Please use this identifier to cite or link to this item:
|Title:||Assessing impacts of operations on fish reproduction in Missouri River Reservoirs|
|Authors:||Ploskey, Gene R.|
Harberg, Mark C.
Power, Greg J.
Stone, Clifton C.
Unkenholz, Dennis G.
|Keywords:||Reservoirs--Missouri River--Environmental aspects|
Fishes--Effects of water levels on--Statistical methods
Fishes--Missouri River--Reproduction--Effect of water levels on
|Publisher:||Environmental Laboratory (U.S.)|
U.S. Army Engineer Waterways Experiment Station.
|Series/Report no.:||Technical Report;EL-93-21|
|Abstract:||This report describes a method for predicting impacts of system-operating alternatives on fish reproduction in six Missouri River reservoirs (Fort Peck, Sakakawea, Oahe, Sharpe, Francis Case, and Lewis and Clark). Effects of seasonal or annual variations in reservoir hydrology on catches of young-of-year (YOY) fish in summer were quantified using correlation and regression analyses. Software was developed that predicts YOY catch and calculates a fish reproduction index (RI) for every possible year in the 93-year period of record (1898-1990) and any operational alternative. The method allows users to evaluate operational alternatives by comparing results from a long chronology of predicted indices. Small sample sizes and poor correlations between YOY fish catch and most fish stocking variables kept researchers from using stocking variables as covariates in regression analyses. Despite data limitations, the number of fingerling walleye stocked apparently is a legitimate covariate. The YOY walleye catch in Lake Sakakawea was adjusted to include only nonstocked YOY as a dependent variable. This adjustment resulted in a much stronger relation between YOY catch and change in area from April through June than when catch consisted of both stocked and naturally produced walleye. Correlation of YOY catch with weather variables yielded few consistent or useful results, and weather variables were not used in regression analyses. Correlation and regression analyses using hydrologic variables derived from daily data provided little or no improvement in predictive capability over variables derived from end-of-month data. Many highly significant relations were found by regressing the geometric mean catch of YOY fishes on hydrologic variables derived from monthly data. Four system-operating alternatives were evaluated with an integrated model that pooled and post-processed predictions for all reservoirs and indicator species. Alternatives differed mainly in minimum reservoir elevations in the four largest reservoirs and in inflows to the two run-of-river reservoirs during drought. Two environmental alternatives allowed for seasonal variation in water-level or hydrologic patterns among years. These alternatives, which provided a year of high water to one of the three largest reservoirs on a rotating basis, produced similar reproductive indices in most years. However, the alternative allowing the greatest summer drawdown produced six exceptionally high RI values and yielded more years with above-average indices than the alternative which limited drawdown. These results are significant because a strong year class of fish can persist for about 5 to 8 years, and sport fishes may dominate the catch of anglers for 3 to 5 years. Alternatives that limit annual drawdown are desirable only for severe drought periods when the fish reproduction and reservoir fisheries are both adversely affected by low water. The integrated model depends upon predicted hydrology from 1898 to 1990 to calculate independent variables, so values of some variables were outside the range of data used to derive regression equations. Extrapolation beyond the original data is not a serious problem for the integrated model because predictions for every reservoir and species were standardized to values between zero and one. Consequently, a prediction from a single equation cannot overly bias the composite annual estimate of the RI. Also, the integrated model was designed solely to compare alternatives, not to make quantitative predictions. Extrapolation is of concern for users making predictions of YOY catch. In these cases, input data should be screened, or users must assume that relations are consistent over a wider range of values of independent variables than ever observed.|
|Gov't Doc #:||Technical Report EL-93-21|
|Rights:||Approved for Public Release; Distribution is Unlimited|
|Size:||61 pages/2.890 Mb|
|Appears in Collections:||Technical Report|
Files in This Item:
|TR EL-93-21.pdf||2.96 MB||Adobe PDF|