FISHERY BULLETIN: VOL. 78, NO. 3 



ies, which allowed eventual crude prediction of 

 numerical population behavior, might be roughly 

 analogous to the following field data collection: 

 1) Collecting data on at least adult stock, young-of- 

 year, and yearling densities from 10 fish popula- 

 tions of the same species in similar environments 

 for 18 yr each. 2) Restructuring refuge area 

 habitats for six of the similar populations, perhaps 

 by removing or increasing weed cover, dramat- 

 ically reducing the size of all populations, and 

 collecting appropriate data for an additional 11 yr. 

 Few fishery investigators have the opportunity 

 to carry out such an "experiment" in a field 

 context. Instead, a single population may be 

 studied, under fortunate circumstances, for per- 

 haps one or two decades. Analogous replication is 

 impossible. Since the investigator is (usually) not 

 allowed to actively manipulate population age or 

 size structure, but must instead maintain a pas- 

 sive observer role, data collected in a decade might 

 cover only a small range of juvenile and/or adult 

 stock sizes. Recruited year classes, exposed to 

 perhaps violent fluctuations in environmental 

 factors influencing early life survival, might rare- 

 ly give any indications of a dependency of recruit- 

 ment on adult or juvenile densities in previous 

 years. 



Faced with such constraints on data collection, 

 there seem possible several constructive alterna- 

 tive responses. The general passive approach may 

 be neither appropriate nor effective, and active 

 (experimental) manipulation of populations, forc- 

 ing collection of data not otherwise obtainable, 

 may be required. This approach has been advo- 

 cated by Walters and Hilborn (1976) although it 

 clearly calls for major rethinking of the fishery 

 biologist's role. Second, it is possible that year- 

 class strength and adult and other stock compo- 

 nents during past years may be estimated through 

 data extraction techniques based on simple gross 

 population measures, e.g., from total biomass 

 harvested from commercial species (Walter and 

 Hoagman 1975). Thus, rather than bemoaning the 

 slow pace at which future observations may be 

 gathered, one may consider past fishery data as an 

 untapped reservoir of information suitable for 

 analysis of the dynamics of recruitment. Statis- 

 tical evaluation of relations among such extracted 

 estimates does, however, raise serious analytic 

 and philosophic issues. Finally, comparative study 

 of year-class fluctuations among related species 

 and fisheries holds far more promise for revealing 

 biological mechanisms underlying recruitment 



than is indicated by published literature (Regier 

 1978). 



Since there may be constructive responses to 

 data collection problems, the probably inherent 

 high variability of the stock-recruitment process 

 causes the author greater concern. Although the 

 impact of specific environmental factors may oc- 

 casionally be separated from possible internal 

 biological controls (Nelson et al. 1977) and allow 

 reduction of unexplained variation in year-class 

 strength, it seems unlikely that a single environ- 

 mental variable regularly exerts significant im- 

 pact on year-class strength. Thus, while apparent 

 variation in year-class strength may be reduced 

 under fortunate circumstances, either by account- 

 ing for environmental impacts or by considering 

 all relevant stock components, it seems unlikely 

 that collected data will ever fall neatly along some 

 theoretical curve or surface. In general, expecta- 

 tions for statistical measures of goodness of fit for 

 stock-recruitment relations are probably grossly 

 unrealistic and poor fits should be expected. How 

 one ought to evaluate empirical stock-recruitment 

 fits, when the appropriate standard for compari- 

 son is probably not "100*7^ of variation" or a 

 correlation of 1, is not clear, although attention 

 has already focused on optimal use of unreliable 

 stock-recruitment parameter estimates (Walters 

 1975). The danger of presuming independence of 

 recruitment and population stock components, 

 however, seems far more severe than are errors of 

 estimation and generally unsatisfying statistical 

 analyses. 



It is hoped that the results of these experiments 

 and the demonstration of a complex multistage 

 recruitment process will stimulate renewed inter- 

 est in study of the possible biological determinants 

 of recruitment. That simple stock-recruitment 

 theory may often be biologically inappropriate 

 seems clear But whether more complex and more 

 biologically realistic models of recruitment pro- 

 cesses, with their further demands for data collec- 

 tion, will prove of practical use seems far from 

 clear. In the author's view, at present, a wide gulf 

 separates stock-recruitment theory from practice. 

 More careful consideration of the practical use of 

 this body of theory and more realistic expectations 

 from its use are required if the theory is to achieve 

 its proper role in fishery management. 



ACKNOWLEDGMENTS 



My deep appreciation is given to the many 



576 



