122 



Fishery Bulletin 90(1). 1992 



oocytes or postovulatory follicles. The sexual matur- 

 ity of these females is certain, but some mature females 

 may be excluded if only the first set of criteria are used. 

 Criteria sets 2 to 5, if added to the first set, broaden 

 the maturity definition to include females having 

 ovaries in the earliest stages of vitellogenesis and those 

 showing possible signs of past reproductive activity (/?- 

 or a-stage atresia). Each additional criteria set that one 

 might add to the first set increases the risk that im- 

 mature fish will be classed as mature. Females in set 

 6 are considered to be immature because they have 

 none of the characteristics mentioned in the other five 

 sets. 



Use of p atresia as a possible sign of past reproduc- 

 tive activity seems justified. Females with early yolked 

 oocytes and p atresia (set 2) were larger on the average 

 than those with no /3 atresia (set 3; t 2.45, P 0.015, df 

 293); and females with unyolked oocytes and ft atresia 

 (set 4) were larger than those with only a atresia of 

 the unyolked (set 5; t 7.69, P<0.001, df 709). In addi- 

 tion, the ranking of criteria sets based on our intuitive 

 appraisal of the risk of classification error is largely 

 borne out by the length distributions of the females 

 identified by the criteria set, since mean length de- 

 creased with criteria set number (Fig. 10). 



To estimate the length at which 50% of the Dover 

 sole are mature (ML50) using all six histological 

 criteria sets, we first used a maturity algorithm to 

 estimate the fraction of fish that were mature in a given 

 length-class. This algorithm is a regression method 

 similar to those used to construct age-length keys 

 (Bartoo and Parker 1983, Kimura and Chikuni 1987). 



Figure 10 



Length distribution of female Dover sole Microstomus pacif- 

 mis identified by six sets of histological maturity criteria, rang- 

 ing from criteria set 1 where maturity is certain, to set 6 where 

 all females are considered to be immature (see Table 15). 

 Females captured in November-December in California and 

 Oregon; filled triangles indicate mean length of females. 



This analysis was based on two equations. The first 

 equation was 



Qjji = Qmii qjjm + (1 " Qniji) ^jji 



(Eq. 12) 



where qj | i is the fraction of fish of length-class j in the 

 ith criteria set; qjim = Ijii and qjiim = qj|6 because the 

 criteria set 1 consists of all mature (m) fish and the 

 criteria set 6 consists of all immature (im) fish; and 

 q^ii the overall fraction of mature fish in the ith cri- 



