NOTE Adams and Howard Natural mortality of Sebastes mystinus 



157 



domly throughout each day's sampling. Mortality 

 coefficients from individual counts and daily aver- 

 ages are nearly identical (because the daily sample 

 size is consistent). For the individual counts, the co- 

 efficients of determination (r 2 ) are very small (be- 

 cause counts range from zero to a maximum number 

 each day). However, if the daily averages were used, 

 the coefficients of determination would be high (be- 

 cause the daily variation has been removed), but the 

 probabilities of the tests of regression lines would be 

 much lower (because of fewer degrees of freedom). 



Estimates of daily natural mortality for first-year 

 juvenile blue rockfish range from 0.001 to 0.008 day' 1 

 (Table 1; Fig. 3) over the three sites and four years. 

 The fit of data to the mortality model was good; five 

 of the seven regressions were significant at the 99% 

 level, and a sixth significant at the 92% level. The 

 coefficients of determination were low owing to the 

 large variability in the dependent variable (log-trans- 

 formed density, ranging from zero to the maximum 

 number) on each sampling date. The consistency of 

 all the estimates, ranging over only a narrow span. 



from different years and locations, despite the large 

 variance within the daily sampling, was probably the 

 strongest validation of the mortality estimates. 



Our estimates of natural mortality of juvenile blue 

 rockfish are much lower than published values of 

 natural mortalities for other juvenile fishes (Table 

 2). The values in Table 2 include all juvenile fish 

 mortality coefficients estimated from catch-curve 

 analysis and represent both different types of fishes 

 (i.e. flatfish and roundfish) and different types of 

 habitats (marine, estuarine, and freshwater). The low 

 juvenile natural mortality is consistent with the es- 

 timates of low adult natural mortality based on the 

 extreme longevity of this species (Leaman, 1991). 



Mortality estimates were higher during years with 

 stronger juvenile recruitment (Fig. 4). There was a 

 positive relationship between mortality estimates 

 and the strength of the annual population density 

 estimates, with an eight-fold difference in mortality 

 rate estimates corresponding to a twenty-four-fold 

 difference in density. The relationship of natural 

 mortality with population density seems to increase 



