Shrimp and redfish studies, Bryan Mound brine disposal site off Freeport, Texas, 1979-1981. Volume I(A): Analysis of data on shrimping success, shrimp recruitment and associated environmental variables.

The analyses reported herein address the potential impacts to the Texas shrimp fishery from offshore disposal of brine associated with the U.S. Department of Energy Strategic Petroleum Reserve Program at the Bryan Mound storage site, near Freeport, Texas through the analysis of the historical data base for the fishery (Gulf Coast Shrimp Data) and associated shrimp recruitment and environmental variables. Time series analyses, involving ARIMA modeling and fourier analysis of monthly brown and white shrimp catches in area 19 for the period 1960-1977 were performed. The fourier analysis power spectrum estimates were disappointing indicating that there was a large amount of variability in the phase of the seasonal cycle from year to year for both species. As such, predictive models based on the results of the fourier analysis were not presented. ARIMA models were estimated for both brown and white shrimp catch. The seasonal trend for both species, based on a twelve month cycle, was evident in the results of the analysis. However, the absolute magnitude of the shrimp catch for any month was not predicted closely, indicating that environmental factors not in the model were influencing the size of the seasonal peaks from year to year. Deviations from historical trends on the order of 30-50 percent should be detectable with these models. Environmental variables used in the study were grouped into six categories: (1) river discharge; (2) precipitation; (3) temperature; (4) salinity; (5) winds, tides, and Ekmam transport; and (6) recruitment and postlarval indices. Based on the availability of data, two temporal groups of variables were defined. These were a 10 year data set (1964-1973) and an 18 year data set (1960-1977). Ekman transport, Texas Parks and Wildlife Department (TPWD) Galveston Bay and Matagorda Bay catch/effort, salinity and temperature, and Bureau of Commercial Fisheries - National Marine Fisheries Service (BCF-NMFS) postlarval catch/tow and salinity variables were available only for the ten year period. A stepwise multiple regression procedure in SPSS was utilized to develop predictive equations relating indices of shrimping success (catch and catch/effort) for brown and white shrimp to these environmental and recruitment variables. The analysis scheme was structured in two phases. In the first phase, regressions were run for each shrimping success index for area 18, area 19 and area 19, 11-15 fm depths, with each group of categorical variables (e.g., discharge or recruitment variables), yielding a total of one hundred and thirty eight equations for the 10 and 18 year analyses. The results of these analyses are presented as summary tables in this report, along with appropriate means and correlation tables. From these results, a best fit data set composed of important variables from each categorical group was formed for each dependent variable. Twenty three final best fit regression models were generated and the results are presented in this report in tabular form. For each of these best fit regressions, a plot of the time series of the dependent variable and two most important independent variables and a plot of the observed, estimated, and predicted values for the dependent variable are presented. In the majority of cases, these final equations should be used to test hypotheses concerning the impact of brine discharge on the shrimp fishery. In almost all cases for both species, the 10 year regressions explained a greater amount of the variance in catch and catch/effort than did the regressions for the 18 year data set. This was partly due to the unavailability of certain important variables (e.g., zonal Ekman transport, postlarval catch/tow, TPWD bay catch and associated environmental variables) for the entire 1960-1977 period, but was also due to lower correlations in the 18 year record between the dependent variables and important environmental variables. Possible reasons for this poorer fit over the 18 year period include inaccuracies in the reporting system during the early years and the introduction of other factors (e.g., economic considerations) during the mid 1970's (at the end of the 18 year period). In general, the brown shrimp regressions explained more variance than did those for white shrimp, especially for the 10 year data. Some of the differences are attributable to the fact that several important variables (e.g., TPWD bay shrimp catch and associated environmental variables) were available only for brown shrimp 10 year analyses. These data were not collected in a systematic manner for white shrimp over the entire 10 year (1964-1973) period. Even taking this into consideration, the regressions for brown shrimp were better. The explanation for this probably involves the fact that white shrimp catch in areas 18 and 19, and especially area 19, 11-15 fm depths, contains more spurious variation than does brown shrimp catch in these same spatial strata. Discharge and precipitation variables were, for the most part, positively correlated with white shrimping success indicators while being negatively correlated with brown shrimping success indicators. Lagged precipitation and discharge variables appeared to be more important for predicting white shrimp indices. Wind, tide, and Ekman transport variables proved to be very important in predicting both white and brown shrimp catch and catch/effort, with Ekman transport being more important for brown shrimp than for white shrimp. Ekman transport variables were generally postively correlated with brown shrimp catch and negatively correlated to white shrimp catch. White shrimp catch was more closely related to wind speed and direction, but the ecological basis for these trends are not clear. The results of this study point to these relationships as areas of concern for future studies. For brown shrimp, TPWD bay shrimp catch and postlarval catch were important predictors of brown shrimping success, as were the environmental variables (temperature and salinity) collected with these recruitment data. To assess the importance of fishing effort in predicting shrimp catch, best fit regressions were run with effort as one of the independent variables and the results were compared to the results of the best fit regression analyses without effort. Effort was more important for the 18 year equations than for the 10 year equations due to the better fits with environmental variables and recruitment for the ten year period. The methodology whereby these regressions equations can be used in impact assessment is discussed. The methodology centers on establishing 95 percent confidence limits for predicted shrimp catch or catch/effort from the models, using the suite of pertinent environmental variables for the year to be tested. If the observed value for catch or catch/effort for the year falls outside these confidence limits, the null hypothesis of no significant change in catch or catch/effort due to brine discharge is rejected. This methodology assumes that other environmental factors not considered in the development of the model and not operative during the 1960-1977 period, are not occurring in the year for which impacts are being assessed. Q-mode cluster analysis was used in an attempt to classify good and poor brown and white shrimping years (based on the criteria of catch) using environmental variables that were important in the categorical equations. As expected, the results were better for the cluster analyses based on the ten year data set as compared to the results for the 18 year period, with especially poor results for brown shrimp catch in area 19 for the 18 year period. Analyses which included variables important to white shrimping success and to brown shrimping success generally showed results similar to those from analyses involving only variables important to predicting white shrimping success.