seaturtle.org : MTN : ARCHIVES : Sign In

Why should sea turtle conservationists be concerned about shrimp stock assessments and management? Simply put, prior shrimp stock assessments have been flawed and have encouraged overfishing. Exposure to shrimp fishing effort levels higher than are necessary to maximize shrimp yield per recruit is not a good thing for sea turtle populations. Therefore, the focus of this paper is on overfishing and the apparent flaws in shrimp stock assessments that have contributed to it.

OVERFISHING

Ludwig et al. (1993) noted that there is remarkable consistency in the history of resource exploitation in that resources are inevitably overexploited, often to the point of collapse or extinction. They suggested such consistency is due to the following common features:

1. Wealth or the prospect of wealth generates political and social power that is used to promote unlimited exploitation of resources.

2. Scientific understanding and consensus is hampered by the lack of controls and replicates, so that each new problem involves learning about a new system.

3. The complexity of the underlying biological and physical systems precludes a reductionist approach to management. Optimum levels of exploitation must be determined by trial and error.

4. Large levels of natural variability mask the effects of overexploitation. Initial overexploitation is not detectable until it is severe and often irreversible.

The Committee on Ecosystem Management for Sustainable Marine Fisheries (CEMSMF, 1999) defined overfishing as fishing at an intensity great enough to reduce fish populations below the size at which they could provide the maximum long-term potential (sustainable) yield, or at an intensity great enough to prevent their recovery to that size. From the point of view of fishery stock dynamics, there are two recognized types of overfishing. Growth overfishing occurs when the level of fishing mortality (determined by the amount of fishing effort and factors affecting the fishing power of fishing units) exceeds that which produces maximum sustainable yield or maximum yield per recruit. Trends of reduction in size of individuals in the annual catch coupled with trends of reduction in annual catch per unit effort or total catch are symptoms of growth overfishing. Recruitment overfishing can occur when fishing mortality continues to increase beyond levels that produce growth overfishing until the stock collapses, because the remaining spawners are too few to produce enough offspring to restore the stock. Some shrimp stocks can withstand extended periods of high fishing mortality without producing major concern about recruitment overfishing, but growth overfishing can be a significant economic problem under such conditions (Neal & Maris 1985).

While growth overfishing produces negative socio-economic impacts, the negative impacts of recruitment overfishing, both ecological and socio-economic, are much more severe and either prolonged or permanent. Recruitment overfished stocks either do not recover or they take a very long time to recover. Not only can an important fishery be lost, but also marine ecosystems can be irreparably altered by paucity or loss of important species. Therefore, the CEMSMF (1999) concluded that management agencies should adopt regulations and policies that strongly favor conservative and precautionary management and that penalize overfishing. Unfortunately this has not been the case with shrimp management.

The National Marine Fisheries Service (NMFS) defined shrimp overfishing only in terms of recruitment overfishing (Klima et al. 1990, Nance 1993b, NMFS 1999). Then NMFS (1999) concluded that stocks of brown shrimp (Penaeus aztecus) and white shrimp (P. setiferus) were not overfished under the pre-SFA (Sustainable Fisheries Act of 1996; see CEMSMF 1999) definition of overfishing, and that neither stock was approaching an overfished condition [as so defined]. This was a dangerous course to take, since growth overfishing exacerbates socio-economic hardships experienced by those in the fishery, and the exact timing of recruitment overfishing is difficult if not impossible to predict. The Gulf of Mexico Fishery Management Council's (GOMFMC) Management Plan for the Shrimp Fishery of the Gulf of Mexico contained definitions of overfishing for brown shrimp and white shrimp that were disapproved under SFA guidelines (NMFS 1999).

Some participants in the shrimp industry believe that recruitment overfishing of shrimp stocks is either impossible or highly unlikely. The "conventional wisdom" has been that annual production of shrimp is determined wholly by environmental variables outside the control of management agencies, that the fishery should harvest every shrimp it can each year, and that the stock is not jeopardized in any way by such harvest.

However, the CEMSMF (1999) pointed out that environmental changes can produce effects similar to those of fishing, and that it is often difficult to distinguish them from the effects of fishing. Recognizing that environmental fluctuations exert a fundamental influence on the behavior of marine ecosystems and that they cannot be controlled directly, the committee (CEMSMF 1999) nevertheless stated that uncertainties about effects of environmental variability should not be used as an excuse to continue overfishing.

EVIDENCE OF SHRIMP OVERFISHING

Almost two decades ago, Gulland and Rothschild (1984) stated that a reduction of shrimping effort in the Gulf of Mexico would most certainly lead to economic benefits. They stated further that an increase in effort would be of limited economic value to the fishermen, and could result in increased risk of population collapse or a sustained reduction in production of the population. They suggested that a conservative view must be taken on the potential for biological danger to the stocks. Yet, shrimp fishing effort in the Gulf of Mexico was allowed by management agencies to continue to increase.

Signs of shrimp growth overfishing in the Gulf of Mexico and along the U.S. Atlantic coast have been developing for decades. Size composition has long been recognized as a simple criterion for assessing status of a fishery (Henderson 1972; Ricker 1975). Declining average size of individuals can indicate increasing mortality (usually equated with increased fishing effort) or decreasing growth (usually attributed to overcrowding in overabundant populations, but overcrowding is not likely in heavily fished shrimp stocks). Caillouet et al. (1980) detected trends of reduction in size of brown shrimp and white shrimp in reported annual catches from Texas and Louisiana during 1959-1976. Downward trends in size of these two species in reported annual catches have also been detected in the Gulf of Mexico and on the U.S. Atlantic coast by Caillouet and Koi (1980, 1981, 1983), Nichols (1984), Nance and Nichols (1988), Nance (1989), and Nance et al. (1989). Caillouet and Koi (1980) conducted simulations showing that the ex-vessel value of a given weight of landings could be greatly increased, if the trends toward decreasing size of shrimp in the landings could be reversed. This was true for brown shrimp, white shrimp and pink shrimp (P. duorarum). Although foreign imports of shrimp most certainly played a large role in reducing the real price (price adjusted for inflation) of domestic shrimp through competition (Keithly & Roberts 2000), reduction in size of shrimp in the catch also took its toll on value of the catch (Caillouet & Koi 1980; 1981; 1983).

Declines in catch per unit effort, beginning as early as 1960, have been evident for Gulf of Mexico brown shrimp and white shrimp and have accompanied continued increases in fishing effort (Klima et al. 1990; Nance 1993a; Nance 1999; Neal 1975; Nichols 1984). In addition, the total annual catch of brown shrimp (as well as that of pink shrimp) seems to have declined over the past decade or so. Catch per unit effort for brown shrimp, white shrimp and pink shrimp has been declining for almost four decades (Nance 1999).

Some have argued that shrimp fishing effort has decreased rather than increased over recent years. But, changes in fishing mortality are not always directly proportional to observed changes in fishing effort. Technological improvements in boats, vessels, gear, equipment, and fishing strategies as well as increasing knowledge and skill of fishermen can increase fishing power of the fishing units. Griffin et al. (1997) examined historical trends in standardized fishing effort (nominal fishing effort adjusted for fishing power) for the Gulf of Mexico shrimp fleet. They showed that relative fishing power of the fleet increased from 1965 through 1993. Trends in standardized effort in the inshore and offshore fisheries were generally upward over the same years, but appeared to level off in the offshore fishery in the last seven of those years, and to decline in the inshore fishery in those same seven years. Obviously, it would be important to update this time series analysis to the present.

In its Texas Shrimp Fishery Briefing Book April 2000, TPWD (2000) stated that "A recent 18-month comprehensive review of the shrimp fishery by TPWD once again documented serious overfishing including a continuing long-term downward trend in the population of adult spawning shrimp in the Gulf." The briefing book also stated that "Failure to reverse these trends could lead to an economic and biological collapse of the shrimp stocks." TPWD accepted such trends as warning signs of growth overfishing and proactively recommended reductions in fishing mortality in hopes of forestalling a collapse of these stocks due to recruitment overfishing. Although some doubters suggested that TPWD's evidence was insufficient to justify additional shrimping regulations, the Texas Parks and Wildlife Commission ruled in favor of the additional regulations after hearing and reviewing published testimony.

POTENTIAL FLAWS IN SHRIMP STOCK ASSESSMENTS

The status of penaeid shrimp stocks in the Gulf of Mexico (as well as on the U.S. Atlantic coast) could be worse than indicated by past stock assessments. There are potentially serious, unresolved flaws in the stock assessments conducted by NMFS. Virtual population analyses (VPA), spawner-recruit relationships, recruitment indices, and recruitment overfishing indices are all based on number of shrimp estimated from the weight of catch within size class intervals, using methods that may be statistically biased. Not only may these methods yield biased estimates of the number of shrimp, but also the magnitude of this bias may be size related; i.e., the bias may increase with decrease in size of shrimp. Such biases could have affected prior estimations of spawner-recruit relationships (Gulland & Rothschild 1984; Klima et al. 1990; Nance 1989; Nance & Nichols 1988; Nance et al. 1989), recruitment overfishing indices (Klima et al. 1990; Nance 1993b; Nance 1998), and VPA results (Nance 1989; Nance 1999; Nance & Nichols 1988; Nance et al. 1989; Nance et al. 1994) upon which NMFS' shrimp management recommendations to the GOMFMC have been based. Keep in mind that the annual total catches of shrimp measure in the millions of pounds, so discrepancies resulting from biases in estimating number of shrimp could have dramatic effects on stock assessment results.

NMFS' time series database includes observations derived from both "box-graded" and "machine-graded" shrimp catches. Box grading provides a single average count (number of shrimp per pound) applied to the landed portion of the catch of a shrimp vessel or boat. Machine grading separates the landed portion of the catch into segments sorted into count class intervals set by shrimp processors and influenced by marketing strategies.

Sampson (1994) examined statistical biases in estimating number of fish landed from sample average weight and the weight of fish landed, using samples of equal size. Sample size (number of shrimp taken per sample) is not constant for samples taken from shrimp landings, so Sampson's (1994) statistical estimation methods are not strictly comparable to those used for shrimp. Nevertheless, his paper elucidates the kinds of statistical considerations that are needed to examine potential biases in the estimation methods used for shrimp. The statistical estimation problem has to do with how well the average count (for box-graded catch), or the midpoint count (for machine-graded catch), represents the true mean of a count class interval.

Nichols (1984) and Parrack (unpublished) each presented a method of estimating number of shrimp from weight of catch, but neither method has been evaluated for statistical biases or received adequate peer review. Nichols' (1984) method is based on pounds of shrimp within size class intervals expressed in count. Parrack's (unpublished) method is based on pounds within size class intervals expressed in pounds per shrimp, the reciprocal of count. Nichols' (1984) method has been the one most used in NMFS' stock assessments. The two estimation methods are as follows:

1. Nichols' (1984) method - The number of shrimp in a count class interval is estimated by multiplying the midpoint count of the interval by the pounds in the count class. The method can be portrayed using a simple example. If the count class is 6-10, the midpoint is 8. For a 100 pound catch in this class interval, the number of shrimp is estimated to be 8x100 = 800. If the count class is 60-120, the midpoint is 90. For a 100 pound catch, the estimated number of shrimp is 90x100 = 9000. It is clear from this example that the larger the count, the greater the distortion in estimated number of shrimp, if the estimation method were biased.

2. Parrack's (unpublished) method - The number of shrimp is estimated by dividing the midpoint between reciprocals of the lower and upper limits of a count class into pounds in the count class. Using the same example as above, the reciprocals of 6 and 10 are 1/6 and 1/10, and their midpoint is 2/15 or 0.133. For a 100 pound catch in this class interval, the number of shrimp is estimated at 100/0.133 = 751.88. If the count class is 60-120, the midpoint of the reciprocals of these count class limits is 1/80 or 0.0125. For a 100 pound catch, the estimated number of shrimp is 100/0.0125 = 8000. Not only does this method give results differing from those of the Nichols (1984) method, but again any bias in the estimation method would produce a greater distortion when estimating the number of small shrimp than of large shrimp. Parrack's (unpublished) method is akin to that examined by Sampson (1994), but it is not based on samples of equal size.

Another potential problem with Nichols' (1984) method is that the weight of catch in a count class interval was "assumed to be uniformly distributed by weight between category boundaries" (i.e., count class limits). Although Nichols' (1984) description is not clear, my interpretation is that he assumed that the weight of the catch within a count class interval was uniformly distributed over the interval (i.e., each fraction of the partitioned weight was equal). This is equivalent to assuming, implicitly, that the number of shrimp in each fraction increases in direct proportion to count, from the lower limit (lowest count) to the higher limit (highest count) of the interval. Under this assumption, when count is converted to weight per shrimp by taking its reciprocal, the number of shrimp in each fraction of the weight (partitioned according to Nichols' method) declines logarithmically with increase in weight per shrimp over the interval. Thus, the reciprocal of the midpoint count of the count class cannot accurately represent the mean weight per shrimp in the count class (see Sampson 1994 for discussion). This probably would not be as serious a problem if the count class intervals in NMFS' database were very narrow and of constant width. However, they are neither. Frequency distributions of count and weight per shrimp are unknown, except perhaps for white shrimp (based on old studies). Nichols (1984) converted white shrimp length-frequency distributions to distributions of weight per shrimp or count and used them in an alternative method for estimating number of white shrimp within count class intervals. However, the resulting frequency distributions of weight per shrimp or count were not presented, nor were details of how they were used to estimate number of shrimp.

So, the degree to which Nichols' (1984) or Parrack's (unpublished) methods may be biased remains to be determined. In any case, the distortion in numbers resulting from estimation biases would be greater for small shrimp than for larger ones, and the errors would be cumulated by aggregating the estimates for total catches, catches of recruits only, or catches of spawners only. This brings into question all past shrimp stock assessments based on Nichols' (1984) or Parrack's (unpublished) methods. If a statistically valid and unbiased estimation procedure were developed and applied, it could lead to different estimates of number of shrimp and different stock assessment results.

A solution to this statistical estimation problem will require theoretical considerations related to sampling theory and size frequency distributions of shrimp. It may require fishery-independent sampling, determination of size distributions within "box-graded" and "machine-graded" catches, use of stochastic methods applied to the available catch and size class data, or a combination of these approaches and perhaps others. Until this is done, all results based on Nichols' (1984) or Parrack's (unpublished) methods are questionable.

Additional problems exist regarding the NMFS database used to estimate shrimp numbers from weight of catch:

1. Methods of grading shrimp have changed over the time series covered by the database. Shrimp discarding practices, which influence sizes of shrimp landed, have changed over time. Limits of count classes in the database overlap, since they were determined by grading methods and marketing factors, not statistical sampling methods.

2. Count class limits may not represent the actual range of shrimp sizes within a class interval.

3. Count classes representing the largest and smallest size extremes of landed shrimp often contain limits 0 or 999 (representing "unknown" or infinity), which are unrealistically low or high, respectively, and make some calculations impossible. For example, one cannot calculate the weight of shrimp weighing 0 count, and shrimp of 999 count would weigh only 1/999 or 0.001 pound. Previous investigators have dealt with this problem by replacing 0 or 999 with assumed numerical values that allow the necessary calculations. Even though such adjustments allowed estimation of shrimp numbers, their validity has not been adequately evaluated.

4. The number of unique count classes in the NMFS database has varied over the time series.

5. A box-graded catch is assigned to a particular count class interval when its average count falls within that interval, whether or not the actual range in size of shrimp in that catch falls within that interval.

I must point out that these seven additional data problems also affected my analyses concerning trends in size of shrimp (see references below).

Because of the socio-economic and other consequences of shrimp management strategies based on stock assessments that might be flawed, I believe there is an urgent need for a thorough statistical examination of data problems, estimation methods, and stock assessment methods used by NMFS.

I would expect that if better estimates of standardized shrimp fishing effort and better estimates of numbers of shrimp were available, shrimp stock assessments based on them would show the stocks to be more seriously overfished than they now seem to be. If true, this would indicate an even greater need to reduce shrimp fishing mortality.

RECOMMENDATIONS

1. Marine fisheries are common property resources in the United States, and public funds are used to manage and sustain them. Therefore, shrimp fisheries should be managed for the common good. Participants in the shrimp industry and related industries, conservation organizations, taxpayers, and consumers of shrimp all have vested interests in the wise management of these renewable natural resources. Wise management involves sustaining commercial and recreational fisheries and sea turtle populations now and into the future, to perpetuate and optimize the socio-economic benefits these resources provide for the common good.

2. Because of uncertainties surrounding NMFS' estimation of fishing effort, shrimp fishing mortality, and number of shrimp from the weight of catch, NMFS and the GOMFMC should join with Gulf states in an effort to evaluate and improve estimation methods and shrimp stock assessments.

3. NMFS and state marine fisheries management agencies in the Gulf should take whatever steps necessary to place all of their shrimp data files (containing fishery-dependent and fishery-independent observations), adequate documentation of these data files, and detailed explanations of their estimation and stock assessment methods on their web sites, so the data can be downloaded and evaluated by anyone interested in doing so. They should do the same with data files covering other important fisheries species as well as sea turtles. By data files I refer to computer-compatible files containing original observations, and not summary data, although summary data should also be available for downloading. "Legitimate access to nonsecret government information, information that has been paid for by taxpayers, is the public's right" (GOVERNMENT INFO, Agencies slow in complying with electronic access bill, Houston Chronicle, August 7, 2000).

Caillouet, C. W. & D. B. Koi. 1980. Trends in ex-vessel value and size composition of annual landings of brown, pink, and white shrimp from the Gulf and South Atlantic coasts of the United States. Marine Fisheries Review 42:18-27.

Caillouet, C. W. & D. B. Koi. 1981. Trends in ex-vessel value and size composition of reported May-August catches of brown shrimp and white shrimp from the Texas, Louisiana, Mississippi, and Alabama coasts, and 1960-1978. Gulf Research Reports 7:59-70.

Caillouet, C. W., Jr. & D. B. Koi. 1983. Ex-vessel value and size composition of reported May-August catches of brown shrimp and white shrimp from 1960 to 1981 as related to the Texas closure. Gulf Research Reports 7:187-203.

Caillouet, C. W., F. J. Patella & W. B. Jackson. 1980. Trends toward decreasing size of brown shrimp, Penaeus aztecus, and white shrimp, Penaeus setiferus, in reported annual catches from Texas and Louisiana. Fishery Bulletin 77:985-989.

Caillouet, C. W., Jr., D. J. Shaver, W. G. Teas, J. M. Nance, D. B. Revera & A. C. Cannon. 1996. Relationship between sea turtle stranding rates and shrimp fishing intensities in the northwestern Gulf of Mexico: 1986-1989 versus 1990-1993. U.S. Fishery Bulletin 94:237-249.

Committee on Ecosystem Management for Sustainable Marine Fisheries. 1999. Sustaining Marine Fisheries. Ocean Studies Board, Commission on Geosciences, Environment, and Resources, National Research Council, National Academy Press, Washington, D.C., 16 pp.

Griffin, W. L., A. K. Shah & J. M. Nance. 1997. Estimation of standardized effort in the heterogeneous Gulf of Mexico fleet. Marine Fisheries Review 59:23-33.

Gulland, J. A. & B. J. Rothschild (Editors). 1984. Penaeid Shrimps - Their Biology and Management. Fishing News Books Limited, Farnham, Surrey, England.

Henderson, F. 1972. The dynamics of the mean-size statistic in a changing fishery. FAO Fisheries Technical Paper 116, 16 pp.

Keithly, W. R. & K. J. Roberts. 2000. Economics: contrast with wild catch fisheries, p. 261-277. In: Stickney, R. R. (Editor), Encyclopedia of Aquaculture, John Wiley & Sons, Inc., New York, 1063pp.

Klima, E. F., J. M. Nance, E. X. Martinez & T. Leary. 1990. Workshop on definition of shrimp recruitment overfishing. NOAA Technical Memorandum NMFS-SEFC-264, 21 pp.

Ludwig, D., R. Hilborn & C. Walters. 1993. Uncertainty, resource exploitation, and conservation: lessons from history. Science 260:36.

Nance, J. M. 1989. Stock assessment for brown, white, and pink shrimp in the U.S. Gulf of Mexico 1960-1987. NOAA Technical Memorandum NMFS-SEFC-221, 65 pp.

Nance, J. M. 1992. Estimation of effort for the Gulf of Mexico shrimp fishery. NOAA Technical Memorandum NMFS-SEFSC-300, 12 pp.

Nance, J. M. 1993a. Effort trends for the Gulf of Mexico shrimp fishery. NOAA Technical Memorandum NMFS-SEFSC-337, 37 pp.

Nance, J. M. 1993b. Gulf of Mexico shrimp fishery recruitment overfishing definition Workshop 2. NOAA Technical Memorandum NMFS-SEFSC-323, 12 pp.

Nance, J. M. 1998. Shrimp recruitment overfishing analysis for 1998. National Marine Fisheries Service, Galveston Laboratory, Galveston, Texas. Unpublished Document, 7 pp.

Nance, J. M. 1999. Stock assessment report 1998, Gulf of Mexico shrimp fishery. NMFS SEFSC Galveston Laboratory, Galveston, Texas. Unpublished Document, 13 pp.

Nance, J. M., E. F. Klima & T. E. Czapla. 1989. Gulf of Mexico shrimp stock assessment workshop. NOAA Technical Memorandum NMFS-SEFC-239, 41 pp.

Nance, J. M., E. X. Martinez & E. F. Klima. 1994. Feasibility of improving the economic return from the Gulf of Mexico brown shrimp fishery. North American Journal of Fisheries Management 14:522-536.

Nance, J. M. & S. Nichols. 1988. Stock assessments for brown, white and pink shrimp in the U.S. Gulf of Mexico 1960-1986. NOAA Technical Memorandum NMFS-SEFC-203, 64 pp.

National Marine Fisheries Service. 1999. Report to Congress: Status of Fisheries of the United States. National Marine Fisheries Service, October 1999, 104 pp.

Neal, R. A. 1975. The Gulf of Mexico research and fishery on Penaeid prawns. In: P.C.Yong(Ed.), First Australian National Prawn Seminar, Maroochydore, Queensland, Australian Fisheries Council, Northern Fisheries Research Committee, Australian Government Publishing Service, Canberra. pp.2-8

Neal, R. A. & R. C. Maris. 1985. Fisheries biology of shrimps and shrimplike animals. In: A.J.Provenzano (Ed.), Economic Aspects: Fisheries and Culture, Volume 19 of The Biology of Crustacea, Academic Press, Inc., New York. pp. 1-110.

Nichols, S. 1984. Updated assessments of brown, white and pink shrimp in the U.S. Gulf of Mexico. Unpublished Manuscript of Paper presented at the Workshop on Stock Assessment, Miami, Florida, May 1984, 19 pp.

Parrack, M. L. Unpublished. Some aspects of brown shrimp exploitation in the northern Gulf of Mexico. National Marine Fisheries Service, Southeast Fisheries Center, Miami, Florida, 50 pp.

Ricker, W. E. 1975. Computation and interpretation of biological statistics of fish populations. Fisheries Research Board of Canada Bulletin 191, 382 pp.

Sampson, D. B. 1994. Estimating the number of fish landed from their total weight and a sample average weight. Canadian Journal of Fisheries and Aquatic Sciences 51:2537-2548.

Kemp's ridley sea turtle (Lepidochelys kempii) laying at Rancho Nuevo, Mexico. © Doug Perrine/seapics.com