Basin-scale changes in
The objective of the proposed
study is to develop an understanding of the processes controlling recruitment
of cod and haddock on Georges Bank and cod in the
Implicit in our approach is the hypothesis that recruitment of cod and haddock is determined by variability in survival during the egg and larval stages that is constrained by density- and habitat-dependent factors operating during the early juvenile period. Variability in egg and larval mortality is independent of density and results from a complex interaction of biological and physical factors operating on a range of temporal and spatial scales. While their importance may vary among systems, critical factors include the biomass, condition and age structure of the spawning stock, transport and retention of egg and larval, feeding conditions, growth and predation rates. The physical environment sets the stage for these processes and plays a large part in determining the outcome.
As part of the synthesis
phase of GLOBEC we propose to establish comparative biophysical coupled model
studies for transport and growth of larval and early juvenile fish in the two
marine ecosystems Georges Bank and the Norwegian shelf/Barents Sea (the
northern and southern extremes of the distribution of Atlantic cod; see Figs.
1, 2a,b). The study will be carried
out in collaboration with colleagues at the
Rationale for a comparative study. The synthesis phase of GLOBEC is presently underway in order to fully benefit from regional and national research efforts over the last ten years. Particularly, a comparative study between US and Norway GLOBEC programs (see Sundby, 2005 for a description of the latter) has great synergistic potential. There are several reasons for this:
gadoid stocks of the Georges Bank and the Barents Sea are among the most
explored in the
respect to marine climatology and physical setting,
biological setting in the two regions is different. The Arcto-Norwegian cod mature at an age
of about 6 years, while Georges Bank (GB) cod mature at age 2-3 years. Size at
maturity differs as well as the fecundity. The vertical distribution of fish eggs
and larvae differs considerable because of differences in egg buoyancy and
vertical mixing regimes. In Norwegian waters, C.
finmarchicus nauplii and copepodites dominate the diet
of cod larvae and early juveniles from first feeding to 2-3 month old pelagic
GB the prey species are more diverse and also comprise Pseudocalanus spp.,
Oithona spp., and Centropages spp., with Pseudocalanus spp.
comprising the bulk of the diet during the larval stage. Advection of C. finmarchicus
spite of the different physical and biological settings for the two regions,
the parameterizations of the basic biological processes for application in
individual-based models (IBMs) are similar, and we propose to continue to
develop IBMs to analyze the specific responses in the two regions. Currently, the Norwegian models have a
more realistic and detailed foraging component. The
Hypotheses for the NW
NW Atlantic/Georges Bank
(Atlantic cod/haddock). The
Gulf of Maine/Georges Bank region has a pronounced seasonal cycle of weather,
hydrography, and biology.
Peak spawning of cod and haddock on
The biomass concentration of
copeopd prey on
Environmental factors favoring high prey concentrations on GB in March and April favor rapid larval growth, survival and eventual recruitment. On Browns Bank just to the north, recruitment of haddock was related to the timing of the spring phytoplankton bloom with an early bloom favoring high recruitment (Platt et al. 2003). During the GLOBEC Program, years of high growth and survival of cod and haddock larvae were associated with low salinity in surface waters and high zooplankton prey biomass (Buckley et al. submitted; Mountain et al., in prep). Moreover, interannual differences in prey abundance and larval growth and survival were seen early in the season and persisted through the period of larval drift, suggestive of processes operating on large temporal and spatial scales.
Durbin et al. (2003) outlined a process whereby large differences in the winter phytoplankton abundance in the GOM (
) can influence zooplankton abundance and productivity ultimately reaching GB. An unusual phytoplankton bloom occurred in winter 1999 in the GOM as well as on the Scotian Shelf (Platt et al. 2003). The initial spread of low salinity Scotian Shelf water across eastern GOM in early 1999 apparently resulted in a shallow, surface mixed layer that allowed a bloom to develop. The winter bloom also may have allowed an additional generation of Calanus prior to the normal March-April spring bloom. Zooplankton abundance in the GOM was an order of magnitude higher in 1999 than 2000 or in the previous GLOBEC years. The high abundance of zooplankton in the GOM during 1999 also was observed on GB where larval growth rate and prey biomass were highly correlated (Buckley and Durbin, 2005). Larval haddock survival and recruitment also were high in 1999 on the Scotian Shelf and GB. Gulfof Maine
· Hypothesis 1: strong and early influx of Scotian Shelf water to GOM leads to an early phytoplankton boom with increased zooplankton abundance downstream to Georges Bank resulting in increased larval cod/haddock growth.
Norwegian Shelf/Barents Sea
(Arcto-Norwegian - AN - cod). Spawning
of AN cod occurs in March and April along the mid to northern Norwegian shelf
and the eggs and larvae are carried north along the coast into the
Individual growth of AN cod
larvae and juveniles show functional relationships to biotic and abiotic
parameters (Vikebø et al. 2005). One of the most important effects is
through the influence of temperature on feeding intensity, metabolic rates and
thereby growth (Otterlei et al., 1999).
Similarly, light conditions and wind-induced turbulence are important
abiotic parameters in relation to behavioral responses affecting growth of
larvae (Sundby and Fossum, 1990; Suthers and Sundby, 1996; Fiksen et al. 1998). Indirect effects of variables such as
temperature through lower trophic levels are similarly important, particularly,
the production at lower trophic levels. The copepod species, C.
finmarchicus, is the dominant meso-zooplankton in the Subarctic Gyre
of the northern
· Hypothesis 2: Advection of warm, zooplankton-rich Atlantic water from the
Norwegian Seaonto the shelves ( Barents Sea) results in increased larval cod growth and survival.
The proposed hypotheses in the two ecosystems will help guide our research activities. They integrate local biotic and abiotic effects, while introducing important effects associated with upstream, basin-scale/climate effects resulting from interactions of the biology and physics of the open ocean and shelf regions that we have not been able to explicitly consider to date.
Prior Support and Background
The PIs in the present proposal participated in modeling, field and synthesis efforts in the US GLOBEC Georges Bank Program from 1994-2002. [F. Werner (with J. Runge, co-PI) was funded by NSF grant OCE-9806565: Productivity of Calanus finmarchicus and fluctuations in growth and survival of cod and haddock larvae on
Georges Bank: A synthesis of observations and modeling. Oct 1998-Sept 2001; $264,917. Publications are asterisked in the Reference list.] These studies dealt with circulation modeling, larval fish and zooplankton population modeling, in a spatially-explicit, coupled physical-biological setting. The PIs’ modeling activities focused on historical data and its scientific interpretation. Among other aspects, modeling studies established climatological mean physical fields from the Nova Scotian shelf to southern New Englandand perturbations to them; examined the impacts of these on the life history of specific planktonic species and their interactions; and related these to observations of abundance and distribution of the target species. Herein, we will build on these tools and their extensions. During Phases I-III of the GLOBEC program, we gained considerable knowledge of the physical and biological processes controlling the retention and loss and feeding and growth of cod and haddock larvae on the bank, which we briefly summarize as follows:
zooplankton variability. There was considerable
interannual variability in mean zooplankton abundance on
include increased grazing pressure.
We showed that for late
winter/early spring seasons,
Egg and larval mortality. Mountain et al. (2003; submitted) found that egg and larval mortality varied among years and was related to salinity (Figs. 5a and b); note that 1995 had the highest salinity and the highest mortality. Lapolla and Buckley (2005) used counts of otolith daily rings to determine the hatch date distributions of young-of-year juvenile haddock collected in the fall on GBank. Recruitment appeared continuous rather than episodic. A comparison with hatch date distributions from late stage eggs (Mountain et al. 2003) suggested that individuals hatched early in the season had a strong survival advantage over those spawned later in the season.
growth rates. Recent growth rates estimated from RNA/DNA and water
temperature showed pronounced trends with larval size (Fig. 6a) and Julian Day
(JD). The JD effect was
attributed to an interaction of photoperiod, water temperature and prey
availability (Fig. 6b), all of which increase during the larval period on GB (Buckley
et al. submitted). Empirical
models incorporating larval size and photoperiod explained 48% and 61% of the
observed variability in individual growth rates of cod and haddock
respectively. We used residual
growth rates (the difference between observed growth rate and that predicted
using these models) to examine the affects of environmental variability on
larval growth. Strong seasonal and
interannual differences were seen in larval growth (Figs. 7a and b). Growth rates and residual growth rates
were low in 1995 and higher in 1997-1999.
Most potential prey taxa were more abundant during 1997-1999 than in
1995. Growth rate and residual
growth rate were related to prey biomass concentration (Buckley and Durbin,
2004). Both the prey biomass
concentration and residual growth rate were inversely related to salinity (Figs.
8a and b). These studies
established that on
Larval Fish Trophodynamic Model. To evaluate the variability on the larval cod and haddock feeding environment, we developed an individual based model (IBM) described in Werner et al. (1995, 1996, 2001), and Lough et al. (2005). The core of this model is the standard bioenergetic supply-demand function, in which growth is represented as the difference between the amount of food absorbed by a larva and the metabolic costs of its daily activities (Laurence 1985). The formulation includes: (i) variable composition of prey fields including 13 stages of 4 prey types (C. finmarchicus, Pseudocalanus, Centropages and Oithona) as well as protozoans during yolk-sac stages (Quinlan et al. 1997); (ii) effect of turbulence, swimming behavior and satiation on encounters and ingestion of larval fish and their prey (e.g., Dower et al. 1997, Lough et al. 1997a, MacKenzie and Kiørboe, 1995 and 2000); (iii) light limitation on ingestion rates at low and at high light intensities (Huse 1994, and Gallager et al. 1995, 1996) and (iv) effects of temperature on metabolic costs, ingestion rates and growth (e.g., Laurence, 1978; Buckley et al. 2000).
Feeding and growth of larval cod. Our studies on the larval feeding environment have shown that during late winter/early spring, effects of turbulence on contact rates and post-encounter capture can be significant (Werner et al. 1996, 2001). For late spring conditions Lough et al. (1997a) found that prey aggregations near the pycnocline are sufficient for rapid growth without turbulence-enhanced contact rates. Werner et al. (2000), using broadscale data for key zooplankton species: Calanus, Pseudocalanus, and Oithona (the 3 principal prey species found in the stomachs of cod larvae for this time period; Lough, unpubl. data), found that for March-April 1995 Pseudocalanus provide the highest growth rates for 6-10 mm larval cod but that the inclusion of Oithona and Calanus was necessary for model larval fish growth rates to be comparable to those in the field. Lough et al. (2005) found that the differences in observed growth of cod larvae in 1993 versus 1994 were best explained using a bioenergetics model when the prey was restricted to Pseudocalanus spp. and the larvae followed their weighted mean depth. Growth rate and residual growth rate of cod and haddock larvae were directly related to the biomass concentration of Pseudocalanus spp., but not Calanus (Buckley and Durbin, 2004).
In this proposal, we address the overall hypothesis that recruitment
variability in cod and haddock populations depends on the growth and survival through
the larval stage, constrained by density-dependant
mortality during the juvenile stage. Our approach is to combine detailed mechanistic
models of transport, growth and survival of early life stages of cod in the NW
Atlantic and the
Physical fields. Develop realistic physical conditions hindcast for selected years at basin scales run from a common model and forced by a common set of variables, with increased resolution within regional domains to assess the role of local versus far field forcing in the variability of the advection mechanisms affecting the loss or retention on Georges Bank and Norwegian Shelf, the variability of the temperature and feeding environment, and ultimately the variability in the observed larval fish growth.
Basin scale physical and lower trophic
models are available through our collaboration with our colleagues at IMR. For the physical fields (e.g., currents,
temperature, salinity) we will use the the Regional Ocean Modeling System (ROMS
version 2.1) implemented by P. Budgell (IMR,
The domain of the present (IMR) model’s
implementation extends from 20°S in the South Atlantic northwards through
The forcing fields are the daily mean fluxes from the NCEP/NCAR Reanalysis (Kalnay et al., 1996) with adjustments to the applied fluxes dependent upon model surface conditions (Bentsen and Drange 2000). Precipitation is obtained from the NCEP/NCAR Reanalysis and evaporation was derived from the latent heat flux values. Freshwater runoff was obtained by combining the NCEP/NCAR Reanalysis accumulated surface runoff over land areas with the Total Runoff Integrated Pathways (TRIP) discretization and network routing of Oki and Sud (1998).
Tides were not included in this simulation, but they will in the next
version, which will also be global. The global model will still be focused on
A lower trophic model is run off-line
(i.e., using the physical flow fields after these have been computed; M.
Skogen, pers. comm.) to estimate nutrients, diatom and flagellate production (i.e.,
large and small phytoplankton) over the model domain. Zooplankton is not computed, but
is specified externally as a grazing term on the primary producers. Model results for selected years, based
Zooplankton (larval prey fields). Data collected during selected years will be used to examine the space-time variability of the larval fish feeding environment. The distribution and evolution of the zooplankton fields will be specified based on the observed structures. If available, we will include evolving prey (zooplankton) fields computed by Davis et al. in their proposal (letter attached) for the NW Atlantic, with similar prey fields provided for the Norwegian System computed by G. Huse, Ø. Fiksen at U of Bergen and their colleagues at the IMR.
Larval fish individual based trophodynamic models – estimates of growth and survival. We will enhance the larval fish trophodynamic models to include revised formulations for foraging and prey selection, behavior, recent findings of larval sensitivity to light intensity and wavelength, and a re-examination of the haddock bioenergetics. The larval bioenergetics and foraging formulations used in most earlier cod and haddock IBMs were largely based on work by Laurence and Beyer (Laurence 1985). We recently completed a major revision of the cod model (Lough et al. 2005) that incorporated light limitation, recent data on cod metabolism (i.e., Finn et al. 2002; Peck et al. submitted), and implemented a more realistic mechanism to model temperature dependence. We now track stomach contents and larvae feed only when space is available in the gut. Digestion is a linear function of temperature. Modeled growth rates better match those observed in the field and the effect of temperature is dynamic, yielding temperature maxima for growth that are dependent on prey availability (lower at low prey abundance and higher at high prey abundance).
Fiksen and MacKenzie (2002)
published a process-based model for foraging in larval cod that incorporated
process important to prey selection and environmental regulation of
feeding. We will combine our
bioenergetics model with the foraging component from Fiksen and MacKenzie
(2002) to examine the apparent differences in diet composition and preference
of cod on GB and
Predation will be implemented in the model as a function of larval size and water temperature. This base mortality rate will be scaled depending upon the distribution and abundance of potential predators in different zones as the eggs/larvae advect through them or as they develop in the waters along side them. Because we are interested in understanding how and where population regulation may take place, we will require spatially explicit predation mortality fields for larval fish. The broadscale surveys (Madin et al. 1996; Sullivan and Meise 1996) will provide 3D maps of predator abundance, along with maps of predation rate (% removed day-1 m-2) on particular prey types; data on vertebrate predators is available from the Coastal Ocean Program predation studies, e.g., Garrison et al. (2000 and 2002).
Recruitment models – full life-cycle representations. We will extend the results of the preceding objectives by linking them to the dynamics of adult stocks to complete the life-cycle representation of the target fish species. First, we will develop recruitment models for cod and haddock on GB with multiplicative and interactive environmental terms using the state-space approach to analysis of multiple time series. Second, we will involve the development of forward projection models for cod and haddock in which the specific mechanisms explored in this proposal affecting growth and survival will be incorporated into stochastic simulations. Simple hybrid recruitment models similar to those developed by Bailey et al. (2003) for walleye pollock and Pacific hake will also be developed and evaluated.
Selection of Years of Emphasis
NW Atlantic/Georges Bank. During the GLOBEC field program, 1995-1999, the largest year class of haddock was observed (1998) on Georges Bank in the past 20 years, although still larger year classes were produced later. The 1999 year class of haddock was also unusually high on the southern Scotian Shelf (Platt et al. 2003). Cod failed to produce a strong year class during the 90s despite management measures taken to increase spawning stock biomass of both cod and haddock. SSB on GB fell to minimum levels in 1993 and 1995 in haddock and cod respectively, before gradually increasing. Based on the monthly GLOBEC broadscale surveys of fish eggs and larvae, the high recruitment of haddock was attributed to higher survival of larvae, approximately the first month from hatching (Mountain et al., in prep). The greater survival of larvae may be related to the better quality and larger eggs produced by the older spawning year classes entering the population. First-time spawners are known to have high egg mortality, with >50% loss due to inferior eggs.
Environmental conditions during the pelagic period also may have contributed to the greater growth and survival of larvae in 1998 and 1999. What is the role of environmental factors versus spawning stock size/composition that led to the greater survival and recruitment of haddock? Why did haddock larvae survive better than cod larvae when they generally spawn in the same area and time and feed on similar prey? The spawning curve of haddock over the GLOBEC field years extends from January to June with peak spawning occurring between February and May. Peak spawning can be very sharp, limited to essentially one month and can occur any month during February-April. In 1998, peak spawning was broader than in other years, extending over the February-April period. Egg production was not greater for haddock in 1998, but survival through the egg and larval stage was significantly greater. In 1999, haddock egg production was higher than 1998 but egg survival was lower. For cod, spawning can occur between December and June with peak spawning occurring in February or March. The 1997 season had the highest abundance of eggs produced but the lowest egg survival.
Our strategy is to look specifically at three candidate years from the GLOBEC time series, 1995, 1998, and 1999, where there is sufficient data to compare and contrast haddock and cod growth and survival during years of high and low survival/recruitment, and where environmental conditions also provide the greatest contrast of physical conditions/processes in order for us to better understand the possible recruitment mechanisms. Characteristics of these years include:
· The 1995 season had low haddock and cod survival during a warm year where Scotian Shelf intrusion was observed in March and the shelf/slope front moved on-bank to the 60-m isobath during May. What effect did these large physical events have on the residing populations?
· The 1998 field season recorded minimum salinity due to the intrusion of Labrador Slope Water which was observed in the Northeast Channel and eventually came on to Georges Bank from the Gulf of Maine. High Calanus abundance also was noted that year from the broadscale surveys, as well as high haddock survival.
· The 1999 season was warmer, more stratified, and an earlier Calanus bloom was noted which may have led to a 3rd generation (Durbin et al. 2003). Contrasting 1999 with 1998, where haddock survival also was high, will serve to check the hypotheses relating recruitment to secondary production.
Synoptic variation of physical processes for these periods will be provided from the basin scale model. We will also examine variation in life stages of Pseudocalanus, Calanus, and to the extent possible, Oithona that serve as prey for the development stages of fish larvae found on the bank.
Norwegian Shelf. Our Norwegian colleagues have examined several years in their studies. One particular case is their investigation of the direct effects of transport and temperature on distribution and growth of AN cod in 1985 and 1986. These years were chosen because of their differences both in distribution and the individual weight of the O-group cod. Also, the total abundance of O-group cod was higher in 1985 than in 1986, and the center of biomass was displaced further west. The distribution in 1985 covered a larger area than in 1986 and the average length of O-group cod, and therefore the weight (Ellertsen et al., 1989), were significantly lower in 1986 than in 1985 (Ottersen and Loeng, 2000). They used a circulation model combined with individual based larval and juvenile models. The models simulated transport of larvae and pelagic juveniles for the two years (Fig. 11), while keeping record of the individual temperature histories enabling calculation of temperature dependent growth, from spawning until settlement at the nursery grounds.
The present simulations reproduced the variations in horizontal concentration of these two years, though differences in average weights are not: the year 1985 is known to have significant higher average weight (and length) of O-group cod than 1986 (Ottersen and Loeng, 2000). The main reason this is not captured in the simulations is thought to be because the lateral boundary conditions are taken from climatology and do not contain inter-annual variations; their inclusion would also enable estimation of year-to-year variations in fish recruitment (Sundby et al. 1989) and juvenile condition (Ottersen and Loeng, 2000). This topic is planned for continued study. A more sophisticated individual based model where growth depends on food availability, turbulence and light, in addition to transport and temperature, is also envisioned. This will enable us to assess the trade-off between feeding conditions and temperature.
The work to be carried out by
the US PIs and our Norwegian colleagues are fundamentally the same, with variations
in emphasis of the specific domain, and particular field years. Focusing on the activities of the
Task 1 - Implement the model solutions for higher resolution NW Atlantic simulations (Werner and Mountain). Initially we will focus on the 1990’s GLOBEC field years and compute the physical fields for 1995-2000. The basin-scale ROMS will be run with focus on the domain approximately corresponding to Fig. (10). We will also compute the lower trophic level response offline and store the data for inclusion in the analysis of the interannual variability. Werner will spend three months (in the Spring of ‘06) at IMR during his Research and Study Leave from UNC to complete this work in collaboration with P. Budgell. Comparison of observed and modeled NW Atlantic hydrographic and flow fields will be an essential part of this effort.
Task 2 – Compile prey fields from observations (Durbin, Lough and Buckley). Existing data in the URI Oracle database will be augmented with additional data from the process studies. Mapping of the staged-data onto the model domain will proceed as in previous efforts (e.g., Runge et al. 2000). The zooplankton data will be used in estimating (using IBMs) the growth of larval cod for the GLOBEC years.
Task 3 – Develop and Implement
“Holistic” models (Buckley, Lough , Mountain and Werner). In modeling larval haddock growth and survival Heath
and Gallego (1997, 1998) described a “holistic approach” in which
temperature serves as a proxy for a wide range of factors which presumably
affect the growth of larval fish in the sea. They coupled a hydrodynamic
particle-tracking scheme with a logistic growth model incorporating a parabolic
temperature term derived by Campana and Hurley (1989) for haddock larvae on
Browns Bank in the NW Atlantic.
They used this model to investigate the effects of the physical
environment on recruitment process of
Task 4 – Augment and Implement larval fish bioenergetics models (Buckley, Lough and Werner). There are several published bioenergetics models for larval cod growth. Most are based on the early work by Beyer and Laurence (Laurence 1985). Results from our latest version (Lough et al. 2005) are in good agreement with observed growth on GB, capturing the dynamic interaction between temperature and prey availability. We will determine if our GB model reproduces observed growth in the Norwegian Sea system; augment and couple our model to the Fiksen and MacKenzie (2002) process-based model for foraging; and re-evaluate our treatment of search volume and turbulence based on recent findings (Galbraith et al. 2004). Starvation mortality will be implemented using a death barrier. Predation mortality will be implemented in the model using a specified probability of death based on larval size and water temperature (Winemiller and Rose, 1993). A scalar will be included to account for spatial and temporal differences in predator abundance.
Task 5 – Compile a life table for cod and
haddock (Lough, Buckley and Mountain).
Task 6 – Development of proxies for retention, growth and survival (all PIs). Extension of our recruitment modeling to data-poor years requires the development of indices or proxies for key processes. Rather than relying on time-series data from a few fixed sites, we will use indices based on synoptic variation in the basin-scale and regional model runs. For example, egg and larval retention may be indexed as the deviation from the average retention time of water on the bank between February and May. Surface salinity in the GOM/GB region appears to be a useful proxy for larval feeding conditions, growth, and survival. Stratification and the timing and intensity of the winter/spring diatom bloom (from the NPZ model) will also be evaluated as proxy for feeding conditions and growth. We will evaluate these proxies against the GLOBEC and MARMAP time series. Similar proxies will be developed for the NSS, where temperature is likely to be a useful proxy for feeding conditions, growth, and survival. Mortality during the juvenile stage is most likely dependent on the abundance of vertebrate predators including older cod and haddock. We will use assessment data (NEFSC, 2002) and the NEFSC food habits database to scale juvenile mortality within the range generated from the life table.
Task 7 – Hybrid recruitment models (Fogarty,
Buckley and Lough). The models developed in the preceding sections focus
on key elements of the recruitment process growth and survival during the
early life stages. We will extend these results, and the associated
proxies, by linking them to the dynamics of the adult stock to complete the life-cycle
representation. The classical stock-recruitment relationships entail
consideration of population egg production (knowing the spawning stock biomass)
and factors affecting pre-recruit survival, but typically do not explicitly
include the role of exogenous forcing factors. Extended recruitment
models that do include environmental effects have been developed (e.g., Bailey
et al. 2003). These models typically represent environmental factors as
multiplicative effects. A recent application of this general approach to
cod populations throughout the
Extensive time series of cod and haddock recruitment and adult population sizes are available for
Using the life table described above, we will parameterize simple hybrid recruitment models (Bailey et al. 2003) for cod and haddock. Stochastic simulations will be completed in which the range of mortality for successive early-life stages is adjusted according to environmental conditions, and juvenile mortality is a function of density and predator abundance. This approach does not entail estimation or fitting of time series data to determine the role of environmental forcing but rather uses independent information to set probable ranges.
The model-based synthesis proposed directly addresses GLOBEC4b program goals
to synthesize the research findings of the
Continental shelf and marginal sea ecosystems – such as the GBank and
The proposed study provides an inter-comparison of results from the NWA/Georges
Bank with other
Our study can be integrated with other synthesis efforts that propose to generate the evolving zooplankton fields on GBank during the GLOBEC field years (Davis et al.), consider the basin-scale variation of Calanus (McGillicuddy and Haidvogel et al.), and study NAO-associated changes in ocean circulation patterns and impact on marine ecosystems (Greene et al.).