An Assessment of Atlantic Bluefin Tuna (1994)

The degree of fish movement between fishing grounds in the eastern and western Atlantic Ocean and the degree of genetic mixing in the spawning areas in the eastern and western Atlantic Ocean are both important topics relevant to the management of Atlantic bluefin tuna. The review and reanalysis of tagging data constitute the committee's evaluation of the scientific basis for physical movement or mixing of fish between fishing grounds. These tagging data, however, do not provide information on the degree of genetic mixing.

Tag-recapture data provide the strongest evidence available for transatlantic movement of fish. Atlantic bluefin tuna were marked and released in the western Atlantic Ocean from 1954 in varying numbers until 1987 (Parrack, 1990). The total number tagged by the National Marine Fisheries Service (NMFS) since the inception of the program has been over 15,000. Early releases were primarily of large (giant) fish, although more recently releases have included numerous small fish.

An overview of movement from the west to the east by size class (i.e., from age 0 fish to giant fish), as indicated by tag returns, appears in Suzuki (1990).

Recaptures through 1978 of large fish released in the Bahamas in the 1960s, and of small fish released along the U.S. Atlantic coast, document unequivocally that transatlantic migration occurs (Table 3-1). Transatlantic migrants include nine giant fish taken off the Norwegian coast, one medium fish (recovered as a giant tuna over 10 years later) taken off the southern coast of Spain, and 46 small fish taken chiefly from the Bay of Biscay. Estimates of the total proportion of fish tagged in the western Atlantic Ocean in eastern Atlantic samples (Table 3-1) range from 9.3% for giant fish and medium fish to 2.1% for small fish. Estimates for giant and medium fish differ slightly from those given in Mather (1980), who found a total of 10 western migrants in 126 eastern Atlantic recaptures (10/126 = 7.9%). Suzuki (1990) estimated an "overall" proportion of western-tagged fish in eastern Atlantic samples of 3.2% as of 1988.

Estimates of west to east transatlantic movement are variable and depend on the size of fish tagged and tagging locality. All but one of the giant fish transatlantic recoveries were tagged near the Bahamas, and five of the nine migrants were recovered in Norwegian waters in the same year. These results indicate that giant bluefin tuna can travel long distances; only one of the remaining 91 giant tuna or medium fish tagged in the western Atlantic Ocean was recovered in the eastern Atlantic Ocean. This fish was tagged near Bermuda and recovered 10 years later. Mather (1980) hypothesized that the transatlantic migration of

giant fish was not part of an annual pattern but rather represented irregular migration of varying numbers of fish in different years. This hypothesis deserves consideration because bluefin tuna tend to follow ocean currents (Sella, 1929, cited in Mather, 1980) and because of the proximity of the Bahamian tagging sites to the Gulf Stream. This possible transatlantic "route" for giant fish is strikingly similar to the one for larvae proposed by Murphy (1990). Finally, Mather (1980) also noted that the transatlantic movement of small bluefin tuna was variable and that virtually all transatlantic movements of small fish were from the Mid-Atlantic Bight to the Bay of Biscay. Mather concluded that the west to east transatlantic movement of small fish might be due to unusual stimuli and that the great majority of Atlantic bluefin tuna remain on one side of the Atlantic Ocean or the other. Alternatively, the "nonrandom" movement of small bluefin tuna from the Mid-Atlantic Bight to the Bay of Biscay may reflect the importance of both nursery areas for small bluefin tuna.

Eastern Atlantic/Mediterranean bluefin tuna have been tagged and released since 1911, although most of the tag and release data are from 1957 to the present (Cort and Liorzou, 1990c). As in the western Atlantic Ocean, most of the early releases were giant tuna, with the number of small fish released and recovered increasing over the past two decades. Numbers of fish released and recaptured from the eastern Atlantic Ocean and Mediterranean Sea from 1911 through 1992 are listed in Brunenmeister (1980), Cort and Liorzou (1990c), and Cort and de la Serna (1993). A synopsis of these listings is shown in Table 3-2. The proportion of eastern-tagged fish among western Atlantic recaptures varies from 0% for giant fish to 4.5% for small fish. Virtually all of the transatlantic migrants recaptured off the U.S. coast were released from the Bay of Biscay, and most were captured between 10 and 20 months after first released. The "total" proportion of eastern fish among western Atlantic recaptures is 4.0% (19/472) after adjustment for the greater number of small fish among recaptures. This estimate is similar to the estimate of 4.4% of Cort and Loirzou (1990c). Estimates of eastern Atlantic fish among Mediterranean recaptures are 11.3% for giant fish and 2.8% for small fish. The proportion of Mediterranean fish among eastern Atlantic recaptures is 2.8% for small fish.

The absence of documented east to west transatlantic movement of large fish is striking and merits further investigation. Considering only small fish, the proportion of transatlantic immigrants in western and eastern Atlantic samples differs: estimates made here suggest an overall proportion of small western fish among eastern Atlantic recaptures of 2.1% and a proportion of 4.5% eastern-tagged fish among western recaptures.

Tagging data suggest that physical movement or mixing of Atlantic bluefin tuna in fishing grounds from opposite sides of the Atlantic Ocean is significant. Thus, the committee reanalyzed the tagging data to estimate rates of transatlantic movement between fishing grounds. This analysis was undertaken to provide rigorous estimates of transfer across the Atlantic Ocean, which has not been attempted previously. Because it is not possible to estimate transfer rates without knowledge of the total mortality rates and the nonreporting rates (tags recovered but not reported), these quantities are also estimated.

The problem of transfer rates is described by the rates of change in the population sizes, N_w for the west and N_e for the east. Equations for the transfer of fish from the west to the east, applicable to fish tagged with a single tag, are:

(1)

(2)

M = annual rate of instantaneous natural mortality plus shedding. ¹

F_w = annual rate of instantaneous fishing mortality in the west.

F_e = annual rate of instantaneous fishing mortality in the east.

T_w = annual rate of instantaneous transfer from west to east.

Integration of this pair of equations provides estimates of the population size at any time t, over the interval for which F_w is constant in the west and F_e is constant in the east. The solutions are given by:

where X = M + F_e and Y = F_w-F_e + T_w

Similarly, the equations for the caches are:

When t is sufficiently large:

When P_w and P_e are the proportions of tags recovered that are reported for the west and east, respectively, then:

(3)

Analogous equations (with the subscripts e and w reversed) are used to estimate the transfer rate of fish from the east to the west. Because the models given by Equations (1) and (2) do not admit roundtrip movement, any such return movement would cause the transfer rates estimated by Equation (3) to be underestimates of movement rates from one side of the ocean to the other. Also, if fishing mortality decreases with age, as it appears to do, the movement rates estimated by assuming a constant fishing mortality will be underestimated.

Tag-recovery information from Corte and de la Serna (1993) (Table 2; which is reproduced below as Table 3-3) and from NMFS archives (file name MRFISH)

was used to obtain values of catch (C). Some descriptions of qualitative features of the tagging data with respect to movement are given below in the Results section. Shedding (L = M -0.14) (0.14 was taken from the literature as the natural mortality rate; ICCAT, 1993) was estimated from these catch data using the method of Chapman, Fink, and Bennet (1965, equations 5, 6, and 7) to estimate L. This method was used rather than that of Bayliff and Mobrand (1972), because shedding rates do not appear to be constant and appear to be insensitive to type I (initial) shedding. Once the estimate of M (L + 0.14) was available, VPA analysis was done on the catches over time to obtain estimates of F.

Tables 3-4, 3-5, and 3-6 show the number of fish tagged in the western Atlantic Ocean (from MRFISH) and the number of recoveries by year after tagging. Year 0 means fish caught in the same calendar year as tagged; year 1 means fish caught in the calendar year following tagging, etc. Table 3-7 (eastern Atlantic Ocean, ICCAT data) and Table 3-8 (western Atlantic Ocean, NMFS data) show the months in which the fish were tagged and the months in which the fisheries captured the tagged fish. Because most tagged fish were small (less than four years old), inferences about fishing times based on these tag recoveries should apply to small fish.

From 1971 to 1978, there was intensive tagging in the west with both single-and double tagged fish (a small number of double tagged fish actually had more than two tags). Table 3-9 shows the number of recoveries from single tagged fish (SS), the number of recoveries of fish with two tags from double tagged fish (DD), and the number of recoveries of fish with one tag from double tagged fish (DS), by year of recovery. Tagging time varied, so data from the west were sorted into periods of one-quarter year; that is, if a tag was recovered with the same quarter as the release, it was placed in quarter 1. If the recovery was made after more than 91 days but before 182 days, it was placed in quarter 2, and so on. Table 3-10 shows the recovery data for recoveries SS, DD, and DS by year of release and quarter of recovery.

The western data from Table 3-10 were used to estimate shedding rates (Table 3-11). Comparison of SS to DD (Db Db vs. Sing in Table 3-11) shows the most variability, and comparison of DD to DS (Db Db vs. Db Sing) shows the most stability. After examining Table 3-11, it was decided to try L = 0.26 for all ages and also to try L(1) = 0.50, L(2) = 0.40, L(3) = 0.30, and L(4+) = 0.20, corresponding to quarter 1, quarter 2, quarter 3, and all quarters 4 or more. These values are not much different from the value of 0.205 estimated by Baglin et al. (1980).

With the shedding rates mentioned above, and the catch vectors for SS and DD in Table 3-11, VPA analysis was performed showing that the two sets of