The Garrison Diversion Unit (GDU) of the Pick-Sloan Missouri Basin program was authorized in 1965, with the purpose of diverting Missouri River water to the James River for irrigation, municipal and industrial water supply, fish and wildlife habitat, recreation, and flood control. The project was reauthorized in 1986, with the specification that comprehensive studies be conducted to address a variety of issues. One of these ongoing studies addresses potential impacts of GDU construction and operation on lands of the National Wildlife Refuge (NWR) system, including Arrowwood and Sand Lake Refuges (the Refuges) on the James River. A number of concerns at these Refuges have been identified; the primary concerns addressed in this report include increased winter return flows, which would limit control of rough fish; increased turbidity during project construction, which would decrease production of sago pondweed; and increased water level fluctuations in the late spring and early summer, which would destroy the nests of some over-water nesting birds.

The facilitated workshop described in this report was conducted February 18-20, 1987, under the joint sponsorship of the U.S. Bureau of Reclamation, the U.S. Fish and Wildlife Service, and the North Dakota Game and Fish Department. The primary objectives of the workshop were to evaluate the feasibility of using simulation modeling techniques to estimate GDU impacts on Arrowwood and Sand Lake Refuges and to suggest enhancements to the James River Refuge monitoring program. The workshop was structured around the formulation of four submodels: a Hydrology and Water Quality submodel to simulate changes in Refuge pool elevations, turnover rates, and water quality parameters (e.g., total dissolved solids, turbidity, dissolved oxygen, nutrients, water temperature, pesticides) due to GDU construction and operation; a Vegetation submodel to simulate concomitant changes in wetland communities (e.g., sago pondweed, wet meadows, deep and shallow marsh); a Fish submodel to estimate changes in abundance or biomass of rough fish (carp, buffalo) and sportfish (northern pike); and a Wildlife submodel to calculate indices of waterfowl abundance or habitat suitability (e.g., for mallards, western grebes, migrating diving ducks, white-faced ibis, egrets, over-water nesters). Submodels considered weekly to monthly changes in pools within a Refuge over a time horizon of 30-50 years.

Based on workshop discussions and past experience with impact analysis modeling, a phased modeling approach was recommended for the James River Refuges analysis. The first phase would involve two modeling efforts. The existing Sand Lake hydrology model, and a similar one developed for Arrowwood NWR, would be validated and used to predict changes on pool elevations and winter inflows to each pool for a variety of GDU alternatives. Outputs from simulations would then be evaluated in terms of potential fish and wildlife impacts. For example, the models could generate indices comparing the magnitude and timing of winter inflows for pre- and postproject conditions; fisheries biologists could then use these indices to better quantify their concerns relative to potential changes in the frequency of rough-fish control. The other modeling effort in the first phase would involve developing a sago pondweed growth model to integrate Refuge monitoring data and existing literature and perhaps to address some questions concerning turbidity impacts. A second phase of simulation modeling would be undertaken only if the initial analyses of hydrologic outputs indicated significant potential problems and if monitoring and research projects had clarified some of the biological and physical processes that cannot be modeled reliably at the present time (e.g., resuspension of sediments by carp, immigration and winter mortality of fish, loss of waterfowl nests due to wave action). The second phase would attempt to develop an integrated impact assessment model.

In order to address some of the biological and physical processes that presently are not well understood, a number of studies and enhancements to the Refuge monitoring program were suggested. The Hydrology and Water Quality workgroup recommended increasing turbidity and dissolved oxygen sampling, dropping expensive analysis of some trace elements, adding more pesticide analysis (including some biological monitoring), and developing better area-capacity data for the Sand Lake hydrology model. The Vegetation workgroup suggested expanding the number of monitoring stations, monitoring photosynthetically active radiation by depth, and modifying the biomass sampling procedure and schedule. Also suggested were additional analyses of existing Refuge monitoring data and additional field studies concerning sago growth under a variety of environmental conditions and effects of rough fish density on sago. A careful examination of Refuge narrative reports was recommended by the Fish workgroup to characterize conditions that led to various rates of winter-kill. Monitoring enhancement related to a better understanding of fish population dynamics included increasing dissolved oxygen monitoring, continuing present monitoring of fish movement upstream from Jamestown Reservoir into Arrowwood NWR, initiating similar efforts for upstream movement into Sand Lake NWR and downstream movements into both Refuges, and augmenting the present gillnetting program (or replacing it) with sampling for population and age/size structure estimates. The Wildlife workgroup suggested estimating the relative density of mallard nests in over-water and wet meadow nesting areas, estimating the number of western grebe nests lost due to wave action, delineating wet meadows on the Refuge vegetation maps, estimating annual tuber consumption by birds, and monitoring insect/macroinvertebrate abundance. The workgroup also suggested research studies to better understand the relationships between food supplies and the growth and survival of ducklings and young grebes.

the workshop discussions also helped identify some suggestions for modifying project features that, if feasible from an engineering and operational standpoint, would reduce impacts on Refuge lands. These suggestions included: designing drains with control structures or small "reregulation" reservoirs to hold winter return flows that might adversely affect rough fish control, spreading construction activities over a number of years to reduce potential impacts of turbidity on sago pondweed in any single year, scheduling construction to occur after the spring sprouting and elongation growth stages to reduce impacts on sago pondweed, and installing "quick acting" control structures at Arrowwood NWR to reduce pool level fluctuations that might destroy nests of some over-water nesting waterfowl.