Lough and O'Brien: Life-stage recruitment models for Gadus morhua and Melanogrammus aeglefinus on Georges Bank 
137 
fects of Atlantic water inflow and primary production 
and found that the combination of inflow of larvae into 
the Barents Sea and high primary production accounted 
for 70% of the variability in cod recruitment. At high 
latitudes, environmental forcing is stronger, productive 
seasons are shorter, and there are fewer trophic links 
than at low latitudes (Kristiansen et ah, 2011). 
Recruitment of cod in the North Sea is known to be 
related negatively to temperature (O’Brien et ah, 2000) 
and positively to plankton abundance. Beaugrand et 
al. (2003) developed a plankton index of larval cod 
survival that was correlated with age-1 recruits. An 
improved plankton index based on three indicators (C. 
finmarchicus, PseucLocalanus spp., and euphausiids) im- 
proved its applicability to other cod stocks (Beaugrand 
and Kirby, 2010). Daewel et ah (2011) explored the 
bottom-up trophic processes on larval cod in the North 
Sea using a larval fish individual based model and a 
three-dimensional ecosystem model to provide the prey 
field for three years. Their most important finding was 
that larval survival was most related to the timing 
and overlap of species-specific prey. Retention of larvae 
within the prey field and temperature also interacted 
to promote growth and survival. At this location colder 
years were more correlated with larval survival because 
the lower wind stress decreases dispersal and delays 
the onset of first-feeding larvae overlapping with their 
prey. On the other hand, top-down control was reported 
by Fauchald (2010), who found North Sea cod recruit- 
ment, together with abundance of Calanus finmarchicus, 
to be negatively correlated with the SSB of herring, a 
main competitor of cod, over a 44-year time series. Her- 
ring stock size appeared to affect cod recruitment over 
decadal periods, whereas C. finmarchicus appeared to 
explain the annual recruitment, although the principle 
prey of larval cod in the North Sea is Pseudocalanus 
spp. (Heath and Lough, 2007). 
Shelf ecosystems near the southern limit of cod and 
haddock distributions have been hypothesized to be 
controlled by bottom-up processes where there is a posi- 
tive correlation between predators and prey (Frank et 
ah, 2007). On these southern banks, water tempera- 
ture is warmer and species richness is higher than on 
more northern shelves. Variability in the spring bloom 
and copepod prey, believed to support greater growth 
and survival of larvae, has been related to the size of 
the haddock year class on Browns Bank (Platt et al., 
2003; Head et al., 2005). Conversely, Friedland et al. 
(2008) did not find any correlation between the timing 
of the spring bloom on Georges Bank and the haddock 
survivor index. However, they did find that the magni- 
tude of the fall bloom before the spawning year class of 
haddock on Georges Bank was correlated with subse- 
quent recruitment — the hypothesis being that enhanced 
conditions for prespawning adults of increased benthic 
productivity, specifically their main prey, brittle stars, 
improved their egg quality and quantity and resulted 
in a higher survival of eggs and larvae. 
Top-down control of recruitment has been implicated 
for the Scotian Shelf (Frank et ah, 2005) and North Sea 
(Fauchald, 2010) when the large predatory fish are re- 
moved and the ecosystem becomes dominated by pelagic 
fish. The demise of cod has allowed the resurgence of 
herring, which prey on the eggs and larvae of cod and 
other species. A similar trophic switching occurred on 
Georges Bank where groundfish (cod, haddock) shifted 
to small pelagic species (herring, mackerel) in the 1980s 
(Link et al., 2009). A Georges Bank food web study 
(Steele et al., 2007) showed the importance of bottom-up 
forcing for fish yields; however, the authors emphasized 
that both bottom-up and top-down processes operate at 
the same time and no single process determined the 
observed patterns for the three fish categories over 
four decades. Mueter et al. (2009) also found evidence 
for bottom-up control of fish yield for the Gulf of Maine 
and Georges Bank region since the 1980s. Mueter et al. 
(2009) compared different ecosystems (Bering Sea, Gulf 
of Alaska, Norwegian Sea, Barents Sea, Gulf of Maine, 
Georges Bank), and concluded that zooplankton can be 
controlled by both bottom-up and top-down processes 
leading to regime shifts caused by direct and indirect 
effects of fishing and climate. 
Toward better forecasting 
Cod and haddock recruitment depend on a unique com- 
bination of spawning stock attributes and environmental 
effects on the early life stages as exemplified in this 
study. Projected recruitment, based on annual egg pro- 
duction derived from VPA SSB-fecundity relationships 
or egg surveys, is less reliable than that derived from 
surveys of the larval stage. Intensive surveys of larvae 
and pelagic juveniles, however, may be sufficiently accu- 
rate for annual forecasts. 
The challenge is to provide easily obtainable indices of 
environmental conditions that will link egg production 
with survival through the larval and juvenile stages. 
There are many factors contributing to mortality rates 
in most years and that are difficult to separate (e.g., 
advective loss, feeding and growth, predation), except 
in years of strong environmental forcing. We now can 
estimate an approximate loss rate of eggs transported 
off Georges Bank, given wind direction, strength, and 
duration, using recently developed circulation models, 
as well as possible upstream (Scotian Shelf) contribu- 
tions of spawned eggs. Estimating the annual average 
larval mortality rate is more tenuous without survey 
abundance data. 
The greatest life-stage mortality can occur during the 
demersal juvenile stage of almost nine months where a 
small change in mortality rate over this long duration 
can have a significant impact on survivors. A simple 
metric for the density-dependent effects of cod was used 
in this study. However, it is unlikely we will be able 
to fully assess predation mortality for the juveniles 
because it is so variable, and it is difficult to quanti- 
tatively survey the recently settled juveniles owing to 
escapement. Auster and Link (2009) and Steele et al. 
(2007) recommended that feeding guilds be considered 
for monitoring as a management resource because sig- 
