592 
Fishery Bulletin 95(3), 1997 
A second difference between the DFRM used here 
and that of Lo et al. (1993) is the way sex ratio and 
active female proportion were estimated and brought 
into the biomass model. In the present study, the 
proportion of all recruited fish that were spawning 
females was estimated with parameter S, whereas 
this proportion was estimated within the R and D 
parameters in the model of Lo et al. (1993). Orange 
roughy spawning takes place in dense aggregations 
that form after the spawners have migrated hun- 
dreds of km from the nonspawners. Therefore, it is 
not possible to estimate the proportion of active fe- 
males to all recruited fish with the trawls done on 
the spawning ground during the spawning season 
(which are used to monitor fecundity reduction). This 
estimation must be done with a separate trawl sur- 
vey over the whole stock area, preferably before the 
spawners have aggregated significantly. In contrast, 
Dover sole spawning appears to be dispersed over a 
wide latitudinal area of the North American west coast 
slope (Lo et al., 1993) and takes place over a long spawn- 
ing season (6 months; Hunter et al., 1992). In Lo et al. 
(1993), the spawners were assumed to be dispersed in 
the same geographic region as the nonspawners dur- 
ing spawning, and therefore sex ratio and proportion 
active components were estimated from the same trawls 
that were used to estimate fecundity reduction. 
There are few other estimates of planktonic egg 
mortality (Z) in the literature for deepwater spawn- 
ers that can be compared with the value (Z=0.7) ob- 
tained for orange roughy in this study. Lo et al. ( 1993) 
fitted a Pareto decay function to Dover sole egg abun- 
dances, in which mortality was allowed to decrease 
with egg age. Initial mortality was 0.63 which halved 
by the age of 1 day. Another western North Ameri- 
can slope species, sableflsh ( Anoplopoma fimbria), 
has been the subject of egg-production studies (Moser 
et al., 1994) and for this species, Z varied between 
0.25 and 0.48, depending on region (it should be noted 
that counts of 1-day-old and 2-day-old eggs were ex- 
cluded because these eggs were undersampled). 
Thus, the few Z estimates that exist for other 
deepwater species indicate variable mortality rates, 
but that mortalities can be quite high. Orange roughy 
egg abundance was estimated by Koslow et al. (1995) 
in their application of the AEPM to orange roughy 
biomass estimation on St. Helen’s Seamount, Tas- 
mania. These authors assumed that mortality in the 
first 28 h after spawning was zero and that the mean 
abundance of 0 to 28 h old eggs equalled N (y This 
assumption was based on their finding no differences 
in abundance among fertilization or 1-cell eggs 
(stages 1 and 2 of the present study) and their sub- 
sequent two stages which were 2 cell and 4-128 cells 
(stages 3 and 4-9, respectively, of the present study; 
see also Grimes et al. 3 ). As mentioned, the differentia- 
tion of stages 1-7 was not possible for the great major- 
ity of the samples in the 1993 Ritchie Bank survey be- 
cause of egg damage. However, the probability that the 
mortality rate estimated from the grouped stages 1-7 
and stages 8, 9, and 10 was zero was low (P=0.10). 
The procedure used to estimate S, for converting 
spawning female biomass to recruited biomass, is 
easily adapted to estimate the proportion of females 
in the stock area that will spawn in the current year. 
For the mid-east coast stock in 1993, S was estimated 
as 0.52 of all mature females (by weight), a result 
similar to that for orange roughy stocks in southern 
Australia (0.45, by numbers; Bell et al., 1992). Wide- 
area trawl surveys of the east coast were also done 
in March-April of 1992 and 1994, 10 and these yielded 
similar values for these proportions (0.49 and 0.42 
by weight). Also, a low proportion of females with 
fully atretic ovaries was found (0.045) over the mid- 
east coast survey area in 1993, again in common with 
the results of Bell et al. (1992). 
A 
B0 ('000 t) 
B 
Figure 10 
Probability distributions of (A) virgin biomass (B 0 ) 
and (B11994 biomass as a percentage of B 0 , from 
the 1994 stock reduction analysis for the mid-east 
coast orange roughy, Hoplostethus atlanticus, fish- 
ery (see Footnote 2 in the text). Each panel shows 
(solid line) the distribution when the analysis was 
done without the DFRM estimate (with only CPUE, 
trawl survey indices, and mean fish length data) 
and (dotted line) the distribution when the DFRM 
estimate was included. 
