APPENDIX F
Microconstituent Analysis
The committee received three documents to review (Calaprice, 1983, 1984, 1985); none has been published in the primary literature. Other material that we received did not contain any critical reviews of the Calaprice study. The reviews that we received tended to restate the generalizations in Calaprice (1985).
The major difficulty with the Calaprice study is that different statistical techniques led to different mixing values, and that repeated analysis on the same material led to different mixing values.
For example, in a pilot study, Calaprice used 36 age 2+ fish from the Bay of Biscay and 39 age 2+ fish from the coast of Virginia and statistically analyzed various sections of the x-ray spectra separately with a stepwise discriminant function analysis. Out of 512 bins (energy levels), the stepwise discriminant function analysis found eight variables that gave 100% discrimination between the eastern and western Atlantic samples. Calaprice concluded that "it was possible to derive equations that could be used to classify individuals as to area of origin." These early results, however, imply that there was no mixing among age 2+ fish from the two areas, which was contradicted by later analyses of samples from the same areas that apparently showed that some mixing was occurring.
One case where Calaprice deals with uncertainties in his estimates is in the analysis of his pilot study mixing of 39 age 2+ bluefin tuna from the western Atlantic Ocean and 36 age 2+ bluefin tuna from the eastern Atlantic Ocean (Calaprice, 1983, Table 3). Mixing from west to east is given as 14.3%, but the 95% confidence levels are 3 to 37%. Mixing from east to west is given as 23.8%, but the 95% confidence levels are 11 to 53%.
TABLE F-1 Estimates of mixing between the two stocks based on discriminant function analysis using jackknife probabilities of group membership of adult bluefin tuna sampled from a variety of locations and at different times.
|
Estimated % mixing |
|
Source of tuna |
Analyzed 1983 |
Analyzed 1984 |
Western Atlantic Ocean |
||
Massachusetts (1980) |
12.7 |
10.8 |
Massachusetts (1982) |
— |
11.9 |
Gulf of Mexico (1983) |
— |
4.2 |
Eastern Atlantic Ocean |
|
|
Gibraltar (1982) |
11.3 |
90.0 |
Gibraltar (1983) |
— |
4.3 |
Tyrrhenian Sea (1982) |
12.9 |
60.0 |
Tyrrhenian Sea (1983) |
— |
5.2 |
Ionian Sea (1982) |
14.3 |
100.0 |
A summary of some other problems is presented in Table F-1 (adapted from Calaprice, 1983, Table 2). Estimated mixing varies from 4 to 100%, depending on when it was analyzed and on the source of the material.
Although there are new pattern recognition techniques that could be applied to the data collected by Calaprice, the committee does not recommend such analysis. A better approach would be to apply new analytical techniques to new samples.
REFERENCES
Calaprice, J.R. 1983. X-ray fluorescence of stock variation in bluefin tuna, status report. pp. 1-42.
Calaprice, J.R. 1984. X-ray fluorescence study of stock variation in bluefin tuna, Third quarterly report. pp. 1-25.
Calaprice, J.R. 1985. Chemical variability and stock variation in northern Atlantic bluefin tuna. ICCAT SCRS/85/36. pp. 222-252.