Francis et al.: Quantifying annual variation in catchability 



303 



bly low and the variation amongst stocks of the same spe- 

 cies was implausibly high. Where possible, this variability 

 should also be examined for CPUE catchabilities (as long 

 as they are comparable — note that we should not compare 

 trawl and long-line catchabilities). We did not make this 

 comparison in our study because it involves adjusting for 

 different reference units of of effort in different CPUE 

 series, which requires specialist knowledge about the indi- 

 vidual fisheries and data sets. 



Concluding comment 



Our analyses have not been able to take account of the 

 practice, in some stock assessments, of using CVs as a 

 measure of the "quality" of a biomass index, rather than 

 its precision. This practice happens when a high CV is 

 assigned to a series (often of CPUE) that is believed not 

 to index biomass well. The intention is to lessen the con- 

 tribution of the series to the assessment. A problem with 

 this practice is that the judgment of quality is subjective, 

 as is the decision as to how high a CV to assign to repre- 

 sent poor quality. It would be very rare that we had suf- 

 ficient information to determine whether the judgment of 

 poor quality was justified, and whether the assigned CVs 

 were appropriate. It may be that some of the assessments 

 analyzed above produced a biomass trajectory that was a 

 very good fit to a CPUE series (suggesting that a low CV 

 should have been used) but that the trajectory was wrong 

 because, in this case, CPUE was not proportional to abun- 

 dance. We cannot distinguish such an outcome from one 

 in which a precise CPUE series indexed abundance well. 

 The practice of assigning CVs subjectively is not desirable. 

 Ideally, we should change the model assumption of propor- 

 tionality between biomass and index rather than inflate 

 CVs. However, we acknowledge that stock assessment is a 

 very pragmatic discipline in which many compromises are 

 necessary, and we hope that the above results will provide 

 practitioners with empirical evidence to support some of 

 their subjective decisions. 



Acknowledgments 



We are indebted to the many people who provided trawl 

 sui-vey and assessment data for this work. Dave Gilbert, 

 Larry Paul, Paul Breen, and Stuart Hanchet offered useful 

 comments on an earlier version of this paper, and Richard 

 O'Driscoll detected what would have been an embarrass- 

 ing data error. We are grateful to Steve Cadrin and an 

 anonymous referee for useful suggestions. This project was 

 partially funded by the New Zealand Ministry of Fisheries 

 under project SAM 1999/01. 



Literature cited 



Arreguin-Sanchez, F. 



1996. Catchability: a key parameter for fish stock assess- 

 ment. Rev. Fish Biol. Fisheries 6(21:221-242. 



Francis, R. I. C. C. 



1992. Use of risk analysis to assess fishery management 

 strategies: a case study using orange roughy (Hoplostcthus 

 atlanticus) on the Chatham Rise, New Zealand. Can. J. 

 Fish. Aquat. Sci. 49:922-930. 



Francis, R. I. C. C, R. J. Hurst, and J. A. Renwick. 



2001. An evaluation of catchability assumptions in New 

 Zealand stock assessments. N.Z. Fish. Assess. Rep. 2001/ 

 1,37 p. 



Millar, R. B.. and R. D. Methot. 



2002. Age-structured meta-analysis of U.S. West Coast 

 rockfish (Scorpaenidae) populations and hierarchical mod- 

 eling of trawl survey catchabilities. Can. J. Fish. Aquat. 

 Sci. 59:383-392. 



Myers, R. A., and N. G. Cadigan. 



1995. Statistical analysis of catch-at-age data with corre- 

 lated errors. Can. J. Fish. Aquat. Sci. 52:1265-1273. 

 Paloheimo, J. E., and L. M. Dickie. 



1964. Abundance and fishing success. Rapp. P-V. Reun. 

 Cons. int. Explor. Mer 155:152-163. 

 Pope, J. G. 



1980. Some consequences for fisheries management of 

 aspects of the behaviour of pelagic fish. Rapp. P-V. Reun. 

 Cons. Int. Explor. Mer 177:466-476. 

 Punt, A.E., T. I. Walker, B. L. Taylor, and F. Pribac. 



2000. Standardization of catch and effort data in a spatially- 

 structured shark fishery. Fish. Res. 45:129-145. 

 Quinn, T. J., II, and J. S. Collie. 



1990. Alternative population models for eastern Bering Sea 

 pollock. Int. North Pac. Fish. Comm. Bull. 50:243-257. 

 Schaaf W. E. 



1975. Fish population models: potential and actual links to 

 ecological models. In Ecological modeling in a resource 

 management framework (C. S. Russell, ed.), p. 211-239. 

 Resources for the Future, Inc., Washington, DC. 

 Winters, G. H., and J. P. Wheeler. 



1985. Interaction between stock area, stock abundance, and 

 catchability coefficient. Can. J. Fish. Aquat. Sci. 42(5): 

 989-998. 



Appendix 



Alternative mean rank calculations 



In this Appendix we describe some alternative (but unfruit- 

 ful) mean rank calculations referred to in the "Results" sec- 

 tion under the subheading "Can we detect years of extreme 

 trawl survey catchability?" 



We tried three variations on the above procedure for 

 identifying extreme years. In each case we were evaluat- 

 ing an alternative hypothesis about the nature of between- 

 species correlations. Each hypothesis leads to a different 

 method of calculating mean ranks (or alternative statis- 

 tics), and we applied the new method to both the survey 

 data, and to simulated data (to calculate threshold values 

 for the new statistics). If the hypothesis were true we would 

 expect to see more extreme years. In fact, we saw fewer 

 extreme years for all of these alternatives. 



First, we repeated the above calculations after omitting 

 species for which the mean CV ( see Fig. 2 ) exceeded 0.4. The 

 idea here is that, for species with high CVs, there is little 

 information in the year-to-year changes in their biomass 



