Traps to Avoid When Modeling Relationships Between Hydropower, Habitat and the Viability of Riverine Fishes

This presentation will highlight traps that this modeler has fallen into and strategies for avoiding them.  First, avoid spending more effort on well-known processes that we know how to model (e.g., flow-physical habitat) at the expense of complex, poorly understood processes (e.g., ecological) with larger effects.  This adds an undeserved veneer of engineering precision to the overall effort.  Second, avoid designing models to address an immediate goal, but rather anticipate future applications.  Third, when conducting PVA, do not give up on model testing because the future is unknown; we have the past and we have space-for-time substitution.  In time, Bayesian methods can describe parameter distributions consistent with historical data.  In space, comparing populations across geographic range can suggest habitat and demographic factors that lead to population growth vs. decline.  Unfortunately, the spatial scope of modeling is often driven by considerations other than the research question at hand (e.g., scope of funding agency), which might encompass only a few populations.  Finally, avoid making models too situation specific.  Rather, treat applied problems in a general way by classifying important aspects of energy activities and generalize their effects.