fraction of females displaying the characteristic 

 divided by the length of the time interval the 

 characteristic remains detectable. Say, from a 

 sample of 10 females, 2 display a characteristic 

 which lasts for 1 day and which indicates that 

 spawning will take place in approximately 1 wk. 

 The daily spawning fraction 1 wk hence will be 1/5. 

 Given this method for estimating spawning 

 fraction the following relationship holds: 



P - Siabc) 



(1) 



where P = production in eggs, 



a = batch fecundity in (eggs)/(unit 

 weight), 



b = fraction spawning (weight of spawn- 

 ing females)/(weight of all mature 

 females), 



c = (weight of females)/! weight of spawn- 

 ing stock), 



S = spawning biomass. 



Spawning biomass can be estimated directly: 



S = P{abc)~\ (2) 



Hunter and Goldberg (1979) examined female 

 northern anchovies for characteristics that would 

 indicate a recent spawning. They found that 

 following spawning follicles of the northern an- 

 chovy go through a sequence of identifiable de- 

 generative stages. The first two stages, which 

 Hunter and Goldberg referred to as day and day 

 1, have durations of 1 day. Stage identification is 

 subject to error. Day-0 follicles can be misiden- 

 tified as day 1; day-1 follicles can be misidentified 

 as day 2 and beyond. The most easily identified 

 stage is day 1. If the spawning fraction, 6, is based 

 on day-1 follicles an adjustment factor, say d, is 

 required in Equation (2): 



P(ab'c)^d 



(3) 



where 6', replacing b, is the observed fraction. 



The adjustment factor is computed by using 

 information on the fraction of day-0 follicles mis- 

 classified as day 1, say do, and the fraction of 

 day-1 follicles correctly classified, say d^. 



d = (c?o + di) 

 Yarid) = VarCffo) + Var(di). 



Estimates based on Hunter and Goldberg's (1979, 

 542 



table 1) blind classification study for d^ and d^ 

 are 5/21 and 16/19 respectively; hence 



d = 1.080 

 Var(d) = 0.016. 



From examination of 195 females taken by mid- 

 water trawl during the time interval 15-27 Feb- 

 ruary 1978, Hunter and Goldberg estimated the 

 observed daily spawning fraction and its variance: 



b' = 0.159 



Var(6' 



4.561 X 10 



Based on the total female weight of nonspawners 

 the estimated batch fecundity and variance ai-e 

 from Hunter and Goldberg (1979, table 6) 



a = 396 eggs/g (or 3.96 x 10^ eggs/t) 

 Var(a) = 886. 



For the time period 18 February-17 March 1978, 

 Zweifel^ estimated daily egg production. From 177 

 plankton samples, northern anchovy eggs and 

 larvae were staged from time of spawning. Esti- 

 mated total numbers at stage were regressed on 

 time. The ordinate intercept, number at time zero, 

 is the estimated egg production: 



2.321 X lO'-'eggs/d 



26 



Var(P) = 1.825 x 10 



If the female to male sex ratio in numbers were 

 1:1 and if the two sexes had equal growth rates in 

 terms of weight then c could be assumed to be 0.5. 

 However, because of conflicting and insufficient 

 evidence neither of these two hypotheses can be 

 supported. Klingbeil ( 1978) demonstrated that the 

 distribution of northern anchovy sexes is hetero- 

 geneous over space and time and that estimates of 

 sex ratio are dependent on the sampling gear. From 

 the purse seine fishery Klingbeil estimated that 

 the ratio of numbers of females to males varies 

 between 1.14:1 and 2.02:1 for 1969-76. From 9yr of 

 midwater trawl data Klingbeil estimated that the 

 sex ratio is 1.03:1. Since midwater trawl surveys 

 cover a wider geographic area and size range of 

 anchovies, they probably provide an estimate 

 closer to that of the true population sex ratio. 

 However, since neither midwater trawl surveys 



'James Zweifel, Southwest Fisheries Center La Jolla Lab- 

 oratory, NMFS, NOAA, P.O. Box 271, La Jolla, CA 92038, 

 pers. commun. 



