NOTE Laidig and Ralston. Use of otolith characters in identifying larval Sebastes spp. 



167 



nucleus (i.e. the otolith core) were noted for each spe- 

 cies. Increment counts were made beginning at the 

 first clearly defined mark that completely encircled 

 the primordium (see also Penney and Evans, 1985). 

 This "extrusion" check forms at parturition (Ralston, 

 unpubl. data) and defines the outer edge of the 

 nucleus; the distance from the primordium to this 

 mark is the nuclear radius (Fig. 1), and the incre- 

 ment immediately following the extrusion check is 

 the first growth increment. 



Data analysis 



The mean radius of the nucleus and the mean width 

 of the first growth increment were compared among 

 species and years. An overall analysis of variance 

 (ANOVA), incorporating separate pairwise £-tests, 

 equivalent to Fisher's least-significant difference, 

 was performed to detect differences in the size of the 

 nuclear radius and the width of the first increment 

 among species and years. We used a two-way facto- 

 rial analysis and calculated least square means 

 (Searle et al., 1980) to evaluate treatment effects 

 among species, years, and the interaction of species 

 and years. A parametric discriminant analysis, as- 

 suming a multivariate normal distribution with 

 pooled covariance matrix, was used to determine the 

 percentage of each species that was correctly classi- 



fied by nuclear radius and first increment width, 

 alone and in combination with each other. 



To further evaluate species-specific differences in 

 nuclear radius and first increment width, the data 

 were pooled over years; however, we recognized the 

 difficulty in isolating the effect of species alone. 

 Sebastes auriculatus and S. saxicola were only 

 sampled in one year; therefore they were not included 

 in the annual variation analysis. For comparison, we 

 treated these single-year studies as the pooled data 

 for the other species. 



A blind test was performed to determine the accu- 

 racy of the otolith characters in distinguishing be- 

 tween the eight rockfish species used in this study. 

 One hundred otoliths representing all eight species 

 (S. auriculatus [n=5]; S. entomelas [n=13]; S. flauidus 

 [n = 15]; S. goodei [n = 14]; S. jordani [n=25]; S. 

 mystinus [n=9]; S. paucispinis [re=ll]; and S. saxicola 

 [n-8]) were given to a reader. No other information 

 (e.g. species, fish length, etc.) about the individual 

 samples was provided. The reader then attempted 

 to identify the correct species of rockfish, using both 

 measured distances and shading patterns. The re- 

 sults of the tentative classification were compared 

 with the actual species identities to determine per- 

 cent agreement. The significance of this result was 

 evaluated against a null multinomial distribution, 

 by assuming assignments at random to species. 



Figure 1 



A Sebastes goodei otolith displaying the characteristic dark inner ring (DIR) around the primor- 

 dium (PR). NE = nuclear edge. 



