682 



Fishery Bulletin 97(3), 1999 



respectively ( Wyllie Echeverria, 1987 ). An additional 

 concern arises from the protracted parturition sea- 

 son exhibited by these species. Interannual differ- 

 ences in birthdate distributions (Woodbury and 

 Ralston, 1991) result in interannual differences in 

 otolith growth during the first year of life. There- 

 fore, measurements of the first annulus were not 

 included in the analysis because they might have 

 biased the data. Only measurements of otolith growth 

 for the years 1980-87, comprising ages 1-4 and 

 otoliths from fish belonging to the 1979—83 year 

 classes (Table 1) were used. 



The results of this study are dependent upon the 

 accuracy of the ages assigned to each otolith. It is 

 not uncommon for ages to vary by one or more years 

 between readers or among readers re-aging the same 

 otolith. The 1983 annulus was observed over a pe- 

 riod of several years during which one additional 

 annulus was deposited each subsequent year for year 

 classes prior to 1983. By using the distinctive 1983 

 annulus (Fig. 1) as a natural tag (MacLellan and 

 Saunders, 1995), I was able to assign an age to each 

 otolith in this study that was accurate. 



To test the validity of the 1983 annulus in this 

 study, twelve otoliths that had previously been aged 

 were re-examined without knowledge of their date 

 of capture. On the basis of only the pattern of annuli 

 widths, 11 of the 12 otoliths were assigned to their 

 previously determined year class. The one misinter- 

 pretation was for an otolith from a fish bom in 1984, 

 which did not have the 1983 annulus. This error would 

 be unlikely in the data used for this study because the 

 otoliths were first aged by the break-and-bum method, 

 in which the year of birth was determined from count- 

 ing all the annuli. The narrow 1983 annulus could then 

 be used to ensure that the correct age was obtained. 



The following additive model (Weisburg, 1993) was 

 used to calculate treatment effects on annual otolith 

 gi-owth for each species over all years. 



42- 



41 



40- 



39- 



38 



37 



100-fm 

 contour 



Eureka 

 ^Cape Mendocino 



Point Arena 



N 



T 



Cordell Bank' 



Bodega Bay 



8 



$^ Francisco 



126 



125 



122 121 W 



124 123 

 Figure 2 



Sampling locations at Eureka, Bodega Bay, 

 and Cordell Bank. 



■"'.;*/ 



^ + «, + /^, + //,. + 4 + f,,/,- 



;/,■/• 



where J ^,, = index of annual otolith growth of a 

 species; 



\.i = mean square-root of annual growth; 



a^ = year "7" fixed effect; 



P^ = age Y fixed effect; 



7,, = port "/?" fixed effect; 



S/ = sex "/" fixed effect; and 



f^ II = normal error term. 



This model assumes no significant interaction 

 among any of the four main effects. This assumption 

 was evaluated in Figures 3 and 4, where annual least- 

 square means are plotted against year for each spe- 

 cies-sex-port combination by using both a combined 

 model with no interaction term for the age-by-year 



