450 



Fishery Bulletin 97(3), 1999 



Render et al., 1995), been identified subjectively if 

 more than one method of classification was used, 

 sometimes with inconsistent results (Trippel and 

 Harvey, 1991). Recently, West (1990) identified mi- 

 croscopic staging coupled with a gonad index as the 

 most objective and quantitative method. 



Gonad indices should be used with caution as mea- 

 sures of sexual maturity and require biologically re- 

 alistic companion analyses (e.g. Koya et al., 1995). A 

 fundamental flaw of ratios, such as gonad indices, is 

 that they either lose or obscure meaningful informa- 

 tion in the numerator and denominator of the ratio 

 (Garcia-Berthou and Moreno-Amich. 1993). Gonad 

 indices also are questionably applicable for asynchro- 

 nous, multiple-spawning fishes (deVlaming et al., 

 1982). There are two major reasons for this. First, 

 because multiple-spawners ripen and shed numer- 

 ous batches of eggs, the ovary weight of an individual 

 fish fluctuates cyclically and often greatly during the 

 spawning season. Second, somatic weight often var- 

 ies independently of ovary weight for individuals. 

 Somatic "condition" (robustness) usually declines 

 monotonically, and sometimes greatly, as somatic 

 energy reserves become progressively depleted over 

 the spawning season, whereas ovary weights cycle 

 high and low as the individual continues to spawn 

 (Delahunty and deVlaming. 1980; deVlaming et al., 

 1982). Variances in GSIs among mature fish are un- 

 avoidably large (e.g. Macer. 1974) because small, 

 ready-to-spawn fish sometimes have higher GSIs 

 than larger fish that have just shed their eggs. Other 

 potential complications include differences in ova- 

 rian cycling frequencies (spawning intervals) be- 

 tween small and large females within populations 

 ( Hunter and Macewicz, 1985 ). Thus even refinements 

 of the conventional GSI such as the "relative gonadal 

 index" (RGI) of Erickson et al. (1985b), in which go- 

 nadal investment is adjusted for size allometry, can 

 be inappropriate for asynchronous, multiple-spawn- 

 ers if the dynamics of somatic and gonadal condition 

 are unknown. Statistical descriptors of sexual ma- 

 turity that are more accurate and robust than GSIs 

 need to be developed (Garcia-Berthou and Moreno- 

 Amich, 1993; Winters and Wheeler, 1996; Jons and 

 Miranda, 1997). 



Our study has several complementary objectives. 

 First, we evaluate four gross metrics ( fish body length 

 and weight, ovary weight, and oocyte size) for de- 

 scribing stages of sexual maturation in two eteline 

 lutjanids (ehu, Etelis carbunculus, and kalekale, 

 Pristipomoides sieboldii) that contribute to the deep- 

 slope bottom fishery in Hawaii. We examine these 

 four characters because they are readily measured 

 and less costly than the more accurate histological 

 staining of ovarian tissue. We also independently 



assess stage of sexual maturity for each species based 

 on histology of ovarian tissue samples from speci- 

 mens collected during peak periods of spawning. In 

 our study, we use discriminant analysis and logistic 

 regression to further classify individual fish as ei- 

 ther immature or mature on the basis of gross mor- 

 phometric and histological criteria. In conclusion, we 

 evaluate the results of gross classifications of matu- 

 rity for ehu and kalekale by assessing how well L^g 

 estimates based on statistical assignments of matu- 

 rity with gross metrics agree with L^q estimates 

 based on histological evidence of maturity. 



Materials and methods 



Fish collection and shipboard processing 



Two types offish collections were used: hook and line 

 and bottom trawling. For exploitable-sized fish of 

 both species, hydraulic handline gurdies with sizes 

 28 and 34 Izuo circle hooks ( baited with squid strips ) 

 were used to fish near bottom at depths of 60-300 m, 

 on insular slopes and submerged banks located be- 

 tween 22-25=N lat. and 160-170°W long, in the 

 Northwestern Hawaiian Islands (NWHI). Specimens 

 were collected in September 1992; during June, July, 

 August, and September 1993; in August 1996, and 

 in July and August 1997 (Table 1). Most Une-fishing 

 was conducted during daylight hours (0730-1930) 

 aboard the NOAA ship Townsend Cromwell. Addi- 

 tional small kalekale were sampled at 60-65 m off 

 south Molokai, Main Hawaiian Islands, during Au- 

 gust 1992 and September 1993. We also examined 

 additional small ehu caught by bottom trawl at 88- 

 97 m near Kure Atoll (NWHI) in early June 1977 

 (Table 1). All specimens examined were collected 

 during the protracted summer spawning period of 

 eteline snappers in Hawaiian waters (Ralston, 1981; 

 Everson, 1984; Everson et al., 1988). Ehu are thought 

 to be serial spawners (Everson, 1984), and oocyte size 

 distributions within ovaries suggest that Hawaiian 

 populations of both species comprise asynchronous 

 spawners.- 



Once captured, fish were placed immediately on 

 ice. Within an hour of capture, fork length (FL) to 

 the nearest 0.1 cm and ovary-free body weight 

 (OFBW, in g) was estimated for each fish. 



Ovary weight ( OW) of each specimen that appeared 

 female was estimated ±0.1 g (if <5g), ±1 g (if 5-100 

 g) and ±10 g (if >100 g). A tissue plug (1-50 g, de- 



- 1994. Honolulu Laboratory. Southwest Fish. Sci. Cent.. Natl. 

 Mar Fish. Serv., NOAA, Honolulu, HI 96822-2396. Unpubl. 

 data. 



