Zeldis et al.: An estimate of biomass of Hoplostethus atlanticus 
587 
A B 
Figure 5 
Contours of dynamic height (dynamic meters) at 100, 400, and 800 m in the 
hydrographic survey area (A-C). The arrows on the contours show inferred 
current directions. The large filled circles indicate stations used in the analy- 
sis (where cast depths were > the 1,000 m reference depth). Panel (D) shows 
depth profiles of current velocity (cm/sec) at the symbols in A-C, flowing in the 
direction of the arrows through each symbol. In A, the plankton survey area is 
shown for reference and the start and end positions of the buoy deployment are 
shown by the beginning and end of the open-headed arrow. 
spent proportions followed at ap- 
proximately 10-15 day intervals. 
Serial spawning was evident in that 
partially spent, hydrated, and ovu- 
lated proportions became fairly con- 
stant during a 10-day period 
(roughly 25 June-5 July) when ova- 
ries of the fish were developing 
among these stages. Hydration did 
not appear to be associated with 
imminent spawning, because sig- 
nificant proportions (0.15) of hy- 
drated fish were present about 5 
days before planktonic eggs were 
first caught on 15 June. However, 
the first appearance of significant 
proportions (>0.05) of ovulated fish 
(14 June) and planktonic eggs (15 
June) nearly coincided. 
The decline in R t (Fig. 8) or the 
daily fecundity per female weight, 
D, was 787 eggs/(kg x day) (CV= 
0.11; Table 4). 
Ritchie Bank spawning female 
biomass 
When N 0 was divided by D , the es- 
timate ofB ,was 14,000 tons (Table 
4), with C v = 0.50. 
Biomass of recruited orange 
roughy in the mid-east coast 
stock 
The factor S was estimated to be 
1.77, with CV = 0.03. However, his- 
tology showed that for 4.5% of the 
females identified as spawners in 
the wide area trawl survey, their 
entire exogenous vitellogenic oocyte 
complement was actually in the process of atresia (Bell 
et al., 1992). These fish were indistinguishable macro- 
scopically from fish that would spawn successfully and 
did not appear to cluster in any particular area of the 
trawl survey. The factor S was scaled upward to ac- 
count for these fish, resulting in S = 1.85 (Table 4). 
The resulting estimate of B rec was 26,000 t (Table 
4), with CV = 0.50. The bootstrap procedure used may 
have slightly overestimated the CV of S because it 
did not fully take into account the stratum structure 
of the wide-area survey. However, this overestima- 
tion is of little importance because the CV of S was 
so much smaller than that of the other components 
that made up the CV of B rec (Table 4). 
Bias due to turnover 
An important potential bias in the DFRM arises from 
the fact that the method is sensitive to turnover of 
fish on the spawning ground. For example, if fish 
arrive, complete spawning, and leave the trawl sur- 
vey area before or after the trawl survey period, fe- 
cundity will be undetected and biomass will be un- 
derestimated. These effects were unlikely, however. 
Almost no eggs were caught in the first subsurvey 
(Fig. 3A), and very few spent fish were detected in 
trawls on Ritchie Hill until 2 July (Fig. 7). This find- 
ing suggested that no spawning was completed be- 
fore the trawl survey began (8 June). In addition, 
