Ralston and Howard: Year-class strength and cohort variability in Sebastes mystmus and 5. flavidus 



713 



40 



39 



3 



38 



37 



M I I 



Granite Canyo 



i i i i i i i f i 



125 124 123 122 



Longitude (°W) 

 Figure 1 



Map of the study area showing the location of trawl stations and diving 

 sites where direct underwater observations of settled juvenile rockfish were 

 made in Mendocino and Sonoma counties (1983-921. Temperature data 

 were collected daily at Bodega Bay, the Farallon Islands, and Granite 

 Canyon. 



ward direction during the transect. Af- 

 ter 1-3 counts the observer made right- 

 angle changes in direction, which re- 

 sulted in thorough coverage of the study 

 area. The number of daily counts at a 

 site ranged from 10 to 35 ( x =18.8). 



Counts were excluded from data 

 analyses when it was obvious that the 

 distribution of juveniles was influenced 

 by unusual conditions. For example, 

 sampling sometimes coincided with a 

 period of convergence when food-rich 

 oceanic waters moved into nearshore 

 surface layers. The distribution of ju- 

 veniles was very different at these 

 times, as they ascended into the upper 

 2 m of the water column to feed. 



Annual indices of settled juvenile 

 abundance were calculated separately 

 for blue and yellowtail rockfish in 

 Mendocino and Sonoma counties. Be- 

 cause variances increased with the 

 means, individual strip-transect counts 

 were first log-transformed to stabilize 

 the variance. The annual index was 

 then simply calculated as the mean of 

 all counts, e.g. 



1 



Isct=-Y l l°Sel-Cisct + 1 l 



where I sct is the index for species s in 

 county c in year t , and n is the number 

 of counts (C lsct ) conducted; the sampling 

 precision of the index is given by the 

 standard error of the mean. 



the abundance of young-of-the-year juveniles was as- 

 sessed along haphazard transects by a series of timed 

 1-minute counts that covered approximately 20 m. 

 Observers maintained a constant swimming speed, 

 gazing ahead at all times during the counts. 

 Transects started on the outside edge of the kelp bed 

 and followed a series of arbitrary compass headings 

 covering the offshore portion of the study site. After 

 completing counts in deeper habitats, observers pro- 

 gressed into shallower water. 



Species and number of juveniles observed each 

 minute were recorded on a plastic slate with the aid 

 of a watch fastened in the upper corner to monitor 

 time. Observers swam 2 m off the bottom and counted 

 young-of-the-year rockfishes within 3 m in any for- 



Interannual variability in year- 

 class strength 



As in many other species, recruitment in rockfish is 

 highly variable and is described well by the log-nor- 

 mal distribution (Bence et al., 1993; Fogarty, 1993). 

 An accepted way to portray relative levels of varia- 

 tion among different sets of data is through use of 

 the coefficient of variation (CV). The CV of the log- 

 normal distribution is unusual, being independent 

 of the mean and equal to [exp(cr 2 ) - 1] 1/2 , where s is 

 the variance of logarithms of the log-normally dis- 

 tributed variable (Johnson and Kotz, 1970). 



To compare and contrast levels of variation in rock- 

 fish year-class strength at specific life history stages, 

 CVs of annual time series were calculated. Because 

 individual annual abundance statistics were usually 



