Status and assessment of large pelagic resources

Introduction

For management purposes, the large pelagic fish stocks of the region should be considered in two separate groups:

• large coastal pelagics, including wahoo, blackfin tuna and mackerel species Scomberomorus spp; and

• more widely distributed oceanic species, such as yellowfin tuna, skipjack tuna, swordfish, billfish and others.

All of these stocks are transboundary, moving in and out of the EEZs of all or most of the coastal states and extending into international waters. However, the large coastal pelagics are largely confined to the WECAFC area and could therefore be managed by a regional arrangement that remains to be established. In contrast, the distribution of large oceanic pelagics goes beyond the WECAFC area and in some cases is trans-Atlantic. Thus fisheries for these stocks need an arrangement with much wider jurisdiction, such as that provided by ICCAT.

The relative importance of large pelagic fisheries varies considerably among CARICOM states (Table 1). They are most important in the small-island states of the eastern Caribbean, particularly the southern states of Barbados, Grenada, Saint Lucia and Saint Vincent and the Grenadines.

The species composition of the reported landings shows the high relative importance of the large coastal pelagic species with regional distribution, such as dolphinfish, blackfin tuna and Scomberomorus spp. (Table 2).

TABLE 1

Landings (1 000 tonnes) reported to FAO by species group for CARICOM countries for the period 1990-1999

TABLE 2

Average annual landings (tonnes) reported to FAO of coastal and oceanic pelagics for 1990-1999 for CARICOM countries

Assessment of large pelagic stocks of importance to CARICOM countries

ICCAT assessments

ICCAT regularly assesses a number of the species of interest to CARICOM countries, including all large tunas and billfish. However, these assessments are either Atlantic-wide or cover half the Atlantic, depending on how ICCAT has defined the stocks for each species (Table 3). This paper only discusses assessments of those ICCAT stocks that occur within the waters of CARICOM countries.

Northern Atlantic swordfish

This stock occupies Atlantic waters north of 5°N, excluding the Mediterranean. The last assessment of this stock was conducted in 1999 and used both sequential population analysis and non-equilibrium production models to estimate stock status. Maximum sustainable yield (MSY) was estimated at 13 400 tonnes (range 7 600-15 900 tonnes). Fishing mortality in 1999 was 1.3 times the level that would produce MSY, thus the stock was suffering overfishing. In spite of the high fishing mortality, in recent years the stock seems to be slowly recovering, and stock size in 1999 was estimated to be at 65 percent of BMSY, the biomass that would produce MSY. Recovery seems to be the result of both management intervention (catch limits) and the appearance of two recent episodes of high recruitment (ICCAT, 1999). ICCAT has recently concluded that catches must be lower than the current levels to ensure a recovery (ICCAT, 2001a).

Atlantic yellowfin tuna

Yellowfin tuna occupies tropical areas throughout the Atlantic. The last assessment of this stock was conducted in 2000 and used equilibrium and non-equilibrium production models to estimate stock status. MSY was estimated at 150 000 tonnes, with estimates ranging from 145 000 to 152 000 tonnes depending on the model and assumption used. In 1999 effective fishing effort was either at or close to the level that would support MSY. The stock is considered fully-fished (ICCAT, 2001a). ICCAT is concerned, however, that the tendency for fishing power to increase, if not controlled or counteracted, will ultimately increase fishing mortality to overfishing levels (ICCAT, 2001a).

Western Atlantic skipjack

Skipjack occupies tropical and subtropical areas throughout the Atlantic. The last attempt to assess this stock was made in 1999, but no estimates of MSY or other benchmarks were provided. Relative abundance indices suggest abundance has been relatively stable since the early 1980s (ICCAT, 1999).

Atlantic blue marlin

Blue marlin occupies tropical and temperate areas throughout the Atlantic. The last assessment of this stock was conducted in 2000 and used non-equilibrium production models to estimate stock status. MSY was estimated at 2 000 tonnes. Fishing mortality in 1999 was four times the level that would support MSY, and the 1999 biomass was 40 percent of the level supporting MSY. Thus the stock is considered to be overfished and suffering overfishing (ICCAT, 2001b). There are large uncertainties associated with these estimates. The depleted status of the stock suggests that harvest levels in the late 1990s continued to lead to overfishing. It is not known if the measures agreed to in 2000 will change this situation (ICCAT, 2001b).

Atlantic white marlin

White marlin occupies tropical and temperate areas throughout the Atlantic. The last assessment of this stock was conducted in 2000 and used non-equilibrium production models to estimate stock status. MSY was estimated at 1 300 tonnes. Fishing mortality in 1999 was seven times the level that would support MSY, and the 1999 biomass was 15 percent of the level supporting MSY. Thus the stock is considered to be overfished and suffering overfishing (ICCAT, 2001b). These estimates are presumed to be highly uncertain, although the uncertainty has not been quantified. The depleted status of the stock suggests that harvest levels in the late 1990s continued to lead to overfishing. It is not known if the measures agreed to in 2000 will change this situation (ICCAT, 2001b).

Western Atlantic sailfish

Sailfish occupy tropical areas of the Atlantic and are more coastal than marlins. Much of the catch in the offshore area is mixed with spearfish, and the two species are often mixed in landing reports. This precluded assessments of the pure sailfish stock until 2001. The only assessment of this stock was attempted in that year and used non-equilibrium production models to estimate stock status. The assessment failed to produce estimates of stock status parameters from production models. Relative abundance estimates, however, suggest that abundance fell dramatically in the 1960s and has not increased much since. Current catches seem sustainable, but it is not known how far the current levels are from MSY (ICCAT, 2001b).

Other assessments

Other than ICCAT, there is no organization that conducts stock assessments for pelagic species at the level of the Caribbean region or larger. WECAFC collates information on fisheries assessments through its Scientific Advisory Group, but the group does not have the capacity to perform assessments (FAO, 2001a). FAO provides landing trends for each species in the western central Atlantic region of the FAO database on landings, but does not provide assessments of fishery status from such trends (FAO, 1997b). The only other assessments of pelagic stocks are those conducted by individual countries or by subregional organizations such as CFRAMP. For an overview of historical knowledge on pelagic fish in the western central Atlantic, see Mahon (1996a).

Dolphinfish

Oxenford and Hunte (1986) hypothesized that there are two stocks of dolphinfish in the western Atlantic, with a boundary near Puerto Rico. However, recent studies indicate that this hypothesis is not correct. Based on mitochondrial DNA variation, Wingrove (2000) concluded that dolphinfish throughout the western central Atlantic, including the Gulf of Mexico, belong to a single genetic stock. An age and growth study by Rivera and Appeldoorn (2000) did not find the growth differences near Puerto Rico predicted by the two-stock hypothesis. Off Venezuela, the period of peak availability of dolphinfish to fisheries is not consistent with the northward migration of the southern stock proposed by Oxenford and Hunte. In assessments by the United States of America, Gulf of Mexico dolphinfish have been considered part of the Atlantic stock, mainly because of lack of evidence to the contrary (Prager, 2000).

Prager made a preliminary assessment of the northern stock of western Atlantic dolphinfish exclusively from United States (US) data for 1986-1997. He estimated that in 1998 it was not overfished because biomass was 150 percent of B_MSY. Similarly, Prager determined that fishing mortality in 1997 was 50 percent less than F_MSY (the fishing mortality rate that achieves equilibrium MSY), thus dolphinfish were not undergoing overfishing. The estimated MSY for this stock is 12 200 tonnes (80 percent confidence interval is 8 500~21 100). Relative abundance indices do not suggest any decline in abundance, rather the opposite, with biomass increasing significantly from the late 1980s to the early 1990s. Prager acknowledges that these population indices are uncertain and raises the possibility that this increase is an artefact. The mean size of dolphinfish caught has not declined historically either.

Mahon and Oxenford (1999) tried using yield per recruit analysis to define benchmarks for Caribbean dolphinfish. They concluded that life history traits of this species suggest that yield per recruit could be maximized at exploitation rates that would lead to unacceptably low spawning stocks. They therefore suggested that yield per recruit analysis should not be used as a source of benchmarks for the species.

Parker et al. (2000) performed an assessment of the dolphinfish stock in eastern Caribbean waters and used a combination of length-based models (length-based catch curve and length-based virtual population analysis (VPA)). Their results suggest that fishing mortality is much greater than f_MSY (fishing effort corresponding to F_MSY) and that, as a result, catches from the stock are much lower than MSY. These results are highly uncertain and dependent on growth parameters not yet well estimated. They may also be biased, because there is no strong indication that the eastern Caribbean population is a separate stock from the stock present in the rest of the Caribbean. Thus the apparent high mortality estimated in the assessment may reflect a migration of fish out of the eastern Caribbean.

King mackerel

Two migratory groups are considered to occupy US coastal waters of the south Atlantic and Gulf of Mexico. These groups are managed as separate stocks, and quotas are set individually for each of them. According to VPA-derived estimates, the Atlantic migratory group is not considered to be suffering overfishing, because static SPR^[15] was estimated at 36 percent. Transitional SPR is estimated at 39 percent, thus the stock is not overfished either. According to the National Oceanic and Atmospheric Administration (NOAA), US Department of Commerce, current landings have averaged 3 000 tonnes per year in recent times (NOAA, 1998). By contrast, the Gulf of Mexico migratory group was more recently assessed as continuing to suffer overfishing and being overfished (Legault, Ortiz and Scott, 2000). Benchmark estimates for B/B_MSY and F/F_MSY were, however, close to the targets set by the Gulf of Mexico Management Council. Current landings average about 4 000 tonnes per year.

Spanish mackerel

Two migratory groups are considered to occupy US coastal waters of the south Atlantic and Gulf of Mexico. These groups are managed as separate stocks, and quotas are set individually for each of them. According to VPA-derived estimates, the Atlantic migratory group was not considered to be suffering overfishing in 1997, because static SPR was estimated at 42 percent. Transitional SPR was 40 percent, thus the stock was not overfished either. Landings averaged 4 500 tonnes per year (NOAA, 1998). Similarly, the Gulf of Mexico migratory group was assessed as not suffering overfishing, and static SPR was estimated at 47 percent. The stock was not overfished either, with transitional SPR estimated at 35 percent. Current landings average less than 1 400 tonnes per year (NOAA, 1998).

Wahoo

The United States National Marine Fisheries Service (NMFS) within NOAA has not done any assessment or defined stocks of wahoo in US waters. The Gulf of Mexico Fishery Council uses estimates of average annual landings as a proxy for MSY. Current estimates of this proxy MSY oscillate between 650 and 750 tonnes (Michael Prager, NMFS, personal communication). George, Singh-Renton and Lauckner (2000) assessed the wahoo stock in eastern Caribbean waters using a combination of length-based models (length-based catch curve and length-based VPA). Their results suggest that fishing mortality is much greater than f_MSY and that, as a result, catches from the stock are much lower than MSY. These results are highly uncertain and dependent on growth parameters not yet well estimated. They may also be biased, because there is no strong indication that the eastern Caribbean population is a separate stock from the stock present in the rest of the Caribbean. Thus the apparent high mortality estimated in the assessment may reflect a migration of fish out of the eastern Caribbean.

TABLE 3

Stocks identified by ICCAT for species considered in this analysis

Source: ICCAT Standing Committee on Research and Statistics (SCRS) detailed reports for each species.
^a Italics indicate stock fished in waters of CARICOM countries.

Trends in landings as reported to FAO

Introduction

Landings data are the basis for all fishery assessments, and yet these data are not always available for many fishery resources. The monitoring of landings is the responsibility of the fisheries department of each country. However, such data are not often easily accessible at the regional level because there is no institutional arrangement among countries that ensures a common database for the entire Caribbean region.

There are international organizations that have comprehensive databases of landings of pelagic fish that encompass the Caribbean region: the FAO worldwide landing database of marine fisheries production and the ICCAT landing database for pelagic fish of the Atlantic. Not all members of the Caribbean basin are part of ICCAT, although they are all asked to report catches to the organization. In addition, ICCAT actively collates data only on offshore pelagic fish and has limited information on coastal pelagic fish. The FAO database is therefore the only comprehensive database for all pelagic species landed in the Caribbean region.

FAO database

The FAO database describes the annual volumes of nominal fish catches (measured in metric tonnes of live weight) as classified by country or FAO major fishing area. The fish catches are presented at various taxonomic levels - commonly family, genus or species. The database currently provides annual statistics representing 50 years of catch landings data from 1950 to 2000. These data describe catches in approximately 245 countries or land areas comprising 27 FAO major fishing areas (www.fao.org).

The statistics in the FAO database reflect landings data reported by individual countries for their commercial, industrial and small-scale fisheries in both coastal waters and high seas. However, data from recreational fisheries are not reported. Many countries are members of FAO regional fishery commissions and are required to report data from their flag vessels fishing in the region to the commissions. In the Caribbean region, the FAO database is the only source that contains data for all fished species.

The database is not without its limitations. Its statistics are necessarily influenced by a country’s capability and willingness to accurately collect, process and report data from its fisheries sector. Moreover, as these abilities vary widely between countries and regions, care must be taken when making comparisons using the data. The statistics can be susceptible to underreporting or non-reporting of discarded catches, harvests that are misreported and failures to include illegal fishing data. A further limitation of the data is that catch is not always reported at the species level. Data can be fragmentary for some species or taxonomic groups within a region, because there is often a lack of uniformity among countries in reporting formats (e.g. taxonomic divisions used or species reported). Furthermore, in certain regions the small-scale fisheries sector is underrepresented in the data.

Landings data were extracted from the Capture Production database in FAO FISHSTAT+ for the period 1970-1999 for catches of all species and countries in the western central Atlantic region. For the purpose of this study, species were grouped into three species groups: unclassified, not pelagic and pelagic. Pelagics were further divided into “offshore”, “coastal”, “small” and “other”; only the first two pelagic groups contain species of interest to this study (Table 4). These groupings were designed to satisfy the objectives of the project and, although related to ecological classifications, they do not necessarily reflect these strictly. Some species were included in the group of “other pelagics” because there were too few data to properly treat them in the analysis.

Given the focus of the project on CARICOM fisheries, countries were placed in country groups. They were also classified into “environment” categories of countries that had similar types of fisheries and thus similar species composition in their landings (Table 5). The “continent” category includes countries on the continent of South America, in Central America and an island country within the continental shelf (Trinidad and Tobago). “Small Antilles” includes countries in the southern Caribbean with small islands and platforms. “Large Antilles” includes countries in the northern Caribbean with extensive shallow water habitat because of the size of the island or the presence of extensive coralline platforms (Bahamas, Turks and Caicos Islands). Three countries, Mexico, the United States and Venezuela, were considered to be in their own category because of the size and complexity of their fisheries.

TABLE 4

FAO taxonomic groups included in each of the species-group categories considered in this study^a

a The study deals in depth with only the first two groups.
b Not elsewhere included.

TABLE 5

Country groupings used in the analysis of FAO catches

Estimation of landings by species group

The analyses only attempt to reconstruct the composition of reported landings; we have made no attempt to estimate unreported ones. Two different assumptions were made, generating two estimates of landings by species and species group. These two assumptions represent, in a sense, two extreme scenarios, and the likely composition of the reported landings falls somewhere between the two.

Assumption 1. All taxonomic categories are reported at the lowest category available in the FAO reporting system, therefore estimated landings equate reported landings. This is the standard assumption made in most analysis of FAO data. For the purpose of this study it implies among other things that:

• “marine fish nei” landings contain no landings of pelagic fish available at the species level on the FAO database;

• “tunas nei” landings contain no landings of bluefin or yellowfin tuna or any other species available at the species level on the FAO database; and

• “mackerel-like fish nei” landings contain no landings of Spanish mackerel, king mackerel or any other species available at the species level on the FAO database.

Thus if a country does not report landings of yellowfin tuna, even if it does report landings of “tunas nei”, this assumption results in an estimate of zero for yellowfin tuna landings.

Assumption 2. Taxonomic categories that group lower taxonomic categories have the same species composition as that of the landings reported at lower taxonomic levels. This assumes that there is no preferential reporting for any species, including the most valuable species such as tunas. For the purpose of this study it implies among other things that:

• “marine fish nei” landings contain a proportion of pelagic fish equal to the proportion that can be calculated from landings reported at lower taxonomic levels;

• “tunas nei” landings contain a proportion of landings of bluefin tuna, yellowfin tuna and all other tunas equal to the proportion that can be calculated from landings reported at specific levels; and

• for taxonomic groups of fish that have at least one representative characterized at the species level, such species are the only ones landed. Therefore all “tunas nei”, all “mackerel nei” and all “marlins and sailfish nei” represent landings of one of the species of these groups represented in the FAO database.

Assumption 1 does not require additional rules for estimation of landings. Assumption 2, however, is constrained by the availability of detailed information on landings composition at low taxonomic levels. For many countries such information is not available. The assumption was therefore modified as follows:

Assumption 2a. Taxonomic categories that group lower categories have the same species composition as that of the landings reported at lower levels, as estimated from the landings of all countries in each “environment group” for the entire time period 1970-1999.

Thus the composition of “marine fish nei” for all countries in the small Antilles was estimated from the landings reported at lower taxonomic levels for all small Antilles countries. Note that as Mexico, the United States and Venezuela each constitute their own group, their corrected landings are only estimated from the ratios obtained for their own country.

In order to check whether assumption 2 is grossly violated or not, we analysed the country-by-country landings and looked at any evidence that some groups of fish tended to be better reported than others because of their value.

Estimation of landings by species

The method described above can be used to estimate species-specific landings according to the two hypotheses presented. Assumption 2a needs to be extended to lower unclassified groupings such as “mackerel-like fish nei” or “tuna-like fish nei”. Below, we used this method to estimate the landings of yellowfin tuna according to the two hypotheses. The specific estimate made for yellowfin landings L_2a for assumption 2a then becomes:

where L_t is the landings apportioned to yellowfin tuna from the “tuna-like fish nei”^[16] category estimated as:

where L_u is the landings apportioned to yellowfin tuna from the “marine fish nei” category estimated as:

where L_{offshorepelagics} are the landings apportioned to offshore pelagics from the unclassified marine fish, as estimated in the previous section, L_i are the reported landings of all offshore pelagics identified at the species level, L_billfishnei are the reported landings of "marlins, sailfish nei" and L_y are the reported landings of yellowfin (hypothesis 1). These formulae are used for each group of countries within the same country environment group.

Results

Estimation of landings by species group

There is no relation between the reported landing ratios of pelagic/not pelagic and the country’s proportion of landings reported as unidentified (marine fish nei). Thus, at the country level, countries that have a greater proportion of pelagic landings do not necessarily report fewer landings of unclassified marine fish (Figure 1).

The proportion of identified landings in each species group differs among environment groups. For coastal pelagics and offshore pelagics, this proportion tended to increase over the three decadal periods considered in the analysis (Figure 2). The change, however, is not large and thus supports our decision to estimate a single correction factor for apportioning unclassified landings for each pelagic species group.

Estimates of the species-group composition of landings by country group obtained from the two assumptions differ substantially (Figure 3). The contribution to the total western central Atlantic area landings of offshore pelagics from CARICOM countries is 20 percent according to assumption 2a and only 4 percent according to assumption 1. Similarly the contribution of CARICOM countries to the total landings of coastal pelagics is 22 percent according to assumption 2a and only 11 percent according to assumption 1.

The total landings of pelagics for the region are larger for assumption 2a. Although the total unclassified landings for the region only represent 17 percent of total landings, assumption 2a leads to large increases in estimated landings for pelagic groups as compared to the estimates obtained from assumption 1 (Table 6a). For offshore pelagics landings are about 50 percent greater, and for coastal pelagics 90 percent greater, for assumption 2a than for assumption 1.

FIGURE 1
Proportion of landings reported as unidentified (marine fish nei) as a function of the ratio (log scale)
of pelagic landings to not pelagic

CARICOM countries reported 79 percent of their historical landings as unclassified marine fish. As a result, apportioning such landings to one of the pelagic groups increased estimated landings by between 100 and 600 percent (Table 6b). Estimated landings of coastal pelagics from CARICOM are 93 000 or 344 000 tonnes depending on whether assumption 1 or 2a is used. Similarly, estimated landings of offshore pelagics range from 65 000 to 443 000 tonnes for assumption 1 or 2a respectively. Historical trends of yearly landings for these two groups also differ greatly, highlighting the uncertainty generated by the presence of large amounts of unclassified landings. Annual landings of coastal pelagics at the end of the 1990s could be either 5 000 or 15 000 tonnes per year, depending on the assumption made about unclassified landings (Figure 4). Similarly annual landings of offshore pelagics at the end of the 1990s could be either 5 000 or 22 000 tonnes (Figure 5). At the country level these differences vary depending on the relative amount of landings reported as unclassified marine fish. For some countries such as Barbados, the estimated landings following both assumptions are rather similar (Figure 6); for others, Saint Lucia for example, landings are similar since 1995, but very different prior to that (Figure 7).

TABLE 6

Estimated landings (1 000 tonnes) for the period 1970-1999 by species group, according to two different assumptions about the species composition of unclassified landings

FIGURE 2
Changes by decade in the correction factor used to apportion landings of unclassified fish to each
of the pelagic species groups following assumption 2a

FIGURE 3
Composition of landings by species group in FAO area 31 and by country group following two assumptions related to the composition of marine fish nei landings

First and second charts, offshore pelagics; third and fourth, coastal pelagics. Charts 1 and 3, assumption 1; 2 and 4 assumption 2a.
Note how the contribution to the landings of offshore and coastal pelagics from CARICOM countries is much greater according to assumption 2a.

FIGURE 4
Estimated landings (1 000 tonnes) of coastal pelagics by CARICOM countries depending on the assumption made about the composition of unclassified landings of marine fish nei

FIGURE 5
Estimated landings (1 000 tonnes) of offshore pelagics by CARICOM countries depending on the assumption made about the composition of unclassified landings of marine fish nei

Estimation of landings by species

As an example of the effects of different assumptions about the species composition of unclassified landings, we estimated landings of yellowfin tuna, the most common of all offshore pelagics in the CARICOM region. Landings of yellowfin tuna for the 1970-1999 period again differ substantially depending on the assumption made. The largest percentage increase in landings found by using assumption 2a corresponds to CARICOM countries, >100 percent, but the overall increase for the total region is much less, 19 percent (Table 7).

The historical landings trend for yellowfin tuna from CARICOM countries is therefore very uncertain (Figure 8). A large portion of this difference is due to apportioning large parts of the unclassified catch of fish from Guyana and Suriname to yellowfin tuna. This is not likely to be correct and is due to the heterogeneity of

TABLE 7

Estimated landings (1 000 tonnes) for the period 1970-1999 of yellowfin tuna by country group, according to two different assumptions about the species composition of unclassified landings^a

^a A large part of this, 72 000 tonnes, is landings of marine fish nei from Guyana and Suriname, probably wrongly apportioned to yellowfin tuna (see text).

FIGURE 6
Estimated landings (1 000 tonnes) of coastal and offshore pelagics in Barbados depending on the assumption made about the composition of unclassified landings of marine fish nei

FIGURE 7
Estimated landings (1 000 tonnes) of coastal and offshore pelagics in St Lucia depending on the assumption made about the composition of unclassified landings of marine fish nei

the environments present in the countries included in the “continent” category. Again for specific countries, such as Barbados, which report little unclassified catch, landing trends for both hypotheses are very similar. For other countries, such as Saint Lucia, which report a large portion of their catch as unclassified, the difference between hypotheses is much larger (Figure 9).

Much of the apportioning of unclassified catch to yellowfin tuna comes from the practice of apportioning some marine fish nei to yellowfin tuna. In only a few countries (e.g. Trinidad and Tobago) is it the result of apportioning tuna nei to yellowfin tuna (Figure 10).

FIGURE 8
Estimated landings (1 000 tonnes)of yellowfin tuna for CARICOM countries depending on the assumption made about the composition of unclassified landings of marine fish nei and tuna nei

FIGURE 9
Estimated landings (tonnes) of yellowfin tuna in Barbados and St Lucia depending on the assumption made about the composition of unclassified landings of marine fish nei and tuna nei

FIGURE 10
Estimated landings (tonnes) of yellowfin tunaa for CARICOM island countries for the entire time period (1970-1999) according to hypothesis 2a

^a Landings are separated depending on whether they were identified as yellowfin tuna (solid), apportioned to yellowfin tuna from the catch of “marine fish nei” (stippled) or apportioned to yellowfin tuna from the catch of “tuna nei” (cross-hatched).

^b Montserrat

Discussion

In spite of their limitations, FAO landings data remain the most reliable source of fisheries information worldwide. Some of the limitations of these data are the result of the intergovernmental arrangements that created FAO, which require that the Organization rely on Member Nations to provide accurate and precise data. Some limitations can be overcome through statistical analysis and the incorporation of other sources of data. Watson and Pauly (2001) document attempts to quantify these limitations by identifying an alternative hypothesis for estimating landings by species.

Some information on the species composition of unclassified landings may already be available for some countries and could be used to correct the results presented in the current analysis. It is likely, for instance, that a large portion of the landings of marine fish nei in the Lesser Antilles is made up of reef fish. While for the Guianas-Brazil region, a large portion of these landings may be groundfish.

One could proceed country by country to allocate the nei to one of the aggregate groups based on a structured set of criteria, as was done by Watson and Pauly for global landings. Alternatively, interviews with the fisheries division of each country may help develop a country-specific apportioning algorithm for unclassified landings. Assumption 2a implies that the composition of unclassified landings does not change over time. This is not true, of course, and the proportion of unclassified fish apportioned to offshore and coastal pelagics following assumption 2 changes by decade. The changes are greater for countries in the continent group, and much smaller for countries from the Antilles. This may partially explain the large catches of pelagics apportioned to Guyana and Suriname under assumption 2a. As mentioned earlier, it is likely that these countries should be included in their own group, separate from the other countries on the South American continent. Unfortunately, much of the catch of these countries is unclassified, and thus there is not a lot of information with which to apportion catches to species.

The apportioning of unclassified catches to recalculate the yellowfin tuna catch demonstrated that the uncertainty associated with species-specific landing estimates is very great for many countries of the region. It is interesting, however, that some countries (e.g. Barbados) have managed to introduce effective monitoring systems for their reported catch. For these countries, the uncertainty in specific landings is small and suggests that such monitoring systems can be implemented effectively in CARICOM countries.