This technical paper deals with the species of tunas and billfishes of greatest commercial importance. The tunas include skipjack, Katsuwonus pelamis, yellowfin, Thunnus albacares, bigeye, T. obesus, albacore, T. alalunga, Atlantic northern bluefin, T. thynnus, Pacific northern bluefin, T. orientalis, and southern bluefin, T. maccoyii. The billfishes include swordfish, Xiphias gladius, Atlantic blue marlin, Makaira nigricans, Indo-Pacific blue marlin, M. mazara, black marlin, M. indica, white marlin, Tetrapturus albidus, striped marlin, T. audax, Atlantic sailfish, Istiophorus albicans, and Indo-Pacific sailfish, I. platypterus. There are many other species of tuna, and several other species of billfish, of lesser economic importance. When general statements are made concerning tunas or billfishes they do not necessarily apply to species other than those listed above. They are often collectively referred to as "tunas and tuna-like fishes" in this report.
IMPLICATIONS FOR STOCK ASSESSMENT
The goal of any fishing management regime is to manipulate fishing mortality rates in order to achieve management objectives and to maintain the fishery within management constraints. The particular objectives and constraints chosen by managers imply that certain research and monitoring activities be addressed so that scientists may evaluate the effectiveness of the management and predict the likelihood of alternative future management outcomes. One component of these objectives and constraints is those associated with Precautionary Approaches to management, namely those associated with management targets and limits relating to overfishing rates and the status of an overfished condition. The scientific obligations to Precautionary Approaches is to determine the status of the stock(s) in question relative to limits and targets, to predict outcomes of management alternatives for reaching the targets and avoiding the limits, and to characterize the uncertainty in both of these. These impose some specific needs for research for stock assessments and monitoring.
A convenient framework to conduct management evaluations is through the use of control rules, for which managers specify variables under their control through some functions related to the status of the stock under a pre-agreed plan for adjusting management actions. Stock assessment research is then conducted to determine the status of the stock(s) in question, to evaluate the likely efficacy of management alternatives, to test the performance of management rules relative to precautionary targets and limits, and to characterize the uncertainty in the scientific advice for management. It is recognized that this involves continuous periodic feedback between managers and scientists with monitoring, re-evaluation, testing, and adjustment of management strategies.
Several suggestions for assessing the status of tunas and tuna-like fishes in relation to the Precautionary Approach are made. Most importantly, models should be developed and research conducted to determine the inputs to those models that appropriately characterize the uncertainty in the status determination. For example, research should lead to the use of estimation procedures with statistical error structures based on comparison among values predicted by the model and values observed in the fisheries. Changes in fishing strategies and their effects on estimation of indices of abundance should be considered, alternative scenarios to account for the problems in interpreting catch rates as indices of abundance should be incorporated into the evaluation of uncertainty; and empirical methods with appropriate evaluation should be employed in data-poor situations until more data are collected and research is conducted which allows the use of more detailed methods of stock assessment.
Quantification of the uncertainty associated with the estimates of the status of a stock relative to a specified reference point is complex, and there is a great need for research to improve and develop the mathematical techniques for doing this. The complexity arises because of the variety of data sources, the numerous types of input required, and the limited knowledge about the underlying processes. Uncertainties that lead to estimation error in stock assessments result from observation error (through measurement and sampling) and model error (through the use of erroneous estimates of parameters and through model mis-specification errors). Several approaches may be taken to integrate uncertainty into the models, but, regardless of the approach used, the quantification of the overall uncertainty of an estimate conceptually involves a four-step process in which (1) a set of hypotheses is developed for each potential source of error; (2) a relative weight or probability is determined for each hypothesis; (3) the likelihood (degree of fit) of the resulting estimate is determined for all combination of the hypotheses; and (4) the results are integrated over probabilities and hypotheses.
In order to do this, there is a need for additional research on approaches for incorporating results from diagnostics such as residual and retrospective analyses into the evaluation of overall uncertainty. Also, it is important to attempt to harmonize and integrate different judgements into a single prior assumption of plausible hypotheses. A comprehensive approach to the quantification of uncertainty is likely to be computationally intensive. If such a process is to be adopted, it is essential that sufficient resources be allocated to the task.
IMPLICATIONS FOR BIOLOGICAL AND ENVIRONMENTAL RESEARCH
Tunas and billfishes have many biological and physiological characteristics that make them unique among marine fishes. Among these are streamlined body shapes and physiological adaptations to permit sustained high-speed swimming and efficient thermoregulation. The latter makes it possible for them to move quickly between surface and deeper waters. All of these characteristics facilitate their highly-migratory nature.
For most tunas there are insufficient measures of many important biological parameters, such as age and growth, stock structure, natural mortality as a function of age, mixing rates, and stock-recruitment relationships. Therefore, for the application of the Precautionary Approach to the tunas, it is important that the uncertainties regarding biological characteristics be identified and taken into account in stock assessment (and management). Substantial amounts of research are clearly required to reduce these uncertainties. Large-scale tagging programs appear to have promise for obtaining better estimates of many of these parameters, including movement patterns relatively to environmental conditions, natural mortality, growth, etc.
There is a growing acknowledgement worldwide that abiotic and biotic environmental changes significantly affect the distributions, and perhaps also the productivity, of various tunas. Thus it is important to determine the nature and extent of the impact of climate variability upon the pelagic ecosystems and the tuna stocks. This natural variability of the pelagic ecosystems should be taken into account as an additional source of uncertainty in stock assessment and management.
In most fisheries for tunas the by-catches tend to be relatively minor compared to those in fisheries for many other species, such as shrimp, but there are some tuna fisheries that take large by-catches. Furthermore, various marine populations, such as albatrosses, turtles, and some sharks, because of their life history characteristics, can be highly vulnerable to small amounts of fishing mortality. As little is known about the biology of many of the by-catch species, or the species compositions and magnitudes of the by-catches in many major tuna fisheries, it is impossible to assess the impact of the fisheries on by-catch populations. It is necessary, therefore, to conduct systematic observer programs to collect reliable information on by-catches. Appropriate indices of vulnerability should be developed across phyla to identify the most sensitive by-catch species discarded by the various tuna fisheries. Simultaneously, it should be accepted that the Precautionary Approach and the FAO Code of Conduct recommend the development of mitigation measures to reduce or to avoid by-catches, even if the potential effects of these by-catches remain unknown. Ad hoc technological research aimed at reducing the amounts of by-catches by fisheries for tunas is thus a promising and recommended field of research.
The recent increase in the use of fish-aggregating devices (FADs) to catch tunas and billfishes is an increasing source of concern because of the small sizes of yellowfin and bigeye tunas taken in association with FADs and the relatively large by-catches. Additional research should be conducted on FADs in the various oceans. The goals of this research should include better understanding of the behaviour of the various species, the sizes of the fish associated with FADs, and determination of the effects, if any, of FADs on the movement patterns and biological characteristics of the fish.
Understanding and predicting how tunas and by-catch species respond to natural and fishery-induced changes is a major challenge for implementing an ecosystem approach into fisheries management. Innovative, large-scale, and multi-disciplinary ecosystem research and ecosystem modelling should be developed at an international scale. The aim of this modelling should be to evaluate better the potential effects on the tuna stocks and the pelagic ecosystems of area-time closures for selected tuna fisheries. The concept of closed strata appears to be promising in the context of the Precautionary Approach, but the potential impact of such management actions should be better evaluated through a comprehensive research program.
IMPLICATIONS FOR DATA COLLECTION AND STATISTICS
Current data collection programs do not all provide complete and accurate sets of data for determining the status of the stocks of tunas and tuna-like fishes, or the impacts of the physical environment and the species that share the habitat of the species in question. If the variations (or uncertainties) associated with each set of data are reported it would help in evaluating uncertainties in the results of the assessments.
Deliberate illegal, unregulated, and unreported fishing is a major source of concern in data collection, particularly since this can cause significant uncertainties in estimating the total catches. In addition, reluctance or incapacity of authorities to collect data and share available data with regional fishery bodies (RFBs) adds more uncertainties in the reported total catches. Failure to account for data coverage or quality, the use of different methods for recording weights or of inadequate size-frequency data where catches are reported in numbers, misidentification of species, and deliberate distortion of reports are common reasons for erroneous data. In logbook programs, the data source is the fishing industry and, while considerable effort can be put into minimizing misreporting and data entry errors, logbooks still record only information required by maritime authorities (locations and times) and activities which are important to the vessel owner, captain, and chief engineer. This usually results in logbooks recording activities associated with the taking of fish which are eventually landed, rather than all of those which are caught, nor on ancillary information required to standardize fishing effort, such as environmental data. Logbook and observer programs must be designed with full attention to data required to evaluate gear efficiency, including information on new technology, the environment, and incidental catches. Finally, implementing logbook systems with small-scale fisheries is frequently impossible. In such cases, a well-designed port-sampling system is necessary to collect the data at the desired level of resolution.
In sample-survey systems, uncertainty may come from inadequate records of fleet composition and activity or poor sampling design. Inaccurate fleet records often arise from systems that register fishing craft, but do not update registries on a regular basis: data on the inactivity of vessels or the removal of vessels from the fleet are not collected, leading to overestimates of fishing effort. Market and consumer surveys rarely produce data that can be reliably used for stock assessment.
With adequate information on fleet composition and activity, logbook data can be treated as a valuable and extensive data sample of the activity of a particular fleet. The data that are collected can be raised to obtain estimates of the total catch and nominal effort, so total enumeration is not necessary. Much care is needed, however, in designing the stratification and sampling programs, and in estimating population level statistics. In suitable circumstances, electronic vessel monitoring system (VMS) reporting can be a reliable way to obtain information on fleet composition and activity and, in conjunction with port sampling, may, to a large extent, supplant logbook reporting as a primary data collection method. Issues of confidentiality appear to be the main constraint to the generalized use of this technology. These issues might best be addressed if monitoring were to be entrusted to RFBs, which, in general, have no enforcement mandate and may be trusted more readily by fishermen.
The RFBs can collect detailed data directly from the industry, or they may be able to obtain such data from national sources. If the data are from national sources, however, they will probably be coded to preserve the confidentiality of individual business enterprises. There are two disadvantages to obtaining the data from national sources. First, it is often necessary to have very detailed data, such as changes in the captains of the vessels, which would probably not be obtained from national sources. Second, the nations may use the records from industry for determining the amounts of taxes to be paid, in addition assessing the fishery, in which case false records may be furnished to the national governments. With constraints required for confidentiality in place at the RFBs, there is no confidentiality-based justification for the withholding of the fine level data needed for stock assessment. Furthermore, exchange of available data with and between RFBs will increase the data available to undertake stock assessments.
For other elements of uncertainty which cannot be quantified, well-designed sampling, tagging, and/or observer programs can provide means of improving data sets but, when these data are available, they should be used to adjust the estimates, rather than simply to quantify the errors. If sampling, tagging, or observer programs cannot be implemented, effort should be directed at improving the accuracy of the data obtained by other means.
While deficiencies in data programs are recognized, and there are ways to improve those programs, there are no universal solutions, so management agencies should carefully determine their research requirements, and, in turn, determine data priorities and cost implications.