Potential climate change impacts on Atlantic cod (Gadus morhua) off the northeastern USA
- First Online:
- Cite this article as:
- Fogarty, M., Incze, L., Hayhoe, K. et al. Mitig Adapt Strateg Glob Change (2008) 13: 453. doi:10.1007/s11027-007-9131-4
- 309 Views
We examined the potential impacts of future climate change on the distribution and production of Atlantic cod (Gadus morhua) on the northeastern USA’s continental shelf. We began by examining the response of cod to bottom water temperature changes observed over the past four decades using fishery-independent resource survey data. After accounting for the overall decline in cod during this period, we show that the probability of catching cod at specified locations decreased markedly with increasing bottom temperature. Our analysis of future changes in water temperature was based on output from three coupled atmosphere–ocean general circulation models under high and low CO2 emissions. An increase of <1.5°C is predicted for all sectors under the low emission scenario in spring and autumn by the end of this century. Under the high emission scenario, temperature increases range from ∼2°C in the north to >3.5°C in the Mid-Atlantic Bight. Under these conditions, cod appear vulnerable to a loss of thermal habitat on Georges Bank, with a substantial loss of thermal habitat farther south. We also examined temperature effects on cod recruitment and growth in one stock area, the Gulf of Maine, to explore potential implications for yield and resilience to fishing. Cod survival during the early life stages declined with increasing water temperatures, offsetting potential increases in growth with warmer temperatures and resulting in a predicted loss in yield and increased vulnerability to high fishing mortality rates. Substantial differential impacts under the low versus high emission scenarios are evident for cod off the northeastern USA.
KeywordsClimate change impactsGulf of MaineCodBottom water temperatures
Fishing has played a vital role in the cultural and economic fabric of the northeastern USA and the identity of many coastal communities is deeply tied to longstanding seafaring and fishing traditions. Combined dockside revenues from oceanic and estuarine commercial fisheries in the northeastern USA exceeded $1.2 billion in 2004. Hoagland et al. (2005) estimated the economic contribution of the seafood industry to the overall economy in the northeast USA to be $10.4 billion in 2003 as a result of direct and indirect effects of commercial fishing activity and satellite industries. Recreational fisheries also result in substantial economic benefits in the region derived from angler expenditures and employment opportunities (Steinback et al. 2004).
We explored the potential consequences of climate change on marine resources of the northeastern USA with particular emphasis on Atlantic cod (Gadus morhua). The demand for salt cod in Europe served as a catalyst for exploration and discovery and there is evidence that Basque fishermen discovered and occupied important cod fishing grounds in North America as early as the 15th century (Kurlansky 1997). Cod has been a mainstay of the commercial fishery in New England since the seventeenth century, and a carved representation of the ‘Sacred Cod’ has hung in the rotunda of the State House of Massachusetts since 1784, signaling the historical importance of this species to the Commonwealth.
Temperature affects virtually every aspect of the biology and ecology of cod, including its distribution (Dutil and Brander 2003; Drinkwater 2005); recruitment success, the number of young cod surviving to a specified age or size (Planque and Frédou 1999; Brander 2000; Drinkwater 2005); feeding (McKenzie 1934, 1938; Rose 2005); and individual growth (Brander 1995; Rätz and Lloret 2003). Cod populations are distributed throughout the North Atlantic in coastal and continental shelf waters characterized by seasonal bottom temperature regimes ranging from less than −1°C to over 20°C and annual mean temperatures from 2–12°C (Dutil and Brander 2003; Drinkwater 2005). Drinkwater (2005) used the 12°C threshold to assess potential loss of thermal habitat of cod throughout the North Atlantic. Increases in temperature have a positive effect on recruitment in arcto- boreal regions where temperatures are close to the lower thermal tolerance limit of cod, and negative effect on stocks inhabiting areas at the upper end of the temperature range (Planque and Frédou 1999; Drinkwater 2005). Stocks inhabiting regions characterized by intermediate temperature regimes showed no detectable effect of temperature.
In this paper we examine the response of cod to temperature changes observed on the Northeast Continental Shelf (NCS) over the last four decades to establish a baseline for assessing potential effects of future climate change on distribution and recruitment. This strategy of forecasting by historical analogy has been employed successfully for evaluating potential climate impacts on marine resources (Glantz 1990), including those off the northeastern USA (Murawski 1993). We then consider predicted regional changes in water temperature derived from three coupled atmosphere–ocean general circulation models under two CO2 emission scenarios and explore potential implications for cod off the northeastern USA. Finally, we develop a temperature-dependent production model for the Gulf of Maine cod to illustrate the interplay of harvesting and environmental change on stock dynamics.
2.1 Cod analysis
We first considered cod abundance and distribution patterns on the NCS as a function of water temperature, depth, and time (year). Standardized research vessel surveys have been conducted in this US region by the Northeast Fisheries Science Center (NEFSC) in the autumn since 1963 and in the spring since 1968 (Smith 2002). A stratified random sampling design is employed with strata defined by latitude and bathymetry. Approximately 350 stations are occupied during each seasonal survey from the Gulf of Maine to Cape Hatteras, off the coast of North Carolina. At each sampling location, the catch of each species is enumerated and weighed and samples are taken for length and age composition, incidence of disease, maturity, and diet composition. Surface and bottom water temperatures and bottom depth are recorded at each station.
2.1.2 Cod production model
2.2 Climate models and emission scenarios
2.2.1 Atmosphere–ocean general circulation models
Atmosphere–ocean general circulation model resolution for the Geophysical Fluid Dynamics (GFDL), Hadley (HadCM3) and Parallel Climate (PCM) models
2.5° × 3.75°
2° × 2.5°
T42 (∼2.8° × 2.8°)
1.125° × 1.125°
1° × 1° (Interpolated tri-polar)
1° × 1° (Interpolated tri-polar)
Top six layers (depth, m)
5, 15, 25, 35, 48, 67
5, 15, 25, 35, 45, 55
13, 38, 64, 90, 118, 147
In recognition of known biases in water temperature and salinity fields in the Northwest Atlantic in the current generation of AOGCMs (Dai et al. 2002), we focused strictly on the anomalies predicted by the models. We first adjusted the model outputs for surface water temperature against observed temperature fields derived from hydrographic observations during NEFSC surveys for five North American geographical sectors (Southern Mid-Atlantic Bight; Northern Mid-Atlantic Bight, Georges Bank; Western Gulf of Maine; and Eastern Gulf of Maine; Fig. 1). From the adjusted surface temperature projections, we estimated bottom temperatures using observed surface to bottom water temperature relationships in each area. Although observations are made throughout the year, the most intensive sampling is in spring and autumn in conjunction with the bottom trawl surveys. Predictive relationships were developed by regressing surface temperature on bottom temperature in the season of interest and then checking for improvement in fit by adding the surface temperature for the other season. The seasonal and region-specific linear fits were applied to model-simulated spring and fall surface temperature anomalies for each region (relative to the 1970–2000 mean temperature) in order to produce predicted bottom temperature anomalies covering the full period of model simulations, from 1900 through 2099 (Fogarty et al. 2007)
2.2.2 Emission scenarios
Future simulations are forced by the IPCC Special Report on Emission Scenarios (SRES; Nakienovi et al. 2000) higher (A1fi), mid-high (A2) and lower (B1) emissions scenarios. These scenarios describe internally consistent pathways of future societal development and corresponding greenhouse gas emissions, and cover a wide range of alternative futures based on projections of economic growth, technology, energy intensity, and population. The A1fi and B1 scenarios used in this study bracket the range of SRES scenarios, and can be thought of as lower and higher bounds that encompass most, though not all, potential non-intervention emissions futures. At the higher end, rapid introduction of new technologies, extensive economic globalization, and a fossil-intensive energy path causes A1fi greenhouse gas (GHG) emissions to grow steadily throughout the century. In the A1fi scenario, CO2 emissions climb throughout the century, reaching almost 30 Gt/year or six times the 1990 levels by 2100. The A2 emissions scenario also reach 30 Gt/year by 2100, although cumulative emissions over the century are slightly lower than the A1fi scenario. Emissions under the B1 scenario are even lower, based on a world that transitions relatively rapidly to service and information economies. Carbon dioxide emissions in the B1 scenario peak at just below 10 Gt/year – around two times the 1990 levels – at mid-century and decline slowly to below current-day levels. Together, these scenarios represent the range of IPCC non-intervention emissions futures with atmospheric CO2 concentrations reaching approximately double and triple pre-industrial levels by 2100: 550 ppm in the B1scenario and 970 ppm in A1fi. Projections for the A1fi and B1 scenarios were used for the GFDL and PCM models, but as HadCM3 did not save any ocean output for its A1fi simulation, we used A2 and B1 instead.
3.1 Cod baseline analyses
3.1.1 Cod distribution
Parameter estimates and model fit statistics for the logistic regression model relating the probability of obtaining cod in NEFSC autumn bottom trawl stations to temperature, depth, and year
P > χ2
3.1.2 Cod recruitment and growth
Model diagnostics for the Gulf of Maine recruitment model with and without consideration of temperature (SSB is Spawning Stock Biomass; AIC is Akaike Information Criterion)
Independent variables represented
Prob > F
SSB (standard Ricker model)
3.2 Bottom water temperature projections
Projected bottom water temperature anomalies (°C) by region and season for low and high emission scenarios for the period 2080–2084
W. Gulf of Maine
E. Gulf of Maine
3.3 Implications for future cod distribution and production
3.3.1 Cod distribution
If the bottom temperature projections under the high emission scenario are realized, it is likely that the 12°C mean annual bottom water temperature threshold for cod (Drinkwater 2005) will be exceeded in the Mid-Atlantic Bight and will approach the threshold level on Georges Bank regions by the end of this century. The mean bottom water temperature in the Gulf of Maine as a whole during the period 1978–2002 was 7.1°C, and if the bottom temperature anomaly projections for this region hold, the 12°C threshold will not be exceeded.
3.3.2 Cod production in the Gulf of Maine
Future climate predicted under increasing anthropogenic emissions of greenhouse gases has the potential to significantly alter the physical structure of the oceans, with direct implications for marine ecosystems and human societies (IPCC 2001; Stenseth et al. 2002; Sarmiento et al. 2004; Behrenfeld et al. 2006). On the NCS, potential changes include warmer water temperatures in all seasons, increased density stratification, changes in the slope water system affecting the continental shelf and Gulf of Maine, and attendant changes in nutrient and plankton dynamics and higher trophic levels (e.g., Greene and Pershing 2001; Pershing et al. 2005). These changes need to be considered in the context of impacts resulting from other human activities such as fishing, pollution, and habitat loss due to coastal development. Climate change can interact with other anthropogenic changes to alter the fundamental production characteristics of marine systems (Stenseth et al. 2002). Climate change will potentially exacerbate the stresses imposed by harvesting and other human activities. Considerable emphasis is now being placed on understanding the causes and consequences of climate change in temperate marine systems (Harvell et al. 2002; Helmuth et al. 2002; Barnett et al. 2005; Drinkwater 2005; Sutton and Hodson 2005; Altieri and Witman 2006) to prepare for anticipated alterations in ecosystem structure and function.
Substantial uncertainty exists in projections of the effects of climate change in the marine environment off the northeastern USA. Existing coupled AOGCMs do not resolve circulation and other important dynamical physical processes on continental shelves, and they exhibit known biases with respect to predicted water mass properties in the northwestern Atlantic Ocean. Nonetheless, these models provide important insights into possible directional changes in important hydrographic features, particularly relative temperature change. We have attempted to reduce the effects of bias by calibrating model output against direct observation for the period of instrumental record, and then considering only the relative changes in predicted properties. We used GCM model output from present-day and future runs to estimate temperature change over time, and then added the predicted difference to the current-day observations. The projections examined in this report must be viewed as provisional estimates until the output from AOGCMs is linked to appropriate shelf sea models. Previous attempts to examine the effects of global climate change on marine resources in this region have utilized projections of atmospheric temperatures (e.g., Drinkwater 2005) or sea surface temperature estimates derived from AOGCMs (e.g., WWF 2005). We have attempted to refine these approaches by also examining potential changes in bottom water characteristics.
A full assessment of potential climate impacts on the marine environment of continental shelves ultimately requires coupling of the results of Ocean General Circulation Models with higher-resolution hydrodynamic models in a fully nested structure. Such coupling properly accounts for responses of the shelf system to the combined effects of forcing at the boundaries with local forcing and adjustments. In the northeastern USA, such a nested approach is necessary to capture features such as topography and vertical mixing which affect circulation and the distributions of temperature, salinity, nutrients and plankton. This need is most pronounced in topographically complex areas such as the Gulf of Maine where a full accounting of factors affecting stratification and convective overturn will be required to understand the direction and magnitude of bottom temperature change.
Under the high emission scenario, cod appear vulnerable to a loss of thermal habitat in the Mid-Atlantic Bight. Thermal habitat losses also are expected on Georges Bank, historically the dominant production region for this species on the NCS. Within the Gulf of Maine, projected bottom temperature anomalies are lower than those for Georges Bank and south and the baseline level is lower. Accordingly, lesser impacts on thermal habitat are expected in this stock area, although some redistribution to cooler water areas within the Gulf can be anticipated, and is subject to other habitat constraints such as substrate preferences.
Consideration of potential impacts of temperature increases on yield and resilience to fishing for the Gulf of Maine indicates that although thermal habitat will generally remain viable, losses in stock productivity are predicted. A negative impact of temperature on recruitment is indicated for this stock, offsetting potential increases in individual growth with increasing temperature and resulting in a predicted loss in yield with increasing temperature. Temperature increases are also predicted to result in reduced resilience to fishing, highlighting the interplay between harvesting and environmental determinants of production. It should be noted that the empirical analysis of temperature effects on cod recruitment and growth presented here is set within a broader context of ecosystem change such as increased pelagic fish biomass which may also affect cod recruitment. It is therefore not possible to fully separate the effects of temperature and other ecosystem changes on cod productivity without further information derived from actively adaptive ecosystem-based management. The results presented here therefore are best considered a guide to the direction and potential magnitude of change in yield if other aspects of cod productivity remain relatively constant (e.g. continued availability of preferred prey resources etc.).
Because cod are comparatively well-studied and at the southern extent of their range in our study area, it is possible to make reasonable assessments of their likely response to temperature change. Marine systems and the life histories of marine organisms are complex, however, and we are at very early stages in our ability to predict climate change effects that integrate across all parts of the ecosystem. We have focused on temperature effects because of the substantial body of information indicating the central role of temperature on cod physiology and ecology. This general approach was adopted in climate impact assessments for cod in other regions (e.g. Clark et al. 2003; Drinkwater 2005). However, temperature is not the sole determinant of cod production. The effects of climate change on other aspects of ecosystem structure and function can be expected to exert additional impacts on cod dynamics and a full assessment of the impact on cod and other resource species will require consideration of overall changes in system productivity under global climate change.
We thank Nick Wolff, Michelle Traver and Loretta O’Brien for data and analyses and Adrienne Adamek for GIS support. The support and encouragement of Erika Spanger-Siegfried, the Union of Concerned Scientists, and the Census of Marine Life are also gratefully acknowledged.