FISHERY BULLETIN: VOL. 80. NO. 4 



ability in April and May salmon catch. The con- 

 clusion drawn from the comparison of early 

 versus late season catch, however, rests on ob- 

 serving the cyclic nature in the late season catch 

 regardless of its cause. 



While we have concluded that an immediate 

 behavioral response is not a likely cause, other 

 related possibilities remain. The observed covari- 

 ation could be caused by an inherently cyclic 

 crab fishery and a negative response of effort in 

 salmon throughout the salmon season (rather 

 than solely in the months of overlap). Further 

 elucidation of the economic question awaits re- 

 sults of an ongoing study of microeconomic be- 

 havior of fishermen. 



Consideration of the life histories of the species 

 and the timing of events implied by the lags in 

 correlation admits the possibility of direct inter- 

 action and dependence of both cycles on environ- 

 mental factors. Oceanographic conditions have 

 been suggested as causes of fluctuations in other 

 fisheries. Wild (1980) recently proposed that a 

 change in sea surface temperature in the late 

 1950's reflects a change in the marine environ- 

 ment that is responsible for the decline in the 

 central California Dungeness crab fishery. He 

 also suggested that changes in sea surface tem- 

 perature were related to fluctuations in the 

 northern California crab catch record. However 

 the actual values of correlation between these 

 processes are not significant. Southward et al. 

 (1975) presented data on cyclic fluctuations in 

 sea temperature and covarying changes in fish 

 population parameters over the past 50 yr in the 

 English Channel. 



Though the observed changes in lags and sen- 

 sitivity to first-differencing may not be related to 

 the causal mechanism, the nature of the covaria- 

 tion between salmon and crab catch does appear 

 to have changed following the decline in central 

 California crab landings. This change is not ex- 

 plained by fishermen switching between species, 

 but could stem from Wild's (1980) proposed 

 change in oceanographic conditions. The de- 

 crease in the period of the cycles in crab abun- 

 dance following the decline is of some interest 

 with regard to the issue of the cause of the decline 

 itself. One of the possible causes of a decrease in 

 period of the cycles is an increase in individual 

 growth rate. This increase in growth rate is a 

 necessary component of one of the potential 

 causes of the decline (Botsford 1981) but is diffi- 

 cult to demonstrate because of the paucity of 

 samples before the decline. 



Possible effects of internal population dynam- 

 ics on the observed behavior are worthy of exam- 

 ination. An interage, density-dependent mecha- 

 nism has been cited as a potential cause of the 

 cycles in crab abundance (Botsford and Wick- 

 ham 1978, 1979). A similar mechanism could be 

 operating on salmon abundance if the several 

 stocks in the fishery were in synchrony. Peter- 

 man (1978) found positive correlations in smolt- 

 to-adult survivorship between several groups of 

 Pacific salmon populations. Populations that are 

 not density-dependent but reproduce only in 

 their final year have also long been known to 

 fluctuate in a cyclic fashion (Bernardelli 1941; 

 Leslie 1945). However, the period of the cycles is 

 equal to the age of reproduction rather than 

 twice the mean age of reproduction as it is in the 

 stock-dependent recruitment case (Ricker 1954; 

 Botsford and Wickham 1978). 



The methods used here could prove useful in 

 other fisheries problems. While time-series tech- 

 niques have been applied to fishery problems, 

 the primary goal has been a final model of the 

 fishery rather than a search for causal relation- 

 ships. The latter approach, the one taken here, 

 has the advantage of leading to models that are 

 based on known causal mechanisms rather than 

 correlations of unknown causal mechanisms. 

 Since the nature of these mechanisms could 

 change significantly (possibly because of a 

 change in fishing policy itself), a policy that is 

 cognizant of them will fare better than one that 

 relies on a statistical description from the 

 past. 



Another analytical time-series technique that 

 we have not used here is the computation of cau- 

 sality as defined by Granger (1969). His special 

 definition of causality is based on whether addi- 

 tion of data from past time on one variable de- 

 creases the error with which another variable 

 can be predicted. The investigations performed 

 here are in the same spirit but do not result in a 

 single quantitative measure of causality. 



While we have demonstrated here a potentially 

 important statistical relationship, we have not 

 uncovered the underlying cause. The ultimate 

 cause, however, is worth pursuing. Its discovery 

 and quantitative description could put salmon 

 and crab management on a firmer basis by sup- 

 plying greater predictive ability. Management 

 could then respond to abundance on a firmer, 

 predictive basis rather than a trial-and-error 

 basis. 



800 



