modelling approach is also developed to relate water flow to fish population characteristics to assess the likelihood of simple relationships between flow and usable area thresholds. Basic predictions of the model are tested against empirical data from a long-term individual-based study of juvenile S. salar and resident brook trout, Salvelinus fontinalis (Mitchell), in West Brook, Massachusetts. For this system, growth rates of both species increased linearly with flow during spring, summer and autumn months and bore no relation to Q95 or wetted-width discontinuities. Winter is identified as a season during which water might be abstracted most safely, but cautiously given sparse knowledge of wild salmonid fish at this time of year. These results, together with the fundamental conceptual problems inherent in usable area-based approaches, suggest that models that relate directly to fish performance outcomes may be more robust as a basis for flow prescriptions. However, this utility will depend strongly on our ability to generalise from a limited set of empirical studies and to use the results of these studies of management actions to inform and improve future models.