2. Method

2.1 Summary of the approach

The method is summarized in this subsection. Because there are significant differences between countries with regard to the interpretation of key terms, definitions are further discussed in Section 2.2. Reference is also made to Annex C, where details of the discard database file structure and a diagrammatic representation of the catch concepts are presented.

2.1.1 Key definitions

The key concepts and definitions are summarized below.

The definition of discards used in this study is adapted from FAO Fisheries Report No. 547 (FAO, 1996b).

Discards, or discarded catch is that portion of the total organic material of animal origin in the catch, which is thrown away, or dumped at sea for whatever reason. It does not include plant materials and post harvest waste such as offal. The discards may be dead, or alive.

Discarding is considered to be an act of volition requiring a decision by fishers to reject or dump the fish. Discards include slipped fish, i.e. fish caught in a net and subsequently released into the sea without being brought on board the vessel. Discards do not include dead corals or empty shells. The release of fish by recreational fishers has not been considered as a discard for the purposes of this study.

Bycatch is the total catch of non-target animals. Discards are not a subset of bycatch since the target species is often discarded.

Discard rate is the proportion (percentage) of the total catch that is discarded.

Catch is used to refer to the “gross catch” as indicated in FAO’s diagrammatic presentation of catch concepts (see Annex C, Figure 3, and Section 2.2.4). Catch includes all living biological material retained or captured by the fishing gear, including corals, jellyfish, tunicates, sponges and other non-commercial organisms, whether brought on board the vessel or not. Plant material is not considered part of the catch for the purposes of this study.

Landings refer to the portion of the total catch brought ashore or transhipped from the vessel. The landings information contained in the discard database is derived from a range of different sources. For a given set of “catch statistics” it may be difficult to determine whether the values are landed weights or the live-weight equivalent of the landings (= nominal catch as used in Fishstat).

Fishery is used as the principal unit of account for the discard database. A fishery is defined as a combination of a fishing area or zone plus a fishing gear plus a target species.

2.1.2 Sources of information

Information on discards and associated catches and landings by fishery was compiled from a broad range of sources. These included papers published in scientific journals, official publications of national fisheries administrations, “grey” or unpublished literature, reports of scientific working groups, catch and discard databases and correspondence and contacts with national and international fisheries experts. Over 3 000 references were compiled in a searchable bibliographic database archived in FAO.

2.1.3 The discard database

Records of over 2 000 fisheries were compiled in a discard database. Each record represents one fishery. The record identifies the fishery in terms of its location, fishing gear used and target species. The key quantitative fields provide the tonnage of the landings and discards for each fishery. Reference fields indicate the source of the landings and discard information and the year(s) to which they refer. A “discard rate” field indicates the percentage of the catch that is discarded (effectively discards as a percentage of discards plus landings). Other fields record additional qualitative information, such as the species composition of the discards and reason for discarding. The structure of the database is presented in Annex C.4.

2.1.4 Raising and key assumptions

The proportion of discards in the catch was obtained from discard studies. The studies were generally based on a sample of the vessels, fishing trips or fishing activities in the fishery. This proportion or discard rate was applied to the total landings of the fishery to raise or extrapolate the tonnage of discards to the level of the fishery. A linear relationship between discards and landings was assumed (see Section 2.4.1 for further discussion of the assumptions). In some cases, notably in small-scale and artisanal fisheries, the proportion of discards in the catch was assumed based on information from similar fisheries.

2.1.5 Verification

Information was checked by the use of multiple information sources for some records, further scrutiny of apparent anomalies (e.g. exceptionally high or low discard rates), by direct contacts with the authors of publications on discards, and by comparisons between extracts from Fishstat and the records. For selected countries the information was checked by requesting verification on the content of the discard database records from the national fisheries authorities or research institutes.

2.1.6 Differences between current and previous estimates

The main difference between the current method and the 1994 estimate is the use of the fishery-by-fishery approach, in contrast with the species or species group approach used in 1994. The information on which the current estimate is based has a substantially broader geographical range and is more representative of the world’s fisheries. The evolution of discard estimates is detailed in Annex B.

2.2 Other definitions and terms used

2.2.1 Other definitions of discards and bycatch

The term “discard” has distinctly different meanings in different jurisdictions, resulting in frequent confusion between “discard” and “bycatch”. This confusion pervades the literature and has resulted in considerable difficulty in the course of the study.

The Nordic workshop (Nordic Council of Ministers, 2003) defined “discard” as:

“the proportion of the catch which is taken on board, or brought to the surface by the vessel and which is subsequently thrown back to sea, dead or dying, or likely to die”.

The definition includes “slipped catches” as discards and is essentially the same as that given above and used in this study.

In contrast, the United States of America Magnuson-Stevens Act (MSA), Section 3(2), (1996) defines bycatch as:

“fish which are harvested in a fishery, but which are not sold or kept for personal use, and includes economic discards and regulatory discards. Such term does not include fish released alive under a recreational catch and release fishery management program”.

This effectively means that bycatch^[6] is equivalent to discards under the Act. This definition has been reinterpreted (NMFS, 1998) in the United States within the context of specific fishery management plans and publications, for example:

“Bycatch: discarded catch of any living marine resource plus retained incidental catch and unobserved mortality due to a direct encounter with fishing gear.”

The Inter-American Tropical Tuna Commission (IATTC) refers to “discards” as commercially important tuna species only (i.e. yellowfin, skipjack, bigeye, bluefin and albacore) that are dumped dead at sea, while “bycatch” is considered to be fish and other animals other than commercially important tunas that are dumped dead at sea.

A recent European Commission (EC) paper (European Commission, 2002a) defines discards as commercial species retained by a fishing gear that have been brought on board a fishing vessel and are thrown back into the sea, effectively ignoring noncommercial species.

Other definitions of bycatch

In Australia’s bycatch policy, the term “bycatch” refers to all non-targeted catch including by-product, discards and the biomass that does not reach the deck of the fishing vessel but is affected by interaction with the fishing gear.

Bycatch is sometimes defined as “discarded catch plus incidental catch” where incidental catch is considered to be retained non-target species. However, if target species (e.g. juveniles) are discarded this may cause some confusion, as target species are not usually considered to be “bycatch”.

Three further terms are used in this study to describe discards, or discard practices.

• Regulatory discards. Catch that is required by regulation to be discarded (from the Sustainable Fisheries Act [SFA], United States).

• Discretionary discards. Catch that is discarded because of undesirable species, size, sex or quality, or for other non-regulatory reasons (NMFS, 1998).

• Highgrading. Discarding of lower value commercial catch to maximize the value of quota. Highgraded discards are part of “discretionary discards” and are common in fisheries managed through individual vessel quotas.

2.2.2 Discard rates

The term “discard rate” used throughout this report refers to the weighted discard rate. The weighted discard rate is derived from the set of complete records for the type of fishery and is the summed discards as a percentage of summed landings plus summed discards.

The term “average discard rate” is the average of the individual discard rates for a set of fisheries. Average discard rates are provided together with their respective standard deviations for several of the major types of fishery.

2.2.3 Fishery and métier

The basic thesis on which the re-estimate is based is that discards are specific to a fishery. The fishery is used as the principal unit of account for the discard database. A fishery is defined as a combination of a fishing area or zone plus a fishing gear plus a target species. The term “fishery” is considered to be equivalent to the French term “métier”.^[7]

A range of analyses can be used to identify fisheries (Pelletier and Ferraris, 2000; Rochet et al., 1994; Laurec, Biseau and Charuau, 1991). Because of the lack of such empirical analyses for many countries and areas, the fisheries listed in the discard database were generally identified on the basis of descriptions of the fisheries sector prepared by the national fisheries administrations, e.g. in national fisheries development or management plans, in national fisheries statistics, or in research reports. Essentially, most of the fisheries listed in the discard database were identified by the competent national fisheries authorities.

Despite the fact that the fishery is an important focus of fishery management, many fisheries administrations do not necessarily compile catch or landings information by fishery. Consequently a substantial number of database entries refer to aggregate or generic fisheries, e.g. “the inshore small-scale, multigear, multispecies fishery”.

Industrial fisheries

Industrial fisheries are large-scale fisheries that use large mechanized fishing vessels as distinct from small-scale and artisanal fisheries. Note that in the EU, the term “industrial fishery” may be used to refer to fisheries for small pelagics harvested for the manufacture of fishmeal.

Small-scale fisheries

This generic term is used in the study to characterize a highly diverse group of fisheries. The definition is essentially country specific, i.e. the country considers the fishery to be “small-scale”. The terms “artisanal fisheries” and “small-scale fisheries” are considered equivalent for the purposes of this study and embrace other categories (e.g. subsistence, traditional, indigenous) as used in national fisheries statistics, or in the fisheries terminology of different countries. It is recognized that the term “small-scale” refers to “scale” rather than the nature of the fishing operation itself, e.g. the family nature of artisanal fisheries.

2.2.4 Other terms used

Landings

Landings values in the discard database are reported as given in the source of reference, except in rare cases such as when lobster or shrimp catches are reported as tail weight. In such cases the reported landings are converted to live-weight equivalent.

It is not always clear whether the mass of landings or catches reported in national fisheries statistics or other sources used is the “gross catch”, the “landings” or the “nominal catch” as per FAO definitions (see Annex C, Figure 3, which gives a comprehensive graphical illustration of the different catch concepts). Fishstat provides statistical information on catches as “nominal catches” by species and country. The nominal catch is the live-weight equivalent of the landings.

No attempt has been made to adjust for additional catches or landings arising from illegal or unreported fishing activities, or for possible inaccuracies in national fisheries statistical information as no adequate information is available at a global level. The principal reasons for this are the lack of any standardized reporting of such catches at global, regional or national level and the inability to resolve conflict with official reports of national fisheries catches.

Target catch

This term refers to catch of a species, a particular size or sex, or an assemblage of species that is primarily sought in a fishery, such as shrimp in a shrimp fishery or mature female fish in a roe fishery. The definition of targeted catch within a fishery is not static, as in a multispecies fishery, the mix of species targeted and caught may change over time.

Incidental catch

This term is used in the context of rare incidents or events such as catches of marine mammals, turtles or seabirds. Incidental catch is generally expressed in numerical terms rather than in terms of weight. Incidental catch is usually discarded and is considered as a discard for the purposes of this report.

Slipped catch

This term is applied to catches (usually purse-seine catches) that are released in the water without being taken on board the vessel. Slipped catches are considered to be discards. Quantities of slipped fish are difficult to estimate.

Trash fish

This term is not generally used in the study but refers to non-commercial or very low-value fish, usually caught by a trawl fishery. Trash fish is usually discarded unless collected at sea, or landed for aquaculture feed or fishmeal manufacture.

Debris

This term is used in the restricted sense of non-organic materials caught during fishing operations. Examples include rocks, sand, mud and plastic bottles. Organic materials such as dead shells, dead coral and plant materials (seaweed) are also considered debris.

Endangered and charismatic species

Endangered species^[8] are those threatened with local or global extinction. Charismatic species,^[9] sometimes referred to as “icon species”, are species that for cultural or religious reasons society accords an existence value substantially in excess of market value (e.g. dolphins, seals, albatrosses).

2.3 The discard database

2.3.1 Structure of the discard database

An inventory of the world’s fisheries was compiled^[10] and a search conducted for quantitative information on landings and discards from each fishery. The information was stored in the form of a master spreadsheet and with numerous supporting spreadsheets. The master spreadsheet is referred to as the “discard database” and contains 33 fields. The field structure is detailed in Annex C.4, Table 33. The fields can be divided into six categories.

• Area. Fields contain information on area, including the FAO statistical area code, national or regional fishery statistical areas and the name of the country.

• Fishery. Fields describe the fishery, including the name of the fishery, the type of gear and the target species.

• Landings. Fields contain information on the landings of the fishery including the quantity of landings in tonnes, year of reference and source of information.

• Discards - quantitative. Fields contain information on discards in the fishery: quantities in tonnes; the basis for estimation of the discards (e.g. observer reports, research survey); the reference to the source of the information; and the year or period to which the discard information refers.

• Discards - descriptive. Fields contain information on the reason for the discards, the measures or policies relating to discards and information on the status of exploitation of the fishery.

• Flags. Fields used either to distinguish particular records (e.g. those referring to incidental catches of marine mammals, or to small-scale fisheries).

The supporting spreadsheets were used to transform the landings and discard information provided in the source material to the formats and units required in the discard database. For example, some studies present discards as numbers of fish of different sizes, requiring a transformation from numbers discarded to weights discarded. The format and content of the subsidiary worksheets vary in relation to the different source materials.

2.3.2 The records in the discard database

There are over 2 000 records in the discard database of which 1 275 contain quantitative information on either landings or discards. The remaining records list fisheries for which quantitative information was not recorded.

Of these 1 275 records, 788 are quantitatively complete, i.e. they contain quantitative information on both landings and discards for a given fishery, 1 274 records contain information on catches, while 839 contain information on discard quantities. Some records are considered to be “duplicates”, i.e. there is more than one record for the same fishery, either for different time periods, or providing information from different authors or sources. Sixty-two records refer exclusively to numbers of marine animals caught incidentally (marine mammals, seabirds, turtles). Excluding duplicates and incidental catch records, 956 records contain catch information, while 755 records contain discard information. Some records are used for summary or checking purposes.

2.3.3 Scope of the database

The primary focus of the study is on commercial and subsistence marine capture fisheries for finfish and shellfish. Records of incidental catch of marine mammals, turtles, seabirds and protected species are included because of the growing impact of the catches of these species on fishing activities. All such incidental catches are considered to be discarded.

The study does not cover freshwater and recreational fisheries. The importance of catches and discards in some recreational fisheries is recognized, but few countries^[11] maintain adequate records. Freshwater species, species that migrate between freshwaters and marine habitats, reptiles, amphibians and aquatic plants have been excluded from Fishstat values and other values used.

Post-harvest waste, such as offal, guts, frames and waste from surimi processing, is not considered a discard. Roe fisheries (e.g. herring, or United States rock sole) may have substantial wastage of males, which are not considered as discards since much of the sorting takes place onshore.

Shark finning

In theory, the practice of shark finning may not be considered different from filleting and gutting. The shark carcass would then be considered as “offal” or waste of a processing operation rather than as a discard. However, in this study, finned sharks are considered to be discards because most of the edible portion is discarded and because of the widespread condemnation^[12] of and legislation^[13] on what is considered a wasteful practice.

No allowance has been made for the quantities of fish killed through interactions with fishing gear that does not result in their capture. These unobserved mortalities may be caused by the impact of trawl gear on the bottom, escapement or drop-out from nets, ghost fishing by lost nets and similar gear inefficiencies (e.g. there are high scallop mortalities associated with scallop dredges).

Both fishers and observers tend to focus on commercial species and recognized animals. There is a tendency to group tunicates, sponges, echinoderms, hermit crabs, worms and corals with jellyfish^[14] and perceive such biomass as debris, rather than as organic material. These non-commercial animals are frequently ignored and not recorded as discards during studies. This biomass tends to be omitted from estimates of discards. Many of these animals also pose practical problems of measurement of the biomass concerned (e.g. jellyfish), but may constitute a significant proportion of the total biomass harvested by trawls (Prena et al., 1999). The literature contains relatively few estimates of invertebrate discards and discards of unusual species such as sea snakes. Because of a lack of information the estimates have made no allowance for such unperceived or unrecorded discards.

2.3.4 References and bibliographic archive

To facilitate checking and updating of the discard database each discard database record contains two bibliographic reference fields indicating: (i) the source of the catch or landings information; and (ii) the source of the discard rate or discard tonnage estimate. These bibliographic references and those used in the text of the report have been compiled in a bibliographic database using a commercial bibliographic software. Electronic versions of many of the reference materials are organized by continent, country and several generic categories in an electronic archive held in FAO Fisheries Department, Fishery Industries Division (FIIT).

2.4 Assumptions and issues related to the method

2.4.1 Assumptions and aggregations

Certain assumptions and aggregations were necessary to prepare the discard estimates.

Correlation between total landings and discards

It is assumed that for a given fishery, during a given period, there is a linear relationship between landings and discards at the aggregate level. In other words, the discard rate of a sample has been applied to the total landings of the fishery to derive the total quantity of discards. This relationship does not necessarily hold true at the level of individual vessel trips or fishing operations,^[15] or in relation to the landings of target species. Furthermore, the linear nature of the relationship is open to question (Trenkel and Rochet, 2001). For further discussion see Section 2.4.3 on “raising”.

Representative sample

Discard rates for a particular fishery are generally based on a sample of discards by particular vessels. The sample discard rates are assumed to be representative of the entire fishery for the purposes of raising (extrapolating) the discards to the fleet or fishery level. While this assumption is essential in order to estimate the quantity of discards from a given fishery, the assumption is open to a range of criticisms (see Annex C, Section 2.6 for further discussion of discard sampling). As the quantity of the landings for which discard estimates have been made (the sample) accounts for 94 percent of the ten-year average of Fishstat nominal catch, it is assumed that the weighted discard rate is a representative discard rate for the global marine catch.

Countries and fisheries with low or negligible discard rates

Based on expert opinion from in-country sources, the fisheries in several countries were assigned a discard rate of 1 or <1 percent (see Annex C.5, Table 35). These countries include the Pacific Island states, the small island countries of the Caribbean and several South Asian and Southeast Asian countries. There are some notable exceptions to the latter category, e.g. the Arafura Sea shrimp fishery (Indonesia) and some Chinese fisheries and trawl fisheries in the Philippines.

In the absence of information to the contrary, fisheries in the following categories were also assumed to have discard rates ranging from <1 to 5 percent: (i) artisanal and subsistence fisheries, in particular those based on coral reef resources and small pelagic species and those based on collection by hand or by divers; (ii) fisheries prosecuted for fishmeal; and (iii) fisheries using factory trawlers where minimum size regulations are not applied.

Comparable fisheries

Fisheries considered to be similar were assumed to have a comparable discard rate, i.e. a known discard rate from one fishery was applied to a fishery considered to be similar. Each assumption is essentially a case-by-case subjective judgement by the author based on personal knowledge of the fisheries, on contacts with experts on the fisheries in question, or on apparent close similarities between fisheries in terms of area, gear, target species, markets and regulations as deduced from the literature on these fisheries. Examples include artisanal reef fisheries, tuna pole and line fisheries for a given ocean, and the set of Celtic Sea demersal fisheries.^[16]

Generic fisheries

In the absence of more detailed information, fish catches/landings were aggregated into generic fisheries, e.g. “south coast artisanal multigear multispecies fishery” or “all industrial trawl fisheries”. It is acknowledged that such groups may contain several different fisheries with different discard rates. With the help of local experts, future discard estimates may achieve a greater level of disaggregation and precision.

Fisheries for tuna and highly migratory species (HMS)

Tuna fisheries, fisheries for HMS and other highly dispersed fisheries for which statistical information has been collected by relevant regional organizations (e.g. by the International Commission for the Conservation of Atlantic Tunas [ICCAT], the Indian Ocean Tuna Commission [IOTC], the Inter-American Tropical Tuna Commission [IATTC], the South Pacific Commission [SPC] and the Commission for the Conservation of Antarctic Marine Living Resources [CCAMLR]) were generally aggregated by ocean or major fishing grounds rather than by flag state (e.g. Western Central Pacific tuna purse-seine fishery). The statistical information collected by the regional fisheries organizations was used as the basis for these discard calculations. This means that vessels from several flag states may be grouped into one fishery and database record. In order to avoid double counting of such catches, tuna and HMS catches were subtracted wherever possible from catches recorded by country in the discard database.

2.4.2 Availability and quality of information

A complete discard database record requires two pieces of information: (i) the total catch or landings by a fishery; and (ii) either the discard rate or the total quantity discarded by that fishery.

Absence of discard information

There is a general absence of quantitative information on discards or discard rates and relatively few countries have made comprehensive assessments of discards. Essentially, many of the difficulties encountered by Alverson in preparation of the 1994 assessment still exist.

Catch/landings information by fishery

At the country level, aggregate statistical information on fish catches is generally published by species, fleet or area, but more rarely by fishery. Few lists of fisheries exist in the published literature, much less the associated quantitative information on catches or landings. Nevertheless, such information is often available in the unpublished internal reports of national fisheries administrations. In many jurisdictions fisheries tend to have an amorphous or fluid definition. This is partly because several different gears may be used, several species may be targeted on a single fishing trip or by a particular vessel, and because the fishery changes over time. Consequently the attribution of catches to a particular fishery may be difficult.

At the global level, FAO nominal catch statistics (Fishstat) are available by area and species (or species group), but not by fleet, fishing gear or fishery. The FAO database of fishing vessels contains information on the numbers of decked and undecked vessels by size class and by type of vessel (e.g. trawler, longliner). The Fishstat (nominal catch) database and the vessel database are independent of each other. Thus, the Fishstat catch information cannot currently be linked to a type of vessel or fishery.

Quality and scope of discard information

Studies on discards rarely refer to the total catch of the fishery studied. Even in peer-reviewed publications, the terms “bycatch” and “discards” are at times used in an apparently equivalent or interchangeable manner, often rendering the information unusable without clarification from the authors. In many of the references cited it is not clear whether the catch values quoted refer to landings, gross catch or nominal catches.

Many discard studies have a narrow focus on the discards of one or few target commercial species, which may be reported in numbers, without the information necessary to convert the discard numbers to weights. Studies frequently ignore noncommercial finfish species and a significant discarded biomass of invertebrates such as tunicates, corals, coelenterates (jellyfish), sponges, echinoderms and other commonly discarded invertebrates.

Information in the published literature is generally incomplete. For example, the average weight of shark fins and the total weight of shark fins landed may be given,^[17] but the average weight of the shark is not provided, nor the estimated weight of shark as a percentage of the total catch. Numbers of fish are often given, but there are no means of converting the numbers to weight. The lack of characterization of the fleet or the difficulty in clearly identifying the fleet to which the discard information refers, creates substantial problems in identifying the corresponding catch/landings by fishery in the fishery statistics of the country or regional fisheries organization, and in the subsequent raising of observed discards to the fleet or fishery level.

Time series

Ideally, an analysis of trends in discarding practices should be based on adequate time series. Selected time series information is presented in Annex A.6 in support of the conclusions of this report. However, there is a general lack of globally representative time series on discards. Existing time series are often short as the observer programmes or the discard studies are often funded as a relatively short-term project rather than as an integral part of the normative fisheries information collection process. Interpretation of time series is further complicated by the need for supplementary information (e.g. changes in regulations, market conditions, catch per unit effort or size of year classes) needed to determine the reasons for changes in discard rates or in the absolute levels of discards.

2.4.3 Variability, sampling and raising

Some of the issues raised above are partly a result of the inherent characteristics of discard information, namely: (i) the high level of variability in discards; and (ii) the inability to correlate discards with other variables.

Variability

Discards reflect the response of the fisher to the changing circumstances of the fishery. The quantity of discards depends on an individual fisher’s decision on where and how to fish, on the results of the fishing activity and on the behaviour and payment of the crew. Discards will tend to vary^[18] in relation to catch composition, seasons, fishing areas, rigging of the fishing gear, market prices, port of landing, duration of the fishing trip, quota regulations, minimum landing size regulations and many other factors. Interannual variation may be linked to the presence of strong year classes of smaller less-marketable fish. Efforts to correlate the volume, composition and temporal or spatial variability of discards with such parameters have poor or mixed results. Fishers’ discard behaviour (see Annex D) is characterized more effectively by game theory than by stable correlations with single, or even multiple parameters. Despite the high variability inherent in individual discarding actions (e.g. by vessel, trip, trawl and season), aggregate (summed) discard volumes tend to provide a relatively accurate estimate of discards.

Sampling

A comprehensive sampling or discard recording programme is required to obtain an accurate estimate of discards. Such a programme can be carried out by on-board observers, on board by fishers, through interviews with fishers or through comparison of landings with a known profile of the total catch. Observer programmes have consistently been shown to provide the most accurate results, although this is not necessarily so if discarding is illegal. However, observer programmes may be costly and may not be appropriate for all types or sizes of vessels. The problems encountered in the design and uses of discard sampling programmes are further addressed in Annex C.2.

Raising

Raising or extrapolating discard estimates obtained from sampling to the level of the fishery or fleet presents a further set of problems. There are two basic options available: to raise as a function of effort or to raise as a function of total recorded catch of the fishery. Effort information is rarely available and catch information often means recorded landings. Raising discard estimates as a function of single target species landings data may result in substantial error (Matsuoka, 1997) as discards will tend to have a weaker correlation with the landings of a single species than with total landings. Target species landings are likely to be a function of the distribution and availability of the target species and may not be correlated (van Beek, 1998) with the temporal and spatial distribution and the size range of the discarded species. Complex models may also be used for raising, e.g. including information on catch composition, minimum landing sizes, year classes, seasons or market prices. The raising of discard estimates is further discussed in Annex C.3.

In this study, total quantities of discards were used if available as such in the cited literature, i.e. if the author had extrapolated from the fleet sampled to the total fishery. In these cases, the sample was more commonly raised by landings, and less frequently raised by effort. In cases where both raising methods were adopted, the mean estimate of discards was used, unless the author stated a preference. Where the raised discards quantity was not provided, discards were raised in linear^[19] proportion to landings, as the only available raising factor.

2.4.4 Analysis

Analysis and interpretation of the discard database encountered several difficulties that may result in inconsistencies and potential sources of errors.

Temporal inconsistency

Every effort has been made to use discard and landings information from the 1994 to 2003 period. For a given database record the information on which a discard rate is estimated and the information on landings for that particular fishery may refer to different years. Landings quantities and discard quantities from different years were summed to provide the respective global totals.

National check-sum gaps

The sum of the catches for fisheries where information is available is frequently less than the total recorded national catch. Assignment of a discard rate to the balance of the catch is problematic and was not attempted (also see Confidence limits on p. 14).

TABLE 1
Generic example of check-sum gap and temporal inconsistency issues

Estuarine and freshwater species

Freshwater species have been excluded from the FAO Fishstat quantities used in the study. Catches of freshwater species in marine and estuarine waters are not readily distinguishable in many catch statistics and may make a significant contribution to catches and discards in countries with large coastal wetlands and estuaries (e.g. Bangladesh, Brazil).

Distant water fishing nations

In the discard database, catches of distant water fishing nations were generally assigned to the coastal state where the fishing takes place. Alternatively, distant water catches were assigned to the flag state. The assignment is dependent on the information available with regard to the fishery or fleet. For example, with respect to a coastal state that has issued fishing licences to a distant water fleet, the name given to the fishery indicates the distant water nature of the fishery, e.g. country: Senegal; name of fishery: EU deepwater shrimp trawl.

Double counting

Double counting may arise as a result of including several records that relate to the same fishery. This occurs when several different studies quantify the discards in a particular fishery, possibly using different approaches, or for different time periods. In general, the most recent value or the value that is (subjectively) judged to be the most accurate has been chosen. Every effort has been made to avoid double counting in calculating total global discards and the corresponding total landings by using a single record for each fishery. Records in which double accounting arises are flagged in the database. All records containing a discard rate (e.g. a time series) are used to estimate mean discard rates for different fisheries.

Database bias

The results of literature searches or Internet searches using a keyword such as “discards” will tend to generate more information on fisheries in which discards are a concern than on fisheries for which discards are not considered problematic. Thus the records and fisheries contained in the discard database may be biased in favour of fisheries with high discards. The inclusion of records of artisanal fisheries with a low assumed discard rate and the use of the fishery-by-fishery methodology may counterbalance this potential bias. Large numbers of relatively minor fisheries are included, whereas the database information is incomplete for some major fisheries. The database is also biased in favour of fisheries for which documentation exists in English, French, Portuguese and Spanish since most literature searches were made in these languages. Internet information and “grey” literature published in other languages, in particular in Arabic, Russian, Japanese, Korean and Chinese, were not comprehensively accessed.

It is not possible to quantify these potential biases.

Confidence limits

Some of the references for individual records provide confidence limits for discard rates or discarded quantities. However, these individual record confidence limits cannot be summed or aggregated across records.

As the sum of the landings in the discard database is equivalent to 94 percent of the ten-year average Fishstat nominal catch, this “sample” represents a substantial proportion of the population of the world’s fisheries. As such, measures of sampling error of the weighted mean result in small upper and lower limits. The range of values for the global estimate is provided (see Annex A.1). The range does not reflect the internal variance of individual records.

Indications of the level variance in the discard estimates are provided for the major types of fisheries (e.g. shrimp trawl, finfish trawl) as standard deviations from the mean discard rate for each of these groups of fisheries (Annex A.2).

Narrow confidence limits are required for stock assessments in some jurisdictions, for example by the EC. The observer coverage required to achieve similar confidence limits for discards may incur substantial costs.

Survival of discards

This study does not address the survival of discards, which has been studied in many fisheries. Among the factors influencing the survival of discards are the depth of fishing, duration of trawls, soak time for lines and nets, and the physiology of the species discarded. In comparison with fish escaping from trawls, those fish escaping from traps tend to have a high survival rate as do releases of live lobster and crab.

Impact of discards

An associated FAO study (Poseidon Aquatic Resource Management Ltd, 2003) has examined aspects of the economic and ecological impacts of discards. These impacts are difficult to distinguish and isolate from the impacts of bycatch and fishing activities. The ecological and economic issues are briefly discussed in Sections 4.5 and 4.6.3 respectively.

Interpretation

The global discard estimate provided in the results section may be misinterpreted, no matter how carefully predicated by caveats. As previously indicated, the records in the database may be biased in favour of a high estimate of discards. The database remains incomplete and discard information on several important fish-producing countries has not as yet been compiled, or is only partially compiled. These countries include the Democratic Republic of Korea, the Republic of Korea and the Russian Federation. The omission of some important fisheries may in itself be a source of bias. Assumptions regarding discard rates applied to certain fisheries will require further verification. Discard estimates may be politically sensitive and imprecise discard estimates can lead to political and other difficulties.^[20] As such, these results must be treated with due caution and interpreted in the appropriate context.

2.4.5 Future updating of the discard estimate

Discard practices may change rapidly as a result of changes in fish stocks, in regulations, in markets, or in any of the multiple factors influencing the behaviour of fishers. The estimates should therefore be repeated at intervals in order to monitor trends in discarding practices and the implementation of the relevant parts of the CCRF.

FAO plans to update the discard estimates periodically from national sources and through regional fisheries organizations. The country-by-country architecture of the discard database and the references associated with each record enable updating, verification, substitution or addition of records by competent experts from each FAO statistical area.