Nick Bouwes, ELR
Much of the fish habitat information used to develop empirical fish capacity models is coarsely resolved, often based on surrogate variables for the actual environmental cues to which fish are responding. For example, correlations between fish density and geomorphic units (e.g., pools, riffles) are commonly used to estimate fish abundance. Fish are likely not responding to the geomorphic units, but rather to spatial patterns of depths and velocities. Additionally, relationships between these coarser habitat features and discharge are difficult to quantify and thus cannot inform evaluations of flow or restoration alternatives. Ecohydraulic fish habitat models have addressed these challenges and successfully predicted fish location and abundance.
Habitat suitability index models (HSI) in conjunction with hydraulic models have been used extensively to evaluate how changes in stream discharge influence the availability of usable microhabitats for several species of fish across multiple life stages. PHABSIM is the most popular of these models. Generally, frequencies of fish use (e.g., observed through redd surveys) of particular depth, velocity, substrate, and occasionally cover values, are used to develop habitat suitability curves. The habitat suitability curves are then used to weigh measured or modeled habitat features in a reach to estimate weighted usable area (WUA). As higher resolution data within reaches become increasingly more feasible to collect, models can describe detailed spatial patterns of microhabitat quantity and quality. Further, the carrying capacity of a modeled reach, an input to life-cycle models used in the Columbia River Basin, can be estimated by dividing WUA by the territory or redd size required by an individual.
A criticism of HSI models is that they are site-specific, making extrapolation to other locations unreliable. For example, if observations of depths used by salmonids to develop habitat suitability curves come from a larger stream, this might incorrectly suggest that salmonids cannot use smaller streams where maximum depths are less than the minimum depths used by fish in the larger stream. A more robust approach is to develop more generalized habitat suitability curves using fuzzy inference systems (FIS). FIS are intuitive, flexible in adjusting model parameters and variables, are more robust with imprecise data, can incorporate expert knowledge, and can represent more complex multivariate relationships than traditional HSI models. When combined with high resolution hydraulic model outputs, FIS-based habitat models also provide a spatially explicit depiction of habitat suitability and an estimate of WUA, which can be used to estimate carrying capacity as described for traditional HSI models above.
Another criticism of HSI models is that they do not include important variables such as temperature and food availability; however, since spawning salmon are no longer feeding while occupying redds, HSI models can provide accurate predictions of potential redd locations. ISEMP and CHaMP have developed both HSI and FIS spawner models that are automated across all CHaMP site surveys. In addition, these models have been built for the CHaMP Workbench, allowing anyone to quickly and easily pull CHaMP data and run the models for project specific evaluations. These models are also being used to help identify configurations of geomorphic unit types that can be obtained from reach typing and therefore estimated across a stream network that best support redd construction.
Ecohydraulic models can be used to identify areas where specific types of restoration will likely be the most cost-effective. Extrapolation of these reach-scale estimates to the network scale will help determine the most cost-effective restoration actions and where to apply them. This approach can be used in life-cycle assessments of salmonid populations following restoration efforts.
Findings and Uses
Fuzzy Inference Systems (FIS) can be used to develop habitat suitability curves that are more robust with imprecise data, can incorporate expert knowledge, and can represent more complex relationships than traditional HSI models; however, HSI models remain a valuable tool for making accurate predictions of potential redd locations.
ISEMP and CHaMP have developed and automated both HSI and FIS ecohydraulic spawner models across all CHaMP site surveys.
Using the CHaMP Workbench, anyone can quickly and easily pull CHaMP data and run the models to evaluate potential improvements to carrying capacity due to specific restoration actions.