2
Adequacy of NMFS Data, Scientific Foundations, and Models

ADEQUACY OF SCIENCE FOR STOCK ASSESSMENTS

Issues

Stock assessment is the scientific and statistical process whereby the status of a marine fish population or subpopulation (stock) is assessed in terms of population size, reproductive status, fishing mortality, and sustainability. NMFS allocates a substantial portion of its personnel and resources to stock assessment, and the resulting assessments form the basis of biological reference points and management decisions, such as quotas, restriction of effort, and closing of seasons or areas to fishing. Previously, study committees sponsored by the National Research Council have conducted reviews of stock assessments, including reviews at the species level (such as Atlantic bluefin tuna, NRC, 1994; Pacific salmon, NRC, 1996b) and at the regional level (such as Northeast marine fisheries, NRC, 1998a), and have reviewed stock assessment methods (NRC, 1998b) and data issues (NRC, 2000a) generally.

Uncertainty

Why is the stock assessment process reviewed and questioned so frequently? As indicated in the Research Council report Improving Fish Stock Assessments (NRC, 1998b), stock assessments are intrinsically uncertain.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 15
Science and Its Role in the National Marine Fisheries Service 2 Adequacy of NMFS Data, Scientific Foundations, and Models ADEQUACY OF SCIENCE FOR STOCK ASSESSMENTS Issues Stock assessment is the scientific and statistical process whereby the status of a marine fish population or subpopulation (stock) is assessed in terms of population size, reproductive status, fishing mortality, and sustainability. NMFS allocates a substantial portion of its personnel and resources to stock assessment, and the resulting assessments form the basis of biological reference points and management decisions, such as quotas, restriction of effort, and closing of seasons or areas to fishing. Previously, study committees sponsored by the National Research Council have conducted reviews of stock assessments, including reviews at the species level (such as Atlantic bluefin tuna, NRC, 1994; Pacific salmon, NRC, 1996b) and at the regional level (such as Northeast marine fisheries, NRC, 1998a), and have reviewed stock assessment methods (NRC, 1998b) and data issues (NRC, 2000a) generally. Uncertainty Why is the stock assessment process reviewed and questioned so frequently? As indicated in the Research Council report Improving Fish Stock Assessments (NRC, 1998b), stock assessments are intrinsically uncertain.

OCR for page 15
Science and Its Role in the National Marine Fisheries Service The uncertainty arises as a result of several factors, including the inherent variability in populations, variability associated with the process of observing populations and measuring them (sampling error), and the uncertainty associated with modeling assumptions (such as model misspecification). Gathering more information on fish stocks is important because it generally leads to greater certainty and confidence, but increasing the amount of data collected does not necessarily solve the problem of uncertainty in assessments. If NMFS could conduct a complete census of the fish in the ocean, scientists and decision-makers would still have to deal with uncertainties of natural variation in populations resulting from processes of birth, death, growth, immigration, and emigration, each of which is affected by environmental factors in ways that are not entirely understood. Furthermore, scientists would still need to base their analyses and managers make their decisions on predictions of future stock size and of how fishing activities will affect the stock. The science and management of marine fisheries can be improved as more data are collected (NRC, 2000a), but uncertainty will still exist because of the nature of the system. Uncertainty can lead decision-makers, stakeholders, and even fisheries scientists to be overly skeptical of predictions made through the stock assessment process. As Colin Clark noted, “any admission of scientific uncertainty only destroys the credibility of the science” (Clark, 1966). That is unfortunate because even uncertain scientific results contain valuable information. Perfect predictions will never be available, but useful predictions can be developed, and less risky management actions can be based on such predictions. Scientists and non-scientists view uncertainty differently, but this does not justify ignoring stock assessment advice, just as it would not justify ignoring weather predictions, national economic forecasts, or health advisory alerts. Data In stock assessments, constant attention must be given to how fisheries data are collected and how they are used. The information content of data can be improved through technological and methodological advances, the use of efficient estimators and design protocols, an increase in the capacity of NMFS to gather and analyze data, and active communication among scientists, managers, and stakeholders. Several aspects of those issues have been discussed at length in previous Research Council reports and are briefly revisited here.

OCR for page 15
Science and Its Role in the National Marine Fisheries Service Sampling Protocols The sampling protocols and methods set up by NMFS initially were of good quality. However, as agency responsibilities changed and sampling methods improved, earlier sampling protocols became less appropriate. Previous reports recommended that standardized and formalized data-collection protocols be established nationwide to allow for greater efficiency in overall sampling design and quality assurance. NMFS has responded to that recommendation by supporting the development of federal-state partnerships in the collection, management, and storage of fishery-dependent data through regional protocols, but progress on standardized protocols has been slow. NMFS has also addressed problems in communicating the rationales for fishery-independent sampling protocols to fishing-industry personnel on sampling cruises and in cooperative research. Fishery-Dependent Data: Bycatch Data Previous reports (NRC, 1998a; 2000a) have discussed aspects of fishery-dependent data collection that could be improved by NMFS. To various degrees, NMFS has been able to address their recommendations. The committee’s review of the recent series of lawsuits found that collection of bycatch data remained problematic and was, in part, the basis of several lawsuits, e.g., cases challenging FMP amendments for Pacific and Northeast groundfish. “Bycatch” refers to fish that are caught but not retained (discards) plus fish killed because of encounters with gear. Bycatch data can be collected in several ways, including vessel logbooks, dealer reports, and observers. Logbook data come from fishers who self-report their catch, dealer reports list the amount of bycatch that is landed and sold when it is not prohibited and when the catch is salable, and observers directly observe the catch of species that are kept and species that are later discarded as they are brought onboard the vessel at sea. Of the three, observers offer independent, unbiased data collection that is especially important for non-targeted or prohibited species. Observer programs are expensive, and their use is mandated for specific fisheries, such as the Georges Bank scallop fishery, and in fisheries that have interactions with protected species. Observer coverage is used for commercial fisheries and is virtually nonexistent for recreational fisheries. In 1999, NMFS spent $9.2 million and the fishing industry another $10 million for observer coverage compared with $28.8 million on fishery-

OCR for page 15
Science and Its Role in the National Marine Fisheries Service independent surveys, $3.9 million on recreational surveys, and $2.8 million on vessel monitoring system (VMS) programs. In the case of Natural Resources Defense Council (NRDC) v. Evans, NMFS and the Groundfish Management Team (GMT) of the Pacific Fishery Management Council (PFMC) used historical studies to ascertain the proportion of bycatch currently taken. Plaintiffs challenged NMFS’s use of the historical data, and these data were found unacceptable by the court. The court made the following points indicating its understanding of trip limit-induced discards: An irony exists in that as fishing allowances are lowered to protect a species, the bycatch percentage increases. Fishing boats continue to catch multiple species of fish at the same time, but they are compelled by regulation to discard a greater percentage of the protected species. As bocaccio and lingcod fishing allowances have decreased in recent years, it is therefore, as both sides agree, a virtual certainty that the bycatch mortality rates for each fish have in turn increased. NMFS admits that it is a “virtual certainty” that their (bocaccio and lingcod) bycatch mortality rates have risen. A Northwest and Alaska Fisheries Center report (Pikitch, 1988) also corroborates evidence that regulatory bycatch increases as landing limits are reduced, and both bocaccio and lingcod have been protected by reduced landing limits. It must follow that bycatch discard has increased since the Pikitch Study was conducted and the 16 percent and 20 percent figures that NMFS has arbitrarily set are no longer accurate, if they ever were. The GMT used values from Pikitch (1988) and its expert knowledge to estimate discard rates for various species. It was those values that the court found to be too low, but there were no more recent data to document the magnitude or even the direction of change in total discards and discard rates. Some discard rates may have risen, some may have declined; consequently, the court concluded that estimates of discards based on past behavior are not relevant for today’s fishery. In fact, estimates from the Pikitch study may have never been accurate estimates of West Coast discard rates, because they were limited to only a few vessels and ports. Why were there no current data to improve critical estimates of regulatory discards? As mentioned previously, the preferred method of collecting discard data is via at-sea observers. The annual cost of each full-time observer may exceed $50,000. The total cost of an adequate observer program for West Coast groundfish may be about $5 million. Total revenue of

OCR for page 15
Science and Its Role in the National Marine Fisheries Service the West Coast groundfish industry (excluding whiting) is about $45 million. Thus the cost of an adequate observer program may easily exceed 10 percent of the gross revenue of the fishery. There is a special “disaster” appropriation of about $2.5 million for a pilot observer program in the NMFS budget, but no data have been obtained yet. Industry suggested other methods of obtaining bycatch data, but the PFMC and NMFS rejected these suggestions. Because at-sea observers were too expensive and other methods were not pursued, no information regarding discards is now obtained. The general lesson learned is that fishery-dependent data on bycatch are severely limited and additional resources are needed to increase the number of at-sea observers to avoid future litigation of this type. Fishery-Dependent Data: Recreational-Catch Data There are often problems with the use of fishery-dependent data for stock assessment. One of the major problems, the lag between data collection and their availability for use in stock assessment, is highlighted in the recreational fisheries. Recreational fisheries present a concern because they form an important and growing component of some marine fisheries. Recreational fisheries are characterized by a large number of people entering marine waters from many access points and individually harvesting only a few fish each. Because of how recreational fisheries operate, differences exist from commercial fisheries in how data are collected, how the fish and the fishers are managed, and consequently how these fisheries should be modeled in contrast with commercial fisheries. The statistical-survey methods (NMFS Marine Recreational Fishery Statistics Survey) that have provided reliable data estimates have been slow, with many months passing between harvest by the angler and complete catch and effort estimation by NMFS. Because the recreational catch makes up only a small portion of most fisheries, this is usually not a point of contention. However, in the summer flounder fishery (see NRC, 2000a) and some other fisheries, the recreational catch is a substantial portion of total catch. The delay in obtaining catch statistics contributed to the court’s sanction of NMFS in North Carolina Fisherman’s Association, Inc. v Daley. The National Research Council (NRC, 2000a) recommended that alternative statistical approaches be developed to provide these data more quickly. NMFS has recently amended the annual quota-setting process to set a cut-off date for catch statistics in early fall to eliminate the need for possible cuts in the quota in mid-spring.

OCR for page 15
Science and Its Role in the National Marine Fisheries Service Fishery-Independent Surveys For most species, fishery-independent surveys offer the best choice for providing a reliable index of fish abundance. Surveys allow formulation of an unbiased statistical design for the collection of fish population data, control over sampling location and intensity, and quality assurance. NOAA research vessels have provided the cornerstone of this data-gathering venture for many years, but less and less funding has been made available to NOAA to support this important function. NOAA fishery research vessels that conduct fishery-independent surveys are aging. The Research Council (NRC, 2000a) recommended that these vessels be replaced or modernized, and that new vessels be acquired to increase NMFS’s capacity to collect high-quality scientific information and to conduct research. Technological Advances in Data Collection Previous NRC reports (NRC 2000a; 1998a) recommended that NMFS evaluate the usefulness of such modern electronic data-gathering devices as electronic logbooks and VMSs in conjunction with the value-added features that each of these offers to fishermen. All vessels in the New England scallop fleet now use VMSs. Environmental groups that have at times sued NMFS have filed briefs in support of NMFS’s required use of VMSs in the Atlantic longline fishery— a requirement under legal challenge by the fishing industry. Ecosystem Data Ecosystem-level information is being gathered through fishery surveys, commercial and recreational catch monitoring, and all the other means of data collection at NMFS’s disposal. Methods must be developed to use this information effectively and to gather broader ecosystem-level information to understand the role of fisheries science in the context of ecosystem management, and to evaluate the role of fishing and the dynamics of individual fish populations in marine ecosystems. The NMFS Ecosystem Principles Advisory Panel recommended the development of fisheries ecosystem plans to modify single-species approaches to incorporate ecosystem attributes, evaluate how trophic interactions and oceanographic processes affect recruitment, document the role of habitat in supporting fisheries and ecosystem productivity, and develop aggregate models that can be used to predict single-species and ultimately multispecies harvest objectives (NMFS, 1999). The committee encourages the development of such plans because they

OCR for page 15
Science and Its Role in the National Marine Fisheries Service address unmet needs under the MSFCMA to supply new means of assessing the effects of fishing on habitat and ecosystems. Database Management The purview of NMFS science involves a great deal of data collection, and this task has expanded greatly in recent years. An avalanche of data now overwhelms existing data collection and management systems. The volume of data is sure to grow, and the technological means for handling it are advancing rapidly. Improving the means for data processing and management should be a central focus for NMFS over the next decade. This is not a trivial task. Many scientists keep track of the data they collect and use, but the data now in use represent several human lifetimes’ worth of work and should continue to be available for future generations to use in managing fish populations. Data management is essential and requires substantial expertise. NMFS should build on its existing capacity to manage data more effectively. In the process, commercial data-management firms should be consulted to obtain real-time value-added data-management advice and products (NRC, 2000a). Scientist-Stakeholder Communication Because scientists are involved in management, they are often viewed as regulators. The regulator-regulated dichotomy often disrupts communication between the two groups (scientists and fishermen), which is unfortunate because a great deal of information could be and should be exchanged. Information could increase understanding and result in better management of the fish resources. NMFS should facilitate greater cooperation among fisheries scientists, regional fishery management council advisory panels, fishery participants, and other stakeholders to improve the quality and efficiency of data collection and create a shared sense of confidence in what the data indicate. That has begun in earnest. In response to previous National Research Council (NRC, 1998a,b; 2000a) recommendations, NMFS developed cooperative research projects with the fishing industry, for example, a cod-tagging project in the Northeast and monkfish surveys.

OCR for page 15
Science and Its Role in the National Marine Fisheries Service Management-Data Interaction It may be clear that data quality and quantity can influence the quality of management (by the FMCs and NMFS), but little consideration is given to how management can influence the quality of data. Some management regimes and changes therein can increase problems in collecting data and monitoring fisheries. The role of management is to maintain an optimal fishery and not necessarily to optimize data collection, but if the results of management actions are not monitored (through data collection), they cannot be improved. Adequate consideration should be given to how management actions will influence the ability to monitor fish populations and to the social and economic effects of those actions. Furthermore, methods should be developed for minimizing bias and data misreporting when a change in management goes into effect. Also related is the issue of allowing a management action to endure long enough to be evaluated. Fishermen complain when management changes too rapidly. This influences science and monitoring as well. Those considerations point to the need to develop management plans for a longer term and with a broader perspective in mind (NRC, 2000a). Modeling Stock-assessment modeling not only provides an important structure for synthesizing information and determining fish abundances but also serves as a useful predictive tool to evaluate alternative management scenarios and the consequences of potential actions before they are implemented. Stock assessment modeling is undergoing rapid development. NMFS has done well at using state-of-the-art methods and even creating new methods for population assessment, but, as with data collection and management practices, its stock assessment modeling can be improved (NRC, 1998b; 2000a). Alternative Models Every model is a simplified representation of a complex system. Different models characterize a fish population in different ways. By using alternative models, we gain a broader understanding of the nature of the system and the behavior of the models themselves than if we apply a single model to a specific dataset. Thus, previous stock assessment and modeling

OCR for page 15
Science and Its Role in the National Marine Fisheries Service reviews have recommended that a range of alternative assessment models be used to improve understanding of assessment biases and consequences (NRC, 1998b; 2000a). Alternative models need not be more complex models. Some alternative models could be constructed on soft-computing principles, such as fuzzy arithmetic, or they could be based on alternate methods of estimation (maximum likelihood versus least squares versus Bayesian methods). NMFS has responded to previous recommendations by initiating broader training for its stock assessment scientists. For example, in the fall of 2001, the NMFS Northeast Fisheries Science Center held a 5-day workshop to introduce AD Model Builder (a program new to the center) to its modelers as an alternative approach to the standard models. In March 2002, the Working Group on Re-Evaluation of Biological Reference Points for New England Groundfish used this approach to revise estimates of stock biomass and biomass targets, an effort in part undertaken to respond to litigation (Conservation Law Foundation v. Evans, 2001) concerning the rebuilding of these stocks (Working Group on Re-Evaluation of Biological Reference Points for New England Groundfish, 2002). Uncertainty in Models Assessment models are used to estimate abundance and to formulate predictions. There is always uncertainty in the processes of estimation and prediction. Uncertainty does not imply that nothing is known; it implies that a range of possibilities exists and that some outcomes are more likely than others. Historically, scientists proposed the most likely scenario garnered from a single point estimate. That did not allow for the evaluation of alternative strategies from the range of estimates. One action may offer a 50 percent probability that a managed stock will recover within a given period, whereas an alternative action may offer a 90 percent probability. Greater effort should be devoted to characterizing the uncertainty associated with stock assessment so that scientists, decision-makers, and stakeholders can assess risk better. Alternative methods for addressing uncertainty, or for assessing the level of confidence in estimates and predictions, should be considered. Several examples have been given in this regard here and in previous NRC reports, for example applying alternative assessment models to evaluate the effects due to model assumptions and structure, using survey variance to appropriately weigh observations going into assessments, providing confidence bounds or posterior distributions for current biomass and total allowable catch (TAC) estimates, determining costs and

OCR for page 15
Science and Its Role in the National Marine Fisheries Service gains associated with biomass levels so that risk can be assessed, and incorporating estimates of variation in prediction simulations used to explore management control rules. Ecosystem-Level Perspective Marine ecosystems are complex. Many marine-ecosystem models have been and are being developed, but they generally are not ready to be used in the day-to-day arena of fisheries management. Single-species models now used in stock assessment provide quantitative predictions of stock size and harvest potential. Multispecies models and ecosystem models that reliably predict the population sizes of targeted and associated species have yet to be developed. The lack of a comprehensive multispecies or ecosystem model for specific marine environments, however, should not preclude the use of ecological and environmental information or models to the extent feasible when population assessments and predictive management models are being developed. As pointed out in the NRC report Sustaining Marine Fisheries (NRC, 1999b), “ecosystem-based management is an approach that takes major ecosystem components and services—both structural and functional—into account in managing fisheries.” Important ecosystem-level factors that consider predator-prey relationships and marine habitats should be considered, even in single-species models that foster precautionary, risk-averse management. Incorporation of ecological and environmental information into management plans, although important, should not be sought at the expense of a reduction in the quality of single-species stock assessments themselves. Harvesting Strategies and Decision Rules Fish-population estimation has been the central focus of stock assessment scientists for many years. When projecting estimates forward, particularly when examining the effect of different potential harvest levels on a fishery, one usually takes an optimal equilibrium. This coupling of the assessment with an equilibrium analysis has become the default approach for defining benchmarks in fishery performance. But, the strategies for making optimal use of a resource have expanded greatly. Even the concept of optimal harvest has been expanded beyond the simple concept of yield to include other biological and socioeconomic criteria. A variety of new approaches have been developed by biological and

OCR for page 15
Science and Its Role in the National Marine Fisheries Service social scientists to model populations so that strategies for harvest can be examined and the consequences of management actions can be explored before FMPs are enacted. The new approaches can evaluate both short-term dynamics and long-term predictions under different harvest and control conditions. Harvest strategies are the means by which a fishery operates on a resource to obtain and control yield. The fishery may operate early or late in the season, harvest fish over a particular size; fish in one area or another, fish with one gear type or another, or operate with less intensity now so that more or larger fish are available later. Harvest strategies can be explored in a theoretical context through modeling and can be compared with other strategies to help form a basis of decision-making. In recognition that a wide variety of potential harvesting strategies and decision rules exist, it is sensible to explore management options. Stock assessment models should be developed and used in conjunction with harvesting strategies and decision rules so that they can be evaluated simultaneously to provide information for sustaining fisheries (NRC, 1998a).