At present the IPHC reduces quotas in the directed halibut fishery by an amount sufficient to offset the impact of bycatch in other fisheries on spawning biomass. The distribution and timing of the necessary quota reductions are controversial, and the staff in 1995 proposed a new method of distributing the quota reductions among areas. This paper points out some difficulties that persist in the 1995 proposal and presents a different solution, which is to make allowance for average long-term bycatch mortality rates when choosing an exploitation rate for each regulatory area. This would eliminate much of the confusion and controversy surrounding the present compensation procedure because it would be much simpler and it would not require explicit quota reductions to compensate for out-of-area bycatch.

The IPHC management strategy for halibut is to apply a constant exploitation rate, presently 0.30 for fully recruited fish. This exploitation rate is based on historical estimates of spawning biomass and subsequent recruitment of 8-year-old halibut to the setline fishery. Since about 1960, other groundfish fisheries have taken substantial numbers of halibut as bycatch, most of them fish younger than age 8, thereby reducing recruitment to the setline fishery. To account for this bycatch, the IPHC has adjusted the recruitment estimates upward, set the exploitation rate at a level that would be appropriate if there were no bycatch, and then used one or another compensation method to lower quotas in the setline fishery by an amount that varies with the amount of bycatch in other fisheries.

The present method of compensation, called adult reproductive compensation (ARC), is to estimate how many eggs the fish in the bycatch would have produced in the absence of bycatch fisheries, and then to reduce the quota in the setline fishery just enough to increase future egg production by that amount. Quota reductions are distributed among regulatory areas in proportion to exploitable biomass. The effect is to maintain spawning biomass and egg production at the levels that would obtain in the absence of bycatch, while recruitment and yield to the setline fishery are lower. The actual exploitation rate is also lower than the target level owing to the compensation procedure.

As the staff has noted, the present method serves its purpose but it has three weaknesses:

(i) Compensation is performed both for pre-recruits and for recruited fish (i.e., fish of legal size). This is appropriate for maintaining egg production, but it means that recruited fish taken as bycatch are not included among the removals that go into the stock assessment, as they should be. While legal fish do not account for a large proportion of the bycatch in number, they account for about half the weight.

(ii) The amount of compensation in each IPHC regulatory area is determined by pooling the coastwide bycatch, calculating the required coastwide compensation, and distributing it among regulatory areas in proportion to exploitable biomass. This is not consistent with the unidirectional migration of juvenile halibut from nursery areas in western Alaska to adult summer feeding areas around the entire Gulf of Alaska.

(iii) The compensation for each year's bycatch is applied right away, even though the impact on egg production will not occur until several years later. (Juveniles in the bycatch have a modal age of five, while 50% maturity of females does not occur until age 11, and egg production continues for years after that.)

In 1995 the staff proposed a new method of reproductive compensation with two improvements. First, fish of legal size in the bycatch would be treated like other removals of recruited fish. That is, they would be included in the stock assessment along with sport catch and wastage of legal fish, and the setline quota in each regulatory area would be reduced pound for pound of legal sized bycatch taken in that area, but there would not be any other kind of compensation. Second, fish below legal size in the bycatch would continue to be compensated as at present, but the compensation would be distributed among regulatory areas according to an explicit model of juvenile migration.

While the treatment of recruited fish under this new procedure is straightforward, the treatment of juveniles has a number of drawbacks, some old and some new. They are:

(i) There are no suitable data for estimating the parameters of the juvenile migration model, and the distribution of bycatch impacts is very sensitive to the parameter values. It is possible to set a plausible range on migration schedules, but to calculate downstream impacts the staff has to make an essentially arbitrary choice somewhere in the range. This part of the compensation procedure would probably be even more contentious than the present method, even though in principle it is more realistic. For example, at recent levels the juvenile bycatch in Area 4 (Bering Sea/Aleutians) requires about 6 million pounds of compensation. Depending on the migration schedule chosen, the amount applied to downstream areas could be anywhere between zero and 4 million pounds, and the staff has no objective, defensible way to choose a particular migration schedule.

(ii) The timing of the compensation is still years in advance of the impact, and there is no feasible way to correct it. Simply lagging the compensation might appear to offer a solution, but as explained by Sullivan et al. (1994), "…such a procedure would entail annual recalculation of the compensation factors based on fishing mortality and exploitable biomass estimates that are annually revised. Lagged compensation would, perhaps more problematically, also require annual projections of harvest and population levels for both the existing stock and the bycaught portion had it remained in the stock under a similar harvest regime." The calculations would be overwhelming and never-ending.

(iii) It is a lot of work. The population model used to calculate bycatch impacts is considerably more complicated than the one used in the stock assessment because it has to keep track of the location and sex of the fish (in addition to the age and length). Building and running this model is a major commitment of quantitative expertise.

(iv) Apart from the difficulties of implementing any specific method, the basic strategy of reproductive compensation is open to question when viewed as part of the harvest policy, which it is. The harvest rate is chosen assuming that future bycatch will be zero, and the stock is compensated for lost egg production due to past bycatch. But future bycatch will not be zero, and its effect on yield is not considered when formulating the harvest policy.

The Commission's management measures have to make some allowance for juvenile bycatch. The present method of reproductive compensation does that in one way for the stock as a whole, although the compensation is generally not distributed correctly and the implications for yield are not considered. It has proved to be extremely controversial because of its allocation effects, and difficult for the staff to explain and defend. The new method proposed in 1995 is open to the same criticisms.

The aim of this paper is to outline an alternative method of compensating for juvenile bycatch that will achieve the Commission's conservation objectives in a more straightforward, reliable, and understandable way. Our proposal is to estimate the average long-term level of pre-recruit mortality due to bycatch, and to incorporate that into the formulation of the harvest policy in each regulatory area.

The strategy underlying reproductive compensation consists of choosing a harvest rate that would be adequate in the absence of bycatch, and then attempting to compensate the stock after the fact for the loss in egg production due to any bycatch that is actually taken. This may be the best option if bycatch cannot be predicted at the time of choosing the harvest rate, and that was felt to be the case when the Commission adopted a compensation strategy in the early 1980s. But an inherent drawback to this method is that the effect of bycatch on yield to the fishery is not considered when formulating a harvest policy-only the effect on egg production is considered.

If bycatch rates can be predicted even approximately, a better strategy is to incorporate their effects on stock dynamics and yield explicitly when evaluating alternative harvest policies, and thus compensate a priori by reducing the target harvest rate. Formulating the harvest policy in this fashion makes it possible to consider the effects of all sources of mortality on both yield and spawning biomass simultaneously, and to choose the one that achieves the best balance of Commission objectives. It is also a much more straightforward and understandable process than compensation, because bycatch mortality is treated as just one more element in the population model used to evaluate alternative harvest policies.

It is of course not possible to predict bycatch now any more than it was in the 1980s, but all that is required is a very rough estimate of the long-term average. The choice of a harvest rate is in fact not very sensitive to the precise value of pre-recruit mortality, so year-to-year fluctuations or even a poor guess will do no harm. Moreover, the effect of pre-recruit mortality is to reduce the slope parameter of the spawner-recruit relationship, and any uncertainty in the estimate of pre-recruit mortality is sure to be minor compared with the overall uncertainty concerning the form and parameter values of the relationship.

For the last twenty years or so bycatch has varied between 10 and 20 million pounds, and for the last several years has been quite steady at about 15 million pounds. Given the determination of both governments to control bycatch, a significant increase above this level seems unlikely. In the other direction, it is hoped that efforts in both countries will succeed in reducing bycatch to a level of about 10 million pounds, but probably not less and perhaps not very soon. It seems fairly safe, therefore, to predict that bycatch will remain at or near the present level for the next several years at least, but will not increase.

What is needed for the harvest rate evaluations is an estimate not of the amount but of the rate of bycatch mortality. We expect the rate to remain approximately the same as at present unless there are major changes in other groundfish fisheries or in bycatch management. Halibut bycatch is incidental to the other fisheries, so we would expect the fishing mortality rate on halibut in those fisheries to remain approximately the same as long as the fishing mortality rates on the target species remain the same, which is generally the policy in both countries.

During the years 1990-94, average annual removals of halibut were, in round numbers (millions of net pounds):

In future the bycatch of legal size fish will be included in the stock assessment, so the issue here is the eight million pounds of juvenile bycatch, about three-fourths of it taken in Area 4.

While the bycatch of sublegals is large, it is not all that large a fraction of the numbers of sublegals in the sea. Figure 1 shows the proportion of halibut taken as bycatch each year by length (based on a coastwide recruitment level of 15 million 8-year-olds). The strong peak around 50 cm reflects the selectivity of the trawl fishery. Up to age 6, the mean length of halibut is about ten times their age, so bycatch mortality of juveniles is concentrated on ages 4, 5, and 6, when it amounts to 2-3% per year. Among older fish bycatch mortality is about 1% per year. (This compares with a natural mortality rate of 20%.) The cumulative bycatch mortality of juvenile halibut is about 10% for the stock as a whole. That is, bycatch reduces coastwide year-class strength at age 8 by about 10%, or 1.5 million fish.

The impact of coastwide bycatch on recruitment to each regulatory area depends on the migration schedule of juvenile halibut. As stated above, we do not know the correct migration schedule, but using the migration model we can calculate the pre-recruit mortality on the stock in each regulatory area for the whole plausible range of migration schedules, and when we do that it turns out that the estimates of total pre-recruit mortality are not very sensitive to the choice of a migration schedule. (See Appendix for details of the calculation.) The table below shows the proportional reduction in recruitment to each regulatory area under migration schedules covering the full range of possibilities considered by the staff in its 1995 proposal: "Bering Slow" is the schedule that maximizes downstream impacts of juvenile bycatch; "Local Fast" is the one that minimizes such impacts; and "Intermediate" is intermediate. Area 2A is not shown; impacts there are similar to 2B.

Table 1. Proportional reduction in recruitment due to coastwide juvenile bycatch.

It is also informative to look at a breakdown of the coastwide impact into components resulting from bycatch in each regulatory area, as in the tables below.

Table 2. Proportional reduction in recruitment due to juvenile bycatch in each regulatory area. ("*" means less than 0.005.)

Reductions due to juvenile bycatch taken in Area 4.

Reductions due to juvenile bycatch taken in Area 3.

Reductions due to juvenile bycatch taken in Area 2C.

Reductions due to juvenile bycatch taken in Area 2B.

The effect of juvenile bycatch taken in areas other than Area 4 is small (less than 10%) and largely confined to the area where the bycatch is taken. The Area 4 bycatch does affect downstream areas, but at most it reduces recruitment by 4% in those areas. For the whole plausible range of migration schedules, therefore, the cumulative pre-recruit mortality is 10% or less in Areas 2 and 3 and about 15% in Area 4.

In practice, using this method would mean building a reasonable level of pre-recruit bycatch mortality, say 10%, into the calculations that are done to evaluate alternative exploitation rates. Figure 2a shows how pre-recruit mortality changes the spawner-recruit relationship. Figure 2b shows the corresponding yield curves. The harvest rate that maximizes yield decreases as pre-recruit mortality increases, but it does not decrease very much over any limited range of pre-recruit mortality rates. Even more important is that on any one of the curves, yield is not very sensitive to the harvest rate in the vicinity of the maximum because they are all nearly flat there. For both reasons, it is possible to choose an exploitation rate that will obtain something very close to the maximum yield over a considerable range of pre-recruit mortality rates. For example, a harvest rate anywhere in the range 0.25-0.39 would provide more than 95% of maximum yield for any pre-recruit mortality rate in the range 0-20%. These numbers are reported only to show the low sensitivity of the yield-maximizing rate to the level of pre-recruit mortality, and therefore to any variations in the rate or errors in the estimate of the average value. A different estimate of the spawner-recruit relationship would produce different numbers, and in recent years the Commission has also taken biomass projections into account and has lowered the target exploitation rate to maintain desired biomass levels.

This solution would be clean and practical for both the Commission and the staff. It would certainly serve the stock as well as any more elaborate scheme (including the 1995 staff proposal), and by avoiding the migration and timing issues entirely it would remove much of the controversy that surrounds the present method.

On the other hand, one of the attractions of the pay-as-you-go system was that it made bycatch more visible in the management process. Under this proposal, only the bycatch of legal sized fish in each regulatory area would appear as a line item in the table of quota recommendations. The bycatch of sublegals would not directly enter the quota calculation because compensation for it would be built into the harvest policy. There would be no explicit quota reductions to compensate for juvenile bycatch in any area. This may be a loss in some respects, but the compensation amount that appeared as a line item was widely misunderstood, and it would be entirely feasible and more informative to provide separate estimates of yield loss due to juvenile bycatch in some prominent place.

Sullivan, P.J., Trumble, R.J., and S.A. Adlerstein. 1994. Pacific halibut bycatch in the groundfish fisheries: effects on and management implications for the halibut fishery. Int. Pac. Halibut Comm. Sci. Rpt. 78: 28pp.

Quinn, T.J. II, R.B. Deriso, and P.R. Neal. 1990. Migratory catch-age analysis. Can. J. Fish. Aquat. Sci. 47:2315-2327.

If we knew the absolute abundance of each stock at age 1, we could calculate bycatch mortality, natural mortality, and migration year by year to determine the cumulative impact of bycatch on each stock. But we have no stock-specific estimates of abundance at age 1. What we do have is stock-specific estimates of abundance at age 8 from the stock assessment. (In recent years, recruitment at age 8 has been about 15 million fish. The stock composition of 8-year-olds is recent years is estimated to be about 30% in Area 4, 50% in Area 3, and 10% each in Areas 2C and 2B-2A.) We can also calculate the coastwide reduction in recruitment at age 8 due to bycatch, and therefore we can estimate what the coastwide recruitment at age 8 would have been in the absence of bycatch. This can be back-calculated to age 1 to give an estimate of total age 1 abundance. Stock composition at age 1 is then estimated iteratively by fitting to the estimates of stock-specific recruitment at age 8 from the stock assessment. The procedure is similar to (but simpler than) the measurement error model of migratory catch-age analysis described by Quinn et al.(1990).