strength of the available stocks. Analysis of 

 variance of the means of logarithms of the total 

 sardine catch in California for the year classes 

 1930 through 1951 gives the following: 



IE 



HI 



32: 



m 



•Significant at a prolialiility of less than 0.1 percent. 



The average catches by age indicate that sardines 

 are not fully recruited until age III (table 1). The 

 increase in tlie mortality coefficient, even after 

 age III, may be attriluited jiartly to incomplete 

 recruitnient after this age ntid partly to an increase 

 in natural mortality. 



T.vBLE 1. — Mean sardine calrh (In), total mortality 

 coefficients, and change in mortality by age for the year 

 classes 19.i0-51 and iyS0-J,4 in California 



Because of incomplete records, the expected 

 catches of eight year classes, 1926-29 and 1951-57, 

 are constructed through their mean catches and 

 a standard catch curve, so that SjAo reduced to 

 zero. When these additional classes are included, 

 the deviations of loganthniic catches of any two 

 age groups in the same season are positively 

 correlated with each other, where the effects of 

 environment and fishing intensity might be ex- 

 pected to be common for all the age groups. The 

 correlation coefficient is higher for combinations 

 of adjacent age groups (fig. 2). On the other hand, 

 correlation coefficients between deviations in two 

 ages from the same year classes range from 

 — 0.690 to 0.307. This variability indicates that 

 change in survival rate is less important than 

 change in availability and exploitation. 



From these observations in regard to assump- 

 tions given in the last two paragraphs of the 



1.0 

 0.5 

 1.0 

 0.5 

 10 

 0.5 

 1.0 

 0.5 

 1.0 

 05 



-L. 



M 



X 



^fe 





3zr 



m 



2r 



w. 2: 



AGE OF FISH 



FuiUKB 2. — Correlation coefficients between index of 

 virtual availability of two ago groups, ba.sod on the 

 total catch in California, 1932-33 through 1957-58 

 seasons. 



preceding section (on fonnulations), it may be 

 assumed tiiat the year-class-dependent change in 

 total mortality coefficients is smaller than tlic 

 seasonal clianges in botli availability and exjiloita- 

 tion and that the migratory pattern differs from 

 age to age but is similar for adjacent ages. The 

 multiple correlation of deviations of age gr()Ui)s 

 III to V in the same season was calculated to be 

 as high as 0.815. These three age groups furnished 

 the greater part of the period under discussion. 



It is well known that the migratory pattern of 

 the Japanese sardine and its fishing grounds are 

 affected by changes in oceanograiihic conditions 

 (Sako, 1939; Shimomura, 1954). Because sufficient 

 information about the environment of the fishing 

 grounds in California waters was lacking, I 

 correlated the temperature anomaly at the [lier 

 of the Scripps Institution of Oceanography for 

 July througli September (just before the fisliiiig 

 season) with annual mean deviations in the catch 

 of the three major age groups for 26 seasons 

 (fig. 3). The correlation coefficient was 0.599, 

 slight Ij' over the l-i)erceiit level of significance. 



Clark and Daugherty (1950) reported the catch 

 j)er unit of effort and total effort of the |)urse 

 seiners in California waters as well as at eacli of 

 the three major ports, San Francisco, Monterey, 

 and San Pedro, for the seasons 1932-33 through 

 194S-49. To eliminate the effects of improvement 



590 



U.S. FISH AND WILDLIFE SERVICE 



