Over the years there have been numerous changes in sampling objectives, geographic coverage, sampling strategies, and the kinds of data collected. As a result, it is not a straightforward task to estimate age and size compositions of the commercial landings in each of the IPHC regulatory areas. A complete history of the sampling programs and a description of the raw data now stored in the IPHC database can be found in Clark et al. (2000).

In summary, during the 1930's the staff measured the fork lengths of a large number of fish and took a size-stratified subsample of otoliths for age determination from each of a sample of landings, mainly from the Goose Island grounds in British Columbia and from Portlock and Albatross Banks off Kodiak Island in Alaska.

Sampling was greatly curtailed during the 1940's by the war and its aftermath, but during the 1950's was resumed and extended as the fishery expanded northward and westward in Alaska. Since the late 1950's the samples have covered commercial catches from all areas.

Beginning in 1963 and continuing through 1990, the samplers did not actually measure the fork length of any fish in the sample. Instead they collected otoliths from all of the fish in the sample, and the fork lengths were later estimated from one or another predictor relating fork length to otolith size.

During the 1980's the growth rates of halibut decreased dramatically, especially in Alaska, and it became apparent that the predictions of fork length based on otolith size were no longer accurate. The staff resumed measuring actual fork lengths in 1991, and insofar as possible corrected the estimates of fork length at age based otolith sizes for the years 1963-1990. For a number of reasons there is no really satisfactory way to make the correction now, so the size estimates from this period will always be questionable.

Estimates of size at age in the commercial landings are essential for stock assessment but are not useful for studying changes in halibut growth. That is because the size composition of commercial landings is strongly affected by the size selectivity of the commerical fishery, which changes over time, in part because of changes in the commercial size limit, which was first set at 26" (66 cm) and raised to 32" (81 cm) in 1973.

This site contains several series of estimates calculated at different times for different purposes:

1. Terry Quinn's estimates of catch in number and mean weight at age, 1935-1982.

IPHC regulatory area definitions have changed since the early 1980's, so the absolute catch in number is no longer correct. Sample data were lacking for some area/years even in 1982. In those cases Quinn substituted data from other area/years; this is apparent in plots of the data. Details are given in the 1983 paper by Quinn et al. cited in Clark et al. (2000).

The fields in the file are: (1) IPHC regulatory area; (2) year; (3) age; (4) estimated catch in number; (5) mean weight in net pounds (headed and gutted). Mean weight is given as "-1.00" for an age group that did not appear in the landings. The last age group (24) is not a plus group; older fish are simply missing from the estimates.

2. Updated, corrected, and consolidated estimates of age and size composition, 1935-1990.

The fields in the file are:

1-3. IPHC regulatory area; year; age. The last age group (25) is a plus group, meaning that it counts fish aged 25 and older. For years where Quinn's estimates of age composition are used, the estimated proportion at age 25+ is given as "-1", denoting a missing value. This is sure to be a very small number, so its omission has no practical effect on the estimates of proportion at age for ages 1-24, but treating it as zero rather than as missing would affect the fit of an age-structured model.

4. Number of otoliths actually aged.

5-6. Estimated proportion at age in the landings; coefficient of variation (CV) of the estimate.

7. Smoothed mean weight at age (smoothed over years within age). This is the only weight estimate for years where sampling data are lacking or corrupt, or where the predictors appear to be very wrong. It is also the best estimate of weight for calculating catch in number for the years 1964-1990, because the estimates for each year are subject to some variation resulting solely from the differences among fork length predictors used at various times.

8-9. Mean weight at age (in net pounds) estimated from the given year's data; CV of the estimate.

The remaining fields contain estimates of size at age, but only for years after 1963, and only for those area/year cells with usable data. In order to allow for pooling, estimates are shown even for area/year/age cells with very small samples, but of course some of these are actually wild estimates. (Only estimates based on some minimum sample size--10 or 15--are actually used in the stock assessment.)

10-11. Estimated mean fork length in cm; CV of estimate.

12-13. Estimated standard deviation of fork length; CV of estimate.

14-15. Estimated mean log fork length; CV of estimate.

16-17. Estimated standard deviation of log fork length; CV of estimate.

Notes on the consolidated estimates.

1. In all fields, a "-1" indicates a missing value. In the case of proportion at age, this will appear for all ages in a year/area with no catch, and it appears for age 25 in all years 1935-1963. Smoothed mean weight at age is always given for ages 5 through 25+, even for area/year/age cells with no catch; it is mostly missing for ages 1-4. In the case of the remaining fields (dealing with size at age), a "-1" appears for all ages for all years 1935-1963. Thereafter it appears for all ages in a given area/year if there were no otolith size data collected, or if the data in the database were judged to be unreliable. It appears for a particular age group in a given area/year if the age group was completely absent from the age samples.

2. There are very few data on otolith size in 2A before 1981, so for those years the 2B estimates of mean weight at age by year were used to calculate the smoothed mean weights at age for 2A.

3. Because of a change in regulatory area boundaries, Quinn's estimates have no catch in Area 4 before 1952, except for a very small amount in 1945. Under the present area definitions, the westernmost part of the old 3B is now the easternmost part of Area 4, and there was some catch in Area 4 in 1946-1951 as well as 1945. In the updated file, the 3B estimates for 1945-1951 were copied to provide estimates for Area 4.

4. For completeness, the updated file contains predicted estimates of fork length for all area/year/age cells with sufficient usable measurements of otolith size. Most of these agree roughly with measured fork lengths in IPHC setline survey data, but the predictions for Area 3 (and presumably 4) in 1963-1967 are clearly too low. For that reason those years were left out of the series used to calculate the smoothed mean weights at age, as were the years 1942-1947 north of Area 2B because it is known that no data were collected on landings from those areas during that period. The stored estimates of length at age for these years and areas should not be used. The year-by-year estimates of weight at age are also unreliable, but the smoothed estimates of mean weight at age should be serviceable.

3. Updated and consolidated estimates of commercial catch at age in number and weight, 1935-1990.