130 
Fishery Bulletin 110(1) 
mortality on average, a rate of 6 %/d was used for all 
years (Lough, 2010). 
Demersal juveniles 
Considering the demersal stage or age-0 group juve- 
niles, density-dependent mortality is considered to be 
important in many prerecruit fish populations. Can- 
nibalism has been thought to be especially significant 
for cod because strong year classes seldom follow each 
other, most likely the result of age-1 fish preying on the 
age-0 group. Moreover, there is direct evidence of cod 
raised in captivity cannibalizing smaller individuals 
(Laurence et ah, 1981; Folkvord et al., 1994). For the 
Georges Bank time series of cod year classes, age-1 
recruits declined by as much as 50-70% from the previ- 
ous year’s age-1 recruits when stocks were high, indicat- 
ing density dependence. Direct evidence of cannibalism 
was observed from gut contents of field-caught cod on 
Georges Bank, especially by fish longer than 100 cm 
(Link and Garrison, 2002). 
Time-series analyses 
Autocorrelation functions with consecutive time lags 
(12 years) using STATISTICA™ vers. 9.1 (StatSoft, Inc., 
Tulsa, OK) software did not indicate any significant 
(P<0.05) annual time lags for either cod or haddock 
over the full time series. However, there was a negative 
correlation (r=-0.50, P=0.14, n = ll) for cod when the 
VPA-derived recruitment at age 1 was lagged one year 
for the MARMAP years (1977-87) that had some large 
year classes. Nevertheless, for the short-time series 
compared in this study, we were interested in the year- 
to-year variability and did not want to remove autocorre- 
lation if evident. Also, note that Mantzouni et al. (2010) 
found that there was no significant autocorrelation for 
the Georges Bank cod survival index ( log [R/SSB] ) for 
the years 1978-2004. Therefore, the first method used 
in this study was to assign a mortality rate of 3.0 %/d if 
the previous year had better than average recruitment; 
if less than average recruitment, a mortality rate of 2.5 
%/d was assigned. 
A second method for estimating density-dependent 
demersal juvenile mortality was by first predicting the 
abundance of the pelagic juvenile stage for each of the 
years where hatching abundance and larval mortality 
rates were available for years 1977-87 (Lough et al., 
2006) and for 1995-99 (Mountain and Kane, 2010). 
The yearly demersal juvenile-stage mortality rate ( Z ) 
was calculated by the exponential loss between the 
beginning of the demersal stage and age-1 recruitment: 
Z = -In (N t /N 0 )/t , (5) 
where N 0 - the abundance of demersal 
juveniles; 
N t = the number of age-1 recruits; 
and 
t = the duration of the demersal 
juvenile stage. 
The percent mortality rate per day was cal- 
culated as 
M (in %/day) = 100 (l-e -Z ). (6) 
Regressions were made for all years and 
separately for MARMAP years (1977-87) and 
GLOBEC years (1995-99). The regressions 
were not highly significant (P>0.05), but all 
showed the same trend of juvenile mortality 
rates increasing with size of the previous 
year’s recruits. The MARMAP years had the 
best correlation because there was higher 
recruitment than during the GLOBEC years 
when the stocks were depressed. The regres- 
sion for the MARMAP years (Fig 6) was used 
to adjust the demersal juvenile cod mortality 
rates in the time series: 
Y = 0.020 (cod age-1 R ) + 1.716, 
r 2 =0.18, n= 10, (7) 
where Y = mortality (%/day); and 
R - previous year’s number of recruits 
( 10 6 ). 
25 
20 - 
0 -I 1 1 •— 1 1 
0.00 0.05 0.10 0.15 0.20 0.25 
Wind stress (Pa) 
Figure 5 
Haddock (Melanogrammus aeglefinus) larval cohort mortality (%/d) 
versus wind stress (Pa) on Georges Bank for 1998-1999. Cohort 
mortality represents the first 15 days of early life from hatch- 
ing to about 6-mm standard length (SL). For the mortality-wind 
relationship, cohort mortality is assumed to be related to feeding, 
which is based on previous studies showing turbulence-increased 
contact between larvae and their prey, generated by the wind, is a 
dome-shaped relationship. Mortality rates declined with wind stress 
until about 0.15 Pa (9 m/s) and then increased after about 0.20 Pa 
(12m/s), consistent with theoretical underpinnings. 
