134 
Fishery Bulletin 110(1) 
of age-1 fish in the time series was not provided with 
an estimate of their variability, but CVs of SSB and 
recruitment are likely to be similar to those derived 
from the ichthyoplankton surveys icf. Brander, 2003). 
Georges Bank age-1 cod and haddock CVs based on the 
terminal year assessment are about 58% (O’Brien et 
al. 1 ; Brodziak et al. 2 ). 
Hindcasting cod and haddock age-1 recruitment from 
the field egg and larval abundances was remarkably 
close to VPA estimates for the GLOBEC years, 1995- 
99, in part owing to the better sampling coverage, com- 
pared with that of the MARMAP years, 1977-87. The 
MARMAP field recruitment estimates for cod were less 
than the VPA recruitment, especially for the 1983 and 
1987 seasons from eggs and 1977 and 1978 seasons 
from larvae. Under-sampling is a possible explanation 
because the MARMAP surveys were only conducted 
bimonthly. Episodic events, such as transport off the 
bank, could be missed with under-sampling. During 
April and May 1987, an unusual westward transport 
of shelf water may have carried a large portion of eggs 
and larvae from Georges Bank to Nantucket Shoals (Po- 
lacheck et ah, 1992). Hindcasting recruitment of had- 
dock for 1996 from both eggs and larvae did not follow 
the trend in the VPA recruitment, and Mountain et al. 
(2003) mentioned a possible loss off the bank owing to a 
strong March wind event. The 1983 and 1987 hindcasts 
also were lower than the VPA recruitment starting from 
either eggs or larvae. However, the hindcasts starting 
with larvae were very close for the 1977 and 1978 sea- 
sons. Haddock recruitment in 1978 was high and both 
retention indices also were relatively high for eggs and 
larvae to remain on Georges Bank that year (Lough et 
al., 2006; Friedland et al., 2008). 
Abundance at hatching (larvae) was a better predic- 
tor of age-1 recruitment than initial egg abundance. 
There is greater uncertainty in egg abundance and mor- 
tality because the early stage eggs are usually under- 
sampled (late-stage eggs are more dispersed) and the 
identification of the early stages is based on the propor- 
tion of cod-to-haddock late-stage eggs. Also, initial egg 
abundances may have been underestimated owing to 
off-bank loss. Mountain et al.’s (2008) wind-transport 
modeled estimate of egg mortality suggested transport 
off the southern side of Georges Bank to be 12-14 %/d, 
leaving about 6-8 %/d on-bank mortality due to preda- 
tion or morbidity. 
In the present study there was no correlation between 
initial egg abundance and recruitment, but there was a 
significant positive relationship between initial larval 
abundance and recruitment for both cod and haddock. 
Initial abundance of pelagic juveniles was still positive- 
ly correlated with recruitment, but was less significant. 
These correlations would suggest that recruitment was 
largely established by the larval stage in some years; 
however, Lough (2010) compared the 1986 and 1987 cod 
Table 6 
Coefficients of correlation (r) between the various haddock ( Melanogrammus aeglefinus ) life stage model predictions of recruit- 
ment (R) and the cod virtual population analysis (VPA) for age-1 recruitment (R). Base case models start from either annual 
spawning abundance of eggs or hatching abundance of larvae using the stage mortality rates derived from Marine Resource 
Monitoring, Assessment, and Prediction (MARMAP) or U.S. Global Ocean Ecosystems Dynamics (GLOBEC) surveys. The third 
base case is the abundance of eggs derived from the VPA spawning stock biomass (SSB). The annual survey egg and larval 
mortality rates used for each model can be found in Table 2. Egg- and larval-stage mortality used a function of wind stress is 
designated as “/'(W) Eq.” and refers to the equation number in the text. Pelagic juvenile mortality used a constant 6%/d for all 
models. The demersal juvenile mortality rate was determined by two methods: 1) Demersal juvenile mortality rate of 2.5%/d was 
used if that year was below mean recruitment (R) in the full time series 1978-2004 from previous year’s f?" 1 , or a mortality rate 
of 3.0%/d was used if that year was above mean R from previous year’s R (<> mean R -1 ); 2) Demersal juvenile mortality used as a 
function of the previous year’s R~ x from Equation 7 is designated as f(R~ l ) Eq. 7. Correlations (r) R<0.05 are starred. Associated 
figures for some of the models are referred to in parentheses. 
Life stage 
Base case 
Time 
series 
No. 
parameters 
Egg 
mortality 
Larval 
mortality 
Demersal 
juvenile mortality 
?' 
r 2 
Spawning 
MARMAP 1979-87 
4 
Table 2 
Table 2 
<> Mean R l 
0.38 
0.13 
abundance 
4 
Table 2 
Table 2 
f(R~ l ) Eq. 7 
0.49 
0.24 
GLOBEC 1995-99 
4 
Table 2 
Table 2 
<> Mean R~ l 
0.72 
0.52 
4 
Table 2 
Table 2 
f(R~ l ) Eq. 7 
0.83 
0.69 
(Fig. 8A) 
Hatching 
MARMAP 1977-87 
3 
Table 2 
<> Mean R l 
0.07 
0.05 
abundance 
3 
Table 2 
f(R~ l ) Eq. 7 
0.45 
0.02 
GLOBEC 1995-99 
3 
Table 2 
<> Mean R 1 
0.82 
0.67 
3 
Table 2 
f(R~ l ) Eq. 7 
0.89* 
0.79 
(Fig. 8B) 
Eggs VPA SSB 
MARMAP 
4 
f(W) Eq. 3 
f(W) Eq. 4 
f (R _1 ) Eq. 7 
-0.16 
0.03 
GLOBEC 
4 
f(W) Eq. 3 
f(W ) Eq. 4 
f(R~ l ) Eq. 7 
-0.08 
0.01 
1978-2004 
4 
f(W) Eq. 3 
f(W) Eq. 4 
fiR- 1 )Eq. 7 
0.36 
0.13 
(Fig. 80 
