488 



Fishery Bulletin 97(3), 1999 



9 



tr 



1.5-, 



1 



0.5 







-0.5 



-1 



-1.5 



o 

 o o 



,oO 



900 



— ' 1 ' 1 ' 1 



2,000 4,000 6,000 



Egg production index (billions) 



Figure 5 



Residuals from the regression of ln( recruits) on 

 April mean sea level height at Crescent City. 

 California, versus the ocean shrimp egg pro- 

 duction index from the parent year Both indi- 

 ces were calculated assuming elemental trawl 

 efficiency of 0.50. 



of the stock-recruitment relationship for ocean 

 shrimp. If the 1990 year class was included, the 

 graphs could be interpreted as supporting a dome- 

 shaped stock-recruitment curve, or could even be 

 interpreted as evidence of no relationship between 

 stock and recruitment at all. Given the extremely 

 large fishery impact on egg-bearing females in the 



30- 



20- 



Area 88 



K * 



""i's •'"iiSJii 



X M X „ 





10 



20 



60 



ID 

 O 



^ 40 



20- 



Area 86 



X 



fgii'ihilgiiiiiiiii ,i8. 



10 



20 



ID- 



Area 84 



«« 



SwSx X^XxXB^K* «XX 



Date in April 



Figure 6 



The percentage of egg-bearing shrimp in April samples of 

 the ocean shrimp fishery, for Pacific States Marine Fisher- 

 ies Commission areas 84-88. 1961-97. 



spring of 1989 (Fig 2), in combination with the fact 

 that April has been shown to be a critical period for 

 larval survival, excluding the 1990 data point seems 

 the most conservative approach. With the 1990 year 

 class excluded, a multivariate model incorporating 

 April SLH and the log of the spawner index, was 

 highly significant (P<0.001 1 and explained 79*^ of the 

 variation in log recruitment (Table 2, model 2). The 

 comparable multivariate model, incorporating the 

 egg production index, was also highly significant 

 (P<0.01) and explained 68^^ of the variation in log 

 recruitment (Table 2, model 3). Both models incor- 

 porated April SLH and log spawners or eggs with 

 negative and positive slopes, respectively. A linear 

 regression of the recruitment index on itself, at a lag 

 of one year, was nonsignificant (P>0.05) and showed 

 little evidence of serial autocorrelation, suggesting 

 that the evidence presented in Figures 4 and 5, for a 

 statistical dependence of recruitment on parent stock 

 in ocean shrimp, may be valid. 



