SUCCESS CRITERIA BASED ON A FISHERY SPECIES, PINK SHRIMP
Team members are Joan Browder, Peter Sheridan, Nelson Ehrhardt, Victor Restrepo, Michael Robblee, and James Nance. Phone discussions have occurred, and a meeting was held April 5, 1996. Nelson Ehrhardt and Joan Browder attended the Success Criteria Workshop held at the University of Miami. This version includes an update based on analyses that followed the Workshop.
Florida Bay and associated coastal areas are the nursery grounds for the Tortugas pink shrimp stock. A severe decline in Tortugas pink shrimp catches and catch rates occurred from 1985 through 1992. The pink shrimp declines are roughly correlated with observations of other signs of deterioration in the health of the Florida Bay ecosystem.
Pink shrimp are important both economically and ecologically in South Florida. Until the decline of the Tortugas fishery, pink shrimp was Florida's number one fishery species in terms of value, and the bulk of the landings came from the Tortugas. A major ecological role of pink shrimp is to transform microorganisms (algae and detritus-feeding bacteria and fungi) into substantive food items for larger animals. Pink shrimp are major prey of gray snapper and other game fish species of coastal South Florida. They also are prey of Great White Herons, Little Blue Herons, and other water birds that feed on the shallow flats in Florida Bay.
Pink shrimp are largely an annual crop in that most of the harvest consists of shrimp still in their first year of life. Therefore, pink shrimp harvests are influenced by environmental conditions over a relatively short time period, and there is not a mixing of annual year classes to confound the determination of environmental effects.
Previous studies suggest a relationship exists between pink shrimp catches and freshwater flow to the coast and Florida Bay. Browder (1985) found a statistically significant relationship between shrimp landings and Everglades water levels (Well P35), which she used as an index of freshwater flow to the coast and Florida Bay. Sheridan (1996) predicted pink shrimp landings from 1987-1988 through 1994-1995 using statistical relationships developed each year between shrimp landings and other indices of freshwater flow (water levels at well P38, rainfall, and water releases at the Tamiami Trail). Sheridan's approach was successful not only in predicting the poor landings of 1987-88 (November-October) through 1991-92 but also the increased landings of 1992-1993 and, particularly, 1993-94, which were years of increasingly higher rainfall.
The Tortugas group of pink shrimp appears to be influenced by some variable related to freshwater inflow. Kinne (1971) observed that salinity is a major variable influencing estuarine invertebrates. Preliminary analyses (P. Sheridan, NOAA/Fisheries, Galveston, unpublished) support the idea that salinities in Florida Bay affect pink shrimp recruitment to the Tortugas fishery. The influence probably is on growth and survival. According to recent modeling work, temporal and spatial differences in salinity and temperature in Florida Bay can produce large differences in recruitment from Florida Bay to the fishery a Browder, NOAA/Fisheries, Miami,, unpublished). The model simulated growth and survival as functions of temperature and salinity based on laboratory studies. Temperatures and salinities were from two stations in Florida Bay. In addition to salinity, detritus and nutrient inputs are other functions of freshwater inflow that may influence pink shrimp growth and survival.
Monthly catch and effort data for the Tortugas pink shrimp fishery have been collected since 1960 and are readily available. The data are collected and processed regularly in a cooperative program between the Florida Department of Environmental Protection and the National Marine Fisheries Service.
We propose "recruitment to the fishery" as the ecological indicator based on pink shrimp. The proposed specific measurement is "catch rate, or catch per unit of effort (CPUE)", determined as catch in weight per vessel day fished. We propose that catch rates be calculated separately for small shrimp (68 shrimp or less to the pound) and larger shrimp (the remainder of the catch). In the penaeid shrimp fisheries, shrimp that number less than 68 to the pound are considered the recruits; however, the rest of the catch also consists mainly of young of the year fish, since penaeid shrimp have a short life span. Monthly CPUE in terms of number of shrimp is closely correlated with monthly population estimates determined by virtual population analyses (VPAs) (Ehrhardt et al. 1996), which supports the validity of CPUE as a meaningful index of population size.
Rainfall strongly influences freshwater inflow to the coast, and South Florida's rainfall is highly variable not only from month to month but also from year to year. To be most useful to the restoration effort, an ecological indicator based on pink shrimp must be referenced to a predicted value that changes with the rainfall of previous months (coinciding with the period when shrimp are in their estuarine life stage) so that the effect of water management actions can be distinguished from the effect of rainfall. Otherwise, one would not be able to tell if year-to-year changes in CPUE were simply the effect of annual variation in rainfall. Needed is a dynamic reference based on rainfall. To provide this reference, we suggest that CPUE be predicted by a regression relationship between annual CPUE and monthly rainfall.
In Figures 1 and 2 are annual CPUE predictions for small and large shrimp for biological (July-June) years of the period of record (from Browder in prep.). These predictions were based on monthly rainfall at Royal Palm Hammock (data provided by Everglades National Park) for the January through December period immediately preceding and overlapping the period of the CPUE biological year. Only the months having the strongest effects were included in the equations. (There were 7 monthly independent variables in the small shrimp equation and 5 monthly independent variables in the large shrimp equation.) The equations that produced these figures explained 89% of variation in small shrimp CPUE and 77% of variation in large shrimp CPUE.
The ratio of actual CPUE to predicted CPUE is the index we propose. A ratio greater than one of actual to predicted CPUE is the success criterion. or desired index value. It indicates an improvement in ecological conditions beyond that provided by rainfall alone. We propose that residuals of the CPUE-rainfall relationship (i.e., differences between predicted and actual CPUE provide rough approximations of the effect of water management on CPUE. We recognize that, particularly with regression models initially developed, differences between actual and predicted CPUE may not be purely the effect of water management. Other variables, still to be identified, may need to added to the regression equation to reduce the variation not attributable to rainfall. For instance, in work to date by Browder (in prep.), a dummy variable, or step function, was added to each equation (Figures 1 and 2) to eliminate strong patterns in the residuals. The dummy variable was zero before 1980-81 and one beginning in 1980 in the small shrimp equation and zero before 1981-82 and one after 1981-82 in the large shrimp equation. Recent work by Rice (1996) suggests that the addition of certain months of sea surface temperature as independent variables might further reduce the unexplained variance in CPUE. The method of calculating a dynamic reference related to rainfall should undergo continual refinement as an improved understanding of shrimp ecology in Florida Bay and nearby waters is developed through research.
Annual CPUE data from the fishery will be compared against the success criterion to determine whether there has been an improvement in ecologic conditions.
Figure 3 shows annual catch rates of all shrimp sizes from 1961-62 through 1995-96.
Monthly date were combined into 12-mo periods from July through June. Dates on figures refer to
the starting calendar year for each 12 month period from July through June.
Figure 4 shows catch rates (catch-per-unit-effort, CPUE) of small shrimp (< 68 shrimp/pound) and large shrimp (> 68 shrimp/pound). Note that CPIJE was greater than 500 pounds per vessel-day in every year prior to 1983-84, but, from 1983-84 through 1991-92, CPUE was less than 500 pounds per vessel day in five out of nine years. The difference in landings between early years and the 1983-84 through 1991-92 period was even greater, possibly because low CPIJE discouraged effort. Landings declined sharply beginning in 1985-86 and remained at historic lows through 1992-93, before increasing recently.
Small shrimp CPUE actually was higher during parts of the period 1985-86 through 1995-96 than during the previous 25-yr, whereas large shrimp CPUE was lower. One likely cause of the higher catches of small shrimp was the development, in the early 1980s, of a market for small frozen shrimp. Small shrimp have historically been a relatively small component of total landings.
Fishery data were not gathered until about 1960, after most of the major alterations in South Florida's hydrologic system had been made. It might be possible to predict what catch rates would be if the hydrologic system had not been altered. Hydrologic output from the Natural System Model being used in the South Florida Restoration Effort would provide the input to the predictive models. Using actual rainfall, the Natural System Model estimates what water stages and flow rates would be if there were no canals, levees, and water control structures.
Under any rainfall conditions that occur, CPUE in the Tortugas pink shrimp fishery would be expected to be higher for any given rainfall year if restoration efforts result in a more natural quantity, quality, timing, and location of freshwater flow to pink shrimp nursery grounds in Florida Bay and lower southwest coast estuaries. With restoration of more natural freshwater inflow, actual CPUE should always exceed predicted CPUE for the year~s rainfall. The long term average of future CPUE would be expected to exceed the annual average between 1963-64 and 1980-81.
Index = ACPUE/PCPUE
where ACPUE is the actual CPUE of the year being evaluated and PCPUE is CPUE predicted by rainfall using the regression relationship developed from data of previous years.
The following two hypotheses are provide the theoretical basis for a dynamic target based on rainfall: (1) Tortugas landings (or CPUE) are influenced by freshwater inflow to Florida Bay and the Southwest coast, as approximated by water stage in the Everglades and (2) water stage in the Everglades, which determines the rate of freshwater flow to Florida Bay and the coast, is determined by (a) local (below Tamiami Trail) rainfall, (b) water releases by the South Florida Water Management District across the Tamiami Trail, and (c) stages in Canal 31N, which determine how much water is diverted from Shark Slough to the east coast. Pumpage into Taylor Slough and releases through S-18 on the Canal 111 may also be influencing factors. The proposed hypotheses are testable, and making these tests is a first need in the development of a pink shrimp-based success criteria using landings data. Determination of other factors influencing pink shrimp CPUE is another major need.
"What if" predictions of current landings (or CPUE) from the Natural System Model would provide an ideal dynamic target with which to gauge an indicator variable. The rainfall input to the Natural System Model will have to be brought up to date and updated annually in order for an index based on the Natural System Model to be used regularly. Annual updating of the rainfall data cannot presently be done because of the archaic structure of the model, which makes incorporation of new rainfall data extremely difficult and time consuming. At present, the rainfall data in the model do not extend beyond 1990, although an update through 1995 is in progress.
Continued statistical analyses of additional data in existing time series will help test and improve the success criteria. Landscape-scale modeling of pink shrimp recruitment as functions of environmental variables and habitat would provide useful insight on the areas in which existing variation in environmental variables and habitat could result in the greatest variation in recruitment. Expansion of continuous-recording environmental stations to more locations in Florida Bay and adjacent nearshore areas would be useful. Completion of quality control and refinement of existing environmental data for Florida Bay would help expedite analytical efforts. To support modeling, laboratory studies measuring growth and survival of pink shrimp as a function of temperature and salinity need to be expanded to cover the entire range of temperatures and salinities found in Florida Bay and nearby waters.
Both refinement of success criteria and interpretation of monitoring results will be greatly facilitated by studies relating pink shrimp landings on Tortugas fishing grounds to the densities and overall abundances of pink shrimp juveniles on nursery grounds in Florida Bay and estuaries of the lower southwest coast, including Whitewater Bay.
Updating of the rainfall input of the Natural System Model is necessary to develop a dynamic target for success criteria based on shrimp predictions using output from the Natural System Model.
Exploratory analyses of data by Browder (in prep.) suggest that strong statistical relationships exist between rainfall and CPUE (Figures 1 and 2.) Patterns in residuals in initial models based solely on rainfall suggest that an abrupt change may have occurred in influencing variables other than rainfall in about 1980-81 for small shrimp and about 1981-82 for large shrimp. It is also possible that the relationship between CPUE and rainfall changed at that time, causing the pattern in residuals. With respect to the small shrimp, some of the change in residuals may reflect a change in fishing strategy in response to a change in the market. Regardless of the cause of the change, adding a dummy variable to the equation eliminated the pattern in the residuals and increased the proportion of variation in CPUE that the equation could explain.
To build better future models, it is important to explore the effect of other variables on shrimp CPUE. Rice's (1996) work suggests that the density of juvenile shrimp in Florida Bay is correlated with sea surface temperatures four months previously. He concludes that an affect of temperature on spawning intensity influences the abundance of postlarval shrimp settling onto nursery grounds in Florida Bay. This influence might also account for some of the presently unexplained variation in CPUE. Pink shrimp recruitment is a promising ecological indicator. Further analyses are needed to fine tune the success criteria and properly interpret year to year changes that are observed.
Browder, J. A. 1985. Relationship between pink shrimp production on the Tortugas grounds and water flow patterns in the Florida Everglades. Bull. Mar. Sci. 37:839-856.
Ehrhardt, N. M., C. M. Legault, and J. M. Nance. 1996. Dynamics of pink shrimp recruitment patterns derived from tuned length-based cohort analysis. International Workshop on Crustacean Stock Assessment Techniques. Campeche, Mexico, August 29-31, 1996.
Kinne, O. 1971. Chap. 4.31. Invertebrates. pp. 821-995 in: H. Barnes (ed.) Marine Ecology. Vol. 1, Part 2. Wiley Interscience.
Rice , J. 1996. An analysis of environmental factors influencing juvenile pink shrimp (Penaeus duorarum) abundance in southwest Florida. M.S. Thesis. University of Miami, Coral Gables, Florida.
Sheridan, P. 1996. Forecasting the fishery for pink shrimp, Penaeus duorarum, on the Tortugas grounds. Florida. Fish. Bull. 94:743-755.