454 



Fisher/ Bulletin 97(3), 1999 



20 30 40 



Fork length (cm) 



Figure 1 



Length-frequency (FL, cmi histograms of 

 ehu. Etelis carbunculus (hollow bars, 

 /! = 181), and kalekale, Pristipomoides sie- 

 boldii (solid bars, n=80). 



Size at sexual maturity 



For each species, the length-frequency distribution 

 comprised relatively few individuals <30 cm FL (Fig. 

 1). Hence, our size-at-maturity characterizations 

 might have been affected by the scarcity of imma- 

 ture specimens. 



Ehu On average, ehu in Hawaiian waters mature 

 at an estimated 27.9 cm FL (95% CI=25. 5-30.3 cm) 

 according to histological criteria or 27.8 (25.5-30.1) 

 cm FL based on logistic regression according to gross 

 criteria (r2=0.967 and 0.962, respectively; Fig. 2). The 

 precision of these L^q estimates is reasonable despite 

 the discontinuous lower tail of the size-frequency 

 distribution. Our L^q estimates for ehu provide an 

 adequate objective basis for evaluating gross and 

 histological classifications; however, they should be 

 considered only as preliminary estimates of size at 

 maturity. 



Kalekale A body length-maturity pattern also was 

 strongly evident for this species. Whether based on 

 histological evidence or logistic regression with gross 

 metrics, size at maturity was well described by the 

 logistic model (r^=0.920; Fig. 3). Kalekale in Hawai- 

 ian waters mature at 29.0 (9'5% 01=27.2-30.8) cm 



100 -, 



c 

 S 



Q- 



L50 = 27 8 cm FL (gross cnteria) 

 = 27 9 cm FL (histology) 



Histological chtena • 



Px= 100/(1 + exp-(-8.247f-0,2% FL) 



l^ = 967 



Gross cntena D 



Px- 100/(1 +ei<p-(-7 539+0 271 FL) 



r^ = 0.962 



■T — 



50 



60 



Fork length (cm) 



Figure 2 



Scatterplots of mean percentage maturity by 2-cm 

 length class, for ehu, Etelis carbunculus. Solid circles 

 and solid line represent the observed values and fitted 

 curve based on histological criteria of maturity; hollow 

 squares and dotted line represent the values and curve 

 based on maturity classified by logistic regression on 

 fork length, ovary weight, and oocyte volume. The lo- 

 gistic curves were fitted by nonlinear regression with 

 each class mean weighted by the square root of its count 

 (see Fig. 1). 



FL, according to histological and gross criteria 

 (Fig. 3). 



Discussion 



Gross metric criteria 



For both ehu and kalekale, the proportional weight 

 of ovaries increased allometrically with body length 

 (coefficients of 6j » 3; Table 2), which is typical for 

 asynchronous multiple-spawners (Reiss, 1987). Our 

 results further confirm that gonad indices are un- 

 suitable for identifying maturity stages in asynchro- 

 nous multiple-spawners because the rate at which 

 ovary weight increases with body size differs among 

 fish of different maturation stages (deVlaming et al., 

 1982; Erickson et al. 1985b). Body size, ovary size, 



