|
<br />1126
<br />
<br />T, C, GRAND ET AL
<br />
<br />channel habitats during the summer and fall, and, therefore, provide growth (and presumably survival) benefits to
<br />young fi.sh, It is hypothesized that within-day flow fluctuations caused by hydropower operations can directly affect
<br />the suitability of such habitats by altering (1) water temperature (via the mixing of main channel water into
<br />backwaters) and (2) backwater geometry (i.e, volume, depth and wetted area), In addition to directly affecting fish
<br />growth and bioenergetics (Elliott, 1982), water temperature changes associated with flow fluctuations are also
<br />likely to affect the production of invertebrate prey (Mackey, 1977; Huryn and Wallace, 1986). Invertebrate
<br />production could also be influenced by changes in backwater geometry (Osmundson et a!., 2002), especially ifflow
<br />fluctuations result in periodic desiccation of the shallowest parts of the habitat (e.g, Blinn et at"~ 1995).
<br />Despite the impOltance of backwater habitats to juvenile fishes, there are no established approaches for
<br />modelling how river management affects them. Conventional instream flow methods (e.g, the Instream Flow
<br />Incremental MethodologylPhysical Habitat Simulation System, IFIMIPHABSIM; Bovee, 1982; Bovee et aT., 1998)
<br />do not represent the indirect effects of flow and flow variability that may be most important in these habitats.
<br />Existing temperature models (e,g. Culberson and Piedrahita, 1996), while useful for predicting changes in water
<br />temperature in still-water habitats, do not consider the effects of periodic contact with flowing water on water
<br />temperature or habitat geometry. Thus, as part of a more comprehensive modelling project examining the effects of
<br />flow fluctuations on juvenile Colorado pike minnow growth and survival in backwater habitats of the Green River,
<br />Utah, we developed a physical habitat model to predict the effects of mainstem flow variation on backwater
<br />temperature, geometry and food availability. This model combines a cell-based model of backwater geometry, a
<br />pond-based temperature model and a model of invertebrate production,
<br />Here, we describe (1) our physical habitat model, (2) cal ibration of the backwater temperature component of the
<br />model, (3) sensitivity analysis of the invertebrate production model and (4) results of a flow-fluctuation modelling
<br />experiment. The primary objectives of the flow-fluctuation experiment are to predict the effects of within-day flow
<br />iIuctuations on (1) backwater temperature and wetted area; (2) the proportion of a backwater's mean daily volume
<br />exchanged with the mainstem and (3) the availability of invCl1ebrate prey,
<br />
<br />MATERIALS AND METHODS
<br />
<br />Habitat model oven1iew
<br />
<br />We modelled each of six backwaters habitats as individual backwaters connected at one end to a section of the
<br />mainstem of the Green River, Utah (Figures 1 and 2), An hourly time step was used to update external driving
<br />variables (mainstem flow, mainstem water temperature, air temperature, cloud cover, wind speed and humidity);
<br />food production was modelled on a daily basis, The production of invertebrates was modelled as a function of mean
<br />daily water temperature in each backwater and mean annual invertebrate biomass (see Figure 3 for a visual
<br />depiction of the model's schedules), In our model, the availability of invertebrates for consumption by fish depends
<br />on both daily production, and the magnitude and frequency of water exchange between the backwater and the
<br />mainstem.
<br />Our model was implemented using Swarm (Swann Development Group, New Mexico, USA; www,swaml,org),
<br />a software library for agent-based simulation (Minar et at"~ 1996), Swarm provides tools for observing changes in
<br />simulated habitat during modellUns and for generating question-specific model output. Swarm's graphical tools are
<br />also useful for observing and checking the values of habitat variables on an hourly basis, a feature we used
<br />extensively during the testing and validation stages of model development.]
<br />
<br />Study area
<br />
<br />Our st.udy area was a 6-km segment of the Green River located approximately 240 km downstream of Flaming
<br />Gorge Dam within the Omay National Wildlife Refuge in Utah (Figure 1). Within this reach, the Green River has a
<br />low gradient (0.3 rn/km), a primarily sand substrate and includes numerous backwater areas that are used as nursery
<br />
<br />'Computer code available from authors upon request.
<br />
<br />Copyright X> 2006 John Wiley & Sons, Ltd,
<br />
<br />River Res, Applic, 22: 1125-[ 142 (2006)
<br />DOl: lO,1002/rra
<br />
|