ICES Journal of Marine Science: Journal du Conseil

Drinkwater, Kenneth, F;Mueter, Franz, J;Saitoh,, Sei-Ichi

doi: 10.1093/icesjms/fsy179pmid: N/A

Abstract An international Open Science Meeting entitled Moving in, out, and across the Subarctic and Arctic marine ecosystems: shifting boundaries of water, ice, flora, fauna, people, and institutions, took place 11–15 June 2017 in Tromsø, Norway. Organized by the Ecosystem Studies of Subarctic and Arctic Seas programme and cosponsored by the International Council for the Exploration of the Sea and the North Pacific Marine Science Organization, the primary aim of the meeting was to examine past, present, and future ecosystem responses to climate variability and ocean acidification (OA) and their effect on fishing communities, the fishing industry and fisheries management in the northern Pacific and Atlantic oceans and the Arctic. This symposium issue contains several papers from the meeting covering topics from climate and OA, ecosystem responses to environmental change, and fisheries management including: a synthesis of the ecosystem responses to the AMO-linked cold period of the 1970s and 1980s; a novel approach to understand responses to OA in northern climes using natural carbonate chemistry gradients, such as CO2 vents, methane cold seeps, and upwelling area; the possibility that warm temperatures are allowing two generations of Calanus finmarchicus per year to be produced; a new hypothesis suggesting that in areas where sea ice disappears there could be an increase of fish species with swim bladders; results from laboratory experiments on the effects of temperature and food on Arctic and boreal fish larvae; the application of ecosystem-based management in northern regions; and a description of the United States National Oceanic and Atmospheric Administration approach to marine conservation and how it affects fish populations and fisheries. Introduction The 3rd international Open Science Meeting (OSM) organized by the Ecosystem Studies of Subarctic and Arctic Seas (ESSAS) programme and cosponsored by the International Council for the Exploration of the Sea and the North Pacific Marine Science Organization, was held in Tromsø, Norway, on 11–15 June 2017. ESSAS is a regional programme of the Global Change Project IMBeR (Integrated Marine Biosphere Research), which is part of Future Earth. ESSAS was originally established under GLOBEC in 2005 as the Ecosystem Studies of SubArctic Seas to investigate how climate change currently affects, and will affect in the future, marine ecosystems of the SubArctic (Hunt and Drinkwater, 2005a, b). To kick off the program, ESSAS held its first OSM on the “Effects of climate variability on subArctic marine Ecosystems” in Victoria, Canada in May 2005 (Hunt et al., 2007). Its second OSM entitled “Comparative studies of climate effects on polar and subpolar ocean ecosystems: progress in observation and prediction,” was held in Seattle, USA, in May 2011 (Drinkwater et al., 2012; Mueter et al., 2012; Curchitser et al., 2015). By 2011, ESSAS had added the Arctic to its geographic sphere of research, especially focusing upon the interactions between the Arctic and Subarctic. To reflect this increasing work in the Arctic, ESSAS incorporated the word Arctic into its name in 2015 but retained the same acronym. The Arctic work has focused on Arctic gadoids (Mueter et al., 2016) and comparisons among seas within the Arctic (Hunt et al., 2013) and between the Arctic and the Antarctic (McBride et al., 2014; Hunt et al., 2016; Murphy et al., 2016). Increases in the air and sea temperatures in the Arctic during the last couple of decades have resulted in a significant decline in summer sea-ice cover in the Arctic Ocean, changes in the timing of ice retreat in the spring and of ice formation in the fall, decreases in the thickness of the ice and the loss of multiyear ice (IPCC, 2013; Screen, 2014). In the Subarctic Seas, there have also been large changes in sea temperatures but with important spatial variability. For example, generally warm conditions have been observed in the Barents and Nordic Seas of the North Atlantic (Smedsrud et al., 2013) while in the Bering Sea, temperature conditions have varied between warm and cold periods with corresponding decreases and increases in winter sea-ice cover, respectively (Stabeno et al., 2012). These changes in the water and ice properties have resulted in changes in the biogeochemistry and ecology of these regions (Hunt et al., 2011; Johannesen et al., 2012) including increased ocean acidification (OA; Browman, 2017) and the expansion northward of many species of plankton and fish (Sigurjónsson, 2016). Growth rates, recruitment levels, and phenology are also changing, resulting in increased abundances of some species and decreases in others. Changes in distribution and abundance of fish populations have resulted in changes in fisheries. For example, in some areas expanding populations have resulted in the development of new fisheries or the expansion of existing fisheries, while the loss or contraction of traditionally harvested stocks in other areas has caused those fisheries to disappear. This has resulted in difficulties with fisheries management based on historical fishing rights, e.g. Atlantic mackerel in the North Atlantic (Hannesson, 2016). These are all issues of concern, especially to northern nations. To address some of these issues, the 3rd ESSAS OSM was entitled Moving in, out and across the Subarctic and Arctic marine ecosystems: shifting boundaries of water, ice, flora, fauna, people, and institutions. Its main objectives were (i) to document changes that have occurred in Subarctic and Arctic marine ecosystems during the past century and the processes that led to these changes, especially those related to climate including anthropogenic climate change and OA; (ii) to compare and contrast the changes and processes in the North Atlantic, North Pacific, and Arctic; (iii) to place what is happening today in a longer-term perspective by examining the paleo-ecology of ecosystems and people in Subarctic and Arctic regions related to changing temperature and sea-ice conditions over time scales of millennia to centuries; (iv) to discuss how future changes are likely to further affect these ecosystems; (v) to determine how humans who depend upon these ecosystems will cope with the expected changes in the goods and services they derive from these ecosystems including the opportunities for commercial fishing; and (vi) to study the consequences of economic and societal pressures to coastal communities and nations as a result of these ecosystem changes. The first day of the OSM included a series of four topical workshops on: Paleo-Ecology of Subarctic and Arctic Seas (PESAS); Climate change impacts on nearshore fish habitats in the Arctic; Using natural analogues to investigate the effects of climate change and OA on northern ecosystems; and Arctic and Subarctic climate change impacts: a transdisciplinary perspective (see Supplementary Material). These were followed by nine theme sessions through the week, including: PESAS; Advection and mixing and their ecosystem impacts; Timing/phenology and match–mismatch: are they critical issues?; Shifting habitats, persistent hot spots; Future Subarctic and Arctic marine ecosystems under climate change; Multiple stressors; Ocean acidification; Science, Policy and Management; and a General Open Session (see Supplementary Material for the program). A total of 187 scientists from 17 countries attended the OSM. Highlights of the articles appearing in this symposium issue The present issue includes 11 papers from the OSM on a variety of topics from climate and OA to fish, fisheries and fisheries management. One of the major climate indices in the Atlantic Ocean is the Atlantic Multidecadal Oscillation or AMO, which has a periodicity of ∼60–80 years. It is defined as the detrended North Atlantic sea surface temperature anomalies from the equator to 60°–70°N. AMO-like variability also extends farther north into the Barents Sea and the Arctic. Several studies have documented the ecosystem responses to the mid-20th Century warm period in the North Atlantic associated with the AMO and the recent warming. Drinkwater and Kristiansen (2018) provide a synthesis of the ecosystem responses to the AMO-linked cold period of the 1970s and 1980s following the rapid cooling in the 1960s. This and other cold periods have received much less attention in the scientific literature. During this period, below average air and sea temperatures, expanded sea-ice cover and reduced Atlantic inflow into the Northeast Atlantic Ocean led to decreased primary production, a general southward expansion of arctic and boreal zooplankton and fish species, and a southward retreat of temperate species. The Atlantic cod fishery off Greenland and Labrador/northern Newfoundland and the Norwegian spring-spawning herring off Iceland and Norway collapsed, driven in part by climate-induced declines in growth rates and recruitment. However, intense fishing also played a role in the collapse of these highly valued fish stocks. At the extreme southerly limits of Atlantic cod, such as the North Sea, this species experienced the opposite response as the cool conditions led to improved growth rates and higher recruitment. Distributional shifts and changed abundances also occurred for benthic species and seaweeds. Following the cold period, as the temperatures warmed in the 1990s and 2000s, the ecosystem mainly returned to conditions like those in the warm mid-20th Century. However, this was not true for some species such as Atlantic cod off West Greenland and Labrador/northern Newfoundland, which never recovered. The authors conclude that the primary mechanism through which temperature acts on fish is through its influence on food availability. Understanding the ecosystem response to the AMO variability that is expected to continue into the future, together with anthropogenic climate change, will allow us to anticipate what changes might occur during any prolonged future cooling period and hopefully lead to better management practices. In addition to climate change, acidification of the world’s oceans is occurring at a rapid rate, with some of the largest changes in the Arctic and other cold water regions. This is expected to have important effects on the physiology of many species and may change the dynamics of some populations and the function and structure of some ecosystems. Our understanding of the effects of OA on Arctic and subArctic ecosystems is limited and has until now relied on short-term single-species laboratory studies. Rastrick et al. (2018) propose a novel approach to understand potential responses to OA in northern climes that has not been undertaken in these regions. They review the use of natural carbonate chemistry gradients in tropical and temperate regions, such as around hydrothermal vents, to learn about long-term acclimation and adaptation to elevated levels of pCO2. The authors suggest future OA-focused field studies and monitoring of organisms around CO2 vents, methane cold seeps, estuaries, upwelling areas and fronts in the Arctic and subArctic that contain gradients of pH, carbonate saturation state, and alkalinity. These, in combination with experimental in situ and laboratory studies, would lead to improved predictions of OA impacts on high latitude species and ecosystems. Aarflot et al. (2018) examine the variability of mesozooplankton in the Barents Sea and its relation to environmental conditions from data collected over 30 years. They find that 80% of the variation is from three Calanus species (C. finmarchicus, C. glacialis, and C. hyperboreus). Whereas all three species co-occur to some degree, C. finmarchicus dominates in the Atlantic waters and C. glacialis in the Arctic waters. Calanus hyperboreus is the least abundant of the three species and has the lowest biomass despite a much larger body size per individual. The biomass of the Arctic C. glacialis has been decreasing over the last two decades or more while C. finmarchicus has been increasing. The authors suggest that these changes are related to warming ocean temperatures and provide additional evidence of the borealization of the zooplankton community in the Barents Sea. They further speculate that the large increase in C. finmarchicus may be because the recent very warm temperatures are allowing two generations per year to be produced, as opposed to the one generation per year in earlier, cooler years. The authors also speculate that the increase in the abundance of the smaller size C. finmarchicus may be detrimental for some higher trophic levels, e.g. fish, marine mammals and seabirds, due to less efficient energy transfer. Skogen et al. (2018) investigate the possible future primary productivity in the Nordic and Barents seas through comparing the results from a global climate model (Norwegian Earth System Model) with that from a higher resolution regional model (NORWECOM.E2E) over the period 2006–2070. The regional model is forced by downscaled physics from the global model under RCP4.5. The Gross Primary Production (GPP) is significantly higher in the regional model as the global model predicts much higher sea-ice concentrations, which reduces light levels and delays the spring bloom by 1–2 months, hence lowering the GPP estimates to below observed levels. The lower GPP in the global model also results in less utilization of nutrients as not all surface nutrients were used up during the production season. Relative to climatology, the global model has a cold (in summer) and saline bias owing to poorly resolved physical processes and oversimplified ecosystem parameterization. Through downscaling, the regional model is, to some extent, able to alleviate the bias in the physical fields, and the timing of the spring bloom is close to observations. However, the summer nutrient minimum is one month earlier than observed. There is no trend in future primary production in either model and the trends in modelled pH and Aragonite are the same in both models. The largest discrepancy in the future projection is in the development of the CO2 uptake, where the regional model suggests a slightly reduced uptake in the future. On the basis of comparisons with observations, the regional model outperforms the global model. Kaartvedt and Titelman (2018) discuss mechanisms related to the variability in the geographical distribution of fish and plankton, including one that has been seldom raised. Fish possessing swimbladders need to reach the surface to take in air, which allows them to control their buoyancy. Since sea-ice coverage limits access to the surface, the authors hypothesize that present and projected continuing reduction in ice coverage might lead to the northward expansion of such fish species and would also impact their zooplankton prey. Another mechanism the authors discuss is the effects of the extreme high-latitude photoperiod. Noting the low abundance of mesopelagic fish in the Arctic Ocean, they suggest that this might be because of poor feeding conditions during winter darkness and light summer nights. If light levels are indeed the main limitation on determining geographical distribution, this would suggest that warming temperatures under climate change may not have any effect on mesopelagic fish in the Arctic. However, if temperatures control their geographical boundaries, an invasion of mesopelagics from the south into the Arctic under warmer conditions may reduce key Arctic copepods through increases in predation rates. Resolving the main mechanism (light vs. temperature) producing geographical extensions or shifts is, therefore, vital to improving projections of future biogeographic boundary changes. Invasions of boreal fish species into the Arctic are projected to occur under increasing sea temperatures and declining sea ice. There are educated guesses on the future of these invasive species as well as resident Arctic fish, but these are subject to large uncertainties due to a general lack of information on issues such as their thermal tolerance and ability to cope with changing trophic interactions. To address such issues, a series of three papers based on experimental laboratory studies of eggs and larvae compares the responses to environmental variability of an Arctic gadid (Arctic cod, Boreogadus saida) and a boreal gadid (walleye pollock, Gadus chalcogrammus). Koenker et al. (2018a) investigate the influence of temperature and food on the energetic condition of the larvae of the two species that is closely associated with mortality rates and, therefore, provides an indicator of overall well-being or fitness of the fish. The authors find that the effect of both temperature and food varies with species and ontogenetically. Condition in first-feeding Arctic cod larvae peaks at colder temperatures (2–5°C) than for pollock (5–12°C). At later larval stages, peak condition for Arctic cod occurs at warmer temperatures (7°C), while for pollock the thermal optimum is not stage dependent. Arctic cod are more sensitive to food ration at first feeding than walleye pollock, however; at later larval stages both species have a negative condition response to low food ration, especially at elevated temperatures (5° vs. 7°C). The lower thermal tolerance of Arctic cod, coupled with a higher sensitivity to food availability indicates that Arctic cod are particularly vulnerable to on-going environmental change. Arctic cod is a lipid-rich keystone species and, therefore, a reduction in their energetic condition during summer has the potential to affect the health of higher trophic levels such as predatory fish, marine mammals and seabirds throughout the Arctic. In a second laboratory study involving the same two cod species, Koenker et al. (2018b) investigate the effects of temperature and food availability on survival and growth of larvae. At low temperatures, Arctic cod larvae are better adapted than walleye pollock in terms of growth and survival but under warmer, high food rations, walleye pollock have the advantage, exhibiting higher growth and better survival. The authors also find that the thermal response in the larvae is both species- and stage-dependent. Laurel et al. (2018) incubated multiple batches of gadid eggs and larvae from laboratory broodstock held under simulated seasonal environmental conditions for the species investigated. Arctic cod eggs and larvae were ∼25–35% larger than walleye pollock with 3–6× more energetic reserves. A low thermal tolerance is similar for both species but Arctic cod have a much lower upper thermal tolerance. While this means that Arctic cod have a much smaller thermal window for survival, they can survive for longer periods in the absence of food than can walleye pollock at cold temperatures. The new information on vital rates from all three studies provides a mechanistic framework for understanding potential spatial-temporal shifts of these gadids at the boundary between the Arctic and subArctic resulting from climatic warming and altered productivity regimes by supporting better population forecasts, species distribution models and biophysical transport models for these species. Eggs and larvae of zooplankton and fish are transported by ocean currents, which influence their spatial distribution and survival. Kvile et al. (2018) use a biophysical model of the North Sea cod (Gadus morhua) to explore the relative importance of model resolution, the vertical behaviour of the eggs and larvae and interannual variability in water circulation and temperature on the distribution and survival of cod. Vertical movement and ocean model resolution both influence the results moderately but their effects differ substantially between years. Generally, higher ocean model resolution has a larger effect than changes in the vertical behaviour of the cod larvae. Merrick (2018) describes the United States National Oceanic and Atmospheric Administration (NOAA) approach to marine conservation and its effects on fish populations and fisheries. The management advice must be strongly science-based. Legislative mandates require that marine resources and their habitats be protected to provide productive and sustainable fisheries, safe sources of seafood, the recovery and conservation of protected resources, and healthy ecosystems. In response, NOAA implemented a four-pronged approach: (i) the development of a national framework for conservation science, (ii) implementation that is region-specific, (iii) development of unbiased, scientific advice, and (iv) scientists acting as advocates and science communicators. This approach has been successful with 92% of managed fish stocks no longer being overfished and 84% of stocks that are assessed being at healthy levels. Forty-three of the latter are stocks that have been rebuilt from low or depleted levels. The author argues that it is vitally important that marine conservation decisions everywhere be science-driven, particularly under climate change. Unprecedented and rapid changes are ongoing in northern high-latitude marine ecosystems, due to climate warming. Species distributions and abundances are changing, altering both ecosystem structure and dynamics. At the same time, human impacts are increasing. Less sea ice opens the door for more petroleum-related activities, shipping and tourism. Fisheries are moving into previously unfished habitats, targeting more species across more trophic levels. Skern-Mauritzen et al. (2018) argue that there is a need for Ecosystem Based Fisheries Management (EBFM) and Ecosystem Based Management (EBM) to take the rapid, climate driven changes more fully into account. Recently, there has been much development in qualitative, semiquantitative and quantitative scientific approaches to support EBFM and EBM. They present some of these approaches and discuss how they provide opportunities for advancing EBFM and EBM in the Barents Sea. The authors propose that advancing EBFM and EBM is more about adding tools to the toolbox than replacing tools, and to use the tools in coordinated efforts to tackle the increasing complexities in scientific support for decision making. Collaborative and participatory processes among managers and scientists are pivotal for both scoping and prioritizing, and for efficient knowledge exchange. Summing up and future work Collectively, the above papers represent a glimpse into some of the important issues that the ESSAS programme is addressing. They represent retrospective analyses to understand climate effects on marine ecosystems at long and short time scales and from basin to local geographic scales within the Pacific, Atlantic, and Arctic oceans. Major work is ongoing to understand the mechanistic processes linking climate and ecological variability, including laboratory studies that provide parameters and vital rate information for use in models. Such models are also being used to develop future climate and ecosystem scenarios under anthropogenic climate change and OA. Finally, an important aspect of ESSAS research is the linking of the research to fisheries and ecosystem-based management. Some future work that is planned: development of climate change and ecosystem scenarios in the transition zone between the Subarctic and the Arctic; paleo-ecology studies linking ocean productivity to the timing of the establishment of human settlements, both prehistoric and historic, and fluctuations in the settlement population levels; studies of the life cycle and the mechanisms controlling the distribution and abundance of Arctic/Polar cod (B. saida); comparisons of management strategies of different nations with respect to their preparedness to meet the challenges of climate change and OA; and an exploration of the use of natural analogues to investigate the effects of climate change and OA on northern ecosystems. Strategically, longer-term goals are (i) to engage in more socio-ecological studies that consider not only the natural environment but also the effect of and on humans, (ii) to quantify the uncertainty in future projections of ecological changes, and (iii) to increase our mechanistic understanding of factors influencing fish population variability in northern regions as input to management of sustainable fisheries. Acknowledgements We thank the local organizing committee of the Open Science Meeting in Tromsø led by Benjamin Planque (IMR) and including Vera Lund (IMR), Håkon Hop (Norwegian Polar Institute), Bodil Bluhm (U. Tromsø), and Paul Renaud (AquaPlan Niva), who were joined by Lisa Maddison (IMBeR IPO) from Bergen. We appreciated the help from the OSM Scientific Steering Committee consisting of Olafur Astthorsson (ICES), Andrey Dolgov (Russia), Naomi Harada (Japan), Alan Haynie (USA), George Hunt (USA), Shin-Ichi Ito (Japan), Gudrun Marteinsdottir (Iceland), Sue Moore (USA), Jean Eric Tremblay (Canada), John Walsh (USA), Paul Wassmann (Norway), and Jinping Zhao (China). We also thank the many cochairs of the sessions and the keynote speakers. Finally, we gratefully acknowledge those organizations that contributed financially to the OSM including ICES, PICES, IMBeR, the International Arctic Science Committee (IASC), the Arctic Science Center in Sapporo, Japan, the FRAM Centre in Tromsø, Norway, the Institute of Marine Research in Bergen, Norway, and the US agencies including NOAA Fisheries, the North Pacific Research Board (NPRB) and the North Pacific Fishery Management Council (NPFMC). References Aarflot J. M. , Skjoldal H. R. , Dalpadado P. , Skern-Mauritzen M. 2018 . Contribution of Calanus species to the mesozooplankton biomass in the Barents Sea . ICES Journal of Marine Science , 75 : 2342 – 2354 . Browman H. I. 2017 . Towards a broader perspective on ocean acidification research . ICES Journal of Marine Science , 74 : 889 – 894 . Google Scholar Crossref Search ADS Curchitser E. N. , Rose K. A. , Ito S-I. , Peck M. A. , Kishi M. J. (Eds.) 2015 . Combining modeling and observations to better understand marine ecosystem dynamics . Progress in Oceanography 138(Part B) : 325 – 584 . Google Scholar Crossref Search ADS Drinkwater K. F. , Kristiansen T. 2018 . A synthesis of the ecosystem responses to the late 20th century cold period in the northern North Atlantic . ICES Journal of Marine Science , 75 : 2325 – 2341 . Drinkwater K. F. , Hunt G. L. , Astthorsson O. S. , Head E. J. H. 2012 . Comparative studies of climate effects on polar and sub-polar ocean ecosystems: progress in observation and prediction. Proceedings of the 2nd ESSAS Open Science Meeting, held in Seattle, USA, 22–26 May, 2011 . ICES Journal of Marine Science , 69 : 1119 – 1328 . Google Scholar Crossref Search ADS Hannesson R. 2016 . Managing Shared Migratory Stocks: The Case of the Atlantic Mackerel in Challenges of the Changing Arctic: Continental Shelf, Navigation, and Fisheries. Series: Center for Oceans Law and Policy , 19 , pp. 559 – 570 . Ed. by Nordquist H. , Moore J. N. , Long R. . Brill Nijhoff , Leiden, Netherlands . Hunt G. L. , Drinkwater K. F. (Eds.) 2005 . Ecosystem Studies of Sub-Arctic Seas (ESSAS) Science Plan. GLOBEC Report No. 19, viii, 60 pp. Hunt G. L. , Drinkwater K. F. (Eds.) 2005 . Background on the Climatology, Physical Oceanography and Ecosystems of the Sub-Arctic Seas. Appendix to the ESSAS Science Plan. GLOBEC Report No. 20, viii+96 pp. Hunt G. L. , Blanchard A. L. , Boveng P. , Dalpadado P. , Drinkwater K. , Eisner L. , Hopcroft R. , et al. 2013 . The Barents and Chukchi Seas: comparison of two Arctic shelf ecosystems . Journal of Marine Systems , 109–110 : 43 – 68 . Google Scholar Crossref Search ADS Hunt G. L. , Coyle K. O. , Eisner L. B. , Farley E. V. , Heintz R. A. , Mueter F. , Napp J. M. , et al. 2011 . Climate impacts on eastern Bering Sea foodwebs: a synthesis of new data and an assessment of the Oscillating Control Hypothesis . ICES Journal of Marine Science , 68 : 1230 – 1243 . Google Scholar Crossref Search ADS Hunt G. L. , Drinkwater K. F. , Arrigo K. , Berge J. , Daly J. , Danielson S. , Daase M. , et al. 2016 . Advection in polar and sub-polar environments: impacts on high latitude marine ecosystems . Progress in Oceanography , 149 : 40 – 81 . Google Scholar Crossref Search ADS Hunt G. , Drinkwater K. F. , McKinnell S. , Mackas D. (Eds.). 2007 . Climate variability and subarctic marine ecosystems. Proceedings of the GLOBEC symposium, held in Victoria, Canada, 16–20 May, 2005. Deep-Sea Research II , 54 : 2453 – 2970 . IPCC . 2013 . Climate Change 2013: The Physical Science Basis: Working Group . Cambridge University Press , Cambridge, United Kingdom; New York, USA . 1535 pp. Johannesen E. , Ingvaldsen R. B. , Bogstad B. , Dalpadado P. , Eriksen E. , Gjosaeter H. , Knutsen T. , et al. 2012 . Changes in Barents Sea ecosystem state, 1970–2009: climate fluctuations, human impact, and trophic interactions . ICES Journal of Marine Science , 69 : 880 – 889 . Google Scholar Crossref Search ADS Kaartvedt S. , Titelman J. 2018 . Planktivorous fish in a future Arctic Ocean of changing ice and unchanged photoperiod . ICES Journal of Marine Science , 75 : 2312 – 2318 . Koenker B. L. , Copeman L. A. , Laurel B. J. 2018a . Impacts of temperature and food availability on the condition of larval Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus) . ICES Journal of Marine Science , 75 : 2370 – 2385 . Koenker B. L. , Laurel B. J. , Copeman L. A. , Ciannelli L. 2018b . Effects of temperature and food availability on the survival and growth of larval Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus) . ICES Journal of Marine Science , 75 : 2386 – 2402 . Kvile K. Ø. , Romagnoni G. , Dagestad K-F. , Langangen Ø. , Kristiansen T. 2018 . Sensitivity of modelled North Sea cod larvae transport to vertical behaviour, ocean model resolution and interannual variation in ocean dynamics . ICES Journal of Marine Science , 75 : 2413 – 2424 . Laurel B. J. , Copeman L. A. , Spencer M. , Iseri P. 2018 . Comparative effects of temperature on rates of development and survival of eggs and yolk-sac larvae of Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus) . ICES Journal of Marine Science , 75 : 2403 – 2412 . McBride M. M. , Dalpadado P. , Drinkwater K. , Godø O. R. , Kristiansen T. , Murphy E. , Subbey S. , et al. 2014 . Krill, climate, and contrasting future scenarios for Arctic and Antarctic fisheries . ICES Journal of Marine Research , doi:10.1093/icesjms/fsu002. Merrick R. 2018 . Mechanisms for science to shape US living marine resource conservation policy . ICES Journal of Marine Science , 75 : 2319 – 2324 . Mueter F. J. , Dawe E. G. , Pálsson Ó. 2012 . Effects of climate and predation on subarctic crustacean populations . Marine Ecology Progress Series , 469 : 191 – 193 . Google Scholar Crossref Search ADS Mueter F. J. , Nahrgang J. , Nelson R. J. , Berge J. 2016 . The ecology of gadid fishes in the circumpolar Arctic with a special emphasis on the polar cod (Boreogadus saida) . Polar Biology , 39 : 961 – 1173 . Google Scholar Crossref Search ADS Murphy E. J. , Cavanagh R. D. , Drinkwater K. F. , Grant S. M. , Heymans J. J. , Hofmann E. E. , Hunt G. L. , et al. 2016 . Linking biological diversity and ecosystem functioning in polar ocean ecosystems . Proceedings of the Royal Society B , 283 : 20161646. Google Scholar Crossref Search ADS PubMed Rastrick S. S. P. , Graham H. , Azetsu-Scott K. , Calosi P. , Chierici M. , Fransson A. , Hop H. , et al. 2018 . Using natural analogues to investigate the effects of climate change and ocean acidification on northern ecosystems . ICES Journal of Marine Science , 75 : 2299 – 2311 . Screen J. A. 2014 . Arctic amplification decreases temperature variance in northern mid-to high-latitudes . Nature Climate Change , 4 : 577 – 582 . Google Scholar Crossref Search ADS Sigurjónsson J. 2016 . Changes in distribution and migration of fish stocks in the Northeast Atlantic Ocean due to climate. In Challenges of the Changing Arctic: Continental Shelf, Navigation, and Fisheries. Series: Center for Oceans Law and Policy , 19 , pp. 405 – 428 . Ed. by Nordquist H. , Moore J. N. , Long R. . Brill Nijhoff , Leiden, Netherlands . Skern-Mauritzen M. , Olsen E. , Huse G. 2018 . Opportunities for advancing ecosystem-based management in a rapidly changing, high latitude ecosystem . ICES Journal of Marine Science , 75 : 2425 – 2433 . Skogen M. D. , Hjøllo S. S. , Sandø A. B. , Tjiputra J. 2018 . Future ecosystem changes in the Northeast Atlantic: a comparison between a global and a regional model system . ICES Journal of Marine Science , 75 : 2355 – 2369 . Smedsrud L. H. , Esau I. , Ingvaldsen R. B. , Eldevik T. , Haugan P. M. , Li C. , Lien V. S. , et al. 2013 . The role of the Barents Sea in the Arctic climate system . Reviews of Geophysics , 51 : 415 – 449 . Google Scholar Crossref Search ADS Stabeno P. J. , Farley E. V. , Kachel B. , Moore S. , Mordy C. W. , Napp J. M. , Overland J. E. , et al. 2012 . A comparison of the physics of the northern and southern shelves of the eastern Bering Sea and some implications for the ecosystem . Deep Sea Research Part II , 65 : 31 – 45 . Google Scholar Crossref Search ADS © International Council for the Exploration of the Sea 2018. All rights reserved. For permissions, please email: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Rastrick, Samuel S, P;Graham,, Helen;Azetsu-Scott,, Kumiko;Calosi,, Piero;Chierici,, Melissa;Fransson,, Agneta;Hop,, Haakon;Hall-Spencer,, Jason;Milazzo,, Marco;Thor,, Peter;Kutti,, Tina

doi:

Kaartvedt,, Stein;Titelman,, Josefin

doi: 10.1093/icesjms/fsx248pmid: N/A

Abstract Climate change drives fish and plankton species ranges toward the poles, often related to warmer waters mediating geographic distributions via changes in vital rates. Yet, the distribution of fish may also be governed by less acknowledged mechanisms. Ice limits access to air for physostomous fish filling their swimbladders at the surface. We hypothesize that release of ice constraints may facilitate northward expansion of physostomes, with implied impact on their zooplankton prey. On the other hand, even in a changing Arctic, the extreme high-latitude photoperiod will persist. The abundance of mesopelagic fish is low in the Arctic Ocean. Feeding conditions may be inferior during the darkness of winter and in light summer nights. If the photoperiod is constraining distributions, biogeographic boundaries of mesopelagic fish may be relatively unaffected by climate change. Alternatively, if low temperatures are their main constraint, we hypothesize that northward extensions in a warmer ocean may be detrimental to key Arctic copepods as we argue that their current success relates to low mortality during overwintering in the absence of mesopelagic fish. It is therefore essential to discriminate the role of the light climate at high latitudes from those related to temperatures for assessing future biogeographic boundaries. Introduction It has been predicted that fishes like mackerel, herring, capelin, and salmon might migrate into a future warmer Arctic Ocean (Huse and Ellingsen, 2008; Christiansen, 2017). Poleward expansion of plankton, fish, and mammals (Beaugrand, 2009; Haug et al., 2017) may relate to more hospitable temperatures, but also to habitat changes like reduction in ice. While reduced ice cover can disfavour ice-associated species (e.g. polar cod, Wyllie-Echeverria and Wooster, 1998; Christiansen, 2017), it may be advantageous for others. For example, as warming reduces sea ice extent and thickness in the Arctic Ocean (Comiso, 2012; Stroeve et al., 2012), horizontally migrating fish may track the receding ice edge, benefitting from enhanced light levels during foraging forays into high-latitude oceans (Varpe et al., 2015; Langbehn and Varpe, 2017). So-called physostomous fishes have an open swimbladder that they normally fill by gulping air at the surface (Harden Jones and Marshall, 1953), and might benefit from the release of a constraining ice cover. However, this issue remains to be assessed in evaluations of a future Arctic Ocean. While the Arctic Ocean is characterized by low temperatures and partly ice-covered waters that are sensitive to warming, it is also defined by an extreme photoperiod that will obviously not respond to climate change (Sundby et al., 2016). Most pelagic organisms depend on the light regime for either food search, predator avoidance, or both. The life of mesopelagic fishes—i.e. fish spending daytime at several hundred metres depth and to a varying degree migrating to the surface at night—is strongly related to the ambient light conditions on temporal and geographic scales (Røstad et al., 2016a, b; Aksnes et al., 2017) and may be particularly affected by the extreme photoperiod at high latitudes (Kaartvedt, 2008). Here, we discuss how ice cover and the photic environment may interact with the effect of warmer waters in mediating future distribution ranges of planktivorous fish. The physostomous fishes herring and capelin are major predators on the key copepod Calanus in upper waters during summer (Hassel et al., 1991; Varpe and Fiksen, 2010), while mesopelagic fish may impose heavy mortality on Calanus during their overwintering at mesopelagic depths (Bagøien et al., 2001). Therefore, a main consequence of any altered habitats of these planktivores may be increased top down forces on the Calanus, and thus the Arctic pelagic food web at large. We do not strive toward an exhaustive review, but rather aim to raise examples and hypotheses (Table 1) regarding the importance of ice and light for future pelagic ecosystems at high latitudes. Table 1. Summary of hypotheses for fish in a future warmer Arctic Ocean and suggested consequences for plankton. Hypothesis Suggested consequence for plankton Ice constrains the geographic distribution of physostomous fish by preventing filling of the swimbladder Lower plankton mortality with ice The release of ice constraints to the future, warmer, ocean may facilitate northward expansion of physostomous fish Increased predation pressure on plankton during summer If constraints related to the photoperiod override that of temperature, biogeographic boundaries of mesopelagic fish may be relatively unaffected by climate change Low size-selective mortality of large Arctic copepods during winter in habitats without mesopelagic fish and little change with warming If cold waters currently constrain mesopelagic fish toward the poles; then mesopelagic fish may invade a warmer Arctic Ocean. Switching antipredator behaviour to schooling in light summer nights may facilitate northward extensions Increased mortality of plankton. High accumulated mortality of large, seasonally migrating Arctic copepods during overwintering at mesopelagic depths Hypothesis Suggested consequence for plankton Ice constrains the geographic distribution of physostomous fish by preventing filling of the swimbladder Lower plankton mortality with ice The release of ice constraints to the future, warmer, ocean may facilitate northward expansion of physostomous fish Increased predation pressure on plankton during summer If constraints related to the photoperiod override that of temperature, biogeographic boundaries of mesopelagic fish may be relatively unaffected by climate change Low size-selective mortality of large Arctic copepods during winter in habitats without mesopelagic fish and little change with warming If cold waters currently constrain mesopelagic fish toward the poles; then mesopelagic fish may invade a warmer Arctic Ocean. Switching antipredator behaviour to schooling in light summer nights may facilitate northward extensions Increased mortality of plankton. High accumulated mortality of large, seasonally migrating Arctic copepods during overwintering at mesopelagic depths Table 1. Summary of hypotheses for fish in a future warmer Arctic Ocean and suggested consequences for plankton. Hypothesis Suggested consequence for plankton Ice constrains the geographic distribution of physostomous fish by preventing filling of the swimbladder Lower plankton mortality with ice The release of ice constraints to the future, warmer, ocean may facilitate northward expansion of physostomous fish Increased predation pressure on plankton during summer If constraints related to the photoperiod override that of temperature, biogeographic boundaries of mesopelagic fish may be relatively unaffected by climate change Low size-selective mortality of large Arctic copepods during winter in habitats without mesopelagic fish and little change with warming If cold waters currently constrain mesopelagic fish toward the poles; then mesopelagic fish may invade a warmer Arctic Ocean. Switching antipredator behaviour to schooling in light summer nights may facilitate northward extensions Increased mortality of plankton. High accumulated mortality of large, seasonally migrating Arctic copepods during overwintering at mesopelagic depths Hypothesis Suggested consequence for plankton Ice constrains the geographic distribution of physostomous fish by preventing filling of the swimbladder Lower plankton mortality with ice The release of ice constraints to the future, warmer, ocean may facilitate northward expansion of physostomous fish Increased predation pressure on plankton during summer If constraints related to the photoperiod override that of temperature, biogeographic boundaries of mesopelagic fish may be relatively unaffected by climate change Low size-selective mortality of large Arctic copepods during winter in habitats without mesopelagic fish and little change with warming If cold waters currently constrain mesopelagic fish toward the poles; then mesopelagic fish may invade a warmer Arctic Ocean. Switching antipredator behaviour to schooling in light summer nights may facilitate northward extensions Increased mortality of plankton. High accumulated mortality of large, seasonally migrating Arctic copepods during overwintering at mesopelagic depths Ice as a physical boundary The ice lid blocks access to the surface. Common physostomes at high latitudes like salmonids (e.g. salmon), clupeids (e.g. herring), and osmerids (e.g. capelin) are unable to secrete gas into the swimbladder from the blood and instead fill their swimbladder by gulping air at the surface (Harden Jones and Marshall, 1953; Fahlén, 1968; Blaxter et al., 1979). (Re)filling of the swimbladder would accordingly be restricted in ice-covered waters. The swimbladder wall of herring has a barrier of guanine crystals hampering diffusion rates and allowing for prolonged retention of gas (Blaxter et al., 1979). For salmon, there is some gas leakage trough time, resulting in smaller swimbladder volume and altered swimming behaviour (Korsøen et al., 2009). Yet, clupeids, salmonids, and osmerids may release gas, but for uncertain, and debatable, reasons (Thorne and Thomas, 1990; Nøttestad, 1998; Rudstam et al., 2003; Solberg and Kaartvedt, 2014). Studies from ice-covered fjords and lakes as well as from aquaculture indicate decreased swimbladder volume, behavioural changes, well-fare issues, and even death when access to air is constrained by ice or other physical structures (Table 2, references therein). Because harsh winter weather and ice curb access to high latitude waters, knowledge of marine fish populations under ice is minimal. Yet, based on the limited current data (Table 2) we hypothesize that Ice constrains the geographic distribution of physostomous fish by preventing filling of the swimbladder, and that the release of ice constraints to the future warmer ocean may facilitate northward expansion of physostomous fish. Table 2. Literature on physostomous fish affected by ice or air constraints. Species Location and conditions Change with constrained surface access Reference Sprat (S. sprattus) Oslofjorden, Norway. Upward-looking echo-sounder in temporarily ice covered fjord Altered vertical distribution, smaller swimbladder volume, altered behaviour, and increased, yet unsuccessful searches for air underneath the ice Solberg et al. (2012),Solberg and Kaartvedt (2014) Atlantic salmon (Salmo salar) Aquaculture, submerged cages Deflated swimbladder with time during 25 d experiment. Tilted swimming and deformation in vertebra. Increased swimming near top of cage. Searching for air Korsøen et al. (2009) Atlantic salmon (S. salar) Unable to replace gas lost from the swimbladder Jakobs (1934, cited in Harden Jones and Marshall, 1953) Atlantic salmon (S. salar) Aquaculture submerged cage Negatively buoyant, increased swimming speed, reduced welfare, and increased mortality Fosseidengen et al. (1982), Ablett et al. (1989), and Dempster et al. (2009) Herring (C. harengus) Experiments Fish with artificially emptied swimbladders died when denied access to the surface Blaxter and Batty (1984) Central mudminnow Umbra limi Experiments and field study in lake Facultative air-breathing fish in low oxygen waters searching for air bubbles under the ice Klinger et al. (1982) Grayling Thymallus thymallus Field study lake Acoustically tagged individuals immediately changing vertical distribution with ice cover, swimming in the upper 50 cm Bass et al. (2014) Species Location and conditions Change with constrained surface access Reference Sprat (S. sprattus) Oslofjorden, Norway. Upward-looking echo-sounder in temporarily ice covered fjord Altered vertical distribution, smaller swimbladder volume, altered behaviour, and increased, yet unsuccessful searches for air underneath the ice Solberg et al. (2012),Solberg and Kaartvedt (2014) Atlantic salmon (Salmo salar) Aquaculture, submerged cages Deflated swimbladder with time during 25 d experiment. Tilted swimming and deformation in vertebra. Increased swimming near top of cage. Searching for air Korsøen et al. (2009) Atlantic salmon (S. salar) Unable to replace gas lost from the swimbladder Jakobs (1934, cited in Harden Jones and Marshall, 1953) Atlantic salmon (S. salar) Aquaculture submerged cage Negatively buoyant, increased swimming speed, reduced welfare, and increased mortality Fosseidengen et al. (1982), Ablett et al. (1989), and Dempster et al. (2009) Herring (C. harengus) Experiments Fish with artificially emptied swimbladders died when denied access to the surface Blaxter and Batty (1984) Central mudminnow Umbra limi Experiments and field study in lake Facultative air-breathing fish in low oxygen waters searching for air bubbles under the ice Klinger et al. (1982) Grayling Thymallus thymallus Field study lake Acoustically tagged individuals immediately changing vertical distribution with ice cover, swimming in the upper 50 cm Bass et al. (2014) Table 2. Literature on physostomous fish affected by ice or air constraints. Species Location and conditions Change with constrained surface access Reference Sprat (S. sprattus) Oslofjorden, Norway. Upward-looking echo-sounder in temporarily ice covered fjord Altered vertical distribution, smaller swimbladder volume, altered behaviour, and increased, yet unsuccessful searches for air underneath the ice Solberg et al. (2012),Solberg and Kaartvedt (2014) Atlantic salmon (Salmo salar) Aquaculture, submerged cages Deflated swimbladder with time during 25 d experiment. Tilted swimming and deformation in vertebra. Increased swimming near top of cage. Searching for air Korsøen et al. (2009) Atlantic salmon (S. salar) Unable to replace gas lost from the swimbladder Jakobs (1934, cited in Harden Jones and Marshall, 1953) Atlantic salmon (S. salar) Aquaculture submerged cage Negatively buoyant, increased swimming speed, reduced welfare, and increased mortality Fosseidengen et al. (1982), Ablett et al. (1989), and Dempster et al. (2009) Herring (C. harengus) Experiments Fish with artificially emptied swimbladders died when denied access to the surface Blaxter and Batty (1984) Central mudminnow Umbra limi Experiments and field study in lake Facultative air-breathing fish in low oxygen waters searching for air bubbles under the ice Klinger et al. (1982) Grayling Thymallus thymallus Field study lake Acoustically tagged individuals immediately changing vertical distribution with ice cover, swimming in the upper 50 cm Bass et al. (2014) Species Location and conditions Change with constrained surface access Reference Sprat (S. sprattus) Oslofjorden, Norway. Upward-looking echo-sounder in temporarily ice covered fjord Altered vertical distribution, smaller swimbladder volume, altered behaviour, and increased, yet unsuccessful searches for air underneath the ice Solberg et al. (2012),Solberg and Kaartvedt (2014) Atlantic salmon (Salmo salar) Aquaculture, submerged cages Deflated swimbladder with time during 25 d experiment. Tilted swimming and deformation in vertebra. Increased swimming near top of cage. Searching for air Korsøen et al. (2009) Atlantic salmon (S. salar) Unable to replace gas lost from the swimbladder Jakobs (1934, cited in Harden Jones and Marshall, 1953) Atlantic salmon (S. salar) Aquaculture submerged cage Negatively buoyant, increased swimming speed, reduced welfare, and increased mortality Fosseidengen et al. (1982), Ablett et al. (1989), and Dempster et al. (2009) Herring (C. harengus) Experiments Fish with artificially emptied swimbladders died when denied access to the surface Blaxter and Batty (1984) Central mudminnow Umbra limi Experiments and field study in lake Facultative air-breathing fish in low oxygen waters searching for air bubbles under the ice Klinger et al. (1982) Grayling Thymallus thymallus Field study lake Acoustically tagged individuals immediately changing vertical distribution with ice cover, swimming in the upper 50 cm Bass et al. (2014) The largest stock of clupeids in northern waters is the Norwegian spring-spawning herring (Clupea harengus), but their population limit does not seem to intercept with the ice. However, the Pacific herring, Clupea pallassii abounds in ice affected areas, like the Bering Sea and the White Sea (Tojo et al., 2007; Lajus et al., 2007). The Bering Sea holds overwintering habitats of the Pacific herring that appear to be just at the ice edge (Tojo et al., 2007). While cold water lowers metabolism, which may benefit overwintering during the non-feeding period (e.g. Kooka et al., 2007), we speculate that the constraint by ice on access to air may be a factor in defining the overwintering areas per se. The southeastern Bering Sea exhibits extreme variability in sea ice extent (Stabeno et al., 2012), which also fluctuates across the overwintering habitat of herring (Tojo et al., 2007). Distribution and diet of Pacific herring vary largely between cold and warm years, including in relation to ice (Andrews III et al., 2015). However, it remains uncertain if ice per se constrains distributions of the Pacific herring. On the contrary, it has been suggested that ice-covered Alaskan bays may facilitate survival of overwintering juvenile Pacific herring by providing cover from predatory birds and mammals (Lewandoski and Bishop, 2017). Also, herring in the Baltic persists in habitats that regularly become ice covered (e.g. Lamichhaney et al., 2012), though to what extent the ice affects the population is unknown. The apparently only systematic field studies of how ice may hamper access to air in any physostome are for another clupeid, sprat (Sprattus sprattus). These document changes in vertical distribution, smaller swimbladder, altered swimming behaviour including more frequent, but unsuccessful, search for air under the ice, and termination of gas release when their fjord habitat froze over (Solberg et al., 2012; Solberg and Kaartvedt, 2014). In ice-free conditions the sprat seeks out the surface approximately four times per day, indicating the importance of access to new air (Solberg and Kaartvedt, 2014). The sprat straightaway changed to a shallower distribution upon the fjord freezing over, and moved even shallower upon heavy snowfall (Solberg et al., 2012). The distinction between the responses to ice per se and the lower light due to the snow is supported by observations of krill in the same habitat. Krill moved shallower only when snow covered the ice (Vestheim et al., 2014). Presumably, the ice represents a constraint to the fish that it is to some extent able to handle, albeit at an unknown cost. The small capelin (Mallotus villosus) inhabits the circumpolar northern boreal oceans at the margins of cold Arctic waters (Rose, 2005). Capelin dominates the pelagic fish in the Barents Sea (Gjøsæter, 1998). There, and elsewhere, capelin plays a key role in the food web, both as a planktivore and as prey for other fish, marine mammals, and birds (Gjøsæter, 1998; Rose 2005). Capelin performs long migrations and while being able to forage in cold water (−1.5 °C), it requires warmer waters to reproduce (Rose, 2005). Because its distribution extends further to the north and east in the Barents Sea during warm years than in cold years, both its oceanic distribution and its spawning grounds are anticipated to change with warming (Huse and Ellingsen, 2008). While the Barents Sea capelin may be associated with the productive marginal ice zone during summer (Hop and Gjøsæter, 2013), its relation to ice in winter is unknown. In Icelandic waters, capelin may be common near the ice edge also in winter, but for logistic reasons monitoring does not include ice-covered waters (Birkir Bardarson, Marine Research Institute Iceland, pers. com.). It remains unknown if capelin accumulates at an ice border, such that ice is a constraint per se or the border zone represents a favourable habitat, or if observed distributions just represent fringes of a population that extends into ice-covered waters. In summary, both experimental and field studies suggest that hampered access to air may impact physostomes negatively (Table 2). On the other hand, populations of clupeids, osmerids, and salmonids do persist in environments, particularly lakes, which freeze over in winter (Steinhart and Wurtsbaugh, 1999; Jurveliusa et al., 2000; Klemetsen et al., 2003; Dunlop and Riley, 2013; Bass et al., 2014). At present, large-scale ecological consequences of increased access to the surface in a future ocean with reduced ice cover remains unknown. Upward-looking echo sounders represent a powerful tool for studying fish behaviour in ice-free and iced-covered habitats (Solberg & Kaartvedt, 2014). This approach might represent one way to further test how ice affects other physostomes, including in lakes that are more readily accessible than are comparable marine habitats. Photoperiod and light The extreme photoperiod and light climate of the Arctic alter the trade-offs in diel vertical migration in which mesopelagic fish, and other pelagic organisms, exploit the rich pastures of upper waters in shelter of darkness at night and seek refuge in deep, dim waters in daytime. Midnight sun likely limits the options for safe nocturnal foraging by mesopelagic fish in upper layers in summer (Sameoto, 1989; Norheim et al., 2016), and continuous darkness during winter expectedly hampers visual feeding in deep water any time of day. In northern boreal, waters with more equal diel light cycles dark adapted mesopelagic fish hunt even at several hundred metres depth during daytime (Bagøien et al., 2001; Dypvik et al., 2012). Deep scattering layers of mesopelagic fish occur in all oceans, with an estimated global abundance of 10 billion tonnes (Irigoien et al., 2014). However, mesopelagic fish abundance declines toward Arctic waters (Sameoto, 1989; Dale et al., 1999; Sutton et al., 2017). Mesopelagic fish distribute vertically relative to limited bands of light intensities, so-called light comfort zones, both locally (Røstad et al., 2016a, b) and globally (Aksnes et al., 2017). About half of the mesopelagic fish migrate between upper layers and the mesopelagic zone on a daily scale (Klevjer et al., 2016). The light summer nights in the Arctic apparently prevent the light comfort zone of mesopelagic fishes to overlap with abundant prey resources during much of the productive season. The hampering of nocturnal ascent of mesopelagic scattering layers at high latitude in summer concurs with weakening of the backscatter layers northwards (Norheim et al., 2016). Another challenge is the constant darkness during winter, which might deprive feeding conditions for mesopelagic fish in deep water, like on overwintering Calanus. This prediction has not been tested in the Arctic, but finds support in data from the boreal Lurefjorden, which is characterized by high light extinction and particularly dark mesopelagic waters. Lurefjorden is basically devoid of mesopelagic fish, presumably because dark waters prohibits visual search at mesopelagic depths (Eiane et al., 1999). Yet, in the Arctic, there is unexpected biological activity during the polar night, and gut content of visual predators is evident in shallow moon lit waters (Berge et al., 2015). To what extent this relates to any deeper-living mesopelagic fishes is unknown, as is their use of any non-visual search for prey (e.g. Boscarino et al., 2010). In sum, one may hypothesize that if the importance of photoperiod at high latitudes overrides that of temperature, biogeographic boundaries of mesopelagic fish may be relatively unaffected by climate change (Kaartvedt, 2008). However, if distributions are less tightly linked with the optical environment (Siegelman-Charbit and Planque, 2016) and cold waters currently constrain mesopelagic fish toward the poles (Proud et al., 2017,); then mesopelagic fish may invade a future Arctic Ocean. There is some mesopelagic fish (Maurolicus muelleri) switching to schooling in upper layers during light Norwegian summer nights (Kaartvedt et al., 1998; Prihartato et al., 2015). Such behaviour might facilitate further northward extension if other conditions like temperature became more favourable. The glacier lanternfish (Benthosema glaciale) is the most common mesopelagic fish in the northern Atlantic (Gjøsæter, 1973). The glacier lanternfish drifts passively with currents (Kaartvedt et al., 2009). Their occurrence in the Arctic apparently results from immigration via advection (Sameoto, 1989) and is thus not proof of sustainable populations. Yet, the presence of such apparent expatriates may allow for testing the hypothesis of inferior feeding conditions in the Arctic photoperiod both summer and winter by examining their actual stomach contents and body condition. Top-down effects in a future Arctic Ocean To the extent that ice constrains the geographic distribution of physostomous fish, either for overwintering or during summer foraging migrations, changes in their distribution will inflict altered predation pressures on prey. This relates both to summer and winter ice, as any change in location for overwintering may cause spatial changes in predation on plankton in other seasons. In contrast to vertically migrating mesopelagic fish, horizontally migrating fish like herring and capelin may benefit from the extreme Arctic photoperiod in several ways. As ice melts, the optical habitat may be enhanced allowing for more efficient prey detection (Varpe et al., 2015; Langbehn and Varpe, 2017). Also, these fishes school and use rapid swimming as antipredator strategies (Vabø and Nøttestad, 1997; Crook and Davoren, 2014), allowing them to inhabit upper waters and forage visually throughout the day. Such benefits connected with the high latitude photoperiod providing 24 hours daily for visual search in summer are for example manifested in high growth rates of juvenile cod under midnight sun (Suthers and Sundby, 1996). Visually searching fish select larger prey organisms as these are more easily detected (Brooks and Dodson, 1965). Hence, large Arctic copepods like Calanus would be vulnerable with distributions of planktivorous fish being shifted poleward. Varpe et al. (2015) predicted changed selection pressure on copepods and that large prey would be most impacted by any increased fish predation. This would have enormous ramifications, since the large and lipid-rich Calanus copepods of the Arctic are central in the energy transfer of the marine ecosystem (Steen et al., 2007; Leu et al., 2011; Jónasdóttir et al., 2016). While summer predation may increase on Calanus by horizontally migrating fishes like herring and capelin, any expansion of mesopelagic fish into the Arctic likely will increase predation during winter. Among the main characteristics of marine ecosystems at high latitudes is the seasonal vertical migration of Calanus for overwintering at mesopelagic depths and beyond. Depending on species, Calanus overwinters at several hundred metres to > 1000 m while sustained by stored lipids (Falk-Petersen et al., 2009; Jónasdóttir et al., 2016). Although phenotypic plasticity is large, Calanus finmarchicus normally goes through 1 diapause period (Jónasdóttir et al., 2016), while the larger Calanus glacialis and largest Calanus hyperboreus requires 1–2 and at least 2, respectively. Each overwintering spans 4–9 months (Hirche, 1997; Jónasdóttir et al., 2016). With such long time spent immobile at mesopelagic depths, low mortality rate during diapause is a prerequisite for sustainable rich Calanus populations. Studies in boreal Norwegian fjords indicate that mesopelagic fish are more efficient predators on overwintering Calanus than are invertebrate predators (Bagøien et al., 2001). Winter mortality rates are very high for Calanus exposed to mesopelagic fish and much lower without (Bagøien et al., 2001). Furthermore, in dark fjords without mesopelagic fish, plankton are larger (Eiane et al., 1999; Aksnes et al., 2004). Strikingly, the presumably Arctic C. glacialis dominates over its smaller temperate cousin C. finmarchicus (Bucklin et al., 2000; Bagøien et al., 2001; Niehoff & Hirche, 2005) in Lurefjorden, where mesopelagic fish are lacking, despite overwintering temperatures (∼7 ̊C) exceeding the representative upper comfort temperature of C. glacialis (∼5–6 ̊C, Kosobokova, 1999; Hirche and Kosobokova, 2007). A consequent assumption may be that the success of large Arctic copepods relates to strongly reduced size-selective mortality during winter in habitats without mesopelagic fish. A subsequent prediction is that—as in Lurefjorden—large Arctic copepods can prevail also at lower latitudes (higher temperatures) when mesopelagic fish are absent. This can be tested by addressing the Calanus species composition in habitats with and without mesopelagic fish. The mesopelagic waters of an increasing number of Norwegian fjords appear to become dominated by invertebrates at the expense of mesopelagic fish, possibly related to coastal water darkening (Aksnes et al., 2009). If the success of large Arctic copepods indeed is related to limited top-down control at mesopelagic depths during winter due to low abundance of mesopelagic fish, in turn being related to photoperiod rather than temperature (Kaartvedt, 2008), there might be more resilience to warming than suggested in scenarios predicting substitution of larger Arctic forms of Calanus with smaller, boreal cousins (Falk-Petersen et al., 2007; Hirche and Kosobokova, 2007; Kjellerup et al., 2012). On the other hand, if the hypothesis that photoperiod prevents mesopelagic fish from invading the Arctic turns out to be incorrect and mesopelagic fish indeed expand into a warmer Arctic (Proud et al., 2017), implications for the Calanus and the ecosystem at large may be huge. In sum, the fate of planktivorous fish in a future Arctic Ocean likely depends on more factors than temperature per se. Predicting effects of climate change on plankton like Calanus also requires an understanding of the predation risk, both during overwintering in deep water and when foraging in upper waters during the productive season. The prevailing focus is on direct impacts of temperature and toward potential bottom-up processes like feeding or growth of the Calanus per se (e.g. Søreide et al., 2010; Alcaraz et al., 2014), while top-down processes receive limited attention. We here have outlined some potential scenarios for planktivorous fish and subsequent impacts on their prey (Table 1). We hope to see an increase in studies targeting the role of top-down processes in future assessments. References Ablett R. F. , Marr C. R. , Roberts J. D. 1989 . Influence of chronic subsurface retention on swimming activity of Atlantic salmon (Salmo salar) in cold temperate conditions . Aquaculture Engineering , 8 : 1 – 13 . Google Scholar Crossref Search ADS Aksnes D. L. , Nejstgaard J. , Saedberg E. , Sørnes T. 2004 . Optical control of fish and zooplankton populations . Limnology and Oceanography , 49 : 233 – 238 . Google Scholar Crossref Search ADS Aksnes D. L. , Dupont N. , Staby A. , Fiksen Ø. , Kaartvedt S. , Aure J. 2009 . Coastal water darkening and implications for mesopelagic regime shifts in Norwegian fjords . Marine Ecology Progress Series , 387 : 39 – 49 . Google Scholar Crossref Search ADS Aksnes D. L. , Røstad A. , Kaartvedt S. , Martinez U. , Duarte C. M. , Irigoien X. 2017 . Light penetration structures the acoustic scattering layers in the global ocean . Science Advances , 3 : e1602468. Google Scholar Crossref Search ADS PubMed Alcaraz M. , Felipe J. , Grote E. , Arashkevich E. , Nikishina A. 2014 . Life in a warming ocean: thermal thresholds and metabolic balance of arctic zooplankton . Journal of Plankton Research , 36 : 3 – 10 . Google Scholar Crossref Search ADS Andrews A. G. III , Strasburger W. W. , Farley E. V. Jr , Murphy J. M. , Coyle K. O. 2015 . Effects of warm and cold climate conditions on capelin (Mallotus villosus) and Pacific herring (Clupea pallasii) in the eastern Bering Sea . Deep-Sea Research II , 134 : 235 – 246 . Google Scholar Crossref Search ADS Bagøien E. , Kaartvedt S. , Aksnes D. L. , Eiane K. 2001 . Vertical distribution and mortality of overwintering Calanus . Limnology and Oceanography , 46 : 1494 – 1510 . Google Scholar Crossref Search ADS Bass A. L. , Haugen T. O. , Vøllestad L. A. 2014 . Distribution and movement of European grayling in a subarctic lake revealed by acoustic telemetry . Ecology of Freshwater Fish , 23 : 149 – 160 . Google Scholar Crossref Search ADS Berge J. , Daase M. , Renaud P. E. , Ambrose W. G. Jr , Darnis G. , Last K. S. , Leu E. 2015 . Unexpected levels of biological activity during the polar night offer new perspectives on a warming Arctic . Current Biology , 25 : 2555 – 2561 . Google Scholar Crossref Search ADS PubMed Beaugrand G. 2009 . Decadal changes in climate and ecosystems in the North Atlantic Ocean and adjacent seas . Deep-Sea Research II , 56 : 656 – 673 . Google Scholar Crossref Search ADS Blaxter J. H. S. , Batty R. S. 1984 . The herring swimbladder: loss and gain of gas . Journal of Marine Biological Association UK , 64 : 441 – 459 . Google Scholar Crossref Search ADS Blaxter J. H. S. , Denton E. J. , Gray J. A. B. 1979 . The herring swimbladder as a gas reservoir for the acoustico-lateralis system . Journal of Marine Biological Association of the United Kingdom , 59 : 1 – 10 . Google Scholar Crossref Search ADS Boscarino B. T. , Rudstam L. G. , Tirabassi J. , Janssen J. , Loew E. R. 2010 . Light effects on alewife–mysid interactions in Lake Ontario: a combined sensory physiology, behavioral, and spatial approach . Limnology and Oceanography , 55 : 2061 – 2072 . Google Scholar Crossref Search ADS Brooks J. L. , Dodson S. I. 1965 . Predation, body size and composition of plankton . Science , 150 : 28 – 35 . Google Scholar Crossref Search ADS PubMed Bucklin A. , Kaartvedt S. , Guarnieri M. , Goswami U. 2000 . Population genetics of drifting (Calanus spp.) and resident (Acartia clausi) plankton in Norwegian fjords . Journal of Plankton Research , 22 : 1237 – 1251 . Google Scholar Crossref Search ADS Christiansen J. S. 2017 . No future for Euro-Arctic ocean fishes? Marine Ecology Progress Series , 575 : 217 – 227 . Google Scholar Crossref Search ADS Comiso J. C. 2012 . Large decadal decline of the Arctic multiyear ice cover . Journal of Climate , 25 : 1176 – 1193 . Google Scholar Crossref Search ADS Crook K. A. , Davoren G. K. 2014 . Underwater behaviour of common murres foraging on capelin: influences of prey density and antipredator behavior . Marine Ecology Progress Series , 501 : 279 – 290 . Google Scholar Crossref Search ADS Dale T. , Bagøien E. , Melle W. , Kaartvedt S. 1999 . Can predator avoidance explain varying overwintering depth of Calanus in different oceanic water masses? Marine Ecology Progress Series , 179 : 113 – 121 . Google Scholar Crossref Search ADS Dempster T. , Korsøen O. , Folkedal O. , Juell J. E. , Oppedal F. 2009 . Submergence of Atlantic salmon (Salmo salar L.) in commercial scale sea-cages: a potential short-term solution to poor surface conditions . Aquaculture , 288 : 254 – 263 . Google Scholar Crossref Search ADS Dunlop E. S. , Riley S. C. 2013 . The contribution of cold winter temperatures to the 2003 alewife population collapse in Lake Huron . Journal of Great Lakes Research , 39 : 682 – 689 . Google Scholar Crossref Search ADS Dypvik E. , Klevjer T. A. , Kaartvedt S. 2012 . Inverse vertical migration and feeding of glacier lanternfish (Benthosema glaciale) . Marine Biology , 159 : 443 – 453 . Google Scholar Crossref Search ADS PubMed Eiane K. , Aksnes D. L. , Bagøien E. , Kaartvedt S. 1999 . Fish or jellies—a question of visibility? Limnology and Oceanography , 44 : 1352 – 1357 . Google Scholar Crossref Search ADS Fahlén G. 1968 . The gas bladder as a hydrostatic organ in Thymallus thymallus L., Osmerus eperlanus L. and Mallotus villosus Müll . Fiskeridirektoratets Skrifter Serie Havundersøkelser , 14 : 199 – 228 . Falk-Petersen S. , Pavlov V. , Timofeev S. , Sargent J. R. 2007 . Climate variability and possible effects on arctic food chains: the role of Calanus. In Arctic Alpine Ecosystems and People in a Changing Environment, pp. 147–166 . Ed. by Ørbæk J.B. , Tombre T. , Kallenborn R. , Hegseth E. , Falk-Petersen S. , Hoel A.H. . Springer-Verlag , Berlin . 427 pp. Falk-Petersen S. , Mayzaud P. , Kattner G. , Sargent J. R. 2009 . Lipids and life strategy of Arctic Calanus . Marine Biology Research , 5 : 18 – 39 . Google Scholar Crossref Search ADS Fosseidengen J. K. , Boge E. , Huse I. 1982 . A survey with rainbow trout and salmon in submersible cages (in Norwegian) . Norsk Fiskeoppdrett , 10 : 24 – 25 . Gjøsæter J. 1973 . Age, growth and mortality of the myctophid fish, Benthosema glaciale (Reinhardt) from western Norway . Sarsia , 52 : 1 – 14 . Google Scholar Crossref Search ADS Gjøsæter H. 1998 . The population biology and exploitation of capelin (Mallotus villosus) in the Barents Sea . Sarsia , 83 : 453 – 496 . Google Scholar Crossref Search ADS Harden Jones F. R. , Marshall B. 1953 . The structure and functions of the teleostean swimbladder . Biological Reviews , 28 : 16 – 82 . Google Scholar Crossref Search ADS Hassel A. , Skjoldal H. R. , Gjøsæter H. , Loeng H. , Omli L. 1991 . Impact of grazing from capelin (Mallotus villosus) on zooplankton: a case study in the northern Barents Sea in August 1985 . Polar Research , 10 : 371 – 388 . Google Scholar Crossref Search ADS Haug T. , Bogstad B. , Chierici M. , Gjøsæter H. , Hallfredsson E. H. , Høines Å. S. , Hoel A. H. , et al. 2017 . Future harvest of living resources in the Arctic Ocean north of the Nordic and Barents Seas: a review of possibilities and constraints . Fisheries Research , 188 : 38 – 57 . Google Scholar Crossref Search ADS Hirche H-J. 1997 . Life cycle of the copepod Calanus hyperboreus in the Greenland Sea . Marine Biology , 128 : 607 – 618 . Google Scholar Crossref Search ADS Hirche H-J. , Kosobokova K. 2007 . Distribution of Calanus finmarchicus in the northern North Atlantic and Arctic Ocean—expatriation and potential colonization . Deep-Sea Research II , 54 : 2729 – 2747 . Google Scholar Crossref Search ADS Hop H. , Gjøsæter H. 2013 . Polar cod (Boreogadus saida) and capelin (Mallotus villosus) as key species in marine food webs of the Arctic and the Barents Sea . Marine Biology Research , 9 : 878 – 894 . Google Scholar Crossref Search ADS Huse G. , Ellingsen I. 2008 . Capelin migrations and climate change—a modelling analysis . Climatic Change , 87 : 177 – 197 . Google Scholar Crossref Search ADS Irigoien X. , Klevjer T. A. , Røstad A. , Martinez U. , Boyra G. , Acuña J. L. , Bode F. 2014 . Large mesopelagic fishes biomass and trophic efficiency in the open ocean. Nature Communications, 5 : 3271 . Jacobs W. 1934 . Untersuchungen zur Physiologie der Schwimmblase. 111 . Luftschlucken und Gassekretion bei Physostome. Zeitschrift für vergleichende Physiologie , 20 : 674. Jónasdóttir S. H. , Visser A. W. , Richardson K. , Heath M. R. 2016 . Seasonal copepod lipid pump promotes carbon sequestration in the deep North Atlantic . Proceedings of the National Academy of Sciences , 112 : 12122 – 12126 . Google Scholar Crossref Search ADS Jurveliusa J. , Liljab J. , Hirvonena E. , Riikonena R. , Marjomäki T. J. 2000 . Under ice density and mobility of fish in winter-seining area of two Finnish lakes as revealed by echo-survey . Aquatic Living Resources , 13 : 403 – 408 . Google Scholar Crossref Search ADS Kaartvedt S. 2008 . Photoperiod may constrain the effect of global warming in arctic marine systems . Journal of Plankton Research , 30 : 1203 – 1206 . Google Scholar Crossref Search ADS Kaartvedt S. , Knutsen T. , Holst J. C. 1998 . Schooling of the vertically migrating mesopelagic fish Maurolicus muelleri in light summer nights . Marine Ecology Progress Series , 170 : 287 – 290 . Google Scholar Crossref Search ADS Kaartvedt S. , Røstad A. , Klevjer T. A. , Staby A. 2009 . Use of bottom-mounted echosounders in exploring behavior of mesopelagic fishes . Marine Ecology Progress Series , 395 : 109 – 118 . Google Scholar Crossref Search ADS Kjellerup S. , Dünweber M. , Swalethorp R. , Nielsen T. G. , Møller E. F. , Markager S. , Hansen B. W. 2012 . Effects of a future warmer ocean on the coexisting copepods Calanus finmarchicus and C. glacialis in Disko Bay, western Greenland . Marine Ecology Progress Series , 447 : 87 – 108 . Google Scholar Crossref Search ADS Klemetsen A. , Knudsen R. , Staldvik F. J. , Amundsen P. A. 2003 . Habitat, diet and food assimilation of Arctic charr under the winter ice in two subarctic lakes . Journal of Fish Biology , 62 : 1082 – 1098 . Google Scholar Crossref Search ADS Klevjer T. A. , Irigoien X. , Røstad A. , Fraile-Nuez E. , Benítez-Barrios V. M. , Kaartvedt S. 2016 . Large scale patterns in vertical distribution and behaviour of mesopelagic scattering layers . Scienific Reports , 6 : 19873. Google Scholar Crossref Search ADS Klinger S. A. , Magnuson J. J. , Gallepp G. W. 1982 . Survival mechanisms of the central mudminnow (Umbra limi), fathead minnow (Pimephales promelas) and brook stickleback (Culaea inconstans) for low oxygen in winter Enviornmental . Biology of Fishes , 7 : 113 – 120 . Google Scholar Crossref Search ADS Kooka K. , Yamamura O. , Andoh T. 2007 . Rate of energy depletion and overwintering mortality of juvenile walleye pollock in cold water . Journal of Fish Biology , 71 : 1714 – 1734 . Google Scholar Crossref Search ADS Korsøen Ø. J. , Dempster T. , Fjelldal P. G. , Oppedal F. , Kristiansen T. S. 2009 . Long-term culture of Atlantic salmon (Salmo salar L) in submerged cages during winter affects behavior, growth and condition . Aquaculture , 296 : 373 – 381 . Google Scholar Crossref Search ADS Kosobokova K. N. 1999 . The reproductive cycle and life history of the Arctic copepod Calanus glacialis in the White Sea . Polar Biology , 22 : 254 – 263 . Google Scholar Crossref Search ADS Lajus D. L. , Alekseeva Y. I. , Lajus J. A. 2007 . Herring fisheries in the White Sea in the 18th–beginning of the 20th centuries: spatial and temporal patterns and factors affecting the catch fluctuations . Fisheries Research , 87 : 255 – 259 . Google Scholar Crossref Search ADS Lamichhaney S. , Barrio A. M. , Rafati N. , Sundstrom G. , Rubin C-J. , Gilbert E. R. , Berglund J. , et al. 2012 . Population-scale sequencing reveals genetic differentiation due to local adaptation in Atlantic herring . Proceeding of the National Academy of Sciences , 109 : 19345 – 19350 . Google Scholar Crossref Search ADS Langbehn T. J. , Varpe Ø. 2017 . Sea-ice loss boosts visual search: fish foraging and changing pelagic interactions in polar oceans . Global Change Biology , 2017 : 1 – 13 . Leu E. , Søreide J. E. , Hessen D. O. , Falk-Petersen S. , Berge J. 2011 . Consequences of changing sea-ice cover for primary and secondary producers in the European Arctic shelf seas: timing, quantity, and quality . Progress in Oceanography , 90 : 18 – 32 . Google Scholar Crossref Search ADS Lewandoski S. , Bishop M. A. 2017 . Distribution of juvenile Pacific herring relative to environmental and geospatial factors in Prince William Sound, Alaska. Deep-Sea Research II, doi: 10.1016/j.dsr2.2017.08.002. Niehoff B. , Hirche H-J. 2005 . Reproduction of Calanus glacialis in the Lurefjord (western Norway): indication for temperature-induced female dormancy . Marine Ecology Progress Series , 285 : 107 – 115 . Google Scholar Crossref Search ADS Norheim E. , Klevjer T. A. , Aksnes D. L. 2016 . Evidence for light-controlled migration amplitude of a sound scattering layer in the Norwegian Sea . Marine Ecology Progress Series , 551 : 45 – 52 . Google Scholar Crossref Search ADS Nøttestad L. 1998 . Extensive gas bubble release in Norwegian spring-spawning herring (Clupea harengus) during predator avoidance . ICES Journal of Marine Science , 55 : 1133 – 1140 . Google Scholar Crossref Search ADS Prihartato P. K. , Aksnes D. L. , Kaartvedt S. 2015 . Seasonal patterns in the nocturnal distribution and behavior of the mesopelagic fish (Maurolicus muelleri) at high latitude . Marine Ecology Progress Series , 521 : 189 – 200 . Google Scholar Crossref Search ADS Proud R. , Cox M. J. , Brierley A. S. 2017 . Biogeography of the Global Ocean’s Mesopelagic Zone . Current Biology , 27 : 113 – 119 . Google Scholar Crossref Search ADS PubMed Rose G. A. 2005 . Capelin (Mallotus villosus) distribution and climate: a sea “canary” for marine ecosystem change . ICES Journal of Marine Science , 62 : 1524 – 1530 . Google Scholar Crossref Search ADS Røstad A. , Kaartvedt S. , Aksnes D. L. 2016a . Light comfort zones of mesopelagic acoustic scattering layers in two contrasting optical environments . Deep-Sea Research I , 113 : 1 – 6 . Google Scholar Crossref Search ADS Røstad A. , Kaartvedt S. , Aksnes D. L. 2016b . Erratum to “Light comfort zones of mesopelagic acoustic scattering layers in two contrasting optical environments” [Deep-Sea Res. I 113 (2016) 1–6] . Deep-Sea Research I , 114 : 162 – 164 . Google Scholar Crossref Search ADS Rudstam L. G. , Parker S. L. , Einhouse D. W. , Witzel L. D. , Warner D. M. , Stritzel J. L. , Parrish D. L. , et al. 2003 . Application of in situ target-strength estimations in lakes: examples from rainbow-smelt surveys in Lakes Erie and Champlain . ICES Journal of Marine Science , 60 : 500 – 507 . Google Scholar Crossref Search ADS Sameoto D. 1989 . Feeding ecology of the lantern fish Benthosema glaciale in a subarctic region . Polar Biology , 9 : 169 – 178 . Google Scholar Crossref Search ADS Siegelman-Charbit L. , Planque B. 2016 . Abundant mesopelagic fauna at oceanic high latitudes . Marine Ecology Progress Series , 546 : 277 – 282 . Google Scholar Crossref Search ADS Søreide J. E. , Leu E. , Berge J. , Graeve M. , Falk-Petersen S. 2010 . Timing of blooms, algal quality and Calanus glacialis reproduction and growth in changing arctic . Global Change Biology , 16 : 3154 – 3163 . Solberg I. , Kaartvedt S. 2014 . Surfacing behavior and gas release of the physostome sprat (Sprattus sprattus) in ice-free and ice-covered waters . Marine Biology , 161 : 285 – 296 . Google Scholar Crossref Search ADS PubMed Solberg I. , Klevjer T. A. , Kaartvedt S. 2012 . Continuous acoustic studies of overwintering sprat (Sprattus sprattus) reveal flexible behavior . Marine Ecology Progress Series , 464 : 245 – 256 . Google Scholar Crossref Search ADS Stabeno P. J. , Kachel N. B. , Moore S. E. , Napp J. M. , Sigler M. , Yamaguchi A. , Zerbini A. N. 2012 . Comparison of warm and cold years on the southeastern Bering Sea shelf and some implications for the ecosystem . Deep-Sea Research II , 65 – 70 : 31–45. Steen H. , Vogedes D. , Broms F. , Falk-Petersen S. , Berge J. 2007 . Little auks (Alle alle) breeding in a High Arctic fjord system: bimodal foraging strategies as a response to poor food quality? Polar Research , 26 : 118 – 125 . Google Scholar Crossref Search ADS Stroeve J. C. , Serreze M. C. , Holland M. M. , Kay J. E. , Malanik J. , Barrett A. P. 2012 . The Arctic’s rapidly shrinking sea ice cover: a research synthesis . Climatic Change , 110 : 1005 – 1027 . Google Scholar Crossref Search ADS Steinhart G. B. , Wurtsbaugh W. A. 1999 . Under-ice diel vertical migrations of Oncorhynchus nerka and their zooplankton prey . Canadian Journal of Fisheries and Aquatic Sciences , 56 : 152 – 161 . Google Scholar Crossref Search ADS Suthers I. M. , Sundby S. 1996 . Role of the midnight sun: comparative growth of pelagic juvenile cod (Gadus morhua) from the Arcto-Norwegian and a Nova Scotian stock . ICES Journal of Marine Science , 53 : 827 – 836 . Google Scholar Crossref Search ADS Sundby S. , Drinkwater K. F. , Kjesbu O. F. 2016 . The North Atlantic spring-bloom System—where the changing climate meets the winter dark . Frontiers in Marine Science , 3 : 28 . Google Scholar Crossref Search ADS Sutton T. T. , Clark M. R. , Dunn D. C. , Halpin P. N. , Rogers A. D. , Guinotte J. , Bograd S. J. , et al. 2017 . A global biogeographic classification of the mesopelagic zone . Deep-Sea Research I , 126 : 85 – 102 . Google Scholar Crossref Search ADS Thorne R. E. , Thomas G. L. 1990 . Acoustic observations of gas bubble release by Pacific herring (Clupea harengus pallasi) . Canadian Journal of Fisheries and Aquatic Sciences , 47 : 1920 – 1928 . Google Scholar Crossref Search ADS Tojo N. , Kruse G. H. , Funk F. C. 2007 . Migration dynamics of Pacific herring (Clupea pallasii) and response to spring environmental variability in the southeastern Bering Sea . Deep-Sea Research II , 54 : 2832 – 2848 . Google Scholar Crossref Search ADS Vabø R. , Nøttestad L. 1997 . An individual based model of fish school reactions: predicting antipredator behaviour as observed in nature . Fisheries Oceanography , 6 : 155 – 171 . Google Scholar Crossref Search ADS Varpe Ø. , Fiksen Ø. 2010 . Seasonal plankton–fish interactions: light regime, prey phenology, and herring foraging . Ecology , 91 : 311 – 318 . Google Scholar Crossref Search ADS PubMed Varpe Ø. , Daase M. , Kristiansen T. 2015 . A fish-eye view on the new Arctic lightscape . ICES Journal of Marine Science , 72 : 2532 – 2538 . Google Scholar Crossref Search ADS Vestheim H. , Røstad A. , Klevjer T. A. , Solberg I. , Kaartvedt S. 2014 . Vertical distribution and diel vertical migration in krill beneath snow-covered ice and in ice-free waters . Journal of Plankton Research , 36 : 503 – 512 . Google Scholar Crossref Search ADS PubMed Wyllie-Echeverria T. , Wooster W. S. 1998 . Year-to-year variations in Bering Sea ice cover and some consequences for fish distribution . Fisheries Oceanography , 7 : 159 – 170 . Google Scholar Crossref Search ADS © International Council for the Exploration of the Sea 2018. All rights reserved. For Permissions, please email: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Merrick, Richard

doi: 10.1093/icesjms/fsx228pmid: N/A

Abstract National Oceanic and Atmospheric Administration Fisheries are responsible for the stewardship of the US living marine resources and their habitat and for providing productive and sustainable fisheries, safe sources of seafood, the recovery and conservation of protected resources, and healthy ecosystems to the nation. Their approach to conservation requires, by legislative mandates, that management be informed by science. It has evolved into a four-step approach to providing this advice: (i) the national framework for conservation science, (ii) region specific implementation, (iii) development of unbiased, scientific advice as required by the framework, and (iv) scientists acting, as appropriate, as advocates and science communicators. This approach has been a conservation success where, e.g. 92% of known managed fish stocks are no longer being overfished and 84% of known stocks are at healthy levels, with the latter including 43 stocks rebuilt from depleted levels. In a changing marine climate, it is all the more important that marine conservation decisions be driven by science. Introduction Conservation of living marine resources (LMRs) is a major issue in the United States because of the enormous impact these have on marine ecosystems as well as on the economies of coastal communities and the Nation. The National Oceanic and Atmospheric Administration (NOAA), a bureau of the US Department of Commerce (Figure 1) is responsible to the American people for fulfilling this task. This responsibility is further delegated to NOAA Fisheries, one of the five line offices within NOAA. This office is committed to a four-fold mission to (i) ensure productive and sustainable fisheries, (ii) provide safe seafood, (iii) recover and conserve protected resources, and (iv) maintain healthy ecosystems and resilient coastal communities. American taxpayers entrust the nearly 3000 NOAA Fisheries employees with ca. $945 million annually to support these missions, which, in return, support 1.6 million jobs and $207.6 billion in sales (NMFS, 2017). Figure 1. Open in new tabDownload slide Simple organizational chart of the NOAA including its parent Department of Commerce and its five line offices (including NOAA Fisheries). Figure 1. Open in new tabDownload slide Simple organizational chart of the NOAA including its parent Department of Commerce and its five line offices (including NOAA Fisheries). The US government approach to marine conservation requires, by legislative mandates, that management be informed by science. In so far as US marine conservation is effective, it is largely because of a four-step approach that has evolved from these mandates: Federal legislative and policy drivers provide a national framework for conservation science. This framework is then implemented to meet regional needs. NOAA scientists and their partners develop unbiased, scientific advice as required by the framework. NOAA scientists act, as appropriate, as advocates and science communicators. Ultimately, these four steps are key to maintaining healthy marine ecosystems and resilient coastal communities. The power of this approach is clear from the conservation successes of the United States. Of the more than 500 commercially and recreationally harvested fish species, 91% of stocks with known status are not being overfished and 84% are at healthy levels. NOAA also protects 117 marine mammal stocks and 157 species listed under the Endangered Species Act [ESA (https://www.fisheries.noaa.gov/topic/laws-policies#endangered-species-act)] as endangered/threatened species, most of which are stable or improving in status. In this article, I will explore in more detail the elements of this four-step approach. I will also provide examples of specific successes. The materials here are drawn from an invited presentation made at the Ecosystem Studies of Subarctic and Arctic Seas (ESSAS) Open Science meeting held in Tromsø, Norway 11–15 June 2017. As with the ESSAS presentation, this paper will not evaluate the success or failures of US management of LMRs, but instead will focus on how science is provided to the managers. US structural approaches to marine conservation The US approach to marine conservation has an overarching framework which provides structure for both the science and management of LMRs. The four key legislative mandates (Figure 2) all call for science-based decisions made using best scientific information available (BSIA). Details on each mandate are available from the NOAA Fisheries. For the sake of brevity, I will focus here on one which is illustrative of the others—the Magnuson–Stevens Fishery Conservation and Management Act [MSA (https://www.fisheries.noaa.gov/topic/laws-policies#magnuson-stevens-act)]. Figure 2. Open in new tabDownload slide The key US congressional mandates and federal policies supporting US marine conservation science, which include the Magnuson–Stevens Act, the MMPA, the ESA, the NEPA, and NOAA’s Marine Aquaculture Policy. Figure 2. Open in new tabDownload slide The key US congressional mandates and federal policies supporting US marine conservation science, which include the Magnuson–Stevens Act, the MMPA, the ESA, the NEPA, and NOAA’s Marine Aquaculture Policy. The MSA is the primary law governing marine fisheries management in US federal waters. First passed in 1976, the MSA fosters long-term biological and economic sustainability of US marine fisheries out to 200 nautical miles from shore. Key objectives of the Act are to: prevent overfishing, rebuild overfished stocks, increase long-term economic and social benefits, and ensure a safe and sustainable supply of seafood. Prior to the MSA, US waters beyond 12 nautical miles were international waters and were fished by fleets from other countries. The 1976 law extended US jurisdiction to 200 nautical miles and established fisheries governance that focussed on eight regional fishery management councils (councils) with representation from the coastal states and fishery stakeholders. The councils’ primary responsibility is development of fishery management plans (FMPs) and to make recommendations regarding regulations that implement Federal management in a given region. When reviewing FMPs, FMP amendments, and regulations, the US Secretary of Commerce (as required by the MSA) ensures that they comply with a number of conservation and management requirements, including the 10 National Standards (NSs) (https://www.fisheries.noaa.gov/topic/laws-policies#magnuson-stevens-act)—principles that must be followed to ensure sustainable and responsible fishery management. NS2 requires that “Conservation and management measures shall be based upon the best scientific information available” (https://www.st.nmfs.noaa.gov/science-quality-assurance/national-standards/ns2_revisions). A key element of policies under NS2 is the determination of BSIA, which allows for the use of science from both governmental and non-governmental sources. The criteria to consider when evaluating BSIA are “relevance, inclusiveness, objectivity, transparency and openness, timeliness, verification and validation, and peer review, as appropriate.” Of particular interest to the scientific community is the peer-review process, which ensures that the inclusiveness, quality, and credibility of scientific information and scientific methods meet the standards of the scientific and technical community. Peer review helps ensure objectivity, reliability, and integrity of scientific information. The US Secretary of Commerce, working with each fishery management council (https://www.fisheries.noaa.gov/insight/fisheries-management-united-states), has established a peer-review process which evaluates the scientific information used for conservation and management of the fishery. The Secretary of Commerce publicly identifies the MSA peer-review process in the Federal Register (The Federal Register is the official journal of the federal government of the United States that contains government agency rules, proposed rules, and public notices.) along with a brief description of the process. Detailed information on the review processes are made publicly available on each council’s website, with each council’s Scientific and Statistical Committee playing a pivotal role in the provision of scientific advice (as well as advice on setting each managed stock’s acceptable biological catch). The policies established under NS2 provide clear guidance for how science is introduced, vetted, and used in the management of fish stocks included in regional FMPs. The result is that under the MSA, US fisheries management is a transparent and robust process of science, management, innovation, and collaboration with the fishing industry. For most stocks, a scientific analysis of the abundance and composition of the stock (a “stock assessment”) is conducted and used to determine if the stock is subject to overfishing or is overfished. Using this scientific data, councils set annual catch limits, and if they are exceeded in a fishing year, predetermined accountability measures provide the mechanisms to mitigate the effects of the overfishing. Since 2011, US domestic fisheries have had measures in place to meet the new requirements, and today, 91% of fisheries are maintaining harvest levels at or below agreed-upon annual catch limits. NOAA’s other legislative drivers for marine conservation (https://www.fisheries.noaa.gov/topic/laws-policies) [i.e. National Environmental Policy Act (NEPA), Marine Mammal Protection Act (MMPA), ESA] provide similar frameworks for introducing science into management. These other acts also charge NOAA with conducting scientific evaluations of management actions, but only the MSA sets up a regional governance structure based on regional councils comprised of stakeholders with responsibility for providing sound management advice to regional managers. Regional implementation of the national framework Given the broad diversity of regional conservations issues (Figure 3), it would be inappropriate to attempt to provide a single “national” response. NOAA Fisheries is effective because it treats these as regional issues within a national framework. The eight fishery management council regional fisheries vary markedly. The North Pacific (Alaska) region is dominated by large commercial fisheries, including some of the largest commercial groundfish and salmon fisheries in the world, with landed value in 2015 of $1.7 billion (NMFS, 2017). In contrast, Gulf of Mexico region fisheries are largely recreational, with the value of commercial fisheries being <$0.2 billion in 2015, most of which is from shellfish (NMFS, 2017). Figure 3. Open in new tabDownload slide Selected NOAA Fisheries marine conservation issues by NOAA Fisheries region, ca 2017. Figure 3. Open in new tabDownload slide Selected NOAA Fisheries marine conservation issues by NOAA Fisheries region, ca 2017. To provide the science and management for these regional interests, NOAA Fisheries has 1400 scientists in six regional science centres (with ca. 25 labs), and each science centre is linked to a companion (but physically separate) regional management office. Science centres are led by science centre directors (who are scientists) who report to headquarters via the NOAA Fisheries Chief Science Advisor and not to the lead regional manager/administrator. The regional knowledge and local contacts allow NOAA Fisheries to provide tailored advice for regional problems (while still providing for cross-regional communication of science and management solutions through headquarters offices). Each of the regional science centres also provides their own unique international connections (e.g. the Northeast Centre interacts most frequently with Europe, while the Southeast Centre maintains strong ties with Caribbean and Gulf of Mexico interests.) The headquarters Office of Science and Technology supports national-level bilateral scientific agreements with a variety of nations and scientific organizations. An additional strength of the NOAA Fisheries approach is that the overall scientific portfolio is directed by a single national Chief Science Advisor and Science Board. The Board consists of the directors of the six regional science centres, the director of the national Office of Science and Technology, and the three national-level senior scientists (stock assessment, ecosystems, and economics). In this capacity, the Chief Science Advisor acts as a chief executive, and the Science Board acts as a board of directors to manage and coordinate the regional science portfolios with the national framework. Each science centre, though regionally focussed, is staffed to provide scientific support on all of NOAA Fisheries’ legislative mandates. Most of the scientific advice utilized by NOAA Fisheries managers is produced by the NOAA Fisheries science centres. However, additional direct scientific support is also provided by NOAA’s partners (e.g. academic cooperators). This is particularly important role for NOAA’s Cooperative Institutes (http://ci.noaa.gov/) and Cooperative Science Centers (http://www.noaa.gov/office-education/epp-msi/csc), both of which represent consortia of universities and research institutes. Finally, the administrative framework for presentation of BSIA to NOAA managers provides opportunities for other scientists to provide scientific advice, particularly with respect to stock assessments. Ensuring quality science and scientists Conservation places a heavy responsibility upon NOAA scientists to provide the best possible science. Because of NOAA Fisheries’ regulatory role, decisions are frequently made that are the subject of considerable controversy. NOAA is the most-frequently litigated Federal agency, largely because of its regulatory role. With the legislative drivers and supporting policies in place, it has been difficult for adversaries of NOAA’s regulatory decisions to contest the use of science-based management. However, the need to discredit the science to win a lawsuit has led to attacks on the quality of science and on NOAA scientists. Concerns also exist that science may be manipulated to meet political exigencies or that scientists may be prohibited from expressing their scientific opinions if they conflict with agency policy. This was part of the reason that NOAA Fisheries decided ca. 1999 to separate administration of the science centres from the regulators in the regional management offices, thereby eliminating potential conflicts of interest between science and management at the regional level. All of this has led to the need for strong procedures and policies within NOAA that ensure the continued provision of high quality, unbiased scientific advice including: a NOAA Fisheries Science Quality Assurance Process (SQAP) involving independent peer review of science at all levels, and a NOAA Scientific Integrity Policy designed to guide (and protect) scientists in the ways they express their scientific advice. NOAA Fisheries has developed a SQAP that provides high-quality science as well as prohibits tampering with results. This has produced the most comprehensive scientific review process in NOAA (Figure 4). The NOAA Fisheries SQAP has four levels of review: Figure 4. Open in new tabDownload slide The four elements of NOAA Fisheries Science Quality Assurance Process, including both internal and external [e.g. NAS] review (https://www.st.nmfs.noaa.gov/science-quality-assurance/indexs). Figure 4. Open in new tabDownload slide The four elements of NOAA Fisheries Science Quality Assurance Process, including both internal and external [e.g. NAS] review (https://www.st.nmfs.noaa.gov/science-quality-assurance/indexs). internal review of individual scientist’s fundamental research communications (http://www.nmfs.noaa.gov/op/pds/documents/04/04-113.pdf), which is aligned with NOAA’s policy on review of research communications, and the US Government Information Quality Act; peer review of individual stock assessments (https://www.st.nmfs.noaa.gov/science-quality-assurance/MSA-peer-review-processes/index), which aligns with the US Government’s Office of Management and Budget’s Peer Review Standards (http://www.cio.noaa.gov/services_programs/pdfs/OMB_Peer_Review_Bulletin_m05-03.pdf); annual reviews of each science centre’s research portfolios (https://www.st.nmfs.noaa.gov/science-program-review/index); and NOAA Scientific Advisory Board and National Academy of Sciences (NASs) reviews of cross-cutting, national level topics. NOAA Fisheries employs these nested levels of review for assurance that it is doing the right science, doing it properly, and then applying it appropriately in the management realm. All science destined for public dissemination, whether in peer-reviewed journals or simply as part of a fish stock assessment, is required by NOAA Fisheries to be reviewed as a fundamental research communication. However, in the case of fish and marine mammal stock assessments, a second level of review which focuses on the assessment itself is also required. These are performed through the NS2 peer-review process for fish (see following discussion) and the scientific review groups established under the MMPA for marine mammals. In the case of science with significant impacts (e.g. the implementation of a new national protocol for measuring recreational fishery catches), NOAA Fisheries may also ask for a higher-level review, which can be accomplished through groups such as the NASs. Because all of these reviews focus on single topics or manuscripts, they miss potential improvements of larger science portfolios within the agency’s individual science centres. Annual external reviews of the quality of each centre’s portfolio were implemented beginning in 2012, such that all centres are reviewed each year on the same topic (e.g. in 2017, all centres’ economics and social science programmes were reviewed). The importance of these four levels of review has been especially important with the US Congress, where representatives and senators have called for numerous investigations of NOAA science by the NAS and investigative arms of the Federal government (e.g. the US Government Accountability Office, and the Department of Commerce’s Inspector General). Without exception, all have found the science to be sound. A separate but equal element of NOAA science policies is the NOAA Scientific Integrity Policy. This policy provides a valuable complement to the review programme because it provides clear guidelines on how NOAA scientists can talk about their science. Basically, a NOAA scientist is free to (and is expected to) publish and speak about all of his/her scientific results. At times, these results may be inconsistent or appear to be inconsistent with NOAA Fisheries policy. The Policy provides guidance to NOAA scientists in how to do this in an appropriate way (e.g. by providing a disclaimer that the article represents the opinion of the author(s) and not NOAA Fisheries). The role of advocacy for government scientists The reality is that successful conservation science, as defined by the Agency’s mandates, is actually accomplished through the work of individual scientists and managers. It should go without saying that government scientists must be viewed as the vendors of unbiased, neutral science. That is the reason for many of the oversight activities that NOAA has instituted to ensure sound scientific advice. Nonetheless, there remains a role for government scientists to act as advocates for conservation. Scientists should be committed to not only owning their science, but to getting others to also own it. Important conservation issues demand this, because scientists are frequently in the position of changing a culture through what amounts to “disruptive science”. Here are two quick examples (one small and one large) where scientists acting as advocates have provided the vision necessary to achieve major conservation successes. The first example is conservation of the vaquita (Phocoena sinus). This may presently be the world’s rarest marine mammal, which is on the edge of extinction. The 2016 Joint Mexico–United States survey of the population (endemic to the Gulf of California) found < 30 animals remaining, which appears to be a decrease of 50% since the 2015 survey (Morrell, 2017). This porpoise is often caught and drowned in gillnets used by illegal fishing operations in marine protected areas within the Gulf. If not for the efforts of a small group of Mexican and United States scientists who have repeatedly raised the alarm that the population is precipitously declining and that the illegal bycatch needs to be addressed, it is likely that the species would go the way of the baiji or Chinese river dolphin (Lipotes vexillifer), which is now considered functionally extinct (Turvey et al., 2007). The same US scientists worked with Chinese scientists to raise alarm over the status of baiji, but their efforts came too late. By 2007, there was sufficient notice of the baiji’s imperilled status for the conservation community to fund a new survey, but no animals were found over a 6-week survey. These same US scientists learned from the baiji work that they needed to raise the alarm with the Mexican and US governments and to also communicate this information to the public through lay literature if the declining vaquita population was to be saved. As such, they have worked cooperatively over the past decade to monitor population status and have also worked with the two governments and the conservation community to address the illegal fishery. At present, there is hope that the remaining vaquita can be taken into captivity, while their habitat issues (i.e. illegal fishing) are dealt with by the Mexican government (Goldfarb, 2017). A second example of how a scientist advocate can move conservation forward is the adoption of a formal policy supporting ecosystem-based fishery management (EBFM) by the US government (https://www.st.nmfs.noaa.gov/ecosystems/ebfm/creating-an-ebfm-management-policy). NOAA Fisheries has worked for almost two decades to adopt an EBFM to maintain ecosystems in a healthy, productive, and resilient condition so that they can provide the services that humans want and need (EPAP, 1998; Hilborn, 2004; Pikitch et al., 2004). During this time, fisheries science centres and their academic partners have made significant progress in understanding the scientific underpinnings of EBFM. With this said, NOAA has made little progress adopting a comprehensive, national programme of EBFM until it created a position of Senior Scientist for Ecosystems in 2012. Operationalizing EBFM nationally became a major responsibility of this new Senior Scientist, and having a senior leader focussed on national ecosystem issues resulted in the rapid progress made on producing an actual written policy. NOAA Fisheries has now adopted both an EBFM policy and a roadmap for implementing the policy, largely because of the herculean efforts of dedicated Agency staff (https://www.st.nmfs.noaa.gov/ecosystems/ebfm/creating-an-ebfm-management-policy). The US National Ocean Council has held up this effort as an example for other federal agencies seeking to adopt ecosystem-based management (EBM). Similar progress has been made under the Senior Scientist for Stock Assessments in the development of the “Next Generation Stock Assessment Improvement Plan” and in the development of national policies for the prioritization of stock assessments. Tasking senior leadership with the responsibility of implementing national policies seems to be a good example of how advocacy can achieve positive results in government. Scientists as communicators It has taken more than a passion for conservation science to make these scientists effective advocates. To move conservation forward, they must also communicate with stakeholders and partners, especially with non-scientists. Scientists are accustomed to identifying the quality (precision) of the science they provide. But, we should pay equal attention to whether the information we are providing is what the managers, partners, and stakeholders really need (accuracy or the “right science”). Doing the latter is problematic, particularly because scientists often have a hard time engaging non-scientists in effective dialogue to identify scientific questions and needs. Although scientists may have an unclear vision of what their partners and stakeholders need in the way of scientific advice, the latter often do not know either. As such, it is important to pursue strategies for providing advice that begin by initiating this dialogue. Engagement with stakeholders is a key element of integrated ecosystem assessments as defined both by ICES (as a specific action area) and by NOAA (https://www.integratedecosystemassessment.noaa.gov; Figure 5, see in particular Step 1: Define EBM Goals and Strategies.). Jointly defining goals and targets with a suite of interested parties and stakeholders greatly improves the likelihood that the process will provide useful advice. Similarly, NOAA Fisheries’ strategy for providing scientific advice to managers dealing with climate change has been initiated through a similar joint conversation (Figure 6). There the explicit “end game” is climate informed reference points produced through management strategies driven by managers, but supported by Agency scientists. Figure 5. Open in new tabDownload slide Conceptual model for NOAA's approach to integrated ecosystem assessments. The process begins with the definition of EBM goals and targets and continues through the evaluation of management strategies (https://www.integratedecosystemassessment.noaa.gov/). Figure 5. Open in new tabDownload slide Conceptual model for NOAA's approach to integrated ecosystem assessments. The process begins with the definition of EBM goals and targets and continues through the evaluation of management strategies (https://www.integratedecosystemassessment.noaa.gov/). Figure 6. Open in new tabDownload slide Hierarchy of elements comprising NOAA Fisheries Climate Science Strategy (https://www.st.nmfs.noaa.gov/ecosystems/climate/national-climate-strategy). Figure 6. Open in new tabDownload slide Hierarchy of elements comprising NOAA Fisheries Climate Science Strategy (https://www.st.nmfs.noaa.gov/ecosystems/climate/national-climate-strategy). Conclusion The successes that NOAA Fisheries has had in providing science for conservation are based on a combination of: legislative mandates and NOAA policies focussed on science, regional implementation of the mandates and policies, science quality assurance, and individual scientists acting as advocates and communicators. Using this rubric, NOAA has addressed a multitude of conservation issues over the past two decades where scientific advice has driven the process: ending overfishing for most US fish stocks; rebuilding 41 formerly overfished fish stocks to healthy levels; reducing the incidental mortality of marine mammal species from anthropogenic sources through application of formal and informal take reduction processes; developing and implementing a transparent and effective process for providing the scientific advice to facilitate the listing, delisting, and downlisting of species under the ESA; making climate change impacts an explicit consideration in NOAA Fisheries decision making; and implementing EBFM. These have all been major conservation issues—all driven by scientific advice—all with a potential for major conflicts. Most have been successes (though some are still works in progress). Many of these have had significant economic and social impacts and several have resulted in litigation. In almost every case where science has been called into question, NOAA Fisheries and its sound science has prevailed. Acknowledgements I thank the many scientists and managers within NOAA Fisheries who have made the Agency a conservation success story. I also thank Cisco Werner, Jon Hare, Jason Link, and Michael Simpkins, as well as three anonymous reviewers, for their comments on an earlier version of this paper. Finally, I thank the Norwegian Institute of Marine Research for providing travel support to the 2017 ESSAS meeting. The scientific results and conclusions, as well as any views or opinions expressed herein, are those of the author and do not necessarily reflect those of NOAA or the Department of Commerce. References EPAP (Ecosystem Principles Advisory Panel) . 1998 . Ecosystem based fishery management: A report to Congress by the Ecosystem Principles Advisory Panel. US National Marine Fisheries Service, Silver Spring, MD. 62 pp. Goldfarb B. 2017 . Scientists-mull-risky-strategy-save-world-s-most-endangered-porpoise. Science. Available from http://www.sciencemag.org/news/2016/07/scientists-mull-risky-strategy-save-world-s-most-endangered-porpoise Hilborn R. 2004 . Ecosystem-based fisheries management: the carrot or the stick? Marine Ecology Progress Series , 274 : 275 – 278 . Google Scholar OpenURL Placeholder Text WorldCat Morrell V. 2017 . World’s most endangered marine mammal down to 30 individuals. Science, 355: 558–559. NMFS (National Marine Fisheries Service) . 2017 . Fisheries Economics of the United States, 2015. US Department of Commerce, NOAA Technical Memorandum NMFS-F/SPO-170. 247 pp. Pikitch E. K. , Santora C., Babcock E. A., Bakun A., Bonfil R., Conover D. O., Dayton P. et al. 2004 . Ecosystem-based fishery management . Science , 305 : 346 – 347 . Google Scholar Crossref Search ADS PubMed WorldCat Turvey S. T. , Pitman R. L., Taylor B. L., Barlow J., Akamatsu T., Barrett L. A., Zhao X. et al. 2007 . First human-caused extinction of a cetacean species? Biology Letters, 3: 537–540. Published by International Council for the Exploration of the Sea 2017. This work is written by a US Government employee and is in the public domain in the US. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) Published by International Council for the Exploration of the Sea 2017. This work is written by a US Government employee and is in the public domain in the US.

Drinkwater, Kenneth, F;Kristiansen,, Trond

doi: 10.1093/icesjms/fsy077pmid: N/A

Abstract Following rapid cooling in the 1960s, much of the North Atlantic Ocean was characterized by a cold period during the 1970s and 1980s. This cold period was part of the multidecadal variability in sea surface temperatures known as the Atlantic Multidecadal Oscillation or AMO, which has a period of ∼60–80 years. During this cold period, below average air and sea temperatures predominated, increased ice cover was observed in those northern regions with seasonal sea ice, and evidence was found of reduced Atlantic inflow into the Northeast Atlantic Ocean. The ecological responses included a reduction in primary production and geographic shifts in zooplankton species. Also, there was a general southward expansion of arctic and boreal fish species and a retreat of the temperate species. Major fish stocks such as Atlantic cod off Greenland and Labrador/northern Newfoundland, as well as the Norwegian spring-spawning herring, collapsed commercially. These collapses were partly driven by climate-induced declines in growth rates and recruitment survival, as well as fishing. In contrast, in the more southern range of Atlantic cod, such as the North Sea, the opposite response occurred as the cool conditions led to improved growth rates and higher abundance. Long-term measurements in the English Channel documented the replacement of several warm-water species with more northern cold-water species. Benthic and nearshore species also underwent distributional shifts and changing abundances. Comparisons with the responses to the warm periods suggest that following the cold period of the 1970s and 1980s, the ecosystem in the 1990s and 2000s returned to conditions akin to what they were in the previous warm period of the 1930s–1950s. However, there were some notable exceptions, such as the continued low abundance of Atlantic cod off West Greenland and Labrador/northern Newfoundland. Introduction The surface waters of the North Atlantic have undergone low-frequency temperature variability with a period of ∼60–80 years since at least the 1730s (Saenger et al., 2009). This signal has been termed the Atlantic Multidecadal Oscillation or AMO (Kerr, 2000). Although many prefer the term Atlantic Multidecadal Variability or AMV (e.g. Park and Latif, 2008; Häkkinen et al., 2011), in this article we have maintained the AMO terminology as it has been the one most frequently used in ecological papers. AMO indices have been defined by Delworth and Mann (2000), Enfield et al. (2001), and Sutton and Hodson (2005) among others, all of which are related to the annual mean sea surface temperature (SST) averaged over the North Atlantic after the linear trend has been removed and a low frequency filter applied. Slight differences arise between the indices due to differences in the area over which the averages are estimated, the method of detrending, and the filtering involved. A positive AMO index is associated with above average North Atlantic SSTs and a negative index with below average temperatures. This multidecadal signal has also been observed in subsurface waters down to at least 400- to 500-m depth (Antonov, 1993; Frankcombe et al., 2008). The signal is not only limited to the North Atlantic as it been shown to extend into the Arctic (Chylek et al., 2009; Drinkwater et al., 2014). In the last century, waters throughout most of the North Atlantic were generally cool in the late 1800s and early 1900s, warm during the 1930s to the 1960s, cool in the 1970s and 1980s and have returned to warm conditions since the 1990s (Kushnir, 1994; Sutton and Hodson, 2005; Alexander et al., 2014; Figure 1). The AMO signal accounts for ∼60% of the warming in North Atlantic SSTs from 1970 to 2000 (Polyakov et al., 2010). Figure 1. View largeDownload slide Monthly detrended AMO index for the period 1856 to September 2017 obtained from https://www.esrl.noaa.gov/psd/data/timeseries/AMO/. Values are based on SSTs averaged over 0°–70°N. Smoothed line shows 10-year running means. Figure 1. View largeDownload slide Monthly detrended AMO index for the period 1856 to September 2017 obtained from https://www.esrl.noaa.gov/psd/data/timeseries/AMO/. Values are based on SSTs averaged over 0°–70°N. Smoothed line shows 10-year running means. Although it has been suggested that the AMO variability is related to changes in the large-scale Meridional Overturning Circulation (Latif et al., 2004, 2006; Klöwer et al., 2014), this is still much in debate. Several other mechanisms have been proposed, e.g. volcanic activity (Otterlå et al., 2010), natural aerosols (Booth et al., 2012), atmospheric blocking (Häkkinen et al., 2011), solar forcing (Knudsen et al., 2011), random atmospheric variability (Griffies and Bryan, 1997) and atmosphere–ocean feedbacks (Timmermann et al., 1998). The AMO index is correlated with air temperatures and rainfall over much of the Northern Hemisphere, including North America, Europe, and the Sahara (Sutton and Hodson, 2005; Knight et al., 2006). For example, the two most severe droughts of the 20th century in the United States, the dust bowl of the 1930s and the 1950s drought occurred during the positive AMO between the 1920s and 1960s. In contrast, extreme rainfall fell during this time in Florida and the Pacific Northwest. Hurricane activity has also been linked to changes in surface temperatures that are in phase with the AMO variability with more and stronger storms during warm periods (Goldenberg et al., 2001). A few studies have provided reviews of multidecadal variability from physical changes to ecosystem responses in the North Atlantic. Mann and Lazier (2006) provided several examples of linkages between low-frequency climate variability and ecosystem changes. Oviatt et al. (2015), in a study focussed on global decadal variability, also discussed multidecadal variability, including in the North Atlantic. Earlier, Cushing and Dickson (1976) discussed physical changes and biological responses in the North Atlantic to both the warm (1920s–1950s) and cold (1960s–1980s) periods, consistent with the AMO variability. More recently, Edwards et al. (2013) found a correlation between the AMO and long-term plankton data from the continuous plankton recorder (CPR). There have been strong ecological responses, especially of fish populations, linked to the AMO variability (e.g. Alheit et al., 2014; Nye et al., 2014). By far the largest number of temporally focussed studies have been associated with the warm periods, i.e. both the mid-20th Century warming and the recent warming. The former drew the attention of many fisheries scientists in the 1920s and 1930s as several commercial fish stocks in the high latitude regions of the North Atlantic underwent large distributional changes and recorded increased production (Jensen and Hansen, 1931; Sæmundsson, 1934; Jensen, 1939; Tåning, 1943,1949). This led to an ICES-sponsored conference on climate change in 1948 (ICES, 1949), which would have been held at least a decade earlier if WWII had not interfered with the original conference plans. Several publications later documented some of the ecological responses to this warming event, principally on fish populations, including papers by Beverton and Lee (1965), Russell et al. (1971), Cushing and Dickson (1976), and Cushing (1982), and more recently by Brander et al. (2003), Hawkins et al. (2003), Southward et al. (2004), Drinkwater (2006), and Sundby and Nakken (2008). It was not only fish populations that responded to this early 20th Century warming but there is some evidence that phytoplankton and zooplankton increased during the warm period in the northern regions and may have been partly responsible for the distributional and abundance changes of the fish stocks (Hawkins et al., 2003; Southward et al., 2004; Drinkwater, 2006). Large numbers of publications have been documenting responses in the North Atlantic and other oceans to the recent warming. For example, responses include northward distributional shifts in plankton (Beaugrand et al., 2002) and changes in fish distribution, growth, and timing of migration and reproduction (e.g. Crozier and Hutchings, 2014; Pecl et al., 2017). Fewer publications have attempted to document the extent of the physical oceanographic and ecological changes in the northern North Atlantic that accompanied any of the cold periods associated with the AMO. Therefore, the first objective of this article is to provide a synthesis of the changes to the marine ecosystems of the northern North Atlantic during the cold period of the 1960s–1980s. For this we draw upon the numerous publications that have discussed either regional oceanographic and biological responses during this time or dealt with specific species responses within the North Atlantic. Our second objective is to compare the changes that occurred during this cold period with those observed during the preceding and following warm periods. This should help to determine if the responses during the cold period are simply the opposite of what happens during the warm periods. In the next section, we examine the atmospheric and physical oceanographic changes that occurred during the cold period. Our main study area is the northern North Atlantic from the Arctic to ∼30°N (Figure 2). This is followed by a discussion of the ecological responses to the changes in the physical oceanography. The final section discusses the results and compares the ecosystem responses during warm and cold periods. Figure 2. View largeDownload slide The study area of the North Atlantic Ocean. Figure 2. View largeDownload slide The study area of the North Atlantic Ocean. Physical changes Atmosphere Following the warm period that extended over much of the North Atlantic from the 1920s to its peak in the 1940s–1950s, air temperatures declined quickly in the 1960s resulting in below average temperatures in the 1970s and 1980s (Lamb, 1982; Johannessen et al., 2004). The precise timing of the decrease and its amplitude varied spatially. Cooling began first in the Arctic during the 1940s and 1950s and reached minimum temperatures there in the late-1960s (Chylek et al., 2009). This cooling was accompanied by increases in the areal coverage of snow and ice, which probably intensified the cooling through feedback mechanisms. Farther south, the cooling occurred later, starting in the mid-1960s and were minimum in the early to mid-1970s (Chylek et al., 2009; Johannessen et al., 2004). This cooling resulted in glaciers in Europe advancing, reversing the receding trend that had occurred for several decades during the earlier warm period (Lamb, 1982). Similar advances of glaciers off West Greenland were observed during the cold period of the 1960s to the 1980s (Lloyd et al., 2011). The cause of this atmospheric cooling is uncertain. One contributing factor might have been the reduction in the available solar energy, which was estimated globally at ∼4% from the 1940s to the 1960s (Budyko, 1969). This reduction was caused by lower atmospheric transparency, mostly attributed to volcanic dust particles, although an increase in atmospheric dust due to human activity was also believed to have played a role. Budyko (1969) estimated that this 4% reduction was enough to account for the observed global change in air temperature. On a more local scale, a 30% reduction in solar radiation reaching the sea surface at the Shetland Islands was observed between 1948 and 1965, especially in spring, due to increased cloud cover (Cushing and Dickson, 1976). During the cooling period, there was a slight equatorward shift in the axes of the winds and pressure systems. Westerly winds weakened while outbreaks of northerly winds occurred with increasing frequency over the Nordic Seas. The latter winds were associated with the build-up of a strong persistent pressure ridge over Greenland, especially in winter (Lamb, 1972; Cushing and Dickson, 1976). This increase in northerly winds brought cold Arctic air southward thereby further contributing to the decline in winter air temperatures over the Atlantic Ocean. Ocean temperatures Cooler air temperatures and the stronger northly winds contributed to decreasing ocean temperatures throughout much of the North Atlantic during the 1960s and 1970s as reflected in the AMO index (Figure 1). Sea temperatures north of 30°N generally remained cool through to the mid- to late-1980s (Kushnir, 1994). This decrease occurred in the upper layers of the ocean to about 500 m (Antonov, 1993). The cold period occurred roughly at the time when the Atlantic Meridional Ocean Circulation (AMOC) is thought to have slowed down thus transporting less warm water to the north (Rahmstorf et al., 2015). (The AMOC is the large-scale Atlantic Ocean circulation pattern that consists of warm, salty water in the upper layers flowing north and colder, deep waters flowing south.) At the same time as the slowing down of the AMOC, there was increased southward flow out of the Arctic that further contributed to the cooling and freshening of the waters, especially off East Greenland (Cushing and Dickson, 1976; Lamb, 1982). Although there was general cooling throughout much of the North Atlantic at that time, there still was large spatial variability in the timing and intensity of the cooling (Figure 3). The cooling appears first in the north and east in the Arctic and Barents Sea and several years later in the Labrador Sea (Figure 3). In an investigation of the subsurface (0–400 m) temperatures south of 60°N, Frankcombe et al. (2008) found westward propagation of the low-frequency temperature anomalies. These were most evident at 300–400 m with a slight phase shift between the surface (0–80 m) and the 300- to 400-m depth layer. They noted a cold anomaly in the eastern part of the basin after 1960 that travelled westwards, reaching the western part of the basin after 1970. Figure 3. View largeDownload slide SST anomaly for the northern North Atlantic averaged over 5-year periods: (a) 1950–1955, (b) 1955–1960, (c) 1960–1965, (d) 1965–1970, (e) 1970–1975, (f) 1975–1980, (g) 1980–1985, (h) 1985–1990, (i) 1990–1995, (j) 1995–2000. Anomalies estimated as 5-year averaged deviations from the 1960 to 1991 climatology using the SODA database (http://www.atmos.umd.edu/ocean/). Figure 3. View largeDownload slide SST anomaly for the northern North Atlantic averaged over 5-year periods: (a) 1950–1955, (b) 1955–1960, (c) 1960–1965, (d) 1965–1970, (e) 1970–1975, (f) 1975–1980, (g) 1980–1985, (h) 1985–1990, (i) 1990–1995, (j) 1995–2000. Anomalies estimated as 5-year averaged deviations from the 1960 to 1991 climatology using the SODA database (http://www.atmos.umd.edu/ocean/). In addition to changes in ocean temperatures, there were also changes in the circulation patterns. The East Greenland current flows southward along the continental slope off East Greenland and carries with it a mixture of the waters flowing out of the Arctic through Fram Strait and recirculating waters from the Norwegian Sea. South of 70°N, part of this current branches off to the east forming the East Icelandic Current. It eventually reaches the area off the northern Icelandic coast. In the 1960s, this current intensified causing sea temperatures to decrease off northern Iceland (Stefánsson, 1969). The stronger East Icelandic Current also pushed the ocean front that lies along the southern sections of the Jan Mayen Ridge farther southeastward into the Norwegian Sea. This front separates colder, fresher water in the Iceland Sea from the warmer, saltier water of the Norwegian Sea. In the Norwegian Sea, the circulation is dominated by northward flowing warm and salty Norwegian Atlantic Current, which is an extension of the Norwegian Atlantic Current (Mork and Skagseth, 2010). In the 1960s, this current weakened (Blindheim et al., 2000) and together with the eastward extension of the East Icelandic current likely contributed to the observed colder and fresher conditions in the Norwegian Sea. In contrast, the warm Irminger Current off the continental shelf of southeast Greenland strengthened (Blindheim, 1967; Hermann 1967; Dickson and Lamb, 1972). The intensification of this current, which flows into the Labrador Sea, may have accounted for the lack of substantial cooling there during the 1960s. Indeed, through most of the 1960s, temperatures in the Labrador Sea increased but salinities decreased, resulting in strong vertical stratification that limited winter convection to shallow depths (Lazier, 1980). The low salinity conditions were attributed to an anomalously high atmospheric pressure cell over Greenland that generated the northly winds along the east coast of Greenland causing more, cold polar water to reach the interior of the Labrador Sea (Dickson and Lamb, 1972). However, by 1972 strong northwesterly winds over the Labrador Sea resulted in renewed deep-water convection with a rapid temperature decline in the upper layers of the Labrador Sea as well as a freshening of the deep Labrador Sea Water (Dickson and Lamb, 1972). These changes in the Labrador Sea occurred at the same time as the deep Greenland Sea waters warmed, became saltier, and thinned resulting in a decrease in deep convection (Dickson et al., 1996). The cold mid-depth temperatures in the Labrador Sea continued until the late 1990s (Curry et al., 1998). On the European shelves, sea temperatures were below normal in the 1970s and 1980s including the North Sea, the Baltic Sea, the Celtic Sea, and the Bay of Biscay and the Iberian Shelf (Alheit et al., 2014). Cooler conditions also prevailed in the English Channel (e.g. Southward et al., 1995; Hawkins et al., 2003; Philippart et al., 2011). In the North Sea, Corten (1986, 1990) suggested that part of this decrease was caused by a sustained reduction of Atlantic inflow into the northwestern North Sea in the 1960s and 1970s. Alheit et al. (2014) showed that the cooling also extended into the Mediterranean Sea. Even in the tropical regions of the central North Atlantic, SSTs cooled during the 1970s and 1980s. This cooling is believed to have contributed to a period of minimum hurricane activity (Nyberg et al., 2007). Sea ice Seasonal sea ice occurs in the northern North Atlantic in the Barents and Nordic seas as well as in the Labrador Sea off Greenland and Labrador with high interannual variability in the areal extent of the ice. Greater ice coverage in the 1960s and 1970s is clearly shown by the difference in the sea-ice coverage in percent ice concentration between 1965–1975 and 1994–2000 for the winter (Jan–Mar) and spring (Apr–Jun) for the northern North Atlantic (Figure 4). The predominance of northerly winds and cold air temperatures caused a large flux of ice out of the Arctic expanding ice farther south and east off East Greenland than usual. The eastward expansion was particularly noticeable in the Odden area of the Greenland Sea (Vinje, 2001; see also Figure 4). The East Icelandic Current transported larger amounts of drift ice eastward from the Greenland Shelf to the northern coast of Iceland (Stefánsson, 1969), in contrast to the warm period of 1948–1963, when little to no drift ice was observed (Malmberg, 1984). The East Greenland Current transports drift ice, called “storis”, to West Greenland. Valeur (1976) presented the number of months “storis” appeared off West Greenland between October and the following September for the years 1899/1900–1971/1972. A minimum in the duration of “storis” was recorded during the warm period of the 1930s to the early 1960s; then gradually increased through to the end of the record. The duration of “storis” based on the detrended data from Valeur (1976) correlates significantly with the AMO (R = 0.72, p = 0.02; Figure 5) supporting the contention of a longer ice duration during the 1960s–1980s cold period. Figure 4. View largeDownload slide Average sea coverage in % for Jan–Mar in (a) 1965–1975, (b) 1994–2000, (c) difference in percentage, and for Apr–Jun for (e) 1965–1975, (f) 1994–2000, and (g) difference in percentage. Figure 4. View largeDownload slide Average sea coverage in % for Jan–Mar in (a) 1965–1975, (b) 1994–2000, (c) difference in percentage, and for Apr–Jun for (e) 1965–1975, (f) 1994–2000, and (g) difference in percentage. Figure 5. View largeDownload slide The relationship between the ice duration of “storis” off West Greenland and the AMO index. A 10-year running mean has been applied to both time series. Figure 5. View largeDownload slide The relationship between the ice duration of “storis” off West Greenland and the AMO index. A 10-year running mean has been applied to both time series. Ecological responses Phytoplankton Phytoplankton production depends upon available nutrients, sufficient light levels, and generally strong enough vertical stratification of the water column to maintain the plankton near surface where light levels are high. If the stratification is too strong however, vertical mixing is suppressed, the surface layer nutrients will quickly be depleted, and production will be limited. In the Arctic, the annual phytoplankton production is generally dominated by a single bloom in late spring or early summer, which occurs in response to increases in mean irradiance in combination with a stable mixed layer formed by freshwater from melting ice. In the subarctic, a stronger and earlier spring bloom occurs compared with the Arctic and is usually in response to increased stratification due to increasing upper layer temperatures, freshwater from melting ice, or increased river runoff. A fall bloom typically occurs when strong winds breakdown the stratification and replenish the nearly depleted nutrients in the upper mixed layer. Farther south in the subtropical Atlantic, the phytoplankton biomass peaks during winter when mixing by winds replenishes the euphotic zone with nutrients. The production peak is much reduced in comparison to high latitude spring blooms. The cold period of the 1960s to the 1980s tended to be one of relatively low phytoplankton production in the North Atlantic. Boyce et al. (2010), in a study of interannual phytoplankton trends in the world’s oceans from published Secchi-depth data and in situ chlorophyll-a measurements between 1900 and the early 2000s, found that phytoplankton production had a strong peak in the North Atlantic in the 1930s to the 1950s. Thereafter, the phytoplankton index declined through the late 1950s and 1960s, remained low (∼25% of the earlier peak) through the 1970s and 1980s and showed signs of increasing in 1990s. They also found that mean chlorophyll levels in the Atlantic were weakly positively correlated with the AMO, indicating a response to basin-wide physical forcing. In the Barents Sea, although long phytoplankton time series are not available, one of the main influences on primary production is thought to be sea ice through limiting light levels (Rey and Loeng, 1985). From modelling studies, primary production was found to be as much as 400% higher in a warm year with reduced ice compared with a cold year with increased ice coverage (Slagstad and Wassmann, 1997). In recent warm years with reduced ice, primary production in the Barents Sea has been observed to have increased, which has been attributed to both higher light levels and a longer production period (Mueter et al., 2009). The longer production period was due to both an earlier start date and later end date owing to the reduced sea-ice season. Additional modelling studies by Skaret et al. (2014) predicted increases in Barents Sea primary production under climate change of around 36%, although their model indicated that most of this production would occur in the ice-free regions in the southeast. Kristiansen et al. (2014) found that the overall modelled phytoplankton production increased by 15–30% by 2099 in the Barents Sea due to increased temperatures and reduced sea-ice (more open water). However, most of the increase in production was attributed to small phytoplankton, while large phytoplankton production decreased. These studies strongly suggest that during the cold period of the 1970s and 1980s, primary production was lower than in the warm periods, and perhaps significantly so. On the shelf off northern Iceland, in cold years when there is reduced inflow of Atlantic water and larger quantities of Arctic Water, spring blooms are less intense due to reduced nutrient levels and stronger vertical stratification (Thordardottir, 1984; Astthorsson and Vilhjálmsson, 2002). During the 1960s to the mid-1980s when such environmental conditions prevailed north of Iceland, the mean primary productivity was indeed observed to be low (Gudmundsson, 1998). After the mid-1980s, as environmental conditions improved, the mean productivity increased by a factor of 2 to 3 (Gudmundsson, 1998). Off West Greenland, Lloyd et al. (2011) studied a 100-year-old time series of benthic foraminiferal data from Disko Bugt. They showed that cold-water (warm-water) forams increased (decreased) in abundance during the cold period of the 1960s to the mid-1980s. They also reported a positive connection between West Greenland ocean temperatures and the AMO. Edwards et al. (2001) found that during the mid-1960s to the early 1980s, large areas of the North Sea exhibited low phytoplankton biomass, weaker spring blooms and shorter growth periods than the long term (1960–1995) means based on the CPR greenness index. In the region to the west of the European continental shelf to 15°W and between 47° and 62°N, the signal was less evident except in the north where there tended to be low values during the cold period, although not necessarily minimum values. The initiation of the spring bloom occurred later by over a month in the North Sea between the 1950s and the 1970s and about 2 weeks later in the area to the west of the British Isles in the Northeast Atlantic (Glover et al., 1972). Cushing and Dickson (1976) noted the spring bloom was delayed by almost 1 month between 1948 and 1968 at the southern entrance to the Norwegian Sea and speculated that it was due to the reduction in solar radiation caused by increased cloud cover. In the English Channel, phytoplankton species composition during the cold period of the 1970s and 1980s resembled that recorded in the previous cold period in the 1920s (Hawkins et al., 2003). These species were representative of more northern waters, in contrast to the warm period (1930s–1950s) when they were more typical of southern waters (Southward et al., 2004). This oscillation between different species under variable temperatures in the English Channel was given the term the Russell Cycle by Cushing and Dickson (1976). Zooplankton The dominant zooplankton throughout much of the North Atlantic is Calanus finmarchicus, with Calanus helgolandicus more abundant to the south in temperate waters in the Northeast Atlantic and Calanus glacialis and Calanus hyperboreous occurring farther north in colder Arctic waters and in the Northwest Atlantic. C. finmarchicus is extremely important as prey for Atlantic cod larvae and early juveniles, especially in the more northern areas (Sundby, 2000; Heath and Lough, 2007). Off West Greenland, zooplankton samples were collected on three transects in June–July from 1950 to 1985 in every year but four. Species identification began in 1956. Pedersen and Smidt (2000) and Pedersen and Rice (2002) showed that zooplankton abundance declined as temperatures fell from peaks during the warm period of the 1950s and early 1960s and remained below normal through to the mid-1980s. These authors also found that the total zooplankton abundance was weakly correlated with temperature, although the abundance of C. finmarchicus alone was not significantly correlated with temperature. Off northern Iceland, zooplankton abundance fell as waters cooled in the 1960s (Astthorsson et al., 1983). This was primarily a result of the decline in C. finmarchicus, which constitutes 60–80% of the zooplankton biomass in spring (Astthorsson and Vilhjálmsson, 2002). Zooplankton biomass levels did not significantly recover off the north coast of Iceland until the warm period of the 1990s (Astthorsson et al., 1983, 2007). The total number of copepods and the total zooplankton biomass in the North Sea and the Northeast Atlantic west of the United Kingdom declined from the 1950s to the 1970s (Glover et al., 1972). In addition, the length of the zooplankton season decreased by over 1 month in the North Sea and ∼3 weeks in the NE Atlantic. Beaugrand and Kirby (2010) used a Plankton Index for the North Sea based on the abundance of C. finmarchicus, C. helgolandicus, Euphausiids, Pseudocalanus spp., the total calanoid copepod biomass, and the mean size of female calanoid copepod, all of which were prey for larval cod. This index showed high values during the 1960s and 1970s, and declined through the early 1980s. From the mid-1980s to the end of their record in the late 2000s, the plankton index was declining or low. The minimum index value was at the end of their time series in 2007. Based on data from the CPR survey, Lindley and Batten (2002) showed a period of low plankton species diversity during the late 1970s and early 1980s in the most southerly region of the North Sea, farthest from the region of Atlantic inflow. Species associated with inflow of oceanic or mixed waters from the Atlantic or shelf to the west and south of Britain increased in abundance or frequency of occurrence. Meroplankton also increased but resident holoplankton and those associated with colder oceanic or mixed waters declined. These changes resulted in an increase in the species richness in the areas in the northwestern North Sea. The evidence for a long-term trend was stronger than relationships between diversity and either the North Atlantic Oscillation (NAO) or variation in position of the Gulf Stream in the western Atlantic (Lindley and Batten, 2002). In a study of long-term variations in Atlantic plankton species in the North Sea, Corten (1999) examined 14 species from samples of the CPR Survey during 1948–1996 that were potential indicators of Atlantic inflow. Using Principal Component Analysis, he classified the species into sub-groups, each of which showed a different variation over time. The group, consisting of Candacia armata, Metridia lucens, and Tomopteris spp., showed temporal variations that appeared related to changes in Atlantic inflow being at low levels in the North Sea from 1965 to 1982, and increased in the following years. In the western entrance to the English Channel, northern cold-water zooplankton species became more abundant after 1962 as ocean temperatures cooled and remained so until warming in the 1980s (Hawkins et al., 2003). These changes were also part of the Russel Cycle. Cushing and Dickson (1976) noted that in 1970, the abundance of macroplankton increased in the channel for the first time since 1930. The cold-water species Sagitta elegans returned to the English Channel in 1972 having been essentially absent since the early 1930s and by 1973 reached abundances not seen since 1931 (Southward, 1974). In addition, the cold-water arrow worm (chaetognath) species Parasagitta elegans (formerly known as Sagitta elegans), increased dramatically during the cold period 1970–1980 while the warm-water inshore species (Sagitta setosa) declined (Hawkins et al., 2003). Other cold-water species replaced the more warm-water zooplankton species (Southward, 1980). The jellyfish species Muggiaea atlantica also increased in abundance beginning around 1960 (Blackett et al., 2014). Although typically residing farther south, its success in the western English Channel during the cooler period was attributed by the authors to the increased abundance of planktonic prey from the northwest and a reduction in competition by European pilchards (Sardina pilchardus). Nearshore coastal species Important changes also occurred nearshore during the cold period. Acorn barnacles on the intertidal rocky shores of southern England have shown distributional shifts with changing temperatures (Hawkins et al., 2003). The succession of cold winters and poor summers between 1962 and 1967 resulted in the decline of southern warm-water barnacle species Chthamalus spp., while the more northern, colder water barnacle, Semibalanus balanoides, increased and became abundant (Southward, 1967). During the 1970s to the mid-1980s, the numbers of S. balanoides fluctuated but since 1985 they have become scarcer while Chthamalus spp. have increased (Southward, 1991; Hawkins et al., 2003). Using data extending back to 1953, Mieszkowska et al. (2014) showed that the out-of-phase variability of these two barnacle species is linked to the AMO. Poloczanska et al. (2008) suggested that the recruitment of S. balanoides is driven by temperature but that the response of Chthamalus spp. is due to competition (or lack of it) with the more dominant S. balanoides. Extensive changes occurred in other components of the rocky shore biota not only in the English Channel but also in the Celtic and Irish seas, with mass mortalities following the extremely cold winter of 1962–1963 (Crisp, 1964). Increases in abundance and range expansions in cold-water species were also observed in these regions during the cold period, such that cold-water species became dominant. Marine plants were also affected. The seaweed Laminaria ochroleuca that had spread over the area of SW England coast during the previous warm period decreased in abundance with the arrival of cold waters in the mid-1960s and remained low for at least a decade (Cushing and Dickson, 1976). Fish and shellfish Fish and shellfish have long-been known to respond to climate forcing (Hjort, 1914; Helland-Hansen and Nansen, 1920; Shepherd et al., 1984), thus it is not surprising that with the downturn in temperatures during the 1960s to the 1980s strong responses resulted. In the Barents Sea, the Atlantic cod (Gadus morhua) spawning stock biomass (SSB) declined by a factor of 2 from a high in the early 1950s to a low in the mid-1960s (Hylen, 2002). Although the SSB rebounded in the late 1960s and early 1970s, it again declined from the mid-1970s to the early 1980s and remained low before increasing in the 1990s. Cod recruitment was more variable during this period but from the mid-1960s to the late-1980s it was generally lower than normal, despite 1969 recording the highest recruitment between 1910 and 1995 (Hylen, 2002). Sætersdal and Loeng (1987) earlier had noted the decline in cod recruitment during the cold period of 1977–1982 as well as the shrinking of its feeding area. By 1982/1983, as waters warmed, the distribution area and production of cod biomass expanded and recruitment rose. Examining a longer time series (1900–1983), Sætersdal and Loeng (1987) found that changes in distribution occurred with adult cod retracting into the southwestern Barents Sea during cold periods. This contrasted with the broader geographic distribution during warm periods. These authors concluded that the temperature in the Barents Sea depended upon the amount of Atlantic inflow, with cold (warm) periods occurring when the inflow was reduced (increased). They also found a high incidence of temporal similarity in the survival success of the stocks of cod, haddock (Melanogrammus aeglefinus) and herring (Clupea harengus) in the Barents Sea. Changes in cod spawning locations off Norway also occur in response to climate variability (Sundby and Nakken, 2008). Barents Sea cod, which spawn along the west coast of Norway, exhibited a slightly more southern spawning distribution during the cold period of the 1970s and 1980s. The latitudinal changes in spawning preference were shown to vary with the AMO (Sundby and Nakken, 2008). Around Iceland, Atlantic cod abundance levels declined rapidly in the 1960s and into the 1970s. The decline was delayed relative to the Barents Sea cod as evidenced by the plot of the changes in the SSB (Figure 6). Figure 6. View largeDownload slide The time series for the Atlantic cod SSB for the Barents Sea, Iceland, West Greenland, and Labrador/northern Newfoundland. Figure 6. View largeDownload slide The time series for the Atlantic cod SSB for the Barents Sea, Iceland, West Greenland, and Labrador/northern Newfoundland. Off West Greenland, cod abundance peaked in the 1940s and 1950s; then declined precipitously in 1960s, concurrent with temperature (Buch et al., 1994). The timing of this decline was further delayed relative to those in the Barents Sea and Iceland (Figure 6). Coincident with the decline in West Greenland cod, its distribution shifted southward (Drinkwater, 2006) and growth rates fell (Rätz and Lloret, 2005). Strong easterly winds during spring from Iceland to Greenland are believed to have contributed to a drift of cod larvae to West Greenland in the 1950s. Together with good local environmental conditions, they drove the subsequent large biomass of cod (Buch 1984). During the 1960s, with changing winds, these acted to prevent cod larvae from Iceland drifting to Greenland. This contributed to a decline in year-class strength of cod off West Greenland but a succession of good year classes off East Greenland. Since the 1970s, catches of adult cod have remained extremely low off West Greenland. The abundance of cod larvae was also relatively low during the 1970s to at least the mid-1980s when survey data were available (Pedersen and Rice, 2002). Coinciding with the decrease in cod off West Greenland, there were increases in the yield of northern shrimp (Pandulus borealis) and Greenland halibut (Reinhardtius hippoglossoides). Indeed, the shrimp fishery replaced cod as the dominant industry in West Greenland and resulted in major changes in some coastal communities, with some prospering while others suffered (Hamilton et al., 2003). On the western side of the Labrador Sea, the Atlantic cod stock inhabiting the waters off Labrador and northern Newfoundland (NAFO areas 2J3KL) declined in abundance during the mid-1960s through into the mid-1970s (Figure 6). As the temperatures declined, the stock moved southward and growth rates declined significantly (Drinkwater, 2002). This stock, also known as Northern Cod, continued to decline in abundance through the 1980s. With continued fishing pressure and poor environmental conditions, by 1992 cod had collapsed commercially and a moratorium on cod catches was imposed by the Canadian government (Lilly et al., 2013). This moratorium caused extreme economic hardship on Newfoundland, its fishers and fish plant workers (Hamilton and Haedrich, 1999). It is interesting to note the similarity in the rate of decline in the cod SSB for the four major cod stocks during the cold period: the Barents Sea, Iceland, West Greenland, and Labrador/northern Newfoundland (Figure 6). Also, equally intriguing is the delay of several years in the timing of the decline of the cod SSB as one moves westward, or more accurately southwestward. There is some evidence that there was a delay in the air temperature decline from north to south as seen in Johannessen et al. (2004) and noted earlier. A delay is also evident in the intensity of the SST cold pool from the Barents Sea to the Labrador Sea (Figure 3). As noted earlier, Frankcombe et al. (2008) found westward propagation of the low-frequency temperature signals that was most evident in the subsurface. Could temperature have played a role in the decline and the temporal delay from east to west in the decline? Frankcombe et al. (2008) indicated that a noisy signal in the near surface layer might prohibit a clear signal of westward propagation from SST data. Indeed, our attempt to match the delay in the SSB declines using Simple Ocean Data Assimilation (SODA) SST data (http://www.atmos.umd.edu/ocean/) from the four regions was mostly unsuccessful, the exception being the delay between the Barents Sea and Labrador/Newfoundland. It remains unclear what role temperature may have played in the decline of these four cod stocks and the temporal delay. Equally dramatic to the declines in the cod stocks were the distribution and abundance changes experienced by the Norwegian spring-spawning herring (C. harengus) in the Norwegian Sea. Their abundance peaked in the 1940s and 1950s but declined rapidly during the 1960s in concert with decreasing sea temperatures and intense fishing pressure (Toresen and Østvedt, 2000). By the end of the 1960s the herring stock had collapsed commercially. During the 1970s and much of the 1980s, the stock remained at very low levels despite a fishing moratorium. Their migratory pattern also changed during this time as the adult herring did not leave the coast of Norway after spawning to feed in the Norwegian Sea, as in previous years, but remained near their spawning sites (Vilhjálmsson, 1997). They later migrated into the Norwegian fjords to over-winter as opposed to over-wintering in the Norwegian Sea as occurred previously. During the warm period, large quantities of herring occupied waters around Iceland and this fishery dominated the Icelandic economy (Vilhjálmsson, 1997). The collapse of the herring fishery during the cold period caused great hardships for Icelandic fishers and was a major blow to the country’s economy (Sigurdsson, 2006). In the northwestern North Sea, during the cold years of 1965–1980, Atlantic pelagic fish species in the northwestern North Sea declined. On the other hand, sprat (Sprattus sprattus), a small forage species of the herring family, increased (Corten, 1999). The author attributed this to a large reduction in the Atlantic inflow into the North Sea. Atlantic herring had been abundant during the cold years at the beginning of the 20th century. Their numbers declined during the warm period in the middle of the century, but did not increase with the return of the cold temperatures. Speculation was that overfishing had reduced the stock below capacity for renewal and Atlantic mackerel (Scomber scombrus), which prefers warmer waters, may have filled the niche formerly occupied by herring. Such a switch in dominant species has been observed elsewhere, e.g. in the Gulf of St. Lawrence (Skud, 1982). Throughout the North Sea, the decline in spring temperatures in the 1950s and 1960s resulted in increased survival of North Sea cod and other gadoids. Beaugrand and Kirby (2010) showed that cod biomass was extremely high during the cold period of the 1960s to the mid-1980s. This was part of the gadoid outburst (Hislop, 1996). The gadoid abundance declined significantly afterwards when temperatures rose. Beaugrand and Kirby (2010) also found that the correlation between temperature and cod at age 1 was less than that of age 1 cod and their prey, suggesting that temperature was likely affecting cod through their food. In the southern North Sea, the populations of European anchovy (Engraulis encrasicolus) and sardines, also called pilchards (S. pilchardus), that were abundant during the warm period of the 1930s to the 1950s, all but disappeared from the region during the cold period (Alheit et al., 2014). Anchovy and sardine catches in the Bay of Biscay and in the Iberian upwelling region declined rapidly after the mid-1960s and remained low through the 1970s and into the 1980s. Alheit et al. (2014) also showed that the migration of these species varied in concert with the AMO, such that they moved northward during the warm phases and southward during the cold phases. In the western English Channel, Russell (1973) noted that the abundance of post larvae of teleostean (ray-finned) fish increased with the advent of cooler temperatures after 1965. During the warm period from the 1930s to 1965 few such larvae were observed in the channel. In contrast to the teleostean fish, the European pilchard declined in the region through the late-1960s from high abundance levels during the warm period of the 1930s to the mid-1960s to near zero during the cold period of the 1970s to the mid-1980s (Southward, 1980; Hawkins et al., 2003). The abundance of pilchard eggs also declined during the cool spell of the 1970s, but increased again post-1985. From 1962 through to the 1980s there was an increase in cold-water fish such as cod, haddock, ling (Molva molva), whiting (Merlangius merlangus), plaice (Pleuronectes platessa), dab (Limanda limanda), and lemon sole (Microstomus kitt) (Hawkins et al., 2003). This resulted in nearly an order of magnitude increase in the number of fishing vessels with the return of these more lucrative species (Southward, 1980). These cold-water species had been replaced in the warm period of the 1930s to the early 1960s by the species such as pilchard, horse mackerel (Trachurus trachurus) and southern sea breams (Family Sparidae) that had lower economic value. By the late 1980s, the cold-water species began to decline with the return of the warm waters. These again were part of what was temperature-mediated habitat switching, which in the English Channel was called the Russell Cycle. However, Edwards et al. (2013) showed that the temperature variability giving rise to the ecosystem changes was similar to the AMO. Discussion As described earlier, significant changes in the marine ecosystem in the northern North Atlantic occurred during the 1960s to the 1980s in a period of general cool ocean conditions (see also Table 1). This cooling was part of the low-frequency basin-scale variability known as the AMO with a 60- to 80-year periodicity. The mechanisms leading to the cooling are believed to involve a reduction in solar energy reaching the sea surface and changes in ocean circulation, including the transport of more cold water southward and less warm water northward. These changes in ocean circulation are believed related to changes in the large-scale wind patterns. Colder air and sea temperatures and more northly winds led to an increase in sea-ice extent and caused the sea-ice edge to move farther southward, and in some locations eastward. The ecosystem changes included generally lower primary production and delayed spring blooms. There were distributional shifts in zooplankton populations with many species typical of northern regions extending their distributions southward. In addition, zooplankton abundance and biomass levels generally declined and biodiversity was reduced although in some areas such as the North Sea and the English Channel, plankton levels increased with the invasion of more northern communities. Nearshore communities were also affected with some notable shifts as the southern boundary of typical northern species moved farther southward. It was not just plankton that showed such trends but plant species as well, such as some seaweed species. Table 1. Summary of physical and ecological changes that occurred during the cold period of the 1960s to the 1980s. Barents Sea Norwegian Sea Greenland Sea Iceland Air temperatures Annual average temperature at Bear Island from late 1950s to late 1960s cooled by near 2°C (Førland and Hanssen-Bauer, 2003) Low air temperatures during the 1960s through the mid-1980s as part of the Nordic Seas (Tuomenvirta et al., 2000) Based on two coastal stations on East Greenland, air temperatures in the 1960s were 1°C cooler than in the 1990s (Mernild et al., 2014) Air temperatures around Iceland fell 1–1.5°C between the mid-1960s to the mid-1970s (Hanna et al., 2006) Winds Stronger northerly winds (Lamb, 1972) Stronger northerly winds (Cushing and Dickson, 1976) Stronger northerly winds (Cushing and Dickson, 1976) Stronger northerly winds (Cushing and Dickson, 1976) Ocean temperatures Cooled; mean 0–200 m temperature on Kola Section fell by about 0.5°C between early 1960s to late 1980s (Loeng and Drinkwater, 2007) Temperatures at 50–200 m at Ocean Weather station Mike fell by over 1°C from early 1960s to 1980 (Blindheim et al., 2000) Cold temperatures from 1967 to 1985 (Plate 1a in Bönisch et al., 1997) SSTs along the northern coast declined by 1–2°C from mid-1950s to late-1960s (Hanna et al., 2006) Ocean circulation Reduced inflow of warm Atlantic waters through eastern entrance based on lower temperatures (Sætersdal and Loeng, 1987) Decreased northward flow of Atlantic water; width of Atlantic water increased seaward in 1960s (Blindheim et al., 2000) Increased southward flow of Arctic Water based on increased northerly winds (following Jónsson, 1991) Increased transport of East Icelandic Current carrying cold Arctic waters (Stefánsson, 1969) Sea Ice Increased areal extent, southern ice edge boundary moved south and east (Figure 4) Increased sea ice in northern Norwegian Sea west of Svalbard (Figure 4) Increased areal extent of ice, ice edge moved eastward (Figure 4) Heavy ice years on the north coast (Malmberg, 1969, 1984) Primary Production Expected to have declined due to lower production because of extensive ice cover and more Arctic water based upon opposite response during warm periods (Slagstad and Wassmann, 1997; Mueter et al., 2009) No available data found No available data found During the 1960s to the mid-1980s, the mean primary productivity north of Iceland was observed to be low (Gudmundsson, 1998) Zooplankton Abundance and biomass thought to have declined during cold period based on low biomass during cold years (Dalpadado et al., 2003) No available data found No available data found Abundance was relatively low off north coast during cold period of the 1960s–1980s (Astthorsson et al., 1983; Astthorsson and Gislason, 1995) Fish SSB of Northeast Arctic cod stock fell by a factor of over 5 between late-1940s to mid-1960s (Figure 6) Collapse of herring stock and herring retreated to confines of the west coast of Norway (Vilhjálmsson, 1997) No available data found Disappearance of herring from Iceland as herring confined to west coast of Norway (Toresen and Østvedt, 2000); SSB of Icelandic cod stock fell by a factor of over 5 between 1960 and mid-1970s (Figure 6) Barents Sea Norwegian Sea Greenland Sea Iceland Air temperatures Annual average temperature at Bear Island from late 1950s to late 1960s cooled by near 2°C (Førland and Hanssen-Bauer, 2003) Low air temperatures during the 1960s through the mid-1980s as part of the Nordic Seas (Tuomenvirta et al., 2000) Based on two coastal stations on East Greenland, air temperatures in the 1960s were 1°C cooler than in the 1990s (Mernild et al., 2014) Air temperatures around Iceland fell 1–1.5°C between the mid-1960s to the mid-1970s (Hanna et al., 2006) Winds Stronger northerly winds (Lamb, 1972) Stronger northerly winds (Cushing and Dickson, 1976) Stronger northerly winds (Cushing and Dickson, 1976) Stronger northerly winds (Cushing and Dickson, 1976) Ocean temperatures Cooled; mean 0–200 m temperature on Kola Section fell by about 0.5°C between early 1960s to late 1980s (Loeng and Drinkwater, 2007) Temperatures at 50–200 m at Ocean Weather station Mike fell by over 1°C from early 1960s to 1980 (Blindheim et al., 2000) Cold temperatures from 1967 to 1985 (Plate 1a in Bönisch et al., 1997) SSTs along the northern coast declined by 1–2°C from mid-1950s to late-1960s (Hanna et al., 2006) Ocean circulation Reduced inflow of warm Atlantic waters through eastern entrance based on lower temperatures (Sætersdal and Loeng, 1987) Decreased northward flow of Atlantic water; width of Atlantic water increased seaward in 1960s (Blindheim et al., 2000) Increased southward flow of Arctic Water based on increased northerly winds (following Jónsson, 1991) Increased transport of East Icelandic Current carrying cold Arctic waters (Stefánsson, 1969) Sea Ice Increased areal extent, southern ice edge boundary moved south and east (Figure 4) Increased sea ice in northern Norwegian Sea west of Svalbard (Figure 4) Increased areal extent of ice, ice edge moved eastward (Figure 4) Heavy ice years on the north coast (Malmberg, 1969, 1984) Primary Production Expected to have declined due to lower production because of extensive ice cover and more Arctic water based upon opposite response during warm periods (Slagstad and Wassmann, 1997; Mueter et al., 2009) No available data found No available data found During the 1960s to the mid-1980s, the mean primary productivity north of Iceland was observed to be low (Gudmundsson, 1998) Zooplankton Abundance and biomass thought to have declined during cold period based on low biomass during cold years (Dalpadado et al., 2003) No available data found No available data found Abundance was relatively low off north coast during cold period of the 1960s–1980s (Astthorsson et al., 1983; Astthorsson and Gislason, 1995) Fish SSB of Northeast Arctic cod stock fell by a factor of over 5 between late-1940s to mid-1960s (Figure 6) Collapse of herring stock and herring retreated to confines of the west coast of Norway (Vilhjálmsson, 1997) No available data found Disappearance of herring from Iceland as herring confined to west coast of Norway (Toresen and Østvedt, 2000); SSB of Icelandic cod stock fell by a factor of over 5 between 1960 and mid-1970s (Figure 6) West Greenland Labrador/Northern Newfoundland West of Scotland North Sea/English Channel Air temperatures Nuuk annual air temperatures fell by 1.5–2°C between 1960 and late-1980s with the cooling primarily in winter (Stein, 2007) Cartwright, Labrador, annual air temperatures declined by 2°C between mid-1950s and late-1980s (Colbourne et al., 1994) Air temperatures in UK coastal waters from 1970 to 1990 were ∼1–1.5°C lower than from 1990 to 2006 (Dye et al., 2013) Air temperatures in UK coastal waters from 1970 to 1990 were ∼1–1.5°C lower than from 1990 to 2006 (Dye et al., 2013) Winds Stronger northwesterly winds after 1972 (Drinkwater, 1996) Stronger northwesterly winds after 1972 (Drinkwater, 1996) Lower wind speeds with less southwesterly winds (Watson et al., 2015) Lower wind speeds with less southwesterly winds (Watson et al., 2015) Ocean temperatures Ocean temperatures in June and November over Fylla Bank declined by 1–2°C from mid-1960s to mid-1980s (Stein and Buch, 1991) Temperatures at 100 m at Stn 27 off St. John’s fell by about 1°C from late-1960s to mid-1970s (Colbourne, 2004) Temperatures in North Scotland Slope Current at 185 m were around 0.5°C cooler in the 1980s (start of record) than in the 1990s and 2000s (Inall et al., 2009) SSTs in 1960s to 1980s were low relative to 1990s (Edwards et al., 2002) Ocean circulation Irminger Current increased in early 1960s (Dickson and Lamb, 1972) No available data found No available data found Reduction in the Atlantic inflow to the North Sea from north of Scotland (Corten, 1999) Sea Ice “Storis” extended farther north than normal, also longer duration (Valeur, 1976) Ice area south of 55°N increased by a factor of 3 (Drinkwater, 1996) No ice No ice Primary Production No available data found No available data found Phytoplankton based on greenness index was relative low from 1960 to mid-1980s compared with later years (Edwards et al., 2001) Phytoplankton based on greenness index was relative low from 1960 to mid-1980s compared with later years (Edwards et al., 2001) Zooplankton Based on June-July samples collected on three offshore transects, zooplankton declined after mid-1960s and remained low in the mid-1980s (Pedersen and Smidt, 2000, Pedersen and Rice, 2002) No available data found Declined based on the CPR zooplankton data (Glover et al., 1972). Declined based on the CPR data (Glover et al., 1972). Fish West Greenland cod stock collapsed, SSB of Icelandic cod stock fell by a factor of 20 between late-1950s and mid-1970s (Figure 6) Northern cod stock collapsed, moratorium imposed 1992; growth rates and recruitment declined (Lilly et al., 2013); SSB of Northern cod stock fell by a factor of 15 between mid-1960s and late-1970s (Figure 6) Atlantic cod west of Scotland was high in the 1980s, which was the beginning of the record and declined thereafter (ICES, 2012) Cold water species increased in abundance, such as cod, part of the Gadoid outburst (Hislop, 1996) while warm water species declined, such as sardines and anchovies; (Alheit et al., 2014) West Greenland Labrador/Northern Newfoundland West of Scotland North Sea/English Channel Air temperatures Nuuk annual air temperatures fell by 1.5–2°C between 1960 and late-1980s with the cooling primarily in winter (Stein, 2007) Cartwright, Labrador, annual air temperatures declined by 2°C between mid-1950s and late-1980s (Colbourne et al., 1994) Air temperatures in UK coastal waters from 1970 to 1990 were ∼1–1.5°C lower than from 1990 to 2006 (Dye et al., 2013) Air temperatures in UK coastal waters from 1970 to 1990 were ∼1–1.5°C lower than from 1990 to 2006 (Dye et al., 2013) Winds Stronger northwesterly winds after 1972 (Drinkwater, 1996) Stronger northwesterly winds after 1972 (Drinkwater, 1996) Lower wind speeds with less southwesterly winds (Watson et al., 2015) Lower wind speeds with less southwesterly winds (Watson et al., 2015) Ocean temperatures Ocean temperatures in June and November over Fylla Bank declined by 1–2°C from mid-1960s to mid-1980s (Stein and Buch, 1991) Temperatures at 100 m at Stn 27 off St. John’s fell by about 1°C from late-1960s to mid-1970s (Colbourne, 2004) Temperatures in North Scotland Slope Current at 185 m were around 0.5°C cooler in the 1980s (start of record) than in the 1990s and 2000s (Inall et al., 2009) SSTs in 1960s to 1980s were low relative to 1990s (Edwards et al., 2002) Ocean circulation Irminger Current increased in early 1960s (Dickson and Lamb, 1972) No available data found No available data found Reduction in the Atlantic inflow to the North Sea from north of Scotland (Corten, 1999) Sea Ice “Storis” extended farther north than normal, also longer duration (Valeur, 1976) Ice area south of 55°N increased by a factor of 3 (Drinkwater, 1996) No ice No ice Primary Production No available data found No available data found Phytoplankton based on greenness index was relative low from 1960 to mid-1980s compared with later years (Edwards et al., 2001) Phytoplankton based on greenness index was relative low from 1960 to mid-1980s compared with later years (Edwards et al., 2001) Zooplankton Based on June-July samples collected on three offshore transects, zooplankton declined after mid-1960s and remained low in the mid-1980s (Pedersen and Smidt, 2000, Pedersen and Rice, 2002) No available data found Declined based on the CPR zooplankton data (Glover et al., 1972). Declined based on the CPR data (Glover et al., 1972). Fish West Greenland cod stock collapsed, SSB of Icelandic cod stock fell by a factor of 20 between late-1950s and mid-1970s (Figure 6) Northern cod stock collapsed, moratorium imposed 1992; growth rates and recruitment declined (Lilly et al., 2013); SSB of Northern cod stock fell by a factor of 15 between mid-1960s and late-1970s (Figure 6) Atlantic cod west of Scotland was high in the 1980s, which was the beginning of the record and declined thereafter (ICES, 2012) Cold water species increased in abundance, such as cod, part of the Gadoid outburst (Hislop, 1996) while warm water species declined, such as sardines and anchovies; (Alheit et al., 2014) Table 1. Summary of physical and ecological changes that occurred during the cold period of the 1960s to the 1980s. Barents Sea Norwegian Sea Greenland Sea Iceland Air temperatures Annual average temperature at Bear Island from late 1950s to late 1960s cooled by near 2°C (Førland and Hanssen-Bauer, 2003) Low air temperatures during the 1960s through the mid-1980s as part of the Nordic Seas (Tuomenvirta et al., 2000) Based on two coastal stations on East Greenland, air temperatures in the 1960s were 1°C cooler than in the 1990s (Mernild et al., 2014) Air temperatures around Iceland fell 1–1.5°C between the mid-1960s to the mid-1970s (Hanna et al., 2006) Winds Stronger northerly winds (Lamb, 1972) Stronger northerly winds (Cushing and Dickson, 1976) Stronger northerly winds (Cushing and Dickson, 1976) Stronger northerly winds (Cushing and Dickson, 1976) Ocean temperatures Cooled; mean 0–200 m temperature on Kola Section fell by about 0.5°C between early 1960s to late 1980s (Loeng and Drinkwater, 2007) Temperatures at 50–200 m at Ocean Weather station Mike fell by over 1°C from early 1960s to 1980 (Blindheim et al., 2000) Cold temperatures from 1967 to 1985 (Plate 1a in Bönisch et al., 1997) SSTs along the northern coast declined by 1–2°C from mid-1950s to late-1960s (Hanna et al., 2006) Ocean circulation Reduced inflow of warm Atlantic waters through eastern entrance based on lower temperatures (Sætersdal and Loeng, 1987) Decreased northward flow of Atlantic water; width of Atlantic water increased seaward in 1960s (Blindheim et al., 2000) Increased southward flow of Arctic Water based on increased northerly winds (following Jónsson, 1991) Increased transport of East Icelandic Current carrying cold Arctic waters (Stefánsson, 1969) Sea Ice Increased areal extent, southern ice edge boundary moved south and east (Figure 4) Increased sea ice in northern Norwegian Sea west of Svalbard (Figure 4) Increased areal extent of ice, ice edge moved eastward (Figure 4) Heavy ice years on the north coast (Malmberg, 1969, 1984) Primary Production Expected to have declined due to lower production because of extensive ice cover and more Arctic water based upon opposite response during warm periods (Slagstad and Wassmann, 1997; Mueter et al., 2009) No available data found No available data found During the 1960s to the mid-1980s, the mean primary productivity north of Iceland was observed to be low (Gudmundsson, 1998) Zooplankton Abundance and biomass thought to have declined during cold period based on low biomass during cold years (Dalpadado et al., 2003) No available data found No available data found Abundance was relatively low off north coast during cold period of the 1960s–1980s (Astthorsson et al., 1983; Astthorsson and Gislason, 1995) Fish SSB of Northeast Arctic cod stock fell by a factor of over 5 between late-1940s to mid-1960s (Figure 6) Collapse of herring stock and herring retreated to confines of the west coast of Norway (Vilhjálmsson, 1997) No available data found Disappearance of herring from Iceland as herring confined to west coast of Norway (Toresen and Østvedt, 2000); SSB of Icelandic cod stock fell by a factor of over 5 between 1960 and mid-1970s (Figure 6) Barents Sea Norwegian Sea Greenland Sea Iceland Air temperatures Annual average temperature at Bear Island from late 1950s to late 1960s cooled by near 2°C (Førland and Hanssen-Bauer, 2003) Low air temperatures during the 1960s through the mid-1980s as part of the Nordic Seas (Tuomenvirta et al., 2000) Based on two coastal stations on East Greenland, air temperatures in the 1960s were 1°C cooler than in the 1990s (Mernild et al., 2014) Air temperatures around Iceland fell 1–1.5°C between the mid-1960s to the mid-1970s (Hanna et al., 2006) Winds Stronger northerly winds (Lamb, 1972) Stronger northerly winds (Cushing and Dickson, 1976) Stronger northerly winds (Cushing and Dickson, 1976) Stronger northerly winds (Cushing and Dickson, 1976) Ocean temperatures Cooled; mean 0–200 m temperature on Kola Section fell by about 0.5°C between early 1960s to late 1980s (Loeng and Drinkwater, 2007) Temperatures at 50–200 m at Ocean Weather station Mike fell by over 1°C from early 1960s to 1980 (Blindheim et al., 2000) Cold temperatures from 1967 to 1985 (Plate 1a in Bönisch et al., 1997) SSTs along the northern coast declined by 1–2°C from mid-1950s to late-1960s (Hanna et al., 2006) Ocean circulation Reduced inflow of warm Atlantic waters through eastern entrance based on lower temperatures (Sætersdal and Loeng, 1987) Decreased northward flow of Atlantic water; width of Atlantic water increased seaward in 1960s (Blindheim et al., 2000) Increased southward flow of Arctic Water based on increased northerly winds (following Jónsson, 1991) Increased transport of East Icelandic Current carrying cold Arctic waters (Stefánsson, 1969) Sea Ice Increased areal extent, southern ice edge boundary moved south and east (Figure 4) Increased sea ice in northern Norwegian Sea west of Svalbard (Figure 4) Increased areal extent of ice, ice edge moved eastward (Figure 4) Heavy ice years on the north coast (Malmberg, 1969, 1984) Primary Production Expected to have declined due to lower production because of extensive ice cover and more Arctic water based upon opposite response during warm periods (Slagstad and Wassmann, 1997; Mueter et al., 2009) No available data found No available data found During the 1960s to the mid-1980s, the mean primary productivity north of Iceland was observed to be low (Gudmundsson, 1998) Zooplankton Abundance and biomass thought to have declined during cold period based on low biomass during cold years (Dalpadado et al., 2003) No available data found No available data found Abundance was relatively low off north coast during cold period of the 1960s–1980s (Astthorsson et al., 1983; Astthorsson and Gislason, 1995) Fish SSB of Northeast Arctic cod stock fell by a factor of over 5 between late-1940s to mid-1960s (Figure 6) Collapse of herring stock and herring retreated to confines of the west coast of Norway (Vilhjálmsson, 1997) No available data found Disappearance of herring from Iceland as herring confined to west coast of Norway (Toresen and Østvedt, 2000); SSB of Icelandic cod stock fell by a factor of over 5 between 1960 and mid-1970s (Figure 6) West Greenland Labrador/Northern Newfoundland West of Scotland North Sea/English Channel Air temperatures Nuuk annual air temperatures fell by 1.5–2°C between 1960 and late-1980s with the cooling primarily in winter (Stein, 2007) Cartwright, Labrador, annual air temperatures declined by 2°C between mid-1950s and late-1980s (Colbourne et al., 1994) Air temperatures in UK coastal waters from 1970 to 1990 were ∼1–1.5°C lower than from 1990 to 2006 (Dye et al., 2013) Air temperatures in UK coastal waters from 1970 to 1990 were ∼1–1.5°C lower than from 1990 to 2006 (Dye et al., 2013) Winds Stronger northwesterly winds after 1972 (Drinkwater, 1996) Stronger northwesterly winds after 1972 (Drinkwater, 1996) Lower wind speeds with less southwesterly winds (Watson et al., 2015) Lower wind speeds with less southwesterly winds (Watson et al., 2015) Ocean temperatures Ocean temperatures in June and November over Fylla Bank declined by 1–2°C from mid-1960s to mid-1980s (Stein and Buch, 1991) Temperatures at 100 m at Stn 27 off St. John’s fell by about 1°C from late-1960s to mid-1970s (Colbourne, 2004) Temperatures in North Scotland Slope Current at 185 m were around 0.5°C cooler in the 1980s (start of record) than in the 1990s and 2000s (Inall et al., 2009) SSTs in 1960s to 1980s were low relative to 1990s (Edwards et al., 2002) Ocean circulation Irminger Current increased in early 1960s (Dickson and Lamb, 1972) No available data found No available data found Reduction in the Atlantic inflow to the North Sea from north of Scotland (Corten, 1999) Sea Ice “Storis” extended farther north than normal, also longer duration (Valeur, 1976) Ice area south of 55°N increased by a factor of 3 (Drinkwater, 1996) No ice No ice Primary Production No available data found No available data found Phytoplankton based on greenness index was relative low from 1960 to mid-1980s compared with later years (Edwards et al., 2001) Phytoplankton based on greenness index was relative low from 1960 to mid-1980s compared with later years (Edwards et al., 2001) Zooplankton Based on June-July samples collected on three offshore transects, zooplankton declined after mid-1960s and remained low in the mid-1980s (Pedersen and Smidt, 2000, Pedersen and Rice, 2002) No available data found Declined based on the CPR zooplankton data (Glover et al., 1972). Declined based on the CPR data (Glover et al., 1972). Fish West Greenland cod stock collapsed, SSB of Icelandic cod stock fell by a factor of 20 between late-1950s and mid-1970s (Figure 6) Northern cod stock collapsed, moratorium imposed 1992; growth rates and recruitment declined (Lilly et al., 2013); SSB of Northern cod stock fell by a factor of 15 between mid-1960s and late-1970s (Figure 6) Atlantic cod west of Scotland was high in the 1980s, which was the beginning of the record and declined thereafter (ICES, 2012) Cold water species increased in abundance, such as cod, part of the Gadoid outburst (Hislop, 1996) while warm water species declined, such as sardines and anchovies; (Alheit et al., 2014) West Greenland Labrador/Northern Newfoundland West of Scotland North Sea/English Channel Air temperatures Nuuk annual air temperatures fell by 1.5–2°C between 1960 and late-1980s with the cooling primarily in winter (Stein, 2007) Cartwright, Labrador, annual air temperatures declined by 2°C between mid-1950s and late-1980s (Colbourne et al., 1994) Air temperatures in UK coastal waters from 1970 to 1990 were ∼1–1.5°C lower than from 1990 to 2006 (Dye et al., 2013) Air temperatures in UK coastal waters from 1970 to 1990 were ∼1–1.5°C lower than from 1990 to 2006 (Dye et al., 2013) Winds Stronger northwesterly winds after 1972 (Drinkwater, 1996) Stronger northwesterly winds after 1972 (Drinkwater, 1996) Lower wind speeds with less southwesterly winds (Watson et al., 2015) Lower wind speeds with less southwesterly winds (Watson et al., 2015) Ocean temperatures Ocean temperatures in June and November over Fylla Bank declined by 1–2°C from mid-1960s to mid-1980s (Stein and Buch, 1991) Temperatures at 100 m at Stn 27 off St. John’s fell by about 1°C from late-1960s to mid-1970s (Colbourne, 2004) Temperatures in North Scotland Slope Current at 185 m were around 0.5°C cooler in the 1980s (start of record) than in the 1990s and 2000s (Inall et al., 2009) SSTs in 1960s to 1980s were low relative to 1990s (Edwards et al., 2002) Ocean circulation Irminger Current increased in early 1960s (Dickson and Lamb, 1972) No available data found No available data found Reduction in the Atlantic inflow to the North Sea from north of Scotland (Corten, 1999) Sea Ice “Storis” extended farther north than normal, also longer duration (Valeur, 1976) Ice area south of 55°N increased by a factor of 3 (Drinkwater, 1996) No ice No ice Primary Production No available data found No available data found Phytoplankton based on greenness index was relative low from 1960 to mid-1980s compared with later years (Edwards et al., 2001) Phytoplankton based on greenness index was relative low from 1960 to mid-1980s compared with later years (Edwards et al., 2001) Zooplankton Based on June-July samples collected on three offshore transects, zooplankton declined after mid-1960s and remained low in the mid-1980s (Pedersen and Smidt, 2000, Pedersen and Rice, 2002) No available data found Declined based on the CPR zooplankton data (Glover et al., 1972). Declined based on the CPR data (Glover et al., 1972). Fish West Greenland cod stock collapsed, SSB of Icelandic cod stock fell by a factor of 20 between late-1950s and mid-1970s (Figure 6) Northern cod stock collapsed, moratorium imposed 1992; growth rates and recruitment declined (Lilly et al., 2013); SSB of Northern cod stock fell by a factor of 15 between mid-1960s and late-1970s (Figure 6) Atlantic cod west of Scotland was high in the 1980s, which was the beginning of the record and declined thereafter (ICES, 2012) Cold water species increased in abundance, such as cod, part of the Gadoid outburst (Hislop, 1996) while warm water species declined, such as sardines and anchovies; (Alheit et al., 2014) Arguably the most noticeable changes were in fish populations. Many species underwent distributional shifts, principally southward, as well as changes in the timing and extent of migration patterns. The southward movement of many Arctic and northern boreal species occurred concurrently with a retraction in the distribution of southern boreal and subtropical species. For some species, such as Atlantic cod along the Norwegian coast, there was a tendency for the spawning to shift southward. Large decreases in the biomass of several commercially important species occurred. Atlantic cod off West Greenland and Labrador collapsed commercially, as did the Norwegian spring-spawning herring in the Norwegian Sea. The changes in these fish populations had significant economic impacts. In West Greenland, there was a shift from a cod-dominated economy to one more dependent upon shrimp (Pandalus borealis). Off Labrador, with the collapse of the northern cod stock, many fishers lost their livelihood, although the most significant impact was on fish plant workers as the cod processing plants closed. Shrimp (P. borealis) and, to a lesser extent, snow crab (Chionoecetes opilio) fisheries developed but their biomass was nowhere near to that of the former cod stock. Also, no fish species appeared to move in to fill the niche formally occupied by cod in the ecosystem off Labrador and Newfoundland. In Iceland, where the herring drove much of the local economy during the warm period, the collapse of the herring fisheries brought hard economic times in Iceland. Not all fish populations declined during the cold period. For example, the Atlantic cod in the North Sea increased under colder conditions. This is due to the response of cod to temperature variability. As shown by Planque and Fredou (1999) and later discussed by Drinkwater (2005), the temperature effect on cod has a domed-shaped response with fish inhabiting colder bottom waters (<5°C) tending to decrease (increase) in abundance as temperature falls (rises), while those in warm waters (>8.5°C) tend to increase (decrease). The influence of temperature is primarily through its direct or indirect effect on the early stages of cod and hence recruitment. The North Sea is at the southern boundary of Atlantic cod on the eastern side of the Atlantic and the waters are relatively warm. The decline in temperatures therefore leads to increased cod biomass. Other demersal fish such as haddock, whiting, and Norway pout also increased in abundance within the North Sea and vicinity giving rise to the gadoid outburst. As temperatures rose in the North Atlantic during the late 1980s and 1990s, ecological conditions largely returned to the ecosystem states observed in the 1950s. Phytoplankton and zooplankton production increased again through much of the northern North Atlantic (Boyce et al., 2010) and their distribution generally shifted more northward (Beaugrand et al., 2002). The Atlantic herring in the Norwegian Sea whose commercial collapse coincided with the cold period, rebounded to biomass levels almost as high as in the earlier warm period (Toresen and Østvedt, 2000). The abundance of Atlantic cod in the Barents Sea also increased as temperatures warmed (Kjesbu et al., 2014). In other regions, however, such recoveries did not occur as waters warmed, e.g. Atlantic cod off West Greenland (Buch et al., 2004) and Labrador/Newfoundland (Lilly et al., 2013), where abundance levels remained, especially compared with the earlier warm period. Based on the definition of a regime shift by Bakun (2004) of a significant ecological change over a wide area that lasts for a relatively long time, the cold period could be considered as a climate-forced “regime shift” in the northern North Atlantic. This would be consistent with Drinkwater (2006) who labelled the warming in the 1920s and 1930s as a regime shift, also based on the definition by Bakun (2004). However, one might also conclude that, as the AMO implies, it is simply natural low-frequency variability or oscillation with a relatively smooth transition and not a true regime shift. In this article, we have focussed on multidecadal variability associated with the ocean-derived AMO index. However, another important basin-scale forcing that must be mentioned is the NAO. Having a dominant temporal scale of order a decade, the NAO is the most prominent atmospheric pattern of variability in the Atlantic region. It is due to a redistribution of atmospheric mass between the Arctic and the Subtropical Atlantic centred on Icelandic Low and Azores High (Hurrell et al., 2003). These pressure systems tend to vary together, i.e. they both strengthen or weaken at the same time, thereby varying the north–south pressure gradient in the mid-latitudes of the North Atlantic. Spatially, the NAO variability produces out-of-phase changes in the winds between the Northwest and Northeast Atlantic. This in turn results in spatially dependent physical oceanographic variability, which affects ocean properties such as hydrographic characteristics, vertical stratification and mixing, heat content, circulation patterns, and sea-ice coverage (Visbeck et al., 2003). These then produce ecological variability from plankton to marine mammals and seabirds (e.g. Drinkwater et al., 2003; Brander, 2005; Beaugrand et al., 2008). Several papers have implicated strong linkages between the NAO and the AMO on multidecadal time scales (see review by Grossman and Klotzman, 2009). For example, long‐term positive (negative) phases of the NAO coincide with the negative (positive) phase of the AMO, typically with a lag of several years (Zhang and Vallis, 2006). The associations were hypothesized to be through the Thermohaline Circulation. Li et al. (2013) showed the NAO leading the AMO by 15–20 years and suggested it was due to the large thermal inertia associated with slow oceanic processes. Oviatt et al. (2015) discussed the interactions between the NAO and AMO. However, further study of these links is needed before we can understand fully the role of the NAO on the AMO and the AMO on the NAO. We do not want to leave the impression that ecosystem changes depend solely on temperature variability, although we believe it does play a major role, directly through physiological effects, and/or indirectly, through its influence on prey, predators or competitors. Until now we have said little about the effects of fishing. By the time of the cooling in the 1960s, fisheries in the Atlantic had expanded significantly with more and larger fishing vessels, the development of large long-distance foreign fishing fleets, and an increase in the number of fishers, all of which contributed to significant impacts on fish populations. Thus, there has been much debate as to whether the observed decline in several fish species, such as cod in the 1960s, was mostly due to fishing or to climate. It is clear to us that both played a significant role in the observed changes in the commercial fish populations, for example in the Atlantic cod stocks off West Greenland and Labrador and the Norwegian spring-spawning herring. Link et al. (2010) examined 19 exploited ecosystems corresponding to upwelling, high latitude, temperate, and tropical marine ecosystems. They covered a range of low- to high-productive areas in the Atlantic and Pacific oceans and the Mediterranean Sea that had been fished at different levels. The authors demonstrated that in most ecosystems, the fish dynamics were largely driven by fisheries (landings) but secondarily by environmental drivers. Thus, it is important to consider fisheries as well as the environment when seeking to understand the dynamics of fish populations. However, the relative consistency of the ecosystem responses to the warm and cold periods associated with the AMO does provide strong evidence of the importance of environmental forcing at low-frequencies in the aNorth Atlantic. It is also important to ask what is the primary mechanism through which temperature acts. Based on the wide spread decline in primary and secondary production during the cold period as well as changes in the production cycle that we have outlined in this article, we conclude that the primary mechanism is likely related to food availability. This is consistent with the findings of Drinkwater (2005). Why study the response to a period of cooling when most scientists and the public are more concerned about anthropogenic warming? We believe that studying the cool period of the 1960s–1980s is still of practical significance, since the AMO signal is expected to continue in the future. This is expected to produce extended periods of cooling but perhaps not as strong as in the past since it will be combined with the anthropogenic warming signal. Studying the response to the cooling, we can anticipate what ecosystem changes might occur especially under prolonged cooling and hopefully this will allow us to better manage our marine resources. Acknowledgements We are grateful to several colleagues for conversions over the years about the AMO and particularly the changes that occurred during the cool period including Svein Sundby, Odd Nakken, Chris Reid, Steve Hawkins, and many others. KD acknowledges the support for this research by the Institute of Marine Research in Bergen and was carried out as part of the IMBeR regional programme Ecosystem Studies of Subarctic and Arctic Seas (ESSAS). We thank Steve Hawkins and an anonymous reviewer for their comments on an earlier version of this article and for providing new references or reminding us of references we had neglected to include. References Alexander M. A. , Kilbourne K. H. , Nye J. 2014 . Climate variability during warm and cold phases of the Atlantic Multidecadal Oscillation (AMO) 1871–2008 . Journal of Marine Systems , 133 : 14 – 26 . Google Scholar Crossref Search ADS Alheit J. , Licandro P. , Coombs S. , Garcia A. , Giraldez A. , Garcia Santamaria M. T. , Slotte A. et al. 2014 . Atlantic Multidecadal Oscillation (AMO) modulates dynamics of small pelagic fishes and ecosystem regime shifts in the eastern North and Central Atlantic . Journal of Marine Systems , 133 : 88 – 102 . Google Scholar Crossref Search ADS Antonov J. I. 1993 . Linear trends in temperature at intermediate and deep layers of the North Atlantic and North Pacific Oceans: 1957-1981 . Journal of Climate , 6 : 1928 – 1942 . Google Scholar Crossref Search ADS Astthorsson O. S. , Gislason G. 1995 . Long term changes in zooplankton biomass in Icelandic waters in spring . ICES Journal of Marine Science , 52 : 657 – 688 . Google Scholar Crossref Search ADS Astthorsson O.S. , Vilhjalmsson H. 2002 . Iceland shelf LME: decadal assessment and resource sustainability. In Large Marine Ecosystems of the North Atlantic , pp. 219 – 243 . Ed. by Sherman K. , Skjoldal H. R. . Elsevier , Amsterdam, The Netherlands . Astthorsson O. S. , Gislason A. , Jonsson S. 2007 . Climate variability and the Icelandic marine ecosystem . Deep-Sea Research II , 54 : 2456 – 2477 . Google Scholar Crossref Search ADS Astthorsson O. S. , Hallgrimssson I. L. , Jonsson G. S. 1983 . Variations in zooplankton densities in Icelandic waters in spring during the years 1961-1982 . Fit Fiskideildar , 7 : 73 – 113 . Bakun A. 2004 . Regime shifts. In Robinson A. R. , Brink K. (Eds.), The Sea , Vol. 13 , pp. 971 – 1018 (Chapter 25). Harvard University Press , Cambridge, Massachusetts . Beaugrand G. 2009 . Decadal changes in climate and ecosystems in the North Atlantic Ocean and adjacent seas . Deep-Sea Research II , 56 : 656 – 673 . Google Scholar Crossref Search ADS Beaugrand G. , Kirby R. R. 2010 . Climate, plankton and cod . Global Change Biology , 16 : 1268 – 1280 . Google Scholar Crossref Search ADS Beaugrand G. , Reid P. C. , Ibañez F. , Lindley J. A. , Edwards M. 2002 . Reorganization of North Atlantic marine copepod biodiversity and climate . Science , 296 : 1692 – 1694 . Google Scholar Crossref Search ADS PubMed Beverton R. J. H. , Lee A. J. 1965 . Hydrographic fluctuations in the North Atlantic Ocean and some biological consequences. In The Biological Significance of Climate Changes in Britain, Symposia of the Institute of Biology , Vol. 14 , pp. 79 – 109 . Ed. by Johnson C. G. , Smith L. P. . Academic Press , London, England . Blackett M. , Licandro P. , Coombs S. H. , Lucas C. H. 2014 . Long-term variability of the siphonophores Muggiaea atlantica and M. kochi in the western English Channel . Progress in Oceanography , 128 : 1 – 14 . Google Scholar Crossref Search ADS Blindheim J. 1967 . Hydrographic fluctuations off West Greenland during the years 1950-1966. Redbook of the International Commission on the Northwest Atlantic Fisheries, Part l V: 86–105. Blindheim J. , Borovkov V. , Hansen B. , Malmberg S.-A. , Turrell W. R. , Østerhus S. 2000 . Upper layer cooling and freshening in the Norwegian Sea in relation to atmospheric forcing . Deep-Sea Research I , 47 : 655 – 680 . Google Scholar Crossref Search ADS Bönisch G. , Blindheim J. , Bullister J. L. , Schlosser P. , Wallace D. W. R. 1997 . Long-term trends of temperature, salinity, density, and transient tracers in the central Greenland Sea . Journal of Geophysical Research , 102 : 18553 – 18518, 571 . Google Scholar Crossref Search ADS Booth B. B. B. , Dunstone N. J. , Halloran P. R. , Andrews T. , Bellouin N. 2012 . Aerosols implicated as a prime driver of twentieth-century North Atlantic climate variability . Nature , 484 : 228 – 232 . Google Scholar Crossref Search ADS PubMed Boyce D. G. , Lewis M. R. , Worm B. 2010 . Global phytoplankton decline over the past century . Nature , 466 : 591 – 596 . Google Scholar Crossref Search ADS PubMed Brander K. M. 2005 . Cod recruitment is strongly affected by climate when stock biomass is low . ICES Journal of Marine Science , 62 : 339 – 343 . Google Scholar Crossref Search ADS Brander K. , Blom G. , Borges M. F. , Erzini K. , Henderson G. , MacKenzie B. R. , Mendes H. et al. 2003 . Changes in fish distribution in the eastern North Atlantic: are we seeing a coherent response to changing temperature? ICES Marine Science Symposia , 219 : 261 – 270 . Buch E. 1984 . Variations in temperature and salinity in West Greenland Water, 1970-82. NAFO . Science Council Studies , 7 : 39 – 44 . Buch E. , Horsted S. A. , Hovgård H. 1994 . Fluctuations in the occurrence of cod in Greenland waters and their possible causes . ICES Marine Science Symposia , 198 : 158 – 174 . Buch E. , Pedersen S. A. , Ribergaard M. H. 2004 . Ecosystem variability in West Greenland waters . Journal of the Northwest Atlantic Fishery Science , 34 : 13 – 28 . Google Scholar Crossref Search ADS Budyko M. I. 1969 . The effect of solar radiation variations on the climate of the Earth . Tellus , 21 : 611 – 619 . Google Scholar Crossref Search ADS Chylek P. , Folland C. K. , Lesins G. , Dubey M. K. , Wang M. 2009 . Arctic air temperature change amplification and the Atlantic Multidecadal Oscillation . Geophysical Research Letters , 36 : L14801. Colbourne E. B. 2004 . Decadal changes in the ocean climate in Newfoundland and Labrador waters from the 1950s to the 1990s . Journal of the Norwest Atlantic Fishery Science , 34 : 43 – 59 . Google Scholar Crossref Search ADS Colbourne E. , Narayanan S. , Prinsenberg S. 1994 . Climatic changes and environmental conditions in the Northwest Atlantic, 1970-1993 . ICES Marine Science Symposia , 198 : 311 – 322 . Corten A. 1986 . On the causes of the recruitment failure of herring in the central and northern North Sea in the years 1972–1978 . Journal du Conseil Permanent International pour l’Exploration de la Mer , 42 : 281 – 294 . Google Scholar Crossref Search ADS Corten A. 1990 . Long-term trends in pelagic fish stocks of the North Sea and adjacent waters and their possible connection to hydrographic changes . Netherlands Journal of Sea Research , 25 : 227 – 235 . Google Scholar Crossref Search ADS Corten A. 1999 . Evidence from plankton for multi-annual variations of Atlantic inflow in the northwestern North Sea . Journal of Sea Research , 42 : 191 – 205 . Google Scholar Crossref Search ADS Crisp D. J. 1964 . The effects of the severe winter of 1962-63 on marine life in Britain . Journal of Animal Ecology , 33 : 165 – 210 . Google Scholar Crossref Search ADS Crozier L. G. , Hutchings J. A. 2014 . Plastic and evolutionary responses to climate change in fish . Evolutionary Applications , 7 : 68 – 87 . Google Scholar Crossref Search ADS PubMed Curry R. G. , McCartnery M. S. , Joyce T. M. 1998 . Oceanic transport of subpolar climate signals to mid-depth subtropical waters . Nature , 391 : 575. Google Scholar Crossref Search ADS Cushing D. 1982 . Climate and Fisheries . Academic Press , London , 373 pp. Cushing D. , Dickson R. R. 1976 . The biological response in the sea to climatic changes . Advances in Marine Biology , 14 : 1 – 122 . Dalpadado P. , Ingvaldsen R. , Hassel A. 2003 . Zooplankton biomass variation in relation to climatic conditions in the Barents Sea . Polar Biology , 26 : 233 – 241 . Delworth T. L. , Mann M. E. 2000 . Observed and simulated multidecadal variability in the Northern Hemisphere . Climate Dynamics , 16 : 661 – 676 . Google Scholar Crossref Search ADS Dickson R. R. , Lamb H. H. 1972 . A review of recent hydro-meteorological events in the North Atlantic sector. International Commission for the Northwest Atlantic Fisheries . Special Publication , 8 : 35 – 62 . Dickson R. , Lazier J. , Meincke J. , Rhines P. , Swift J. 1996 . Long-term coordinated changes in the convective activity of the North Atlantic . Progress in Oceanography , 38 : 241 – 295 . Google Scholar Crossref Search ADS Drinkwater K. F. 1996 . Atmospheric and oceanic variability in the Northwest Atlantic during the 1980s and early 1990s . Journal of the Northwest Atlantic Fishery Science , 18 : 77 – 97 . Google Scholar Crossref Search ADS Drinkwater K. F. 2002 . A review of the role of climate variability in the decline of northern cod. In Fisheries in a Changing Climate , pp 113 – 130 . Ed. McGinn N. A. . American Fisheries Society Symposium 32 , Bethesda, Maryland . Drinkwater K. F. 2005 . The response of Atlantic cod (Gadus morhua) to future climate change . ICES Journal of Marine Science , 62 : 1327 – 1337 . Google Scholar Crossref Search ADS Drinkwater K. F. 2006 . The regime shift of the 1920s and 1930s in the North Atlantic . Progress in Oceanography , 68 : 134 – 151 . Google Scholar Crossref Search ADS Drinkwater K. F. , Belgrano A. , Borja A. , Conversi A. , Edwards M. , Greene C. H. , Ottersen G. , Pershing A. J. , Walker H. 2003 . The response of marine ecosystems to climate variability associated with the North Atlantic Oscillation. In The North Atlantic Oscillation Climatic Significance and Environmental Impact , pp. 211 – 234 . Ed. by Hurrell J. W. , Kushnir Y. , Ottersen G. , Visbeck M. . AGU Geophysical Monograph 134, American Geophysical Union , Washington, DC . Drinkwater K. F. , Miles M. , Medhaug I. , Otterå O. H. , Kristiansen T. , Sundby S. , Gao Y. 2014 . The Atlantic Multidecadal Oscillation: its manifestations and impacts with special emphasis on the Atlantic region north of 60°N . Journal of Marine Systems , 133 : 117 – 130 . Google Scholar Crossref Search ADS Dye D. R. , Hughes S. L. , Tinker J. , Berry D. I. , Holliday N. P. , Kent E. C. , Kennington K. et al. 2013 . Impacts of climate change on temperature (air and sea ). Marine Climate Change Impacts Partnership: Science Review , 2013 : 1 – 12 . Edwards M. , Beaugrand G. , Helaouet P. , Alheit J. , Coombs S. 2013 . Marine ecosystem response to the Atlantic Multidecadal Oscillation . PLoS One , 8 : e57212. Google Scholar Crossref Search ADS PubMed Edwards M. , Beaugrand G. , Reid P. C. , Rowden A. A. , Jones M. B. 2002 . Ocean climate anomalies and the ecology of the North Sea . Marine Ecology Progress Series , 239 : 1 – 10 . Google Scholar Crossref Search ADS Edwards M. , Reid P. C. , Planque B. 2001 . Long-term and regional variability of phytoplankton biomass in the Northeast Atlantic (1960–1995) . ICES Journal of Marine Science , 58 : 39 – 49 . Google Scholar Crossref Search ADS Enfield D. B. , Mestas-Nuñez A. M. , Trimble P. J. 2001 . The Atlantic multidecadal oscillation and its relation to rainfall and river flows in the continental US . Geophysical Research Letters , 28 : 2077 – 2080 . Google Scholar Crossref Search ADS Førland E. J. , Hanssen-Bauer I. 2003 . Past and furture climate variations in the Norwegian Arctic: overview and novel analyses . Polar Research , 22 : 113 – 124 . Google Scholar Crossref Search ADS Frankcombe L. M. , Dijkstra H. A. , von der Heydt A. 2008 . Sub-surface signatures of the Atlantic Multidecadal Oscillation . Geophysical Research Letters , 35 : L19602, Goldenberg S. B. , Landsea C. W. , Mestas N. A. M. , Gray W. M. 2001 . The recent increase in Atlantic hurricane activity: causes and implications . Science , 293 : 474 – 479 . Google Scholar Crossref Search ADS PubMed Glover R. S. , Robinson G. A. , Colebrook J. M. 1972 . Plankton in the North Atlantic-an example of the problems of analyzing variability in the environment. In FAO Marine Pollution and Sea Life , pp. 439 – 445 . Ed. by Ruivo M. , Fishing News (Books), Ltd. West Byfleet , Surrey and London . Griffies S. , Bryan K. 1997 . A predictability of simulated North Atlantic multidecadal variability . Climate Dynamics , 13 : 459 – 487 . Google Scholar Crossref Search ADS Grossman I. , Klotzbach P. J. 2009 . A review of North Atlantic modes of natural variability and their driving mechanisms . Journal of Geophysical Research , 114 : D24107. Google Scholar Crossref Search ADS Gudmundsson K. 1998 . Long-term variation in phytoplankton productivity during spring in Icelandic waters . ICES Journal of Marine Science , 55 : 635 – 643 . Google Scholar Crossref Search ADS Häkkinen S. , Rhines P. B. , Worthen D. L. 2011 . Atmospheric blocking and Atlantic multidecadal ocean variability . Science , 334 : 655 – 659 . Google Scholar Crossref Search ADS PubMed Hamilton L. C. , Haedrich R. L. 1999 . Ecological and population changes in fishing communities of the North Atlantic Arc . Polar Research , 18 : 383 – 388 . Google Scholar Crossref Search ADS Hamilton L. C. , Brown B. C. , Rasmussen O. 2003 . West Greenland’s cod-to-shrimp transition: local dimensions of climatic change . Arctic , 56 : 271 – 282 . Hanna E. , Jónsson T. , Ólafsson J. , Valdimarsson H. 2006 . Icelandic coastal sea surface temperature records constructed: putting the pulse on air–sea–climate interactions in the Northern North Atlantic. Part I: comparison with HadISST1 open-ocean surface temperatures and preliminary analysis of long-term patterns and anomalies of SSTs around Iceland . Journal of Climate , 19 : 5652 – 5666 . Google Scholar Crossref Search ADS Hawkins S. J. , Southward A. J. , Genner M. J. 2003 . Detection of environmental change in a marine ecosystem—evidence from the western English Channel . The Science of the Total Environment , 310 : 245 – 256 . Google Scholar Crossref Search ADS PubMed Heath M. , Lough G. 2007 . A synthesis of large-scale patterns in the planktonic prey of larval and juvenile cod (Gadus morhua) . Fisheries Oceanography , 16 : 169 – 185 . Google Scholar Crossref Search ADS Helland-Hansen B. , Nansen F. 1920 . Temperature variations in the North Atlantic Ocean and in the atmosphere . Smithsonian Miscellandeous Collections , 70 : 408 . Hermann F. C. 1967 . Temperature variations in the West Greenland area since 1950 . International Commission for North Atlantic Fishery Research Document 67 , 59 : 12 . Hislop J. R. G. 1996 . Changes in North Sea gadoid stocks . ICES Journal of Marine Science , 53 : 1146 – 1156 . Google Scholar Crossref Search ADS Hjort J. 1914 . Fluctuations in the great fisheries of northern Europe viewed in the light of biological research . Rapports et Procès-Verbaux des Réunions/Conseil Permanent International pour l’Exploration de la Mer , 20 : 1 – 228 . Hurrell J. W. , Kushnir Y. , Ottersen G. , Visbeck M. 2003 . An overview of the North Atlantic Oscillation. In The North Atlantic Oscillation Climatic Significance and Environmental Impact , pp. 1 – 35 . Ed. by Hurrell J. W. , Kushnir Y. , Ottersen G. , Visbeck M. . AGU Geophysical Monograph 134, American Geophysical Union , Washington, DC . Hylen A. 2002 . Fluctuations in abundance of Northeast Arctic cod during the 20th century . ICES Marine Science Symposia , 215 : 543 – 550 . ICES , 1949 . Climatic changes in the Arctic in relation to plants and animals . Rapports et Procès-Verbaux des Réunions/Conseil Permanent International pour l’Exploration de la Mer , 125 : 5 – 52 . ICES . 2012 . Cod in Division VIa (West of Scotland). Report of the ICES Advisory Committee, 2012. ICES Advice, June 2012. Inall M. , Gillibrand P. , Griffiths C. , MacDougal N. , Blackwell K. 2009 . On the oceanographic variability of the North-West European Shelf to the West of Scotland . Journal of Marine Systems , 77 : 210 – 226 . Google Scholar Crossref Search ADS Jensen A. S. 1939 . Concerning a change of climate during recent decades in the Arctic and subarctic regions, from Greenland in the west to Eurasia in the east, and contemporary biological and geophysical changes. Det. Kgl. Danske Videnskabernes Selskab . Biologiske Meddelelser , 14 : 75 . Jensen A. S. , Hansen P. M. 1931 . Investigations on the Greenland cod (Gadus callarias L.) with an introduction on the history of the Greenland cod fisheries . Rapports et Procès-Verbaux des Réunions/Conseil Permanent International pour l’Exploration de la mer , 72 : 1 – 41 . Johannessen O. M. , Bengtsson L. , Miles M. W. , Kuzmina S. I. , Semenov V. A. , Alekseev G. V. , Nagurnyi A. P. , Zakharov V. F. et al. 2004 . Arctic climate change: observed and modelled temperature and sea-ice variability . Tellus , 56A : 328 – 341 . Google Scholar Crossref Search ADS Jónsson S. 1991 . Seasonal and interannual variability of wind stress curl over the Nordic Seas . Journal of Geophysical Research , 96 : 2649 – 2659 . Google Scholar Crossref Search ADS Kerr R. A. 2000 . North Atlantic climate pacemaker for centuries . Science , 288 : 1984 – 1985 . Google Scholar Crossref Search ADS PubMed Kjesbu O. S. , Bogstad B. A. , Devine J. A. , Gjøsæter H. , Howell D. , Ingvaldsen R. B. , Nash R. D. M. et al. 2014 . Synergies between climate and management for Atlantic cod fisheries at high latitudes . In Proceedings of the National Academy of Sciences , 111 : 3478 – 3483 . Google Scholar Crossref Search ADS Klöwer M. , Latif M. , Ding H. , Greatbatch R. J. , Park W. 2014 . Atlantic meridional overturning circulation and the prediction of North Atlantic sea surface temperature . Earth and Planetary Science Letters , 406 : 1 – 6 . Google Scholar Crossref Search ADS Knight J. R. , Folland C. K. , Scaife A. A. 2006 . Climate impacts of the Atlantic Multidecadal Oscillation . Geophysical Research Letters , 33 : L17706. Google Scholar Crossref Search ADS Knudsen M. F. , Seidenkrantz M.-S. , Jacobsen B. H. , Kuijpers A. 2011 . Tracking the Atlantic Multidecadal Oscillation through the last 8,000 years . Nature Communications , 2 : 178. Google Scholar Crossref Search ADS PubMed Kristiansen T. , Stock C. , Drinkwater K. , Curchitser E. N. 2014 . Mechanistic insights into the effects of climate change on larval cod . Global Climate Change , 20 : 1559. Kushnir Y. 1994 . Interdecadal variations in North Atlantic sea surface temperature and associated atmospheric conditions . Journal of Climate , 7 : 141 – 157 . Google Scholar Crossref Search ADS Lamb H. H. 1972 . Climate: Present, Past and Future - Volume 1. Methuen, London. 647 pp. Lamb H. H. 1982 . Climate, History and the Modern World . Methuen , London and New York . 410 pp. Latif M. , Böning C. , Willebrand J. , Biastoch A. , Dengg J. , Keenlyside N. , Schweckendiek U. et al. 2006 . Is the thermohaline circulation changing? Journal of Climate , 19 : 4631 – 4637 . Google Scholar Crossref Search ADS Latif M. , Roeckner E. , Botzet M. , Esch M. , Haak H. , Hagemann S. , Jungclaus J. et al. 2004 . Reconstructing, monitoring, and predicting multidecadal-scale changes in the North Atlantic thermohaline circulation with sea surface temperature . Journal of Climate , 17 : 1605 – 1614 . Google Scholar Crossref Search ADS Lazier J. R. N. 1980 . Oceanographic conditions at Ocean Weather Ship Bravo, 1964–1974 . Atmosphere-Ocean , 18 : 227 – 238 . Google Scholar Crossref Search ADS Li J. , Sun C. , Jin F.-F. 2013 . NAO implicated as a predictor of Northern Hemisphere mean temperature multidecadal variability . Geophysical Research Letters , 40 : 5497 – 5502 . Google Scholar Crossref Search ADS Lilly G. R. , Nakken O. , Brattey J. 2013 . A review of the contributions of fisheries and climate variability to contrasting dynamics in two Arcto-boreal Atlantic cod (Gadus morhua) stocks: persistent high productivity in the Barents Sea and collapse on the Newfoundland and Labrador Shelf . Progress in Oceanography , 114 : 106 – 125 . Google Scholar Crossref Search ADS Lindley J. A. , Batten S. D. 2002 . Long-term variability in the diversity of North Sea zooplankton . Journal of the Marine Biological Association of the United Kingdom , 82 : 31 – 40 . Link J. S. , Yemane D. , Shannon L. J. , Coll M. , Shin Y.-J. , Hill L. , Borges M. F. 2010 . Relating marine ecosystem indicators to fishing and environmental drivers: an elucidation of contrasting responses . ICES Journal of Marine Science , 67 : 787 – 795 . Google Scholar Crossref Search ADS Lloyd J. , Moros M. , Perner K. , Telford R. J. , Kuijpers A. , Jansen E. , McCarthy D. 2011 . A 100 yr record of ocean temperature control on the stability of Jakobshavn Isbrae, West Greenland . Geology , 39 : 867 – 870 . Google Scholar Crossref Search ADS Loeng H. , Drinkwater K. 2007 . Climate variability and the ecosystems of the Barents and Norwegian Seas . Deep-Sea Research II , 54 : 2478 – 2500 . Google Scholar Crossref Search ADS Malmberg S.-A. 1969 . Hydrographic changes in the waters between Iceland and Jan Mayen in the last decade . Jökull , 19 : 30 – 43 . Malmberg S.-A. 1984 . Hydrographic conditions in the East Icelandic Current and sea ice in North Icelandic waters, 1970-1980 . Rapports et Procès-Verbaux des Reunions Conseil Permanent International pour l’Exploration de la Mer , 185 : 170 – 178 . Mann K. H. , Lazier J. R. N. 2006 . Dynamics of Marine Ecosystems . 3 rd edn, Blackwell Publishing Ltd. , Malden, MA USA . doi: 10.1002/9781118687901. Mernild S. H. , Hanna E. , Yde J. C. , Cappelen J. , Malmros J. K. 2014 . Coastal Greenland air temperature extremes and trends 1890–2010: annual and monthly analysis . International Journal of Climatology , 34 : 1472 – 1487 . Google Scholar Crossref Search ADS Mieszkowska N. , Burrows M. T. , Pannacciulli F. G. , Hawkins S. J. 2014 . Multidecadal signals within co-occurring intertidal barnacles Semibalanus balanoides and Chthamalus spp. linked to the Atlantic Multidecadal Oscillation . Journal of Marine Systems , 133 : 70 – 76 . Google Scholar Crossref Search ADS Mork K. A. , Skagseth Ø. 2010 . A quantitative description of the Norwegian Atlantic Current . Ocean Science , 6 : 901 – 911 . Google Scholar Crossref Search ADS Mueter F. J. , Broms C. , Drinkwater K. F. , Friedland K. D. , Hare J. A. , Hunt G. L. Jr. , Melle W. et al. 2009 . Ecosystem responses to recent oceanographic variability in high-latitude Northern Hemisphere ecosystems . Progress in Oceanography , 81 : 93 – 110 . Google Scholar Crossref Search ADS Nyberg J. , Malmgren B. A. , Winter A. , Jury M. R. , Kilbourne K. H. , Quinn T. M. 2007 . Low Atlantic hurricane activity in the 1970s and 1980s compared to the past 270 years . Nature , 447 : 698 – 701 . Google Scholar Crossref Search ADS PubMed Nye J. A. , Baker M. R. , Bell R. , Kenny A. , Kilbourne K. H. , Friedland K. D. , Martino E. et al. 2014 . Ecosystem effects of the Atlantic Multidecadal Oscillation . Journal of Marine Systems , 133 : 103 – 116 . Google Scholar Crossref Search ADS Öttera O. H. , Bentsen M. , Drange H. , Suo L. 2010 . External forcing as a metronome for Atlantic multidecadal variability . Nature Geoscience , 3 : 688 – 694 . Google Scholar Crossref Search ADS Oviatt C. , Smith L. , McManus M. C. , Hyde K. 2015 . Decadal patterns of westerly winds, temperatures, ocean gyre circulations and fish abundance: a review . Climate , 3 : 833 – 857 . Google Scholar Crossref Search ADS Park W. , Latif M. 2008 . Multidecadal and multicentennial variability of the meridional overturning circulation . Geophysical Research Letters , 35 : L22703. Google Scholar Crossref Search ADS Pecl G. T. , Araújo M. B. , Bell J. D. , Blanchard J. , Bonebrake T. C. , Chen I.-C. , Clark T. D. et al. 2017 . Biodiversity redistribution under climate change: impacts on ecosystems and human well-being . Science , 355 : eaai9214 . Google Scholar Crossref Search ADS PubMed Pedersen S. A. , Rice J. C. 2002 . Dynamics of fish larvae, zooplankton, and hydrographical characteristics in the West Greenland Large Marine Ecosystem, 1950–1984. In Large Marine Ecosystems of the North Atlantic: Changing States and Sustainability , pp. 151 – 193 . Ed. by Sherman K. , Skjoldal H. R. . Elsevier , Amsterdam, The Netherlands , 449 pp. Pedersen S. A. , Schmidt E. L. B. 2000 . Zooplankton distribution and abundance in West Greenland waters, 1950–1984 . Journal of the Northwest Atlantic Fishery Science , 26 : 45 – 102 . Google Scholar Crossref Search ADS Philippart C. J. M. , Anadón R. , Danovaro R. , Dippner J. W. , Drinkwater K. F. , Hawkins S. J. , Oguz T. et al. 2011 . Impacts of climate change on European marine ecosystems: observations, expectations and indicators . Journal of Experimental Marine Biology and Ecology , 400 : 52 – 69 . Google Scholar Crossref Search ADS Planque B. , Frédou T. 1999 . Temperature and the recruitment of Atlantic cod (Gadus morhua) . Canadian Journal of Fisheries and Aquatic Sciences , 56 : 2069 – 2077 . Google Scholar Crossref Search ADS Poloczanska E. S. , Hawkins S. J. , Southward A. J. , Burrows M. T. , 2008 . Modeling the response of populations of competing species to climate change . Ecology , 89 : 3138 – 3149 . Google Scholar Crossref Search ADS Polyakov I. V. , Alexeev V. A. , Bhatt U. S. , Polyakova E. I. , Zhang X. 2010 . North Atlantic warming: patterns of long-term trend and multidecadal variability . Climate Dynamics , 34 : 439 – 457 . Google Scholar Crossref Search ADS Rahmstorf S. , Box J. E. , Feulner G. , Mann M. E. , Robinson A. , Rutherford S. , Schaffernicht E. J. 2015 . Exceptional twentieth-century slowdown in Atlantic Ocean overturning circulation . Nature Climate Change , 5 : 956–480. Rätz H.-J. , Lloret J. 2005 . Long-term variability of growth and recruitment of cod (Gadus morhua) off Greenland . ICES Journal of Marine Science , 62 : 1310 – 1321 . Google Scholar Crossref Search ADS Rey F. , Loeng H. 1985 . The influence of ice and hydrographic conditions on the development of phytoplankton in the Barents Sea. In Gray J. S. , Christiansen M. E. , eds. Marine Biology of Polar Regions and Effects of Stress on Marine Organisms , In Proceedings of the 18th European Marine Biological Symposium Chichester, U.K . pp. 49 – 64 . John Wiley and Sons Ltd , Chichester, UK . Russell F. 1973 . A summary of the observations on the occurrence of planktonic stages of fish off Plymouth 1924–1972 . Journal of the Marine Biological Association of the United Kingdom , 53 : 347 – 355 . Google Scholar Crossref Search ADS Russell F. S. , Southward A. J. , Boalch G. T. , Butler E. I. 1971 . Changes in biological conditions in the English Channel off Plymouth during the last half century . Nature , 234 : 468 – 470 . Google Scholar Crossref Search ADS Sæmundsson B. 1934 . Probable influence of changes in temperature on the marine fauna of Iceland . Rapports et Procès-Verbaux des Réunions/Conseil Permanent International pour l’Exploration de la Mer , 86 (Part I): 1 – 6 . Sætersdal G. , Loeng H. 1987 . Ecological adaptation of reproduction in Northeast Arctic cod . Fisheries Research , 5 : 253 – 270 . Google Scholar Crossref Search ADS Saenger C. , Cohen A. L. , Oppo D. W. , Halley R. B. , Carilli J. E. 2009 . Surface-temperature trends and variability in the low-latitude North Atlantic since 1552 . Nature Geosciences , 2 : 492 – 495 . Google Scholar Crossref Search ADS Sigurdsson T. 2006 . The collapse of the Atlanto-Scandian herring fishery: Effects on the Icelandic economy. In 13th International Institute of Fisheries Economics and Trade (IIFET) Conference, July 11–14, Rebuilding Fisheries in an Uncertain Environment , Portsmouth, UK : http://ir.library.oregonstate.edu/xmlui/handle/1957/43506. Sheperd J. G. , Pope J. G. , Cousens R. D. 1984 . Variations in fish stocks and hypotheses concerning their links with climate . Rapports et Procès-Verbaux des Réunions/Conseil Permanent International pour l’Exploration de la Mer , 185 : 255 – 267 . Skaret G. , Dalpadado P. , Hjøllo S. S. , Skogen M. D. , Strand E. 2014 . Calanus finmarchicus abundance, production and population dynamics in the Barents Sea in a future climate . Progress in Oceanography , 125 : 26 – 39 . Google Scholar Crossref Search ADS Skud B. E. 1982 . Dominance in fishes: the relation between environment and abundance . Science , 216 : 144 – 149 . Google Scholar Crossref Search ADS PubMed Slagstad D. , Wassmann P. 1997 . Climate change and carbon flux in the Barents Sea: 3-D simulations of ice-distribution, primary production and vertical export of particulate organic matter. National Institute of Polar Research . Special Issue , 51 : 119 – 141 . Southward A. J. 1967 . Recent changes in abundance of intertidal barnacles in south-west England, a possible effect of climatic deterioration . Journal of the Marine Biological Association of the UK , 47 : 81 – 95 . Google Scholar Crossref Search ADS Southward A. J. 1974 . Long term changes in abundance of eggs of the Cornish pilchard (Sardina Pilchardus Walbaum) off Plymouth . Journal of the Marine Biological Association of the UK , 54 : 641 – 649 . Google Scholar Crossref Search ADS Southward A. J. 1980 . The western English Channel-an inconstant ecosystem? Nature , 285 : 361 – 366 . Google Scholar Crossref Search ADS Southward A. J. 1991 . Forty years of changes in species composition and population density of barnacles on a rocky shore near Plymouth . Journal of the Marine Biological Association of the UK , 71 : 495 – 3513 . Google Scholar Crossref Search ADS Southward A. J. , Hawkins S. J. , Burrows M. T. 1995 . Seventy years' observations of changes in distribution and abundance of zooplankton and intertidal organisms in the western English Channel in relation to rising sea temperature . Journal of Thermal Biology , 20 : 127 – 155 . Google Scholar Crossref Search ADS Southward A. J. , Langmead O. , Hardman-Mountford N. J. , Aiken J. , Boalch G. T. , Dando P. R. , Genner M. J. et al. 2004 . Long-term oceanographic and ecological research in the Western English Channel . Advances in Marine Biology , 47 : 1 – 105 . Google Scholar Crossref Search ADS Stefánsson U. 1969 . Temperature variations in the north Icelandic coastal area during recent decades . Jökull , 19 : 18 – 28 . Stein M. 2007 . Warming periods off Greenland during 1800–2005: their potential influence on the abundance of cod (Gadus morhua) and haddock (Melanogrammus aeglefinus) in Greenlandic waters . Journal of the Northwest Atlantic Fishery Science , 39 : 1 – 20 . Google Scholar Crossref Search ADS Stein M. , Buch E. 1991 . Are subsurface ocean temperatures predictable at Fylla Bank . West Greenland? NAFO Scientific Council Studies , 15 : 25 – 30 . Sundby S. 2000 . Recruitment of Atlantic cod stocks in relation to temperature and advection of copepod populations . Sarsia , 85 : 277 – 298 . Google Scholar Crossref Search ADS Sundby S. , Nakken O. 2008 . Spatial shifts in spawning habitats of Arcto-Norwegian cod related to multidecadal climate oscillations and climate change . ICES Journal of Marine Science , 65 : 953 – 962 . Google Scholar Crossref Search ADS Sutton R. T. , Hodson D. L. R. 2005 . Atlantic Ocean forcing of North American and European summer climate . Science , 309 : 115 – 118 . Google Scholar Crossref Search ADS PubMed Tåning Å. V. 1943 . Remarks on the influence of the higher temperatures in the northern waters during recent years on distribution and growth of some marine animals . Annales Biologiques , 1 : 76 . Tåning Å. V. 1949 . On changes in the marine fauna of the north-western Atlantic . Rapports et Procès-Verbaux des Réunions/Conseil Permanent International pour l’Exploration de la Mer , 125 : 26 – 29 . Thordardottir H. 1984 . Primary production north of Iceland in relation to water masses in May-June 1970-1989 . International Council for the Exploration of the Sea , CM1984/L: 20, 17. pp. Timmermann A. , Latif M. , Voss R. , Grötzner A. 1998 . Northern Hemispheric interdecadal variability: a coupled air–sea mode . Journal of Climate , 11 : 1906 – 1931 . Google Scholar Crossref Search ADS Toresen R. , Østvedt L. J. 2008 . Variation in abundance of Norwegian spring-spawning herring (Clupea harengus, Clupeidae) throughout the 20th century and the influence of climatic fluctuations . Fish and Fisheries , 1 : 231 – 256 . Google Scholar Crossref Search ADS Tuomenvirta H. , Alexandersson H. , Drebs A. , Frich P. , Nordli P. O. 2000 . Trends in Nordic and Arctic temperature extremes and ranges . Journal of Climate , 13 : 977 – 990 . Google Scholar Crossref Search ADS Valeur H. H. 1976 . Polar ice variations off the Greenland west coast, 1900-1972 . Special Publication of the International Commission for Northwest Atlantic Fishery , 7 : 7 – 27 . Vilhjálmsson H. 1997 . Climatic variations and some examples of their effects on the marine ecology of Icelandic and Greenland waters, in particular during the present century . Rit Fiskideildar , 40 : 7 – 29 . Vinje T. 2001 . Anomalies and trends of sea-ice extent and atmospheric circulation in the Nordic Seas during the period 1864–1998 . Journal of Climate , 14 : 255 – 267 . Google Scholar Crossref Search ADS Visbeck M. , Chassignet E. P. , Curry R. G. , Delworth T. L. , Dickson R. R. , Krahmann G. 2003 . The oceans response to North Atlantic Oscillation variability. In The North Atlantic Oscillation Climatic Significance and Environmental Impact , pp. 113 – 145 . Ed. by Hurrell J. W. , Kushnir Y. , Ottersen G. , Visbeck M. . AGU Geophysical Monograph 134, American Geophysical Union , Washington, DC . Watson S. J. , Kritharas P. , Hodgson G. J. 2015 . Wind speed variability across the UK between 1957 and 2011 . Wind Energy , 18 : 21 – 42 . Zhang R. , Vallis G. K. 2006 . Impact of Great Salinity Anomalies on the low-frequency variability of the North Atlantic climate . Journal of Climate , 19 : 717 – 734 . © International Council for the Exploration of the Sea 2018. All rights reserved. For permissions, please email: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Aarflot, Johanna Myrseth; Skjoldal, Hein Rune; Dalpadado, Padmini; Skern-Mauritzen, Mette

doi: 10.1093/icesjms/fsx221pmid: N/A

Abstract Copepods from the genus Calanus are crucial prey for fish, seabirds and mammals in the Nordic and Barents Sea ecosystems. The objective of this study is to determine the contribution of Calanus species to the mesozooplankton biomass in the Barents Sea. We analyse an extensive dataset of Calanus finmarchicus, Calanus glacialis, and Calanus hyperboreus, collected at various research surveys over a 30-year period. Our results show that the Calanus species are a main driver of variation in the mesozooplankton biomass in the Barents Sea, and constitutes around 80% of the total. The proportion of Calanus decreases at low zooplankton biomass, possibly due to a combination of advective processes (low C. finmarchicus in winter) and size selective foraging. Though the Calanus species co-occur in most regions, C. glacialis dominates in the Arctic water masses, while C. finmarchicus dominates in Atlantic waters. The larger C. hyperboreus has considerably lower biomass in the Barents Sea than the other Calanus species. Stages CIV and CV have the largest contribution to Calanus species biomass, whereas stages CI-CIII have an overall low impact on the biomass. In the western area of the Barents Sea, we observe indications of an ongoing borealization of the zooplankton community, with a decreasing proportion of the Arctic C. glacialis over the past 20 years. Atlantic C. finmarchicus have increased during the same period. Introduction Herbivorous zooplankton plays an important role in the marine pelagic food web converting energy from primary production to food for higher trophic levels in the ecosystem. Copepods of the genus Calanus are predominantly herbivores and the most important zooplankton in the Nordic and Barents Sea ecosystems, largely due to their high abundances and lipid contents (Jaschnov, 1970; Tande, 1991; Melle and Skjoldal, 1998; Søreide et al., 2008; Falk-Petersen et al., 2009). Being a high latitude ecosystem, the Barents Sea is characterized by strong seasonality in light and sea-ice conditions, with large impact on the marine biota. Three Calanus species are common here; Calanus finmarchicus is an Atlantic boreal species, while Calanus glacialis and Calanus hyperboreus are of Arctic origin (Conover, 1988; Tande, 1991; Melle and Skjoldal, 1998). Calanoid copepods are particularly well adapted to fluctuating environmental conditions due to reduced metabolic activity (diapause-like state) in winter when food is low, and capabilities of building large lipid reserves during the growing season. The individual lipid content in these species may be as large as 50–70% of the body weight (Lee, 1975; Scott et al., 2000), which make them valuable food sources for higher trophic levels in the system. Indeed, the calanoid copepods constitute a key part of the diet for many ecologically and economically important fish species in the Barents Sea (Wassmann et al. 2006; Orlova et al., 2011; Dalpadado and Mowbray, 2013). Calanus finmarchicus overwinters in deep waters (>500 m) of the Norwegian Sea, and is advected into the Barents Sea with the Atlantic current when it ascends to surface layers in spring (Skjoldal et al., 1992; Torgersen and Huse, 2005). Advection from the Norwegian Sea is vital for sustaining the population in the Barents Sea (Torgersen and Huse, 2005; Skaret et al., 2014), though local reproduction within the Barents Sea is also important (Kvile et al., 2017). This species generally has a predominantly 1-year life cycle in these waters, with the new generation produced at the onset of the phytoplankton spring bloom (Tande et al., 1985; Melle and Skjoldal, 1998). Calanus glacialis is a shelf species largely associated with Arctic water masses in the Barents Sea, and can have both 1- and 2-year life-cycles (Conover 1988; Tande, 1991; Melle and Skjoldal, 1998). The larger congener C. hyperboreus has in general low abundances in the Barents Sea (Hirche and Mumm, 1992; Melle and Skjoldal, 1998; Arashkevich et al., 2002), with its centre of origin in the deep basins of the Greenland Sea and Baffin Bay where it can have up to a 4-year life cycle (Conover, 1988; Hirche, 1997). Since around 1980, the Barents Sea has experienced a warming trend which has been particularly pronounced during the last two decades (Boitsov et al., 2012; Smedsrud et al., 2013). Warming has led to a northward shift in the spatial distribution of fish communities (Fossheim et al., 2015) and to a marked increase in the amount of krill and cumulative biomass of pelagic species (Eriksen et al., 2016, 2017b). Continued warming has increased the dominance of Atlantic species and negatively impacted the Arctic communities (Hirche and Kosobokova, 2007; Kjellerup et al., 2012; Dalpadado et al., 2014; Fossheim et al., 2015; Frainer et al., 2017). Short-lived species like plankton are expected to show rapid responses to a changing climate (Hays et al., 2005), and changes at the base of the marine food chain may propagate through the system with consequences at an ecosystem scale (Beaugrand et al., 2003; Helaouët and Beaugrand, 2007). Revealing ongoing changes in marine plankton (e.g. Beaugrand et al., 2002) is therefore vital for predicting the future of marine ecosystems in a warmer climate. The Barents Sea zooplankton community has been studied extensively (e.g. Hassel, 1986; Tande, 1991; Unstad and Tande, 1991; Melle and Skjoldal, 1998; Arashkevich et al., 2002). Many studies point to the importance of the Calanus species due to their size, abundance and lipid contents, though few have quantified their contribution to the total mesozooplankton biomass. Furthermore, most studies have analysed samples from a restricted time-period of one or a few years with low seasonal resolution. We explored an extensive dataset of species abundance for C. finmarchicus, C. glacialis, and C. hyperboreus, originating from various research and monitoring surveys in the Barents Sea, conducted by the Institute of Marine Research (IMR), Norway, over a 30-year period. IMR has used a standard method of splitting each zooplankton sample in two halves: one for determination of dry weight (dw) biomass, and the other preserved for species counts (Melle et al., 2004). Our aim was to quantify the relationship between sampled mesozooplankton biomass and estimated biomass of Calanus species in the Barents Sea using the pair-wise samples. We further investigated the spatial patterns of the three species in relation to water masses and bottom topography, and evaluated whether there has been a change in the copepod community concurrent with the recent warming in the area. A transition towards dominance of smaller-sized, Atlantic copepods could affect the lipid structure and energy flow in the ecosystem with consequences for many trophic levels in the food web. Material and methods Zooplankton sampling and analyses The standard procedure for zooplankton sampling at the IMR, Norway, is described in detail in Melle et al. (2004) and Skjoldal et al. (2013). Briefly, samples are divided in two halves with a Motoda plankton splitter, one part for determining the biomass (g dw per m2 or m3), and the other half for species identification and abundance estimation. The biomass subsample is separated into three size fractions using mesh gauzes of 2000, 1000, and 180 µm (for details, see Skjoldal et al., 2013). The second subsample is preserved with buffered 4% formalin solution and stored for later processing. The three Calanus species are identified based on size limits (Supplementary Table S1) and morphological characteristics including shape of the curvature of the coxopodite of the fifth leg (P5) (Knutsen and Dalpadado, 2009), and counted separately for each copepodite stage (CI–CV and CVI females and males). Consistent size-limits have been used throughout the period of the samples used in our study (see Hassel, 1986; Melle and Skjoldal, 1998). The size frequency data typically follow normal distributions for each of the species, with some (and variable) overlap between them, particularly for C. finmarchicus and C. glacialis (Hassel, 1986; Unstad and Tande, 1991; Melle and Skjoldal, 1998; Parent et al., 2011; Gabrielsen et al., 2012). Use of fixed size limits to separate the species is therefore an approximation, and the potential for misidentifications is present, particularly in areas where the species co-occur. Individuals of intermediate size are therefore routinely examined for curvature of the coxopodite to reduce the degree of misidentification from the use of fixed size limits. Data description Sample processing for species identification is labour-intensive, and only a fraction of the samples collected by the IMR are processed (all samples are stored in a long-term repository). Over the years, there has still been an accumulation of processed samples originating from various researches and monitoring surveys. We extracted all samples in the IMR database with data on both mesozooplankton biomass and species abundance from the same sampling stations in the Barents Sea (Tables 1 and 2). When multiple samples had been taken at a station, only one (WP2 gear, bottom to surface haul) was included in this study. In total, we analysed 616 samples covering an extensive geographical area (Figure 1). Samples were grouped into five oceanographic regions based on bathymetry and advection (Table 2), and aggregated into the following seasons: winter (November–March), spring (April–May), summer (June–July), and autumn (August–October). The Fugløya-Bear Island transect (FB transect, grey line in Figure 1) is a standard oceanographic transect in the western region, hereafter called “West”, covered by IMR five to eight times each year. Samples from this transect are regularly processed for species identification, and have consistent seasonal coverage since 1995. Region West therefore contributed a large part (∼70%) to the data analysed in this study. Samples from the 1980s (the Pro Mare programme; Sakshaug et al., 2009) were mainly from the spring and summer period. Table 1. Gear characteristics of the sampling equipment in the dataset. Sampling gear . Net opening (cm) . Mesh size (µm) . Lower sampling depth (m) . Sample unit . Samples (n) . WP2 56 180 100, bottom m−2 569 Juday 80 250, 375 40, 50 m−2 14 Hufsa – 180, 375 30, 40, 50, 100 m−3 28 MOCNESS 100 180, 333 bottom m−3 5 Sampling gear . Net opening (cm) . Mesh size (µm) . Lower sampling depth (m) . Sample unit . Samples (n) . WP2 56 180 100, bottom m−2 569 Juday 80 250, 375 40, 50 m−2 14 Hufsa – 180, 375 30, 40, 50, 100 m−3 28 MOCNESS 100 180, 333 bottom m−3 5 For detailed gear descriptions, see Sameoto et al. (2000), Wiebe and Benfield (2003), and Skjoldal et al. (2013). Upper sampling depth for all gears is surface (0 m). Open in new tab Table 1. Gear characteristics of the sampling equipment in the dataset. Sampling gear . Net opening (cm) . Mesh size (µm) . Lower sampling depth (m) . Sample unit . Samples (n) . WP2 56 180 100, bottom m−2 569 Juday 80 250, 375 40, 50 m−2 14 Hufsa – 180, 375 30, 40, 50, 100 m−3 28 MOCNESS 100 180, 333 bottom m−3 5 Sampling gear . Net opening (cm) . Mesh size (µm) . Lower sampling depth (m) . Sample unit . Samples (n) . WP2 56 180 100, bottom m−2 569 Juday 80 250, 375 40, 50 m−2 14 Hufsa – 180, 375 30, 40, 50, 100 m−3 28 MOCNESS 100 180, 333 bottom m−3 5 For detailed gear descriptions, see Sameoto et al. (2000), Wiebe and Benfield (2003), and Skjoldal et al. (2013). Upper sampling depth for all gears is surface (0 m). Open in new tab Table 2. Overview of regions as defined in this study, and number of samples analysed per region. Region . Latitude (°N) . Longitude (°E) . Bottom depth (m)a . Dominating water massb . Main sampling periodc . Samples (n) per season . West 70–75 15.5–21 266 Atlantic 1994–2016 Summer 65 Autumn 170 Winter 177 Spring 89 South 70–73.5 21–40 317 Atlantic 1983–2016 Summer 9 Autumn 7 Winter 0 Spring 3 Central 74–78 21–38 221 Arctic/mixed 1983–2009 Summer 33 Autumn 15 Winter 2 Spring 6 North 78–82 25–36 211 Arctic/mixed 2005–2016 Summer 1 Autumn 22 Winter 0 Spring 0 East 71–80 41–61 234 Arctic/mixed 1983–1994 Summer 5 Autumn 11 Winter 1 Spring 0 Region . Latitude (°N) . Longitude (°E) . Bottom depth (m)a . Dominating water massb . Main sampling periodc . Samples (n) per season . West 70–75 15.5–21 266 Atlantic 1994–2016 Summer 65 Autumn 170 Winter 177 Spring 89 South 70–73.5 21–40 317 Atlantic 1983–2016 Summer 9 Autumn 7 Winter 0 Spring 3 Central 74–78 21–38 221 Arctic/mixed 1983–2009 Summer 33 Autumn 15 Winter 2 Spring 6 North 78–82 25–36 211 Arctic/mixed 2005–2016 Summer 1 Autumn 22 Winter 0 Spring 0 East 71–80 41–61 234 Arctic/mixed 1983–1994 Summer 5 Autumn 11 Winter 1 Spring 0 Samples were aggregated into the seasons winter (November–March), spring (April–May), summer (June–July), and autumn (August–October). a Mean of sampling stations. b Dominating water mass in samples analysed: Atlantic (T > 3 °C), Arctic (T < 0 °C), mixed (0 °C < T < 3 °C) based on temperature at 50-m depth. c >90 % of samples taken during this period. Open in new tab Table 2. Overview of regions as defined in this study, and number of samples analysed per region. Region . Latitude (°N) . Longitude (°E) . Bottom depth (m)a . Dominating water massb . Main sampling periodc . Samples (n) per season . West 70–75 15.5–21 266 Atlantic 1994–2016 Summer 65 Autumn 170 Winter 177 Spring 89 South 70–73.5 21–40 317 Atlantic 1983–2016 Summer 9 Autumn 7 Winter 0 Spring 3 Central 74–78 21–38 221 Arctic/mixed 1983–2009 Summer 33 Autumn 15 Winter 2 Spring 6 North 78–82 25–36 211 Arctic/mixed 2005–2016 Summer 1 Autumn 22 Winter 0 Spring 0 East 71–80 41–61 234 Arctic/mixed 1983–1994 Summer 5 Autumn 11 Winter 1 Spring 0 Region . Latitude (°N) . Longitude (°E) . Bottom depth (m)a . Dominating water massb . Main sampling periodc . Samples (n) per season . West 70–75 15.5–21 266 Atlantic 1994–2016 Summer 65 Autumn 170 Winter 177 Spring 89 South 70–73.5 21–40 317 Atlantic 1983–2016 Summer 9 Autumn 7 Winter 0 Spring 3 Central 74–78 21–38 221 Arctic/mixed 1983–2009 Summer 33 Autumn 15 Winter 2 Spring 6 North 78–82 25–36 211 Arctic/mixed 2005–2016 Summer 1 Autumn 22 Winter 0 Spring 0 East 71–80 41–61 234 Arctic/mixed 1983–1994 Summer 5 Autumn 11 Winter 1 Spring 0 Samples were aggregated into the seasons winter (November–March), spring (April–May), summer (June–July), and autumn (August–October). a Mean of sampling stations. b Dominating water mass in samples analysed: Atlantic (T > 3 °C), Arctic (T < 0 °C), mixed (0 °C < T < 3 °C) based on temperature at 50-m depth. c >90 % of samples taken during this period. Open in new tab Figure 1. Open in new tabDownload slide Geographical distribution of samples analysed in this study (n = 616). The Barents Sea was divided into five oceanographic regions as defined in Table 2. Outer bounds of the polygons are included as a visual aid. Samples were defined as Arctic (T < 0 °C), Atlantic (T > 3 °C), or mixed (0 °C < T < 3 °C) based on temperature data from 50 m depth. The FB transect, where a large part of the data originates from, is marked with a line. Most of the samples were from near-bottom to surface hauls, though ∼ 10% had shallower sampling depths (Table 1). Samples with a unit of abundance or biomass m−3 were converted to m−2 by integrating over the water column down to the lowest sampling depth. Differences in sampling gear and depth were accounted for in the statistical analyses. Biomass estimation of Calanus species Copepodite abundances of C. finmarchicus, C. glacialis, and C. hyperboreus were converted to biomass estimates using individual weight-at-stage data from the literature (Table 3). The individual weight can vary considerably, by up to an order of magnitude within a copepodite stage (Figure 2). Part of this variation is due to weight increase as individuals grow through a stage between successive moults. There is also systematic variation in relation to thermal habitat, where individuals tend to become larger when they grow at low compared with higher temperature (Campbell et al., 2001; Melle et al., 2014). Mean weights from studies in or near the Barents Sea were considered representative of those for our study region (Table 3). We also performed length measurements on individuals of C. finmarchicus and C. glacialis stages CIV, CV and adult females, to evaluate the propriety of the weight-data employed for estimating species biomass. Based on these measurements we were confident that the weight-data (Table 3) were reasonable (results are available in the Supplementary Material). Table 3. Dry weight (µg) per copepodite stage (CI–CVI female and male) for Calanus spp. used to estimate biomass in this study. Species . CI . CII . CIII . CIV . CV . CVIf . CVIm . References . C. finmarchicus 1.5 4 13 70 250 235 235 Tande (1982), Tande and Slagstad (1992) C. glacialis 8 16 40 185 600 810 600 Hanssen (1997), Hirche and Kosobokova (2003) C. hyperboreus 10 40 140 500 2000 3500 3500 Hirche (1997) Species . CI . CII . CIII . CIV . CV . CVIf . CVIm . References . C. finmarchicus 1.5 4 13 70 250 235 235 Tande (1982), Tande and Slagstad (1992) C. glacialis 8 16 40 185 600 810 600 Hanssen (1997), Hirche and Kosobokova (2003) C. hyperboreus 10 40 140 500 2000 3500 3500 Hirche (1997) See also Figure 2 for an overview of dry weight measurements of C. finmarchicus and C. glacialis from the literature. Open in new tab Table 3. Dry weight (µg) per copepodite stage (CI–CVI female and male) for Calanus spp. used to estimate biomass in this study. Species . CI . CII . CIII . CIV . CV . CVIf . CVIm . References . C. finmarchicus 1.5 4 13 70 250 235 235 Tande (1982), Tande and Slagstad (1992) C. glacialis 8 16 40 185 600 810 600 Hanssen (1997), Hirche and Kosobokova (2003) C. hyperboreus 10 40 140 500 2000 3500 3500 Hirche (1997) Species . CI . CII . CIII . CIV . CV . CVIf . CVIm . References . C. finmarchicus 1.5 4 13 70 250 235 235 Tande (1982), Tande and Slagstad (1992) C. glacialis 8 16 40 185 600 810 600 Hanssen (1997), Hirche and Kosobokova (2003) C. hyperboreus 10 40 140 500 2000 3500 3500 Hirche (1997) See also Figure 2 for an overview of dry weight measurements of C. finmarchicus and C. glacialis from the literature. Open in new tab Figure 2. Open in new tabDownload slide Mean weight (μg ind−1, points in figure) for copepodite stage CV and adult females of (a) C. finmarchicus, and (b) C. glacialis, as reported by the scientific literature (x-axis). (i) Carlotti et al. (1993), (ii) Tande (1982), (iii) Ikeda and Skjoldal (1989), (iv) Scott et al. (2000), (v) Diel (1991), (vi) Hirche et al. (2001), (vii) Gislason (2005), (viii) Båmstedt and Ervik (1984), (ix) Jónasdóttir (1999) (*deep water), (x) Heath and Jónasdóttir (1999), (xi) Runge et al. (2006), (xii) Kjellerup et al. (2012), (xiii) Båmstedt and Tande (1985), (xiv) Hirche (1987), (xv) Hirche and Kattner (1993), (xvi) Hirche et al. (1994), (xvii) Hirche and Kwasniewski (1997), (xviii) Hirche and Kosobokova (2003), (xix) Tourangeau and Runge (1991). Vertical lines show the range of weights, or mean ± SD, when this information has been available. Horizontal lines show the values employed in this study when estimating species biomass for stage CV (dotted) and females (dashed). Physical environment Temperature and salinity profiles from CTD casts from the respective sampling stations were available for most of the dataset. Samples were classified as Atlantic (T > 3 °C), Arctic (T < 0 °C), or mixed (0 °C < T < 3 °C) based on temperature at 50 m, where the core of Arctic water is usually found (Lind and Ingvaldsen, 2012; Lind et al., 2016). Temperature and salinity at 50 m were used as continuous variables in the statistical analyses explaining variance in Calanus sp. biomass (see ii below), and sampling depth as a proxy for bottom depth since some samples were not taken from bottom to surface. Data analyses Statistical analyses were performed to: Estimate the relationship between Calanus biomass (sum of the three species) and the mesozooplankton biomass in the pair-wise samples. Evaluate interspecific differences in biomass between the three Calanus species with regard to key environmental drivers. Analyse inter-annual changes in the Calanus species group regarding species biomass and % contribution to total biomass. For (i) and (ii), we employed the complete dataset with 616 samples (613 samples in (ii) due to missing temperature data from three stations). For (iii), we used summer and autumn data from region West (mainly FB transect) where we had annual observations since 1995. Analyses were performed on log-transformed estimated dw biomass plus a constant (0.01) to enable log-transformation of samples with species absence (zero biomass). Total Calanus vs. mesozooplankton biomass We used Major Axis regression (MA) to estimate the relationship between the observed (log-transformed) mesozooplankton biomass and the estimated total biomass of Calanus spp. This regression technique is suitable for describing the functional relationship between two variables of the same units of measurement when both are subject to observation error (Helsel and Hirsch, 1992; Sokal and Rohlf, 2012). We also performed an ordinary least squares (OLS) regression for comparison with the MA, to evaluate how results would change by the choice of regression model. Calanus biomass at species level OLS regressions with species biomass as response variable was used to evaluate interspecific differences between the Calanus species with regard to environmental factors (temperature, salinity and sampling depth as continuous variables, season as categorical). Data on C. glacialis and C. hyperboreus had considerable zero-inflations as a large portion of the data came from the Atlantic sector of the Barents Sea, so analyses for these species were performed on all samples as well as only presence-data. We also ran separate analyses with presence/absence as a response, using Generalized Linear Models with a binomial distribution. Model selection (i.e. deciding on the optimal models describing estimated biomass at species level) was based on the Akaike information criterion (AIC; Akaike, 1974) which considers the trade-off between model fit and model complexity, and backwards selection (stepwise removal of the least significant term). All analyses were run both on the complete dataset and on data only including samples taken from bottom to surface. To account for differences in sampling gear characteristics like mesh size and net opening, equipment was included as a fixed covariate in the analyses. Due to an overweight of samples from the WP2 sampling gear, this dataset was not suitable for concluding on differences in sampling gear performance. Temporal changes in region West Changes in biomass at species level and changes in the proportion of C. finmarchicus, C. glacialis, and C. hyperboreus in the total mesozooplankton biomass over the period (1995–2016) were analysed with generalized additive models (GAMs) to catch potential non-linear trends in temporal variation. We used a spline based smoother with four degrees of freedom. In analyses of proportions, estimates >1 were set to 1, and analyses were run on arcsine transformed values. All analyses were done in the statistical software packageR (R Core Team, 2016), using the mgcv library for GAMs (Wood, 2017). Results Correlation between Calanus spp. and total mesozooplankton biomass There was a strong correlation between the observed mesozooplankton biomass and the estimated biomass of Calanus species in the samples (r2 = 0.79, p = 0.005) (Figure 3). Results were similar both with the complete dataset and when excluding samples that did not cover the entire water column. The observed biomass spanned a range of about three orders of magnitude, from 0.01 to 48 g dw m−2, with a similar range also for the estimated biomass of Calanus species (0.003–50 g dw m−2). On average, the Calanus species comprised 78% of the mesozooplankton biomass, though this varied between the different regions (see below). Figure 3. Open in new tabDownload slide Observed mesozooplankton biomass and estimated total biomass of three Calanus species in the samples. Samples are shown with symbols by season; winter (November–March), spring (April–May), summer (June–July), and autumn (August–October). The dotted line shows a 1:1 relationship between mesozooplankton and Calanus spp. biomass. Regression results (MA and OLS) are plotted with 95% confidence bands, r2 = 0.79 and p = 0.005 for both regressions. The scatter around the regression line in Figure 3 was approximately one order of magnitude (corresponding to one unit on the log scale). The estimated dw of the three Calanus spp. surpassed the observed mesozooplankton dw sampled at the station (i.e. observations above the 1:1 dotted line in Figure 3) in 19% of the cases. Overestimations occurred in all seasons, both at high and low biomass levels. The MA regression slope was steeper than unity (1.24 on the log-log scale), which means that the % contribution of Calanus species to the observed biomass increased with increasing biomass values. In fact, the regression line crossed the 1:1 line at a log value about 1.5 (32 g dw m−2). The OLS regression had a lower slope (1.1) and did not cross the 1:1 line. OLS in bivariate regressions tends to underestimate the slope of the regression line when both variables are subject to observation error not controlled by the researcher (Sokal and Rohlf, 2012), which may be reflected in our data as well (Figure 3). We therefore conclude that predictions from the MA regression more accurately described the relationship between Calanus spp. and mesozooplankton biomass in the Barents Sea. Hydrographic and spatial differences between Calanus spp. There was considerable variation in the estimated % contribution of each species to mesozooplankton biomass in the water masses defined as Arctic, Atlantic and mixed (large interquartile ranges, Figure 4). However, the water masses were distinctively different regarding which of the three Calanus species that contributed to the mesozooplankton biomass. In Atlantic water, C. finmarchicus constituted a large part of the mesozooplankton biomass whereas C. glacialis had a low contribution to the total. In Arctic water C. glacialis prevailed, with low contribution by C. finmarchicus. Both C. finmarchicus and C. glacialis contributed to the total in mixed water masses. Calanus hyperboreus was generally a small part of the mesozooplankton biomass in all water masses, though relatively more abundant in the Arctic than the other two. Figure 4. Open in new tabDownload slide Estimated proportions of total mesozooplankton biomass for C. finmarchicus, C. glacialis, and C. hyperboreus in different water masses defined as Atlantic (T > 3 °C), Arctic (T < 0 °C), and mixed (0 °C < T < 3 °C). Number of samples (n) from each water mass is indicated in the x-axis labels. The graph presented excludes 12 observations with estimated proportions >200 %. The boxes are divided by the median value, and framed by the upper and lower quartile. The whiskers extend to the first outlier in each direction; other outliers are shown by separate points. Outliers are defined as data points >1.5 times the upper quartile. A summary of biomass estimates and estimated proportions of the three Calanus species in the five regions shown in Figure 1 is available in the Supplementary Material (Supplementary Table S2). The total contribution by the three Calanus species to the mesozooplankton biomass differed across the regions, from ∼50% in the East to >90% in the South. On species level, the % contribution in each area reflected differences between the water masses as illustrated in Figure 4. The West and South regions where Atlantic water prevails was dominated by C. finmarchicus, while C. glacialis was a larger fraction of the total in the North and East regions where Arctic water is present (Figure 5). Both species had a similar contribution to the biomass in the Central region which contains the oceanographic polar front with cooled Atlantic and mixed water masses. Species other than Calanus appeared to have a larger contribution to the mesozooplankton biomass in the North, Central and East regions than in the West and South (Figure 5). The “other” category is usually dominated by species like Metridia spp., Pseudocalanus spp., Microcalanus spp., Oithona spp., Oncaea spp., and Clione limacina (IMR database). Figure 5. Open in new tabDownload slide Estimated proportion of C. finmarchicus, C. glacialis, and C. hyperboreus biomass to total mesozooplankton biomass in different regions of the Barents Sea, based on arithmetic means (g dw m−2) per region. The size of the cakes is proportional to the total mesozooplankton biomass. “Other” represents the total minus the estimated mean biomass of the Calanus species. Winter samples from region West are not included in the figure. The total variation in estimated biomass within the pooled datasets was large, with coefficient of variation (CV) typically greater than one (Supplementary Table S2). CV values tended to be higher at low estimated biomass values and were generally higher for Calanus biomass estimates than for the total mesozooplankton biomass. High CV values suggest a skewed distribution (relative to normal) which is reflected in median values being lower than arithmetic means (by 5–40% for total mesozooplankton biomass, and 20–60% for estimated biomass of C. finmarchicus and C. glacialis). Environmental drivers of Calanus biomass Selected linear regressions based on the AIC and backwards selection, showed that the best model for describing the estimated biomass at species level included season, sampling depth, equipment and temperature (50 m) for all three species (r2 = 0.38 for C. finmarchicus, 0.51 for C. glacialis and 0.31 for C. hyperboreus). Model coefficients with standard errors are available in the Supplementary Table S3. Among the predictors, temperature revealed clear differences between the species (Figure 6a). Calanus finmarchicus had a positive relationship with temperature (p < 0.001), while it was negative for C. glacialis (p < 0.001). Also C. hyperboreus had a negative relationship with temperature (p < 0.001), though weaker than for C. glacialis. Sampling depth was positively related to estimated biomass for all three species (Figure 6b), giving higher Calanus spp. biomass in deep vs. shallow water. The model for C. finmarchicus predicted a higher mean biomass in summer compared with autumn, and lower for winter and spring. For C. glacialis and C. hyperboreus, the models predicted lower mean biomass in winter, spring and summer compared with autumn. Salinity had no significant effect for neither species. These trends were consistent across all datasets (complete, bottom to surface and presence-only data for C. glacialis and C. hyperboreus). Further, binomial models on presence/absence for C. glacialis and C. hyperboreus confirmed the negative relationship of these species with temperature. Figure 6. Open in new tabDownload slide Estimated biomass of the three Calanus species against (a) temperature and (b) sampling depth, with data from equipment WP2 and season autumn. Predictions (straight lines with 95 % confidence bands) are from the linear models log(Calanus sp. dw) ∼ temperature + season + depth + equipment (r2 = 0.38 for C. finmarchicus, 0.52 for C. glacialis, and 0.31 for C. hyperboreus), with mean levels of depth (a) and temperature (b). Temporal changes in region West The total mesozooplankton biomass in June and August in region West showed an increasing trend in recent years (Figure 7a). This coincided with an increase in the medium (1000–2000 µm) and small (180–1000 µm) mesozooplankton size fractions, while the large (>2000 µm) size fraction has decreased since around 2002. GAM analyses on the estimated proportion of the three Calanus species in the corresponding samples revealed a linear decrease in the % contribution to total biomass of C. glacialis over the period (Figure 7b, p = 0.04). Meanwhile, the proportion of C. finmarchicus has increased since the early 2000s (p = 0.003). C. hyperboreus constituted a very small part of the mesozooplankton biomass in region West. Its contribution to the total was generally below 5% except between the years 2002 and 2004 when it had a “peak” contribution (Figure 7b, p = 0.002). Model outputs are available in the Supplementary Figure S1. Figure 7. Open in new tabDownload slide (a) Mean sampled June and August mesozooplankton biomass (g dw m−2) in the Barents Sea, region West, from 1995 to 2016. Figure shows total biomass and biomass divided into three size fractions. (b) Mean estimated proportion (%) of C. finmarchicus, C. glacialis, and C. hyperboreus in the corresponding samples. Error bars show ± the SEM proportion. One potential outlier with estimated proportion of C. finmarchicus >500 % was removed in the figure. The trend lines are results from GAM models with species proportions as response and year as explanatory variable; p = 0.003, 0.04, and 0.002 for C. finmarchicus, C. glacialis and C. hyperboreus, and deviance explained is 10, 4.2, and 12%, respectively. GAM analyses on estimated species biomass over the same period showed increasing biomass of C. finmarchicus since around 2005 (p = 0.05) (see Figure 8b). At the same time, the biomass of C. glacialis decreased (apart from the most recent years), though the trend was not significant at the 0.05 level (p = 0.07). Figure 8. Open in new tabDownload slide Mean biomass (g dw m−2) per stage (CI to CV and CVI female and male) for C. finmarchicus in the western region of the Barents Sea between (a) 1995–2004 and (b) 2005–2016. The figure only displays months which have been consistently sampled over the period. Stage specific contribution to biomass Calanus finmarchicus was a consistently large part of the mesozooplankton biomass in region West, where Atlantic water dominates. Samples from this region revealed that copepodite stages CIV and CV dominated the total species biomass for C. finmarchicus (Figure 8). The new generation consisting of younger copepodites (CI–CIII) appeared in May. However, they comprised a very small part of the estimated total biomass in all months analysed. Stages CIV and CV of the new generation created a seasonal maximum biomass in June–August. Samples from winter months (January, March) indicated that C. finmarchicus overwinters mainly as stage CV in this area. Stages CIV and CV dominated the biomass also for C. glacialis in regions Central, North and East (Figure 9) where this species was a large fraction of the mesozooplankton biomass. Winter samples for C. glacialis indicated overwintering mainly as stage CIV and adults. The younger stages, particularly CIII, had a larger contribution to the total species biomass for C. glacialis during summer and autumn than with C. finmarchicus. The maximum mean monthly estimated biomass of C. glacialis of about 3.6 g dw m−2 was comparable to (but slightly lower than) the maximum biomass of C. finmarchicus apart from the higher values for the latter species after 2005 (Figure 8b). Figure 9. Open in new tabDownload slide Mean biomass (g dw m−2) per stage (CI to CV and CVI female and male) for C. glacialis, with data from the Central, North and East regions considered as most representative for this species. Months with no observations are indicated by NA. Discussion Estimated biomass of Calanus species Calanus spp. are key species at high latitudes spanning from boreal to Arctic ecosystems (Jaschnov, 1970; Conover, 1988; Falk-Petersen et al., 2009). Yet, few studies have quantified the contribution of Calanus species to the total zooplankton biomass. Biomass of Calanus is typically estimated by combining stage-abundance data with mean individual body weights of the respective stages (e.g. Tande, 1991; Hirche and Kosobokova, 2003; Søreide et al., 2008). Using this method, we found a mean biomass of C. finmarchicus around 5 g dw m−2 in June and August (1995–2005) in the western region of the Barents Sea dominated by Atlantic water. After 2005, the biomass of C. finmarchicus has increased. Our estimates for the later years (2005–2016) are in the high end of the range of values reported from other areas. A detailed sampling at Station M in the Norwegian Sea gave a mean biomass of 1.7 g dw m−2 with a temporary maximum of 12.5 g dw m−2 (Hirche et al., 2001). Simulations with a coupled physical–biological model system (NORWECOM) gave a seasonal maximum biomass of C. finmarchicus of 4–5 g dw m−2 in the Norwegian Sea and the Atlantic part of the Barents Sea (Hjøllo et al., 2012; Skaret et al., 2014, see review of estimated biomass of the three Calanus species provided in the Supplementary Table S4). Our estimates for the colder waters of the central, eastern and northern Barents Sea were lower, and similar to values obtained in the same region by Hirche and Kosobokova (2003). Estimated biomass of C. glacialis in the North, Central and East regions was slightly lower than the biomass of C. finmarchicus in the West, with a seasonal maximum around 3.6 g dw m−2. This is comparable to studies of C. glacialis both from the Barents Sea and other areas (Tande, 1991; Madsen et al., 2001; Hirche and Kosobokova, 2003; Daase et al., 2013). Our biomass estimates for C. hyperboreus were 0.1–0.7 g dw m−2 as means for the different areas. These are similar to values reported from the Barents Sea by Tande (1991) and Hirche and Kosobokova (2003). Higher values of up to 4–6 g dw m−2 have been reported from the Greenland Sea (Hirche, 1991; Møller et al., 2006) and Disco Bay (Madsen et al., 2001). Misidentification of Calanus species from the use of fixed size limits (see “Materials and methods” section) may have influenced the results. The most frequent cases of misidentifications are small individuals of C. glacialis wrongly identified as C. finmarchicus (Gabrielsen et al. 2012). A hybrid species is expected to have intermediate prosome lengths (Parent et al., 2012). Species distributions were in our study highly related to water masses; and in Atlantic water where C. finmarchicus dominated, the overall contribution by C. glacialis was low. Co-occurrence between C. finmarchicus and C. glacialis was more prominent in the mixed water masses, and here the potential for misidentification (and possible hybridization) may have been greater. One could expect that increasing water temperatures in the Barents Sea would lead to decreasing size of C. finmarchicus copepodites (Campbell et al., 2001). Albeit a small sample size, the length measurements we performed as part of this study did in fact indicate that C. finmarchicus have become smaller between 1997 and 2010 (Supplementary Table S5). Smaller C. finmarchicus reduces the probability of overlapping in size with its congener C. glacialis. It is also reasonable to expect that warmer conditions would favour the dominance of C. finmarchicus (Kjellerup et al., 2012). We therefore believe that the general trends we observe in this study would be consistent despite the possibilities of species misidentification (due to size overlap and possible hybridization) in our data. Variation in weights of Calanus copepodites Variation in size (weight) can be a considerable source of error and uncertainty in Calanus biomass estimates from species counts. Our Calanus biomass estimates surpassed the observed total biomass in one out of five samples. Responding to the overestimations, we repeated species counts on a selection of samples (formalin preserved) from years with large discrepancies between estimated dw of C. finmarchicus and observed mesozooplankton biomass. The new measurements did, however, not reveal any abundance estimation errors that could explain the biomass overestimations. We believe the overestimations reflect uncertainties in the weight-at-stage data employed when estimating species biomass, as well as variance introduced by subsampling when estimating species abundances (see e.g. Skjoldal et al., 2013). Most studies where Calanus spp. biomass is estimated have used mean weights of copepodite stages from the literature. It is difficult to quantify the uncertainty, but from the variation in mean weights of the older copepodite stages shown in Figure 2 it may be of order 20–30% for C. finmarchicus and C. glacialis, or even larger. In some studies (e.g. Hirche et al., 1991) the weights of individuals have been determined as part of the study, thereby reducing this uncertainty. Size measurements performed on representative material to reveal changes in mean weights over space and time would greatly improve the precision of biomass estimates from zooplankton species abundance data. This may, however, induce a considerable increase in the effort spent on sample analysis. Using some form of plankton-imaging-system (Benfield et al., 2007) may facilitate the approach to make it more practical in routine studies. Calanus spp. as drivers of the mesozooplankton biomass in the Barents Sea Calanus finmarchicus, C. glacialis, and C. hyperboreus are major players in the herbivore zooplankton community of the Barents Sea ecosystem. Our study has shown that Calanus species constitute a major part of the mesozooplankton biomass in all regions of the Barents Sea, and on average around 80% of the total. Large mesozooplankton biomass samples (>16 g dw m−2) were associated with correspondingly large estimated biomass of Calanus species, indicating that biomass “peaks” in the Barents Sea are mainly driven by Calanus spp. The combined biomass of these species explained a major part of the variation in the observed mesozooplankton biomass. Though the total biomass of Calanus spp. contributed in similar proportion to the mesozooplankton biomass across the regions, the highest contribution was observed in regions West and South where there is a high abundance of C. finmarchicus. The proportional contribution of C. glacialis to the zooplankton biomass in its core Arctic water area was lower than the contribution of C. finmarchicus in Atlantic water, and other species than Calanus seem to comprise a larger part of the mesozooplankton biomass here. The larger species C. hyperboreus had a rather low contribution to the mesozooplankton biomass (< 10% in all regions), similar to earlier observations (Melle and Skjoldal, 1998; Arashkevich et al., 2002; Hirche and Kosobokova, 2003). Calanus hyperboreus generally overwinters below 500–1000 m in its core areas (Hirche, 1997), and has probably difficulties in completing a generation cycle in the (relatively shallow) Barents Sea due to its large size and longer life-span making it more vulnerable to predation (e.g. Falk-Petersen et al., 2009; Berge et al., 2012). Our data showed that the contribution of Calanus to the mesozooplankton biomass is lower when the total zooplankton biomass is low (see regression in Figure 3). Considering that a major part of our data was from Atlantic water areas, we believe part of this result can be explained by a seasonal/advective effect of C. finmarchicus. During winter when the mesozooplankton biomass is low, there will be lower concentrations of C. finmarchicus in the inflowing Atlantic water when it has descended (over-winter in deep Norwegian Sea basins) from the surface layers of the advective Atlantic current (Skjoldal et al., 1992). Hence, there will be a lower contribution of Calanus spp. to the total in winter vs. summer periods. A biological explanation is selective foraging by predators. The little auk Alle alle actively selects larger stages of C. glacialis when feeding in the Arctic, and avoids the smaller C. finmarchicus (Karnovsky et al., 2003). Baltic herring has shown size-selective preferences when feeding on copepods (Sandström, 1980), and planktivore fish in the Barents Sea can exert a significant top-down control on their zooplankton prey (Hassel et al., 1991; Stige et al., 2014).). Calanus spp. biomass and hydrography Both this and previous studies (Tande, 1991; Melle and Skjoldal, 1998; Hirche and Kosobokova, 2003) have demonstrated that the contribution of C. finmarchicus and C. glacialis to the zooplankton biomass in the Barents Sea is highly related to which water mass dominates. Weydmann et al. (2014) described temperature and bottom depth as the main drivers for zooplankton variability in the West Spitsbergen Current. Daase et al. (2007) demonstrated similar temperature-relationships as our study for the Calanus species in waters off Svalbard, and related the findings to advective processes. The steep, negative biomass-temperature relationship of C. glacialis in our study reflected large difference in biomass of C. glacialis in Arctic vs. Atlantic water masses. The area of Arctic water in the Barents Sea has been declining over the last few decades (ICES, 2017). This could possibly be associated with a reduction in the habitat (extent and conditions) of C. glacialis in the northern Barents Sea. It has been suggested that C. glacialis will decrease in Arctic areas of the Barents Sea if continuous warming leads to a greater mismatch between phytoplankton production and C. glacialis development due to earlier break-up of the winter ice (Hirche and Kosobokova, 2007; Søreide et al., 2010). The decrease of this species at the southwestern entrance (region West) could reflect a general decline in the core area further north. However, our data from the northern Barents Sea are limited (n = 23; Tables 2) and too heterogenous in time to allow us to examine if this has been the case. This is an important issue from an ecosystem perspective which we plan to address in a future study, using archived samples dating back to the 1980s. Calanus finmarchicus is an expatriate in Arctic water masses, and its reproductive cycle is limited by the low temperature environment (Melle and Skjoldal, 1998; Hirche and Kosobokova, 2007; Ji et al., 2012). Previous studies have also established a positive relationship between C. finmarchicus biomass and temperature (Dalpadado et al., 2003; Daase et al. 2007; Dvoretsky, 2011). High temperatures may indicate higher inflow of Atlantic water and thus larger concentrations of advective organisms like C. finmarchicus (Dalpadado et al., 2003). Furthermore, C. finmarchicus has higher growth rates (Campbell et al., 2001) and augmented egg production (Kjellerup et al., 2012) at increasing temperatures. The optimum temperature for this species appears to be about 6–10 °C based on abundance data over its geographical range (Helaouët and Beaugrand, 2007; Helaouët et al., 2011; Reygondeau and Beaugrand, 2011; Melle et al., 2014). The temperature of the inflowing Atlantic water at the FB transect has been increasing by about 1.5 °C since around 1980 to an annual mean level of about 6–6.5 °C after 2004 (Eriksen et al., 2017b). This may have improved the conditions and expanded the optimal habitat for C. finmarchicus in the southern Barents Sea. The number of generations produced per year by boreal Calanus decreases with increasing latitude (Conover, 1988). Though previous studies have suggested that C. finmarchicus produces one generation per year in the Barents Sea (e.g. Tande et al., 1985; Melle and Skjoldal, 1998), there are indications for a second generation of C. finmarchicus, particularly related to warm periods (Timofeev, 2000; Skaret et al., 2014). A second generation of C. finmarchicus may have contributed to the marked increase in biomass of C. finmarchicus in region West during the most recent period analysed here (after 2005). Coupled with the decrease in Arctic water masses in the Barents Sea is an increase of mixed water with intermediate temperatures of 0–3 °C (Eriksen et al. 2017b). Related to the issue of whether C. glacialis has declined as a response to the ongoing warming is therefore also a question of how the Calanus species are coping with the conditions in the mixed water masses. Temperature-driven stage-duration coupled with food availability and the length of the growth season in these waters, will largely determine the ability of C. finmarchicus to reach diapausing stage over the season (e.g. Ji et al., 2012). Calanus glacialis should persist physiologically at these cool temperatures, as suggested by its dominance in the White Sea (Kosobokova, 1999), though it is an open question as to how changes in ice conditions and water masses will affect the species in the mixed waters. Model predictions by Slagstad et al. (2011) have suggested that the secondary production by C. glacialis and C. finmarchicus combined will decrease in a future warmer climate in the northern Barents Sea, due to a temperature regime that is too warm for C. glacialis and sub-optimal for C. finmarchicus. Concluding remarks Plankton are good indicators of climate change occurring in the oceans (Hays et al., 2005). We have shown that the recent warming in the Barents Sea is likely affecting the composition of the mesozooplankton community, increasing the abundance of Atlantic C. finmarchicus in the west. With increased inflow of Atlantic water into the system, we would not expect these changes to be restricted only to the western area, as both fish species and macrozooplankton have shown responses to the warming in extended areas of the Barents Sea (Fossheim et al., 2015; Eriksen et al., 2017b, Frainer et al. 2017). A transition in the mesozooplankton community in certain areas from dominance of C. glacialis towards the smaller C. finmarchicus could be detrimental for higher trophic levels, particularly the size-selective particulate feeders (e.g. Karnovsky et al., 2003; Hirche and Kosobokova, 2007). Consistent time-series like ours from the FB transect and from the joint Norwegian-Russian ecosystem survey in autumn (Eriksen et al. 2017a) are crucial for revealing ongoing changes in zooplankton communities. Progress of the Calanus species in a future, warmer Barents Sea, particularly changes towards dominance of smaller sized individuals over a larger geographical area, deserves high priority in future research considering the key role of these species in the ecosystem. Acknowledgements We acknowledge technicians, scientists and crew of the various research cruises at the Institute of Marine Research (IMR), Norway, who have participated over the years in collecting and processing mesozooplankton data utilized in our study. We thank two anonymous referees for valuable comments on an earlier version of the article. This study was funded by the Norwegian Research Council project TIBIA (Trophic interactions in the Barents Sea—steps towards an Integrated Ecosystem Assessment, No. 228880) at IMR. References Akaike H. 1974 . A new look at the statistical model identification . IEEE Transactions on Automatic Control , 19 : 716 – 723 . Google Scholar Crossref Search ADS WorldCat Arashkevich E. , Wassmann P., Pasternak A., Riser C. W. 2002 . Seasonal and spatial changes in biomass, structure, and development progress of the zooplankton community in the Barents Sea . Journal of Marine Systems , 38 : 125 – 145 . Google Scholar Crossref Search ADS WorldCat Båmstedt U. , Ervik A. 1984 . Local variations in size and activity among Calanus finmarchicus and Metridia longa (Copepoda, Calanoida) overwintering on the west coast of Norway . Journal of Plankton Research , 6 : 843 – 857 . Google Scholar Crossref Search ADS WorldCat Båmstedt U. , Tande K. S. 1985 . Respiration and excretion rates of Calanus glacialis in Arctic waters of the Barents Sea . Marine Biology , 87 : 259 – 266 . Google Scholar Crossref Search ADS WorldCat Beaugrand G. , Reid P. C., Ibañez F., Lindley J. A., Edwards M. 2002 . Reorganization of North Atlantic marine copepod biodiversity and climate . Science , 296 : 1692 – 1694 . Google Scholar Crossref Search ADS PubMed WorldCat Beaugrand G. , Brander K. M., Alistair Lindley J., Souissi S., Reid P. C. 2003 . Plankton effect on cod recruitment in the North Sea . Nature , 426 : 661 – 664 . Google Scholar Crossref Search ADS PubMed WorldCat Benfield M. , Grosjean P., Culverhouse P., Irigolen X., Sieracki M., Lopez-Urrutia A. et al. 2007 . RAPID: research on automated plankton identification . Oceanography , 20 : 172 – 187 . Google Scholar Crossref Search ADS WorldCat Berge J. , Gabrielsen T. M., Moline M., Renaud P. E. 2012 . Evolution of the Arctic Calanus complex: an Arctic marine avocado? Journal of Plankton Research , 34 : 191 – 195 . Google Scholar Crossref Search ADS PubMed WorldCat Boitsov V. D. , Karsakov A. L., Trofimov A. G. 2012 . Atlantic water temperature and climate in the Barents Sea, 2000–2009 . ICES Journal of Marine Science , 69 : 833 – 840 . Google Scholar Crossref Search ADS WorldCat Campbell R. G. , Wagner M. M., Teegarden G. J., Boudreau C. A., Durbin E. G. 2001 . Growth and development rates of the copepod Calanus finmarchicus reared in the laboratory . Marine Ecology Progress Series , 221 : 161 – 183 . Google Scholar Crossref Search ADS WorldCat Carlotti F. , Krause M., Radach G. 1993 . Growth and development of Calanus finmarchicus related to the influence of temperature: experimental results and conceptual model . Limnology and Oceanography , 38 : 1125 – 1134 . Google Scholar Crossref Search ADS WorldCat Conover R. J. 1988 . Comparative life histories in the genera Calanus and Neocalanus in high latitudes of the Northern Hemisphere . Hydrobiologia , 167–168 : 127 – 142 . Google Scholar Crossref Search ADS WorldCat Daase M. , Vik J. O., Bagøien E., Stenseth N. C., Eiane K. 2007 . The influence of advection on Calanus near Svalbard: statistical relations between salinity, temperature and copepod abundance . Journal of Plankton Research , 29 : 903 – 911 . Google Scholar Crossref Search ADS WorldCat Daase M. , Falk-Petersen S., Varpe Ø., Darnis G., Søreide J. E., Wold A., Leu E. et al. 2013 . Timing of reproductive events in the marine copepod Calanus glacialis: a pan-Arctic perspective . Canadian Journal of Fisheries and Aquatic Sciences , 70 : 871 – 884 . Google Scholar Crossref Search ADS WorldCat Dalpadado P. , Ingvaldsen R., Hassel A. 2003 . Zooplankton biomass variation in relation to climatic conditions in the Barents Sea . Polar Biology , 26 : 233 – 241 . Google Scholar OpenURL Placeholder Text WorldCat Dalpadado P. , Mowbray F. 2013 . Comparative analysis of feeding ecology of capelin from two shelf ecosystems, off Newfoundland and in the Barents Sea . Progress in Oceanography , 114 : 97 – 105 . Google Scholar Crossref Search ADS WorldCat Dalpadado P. , Arrigo K. R., Hjøllo S. S., Rey F., Ingvaldsen R. B., Sperfeld E., van Dijken G. L. et al. 2014 . Productivity in the Barents Sea - response to recent climate variability . PLoS One , 9 : e95273. Google Scholar Crossref Search ADS PubMed WorldCat Diel S. 1991 . On the life history of dominant copepod species (Calanus finmarchicus, C. glacialis and C. hyperboreus, Metridia longa) in the Fram Strait . Reports on Polar Research , 88 : 1 – 113 . Google Scholar OpenURL Placeholder Text WorldCat Dvoretsky V. G. 2011 . Distribution of Calanus species off Franz Josef Land (Arctic Barents Sea) . Polar Science , 5 : 361 – 373 . Google Scholar Crossref Search ADS WorldCat Eriksen E. , Skjoldal H. R., Dolgov A. V., Dalpadado P., Orlova E. L., Prozorkevich D. V. 2016 . The Barents Sea euphausiids: methodological aspects of monitoring and estimation of abundance and biomass . ICES Journal of Marine Science , 73 : 1533 – 1544 . Google Scholar Crossref Search ADS WorldCat Eriksen E. , Gjøsæter H., Prozorkevich D., Shamray D., Dolgov A., Skern-Mauritzen M., Stiansen J. E. et al. 2017a . From single species surveys towards monitoring of the Barents Sea ecosystem. Progress in Oceanography (in press). http://www.sciencedirect.com.pva.uib.no/science/article/pii/S0079661117300642 Eriksen E. , Skjoldal H. R., Gjøsæter H., Primicerio R. 2017b . Spatial and temporal changes in the Barents Sea pelagic compartment during the recent warming . Progress in Oceanography , 151 : 206 – 226 . Google Scholar Crossref Search ADS WorldCat Falk-Petersen S. , Mayzaud P., Kattner G., Sargent J. 2009 . Lipids and life strategy of Arctic Calanus . Marine Biology Research , 5 : 18 – 39 . Google Scholar Crossref Search ADS WorldCat Fossheim M. , Primicerio R., Johannesen E., Ingvaldsen R. B., Aschan M. M., Dolgov A. V. 2015 . Recent warming leads to a rapid borealization of fish communities in the Arctic . Nature Climate Change , 5 : 673 – 676 . Google Scholar Crossref Search ADS WorldCat Frainer A. , Primicerio R., Kortsch S., Aune M., Dolgov A. V., Fossheim M., Aschan M. M. 2017 . Climate-driven changes in functional biogeography of Arctic marine fish communities . Proceedings of the National Academy of Sciences of the United States of America , 114 : 12202 – 12207 . Google Scholar Crossref Search ADS PubMed WorldCat Gabrielsen T. M. , Merkel B., Søreide J. E., Johansson-Karlsson E., Bailey A., Vogedes D., Nygård H. et al. 2012 . Potential misidentifications of two climate indicator species of the marine arctic ecosystem: Calanus glacialis and C. finmarchicus . Polar Biology , 35 : 1621 – 1628 . Google Scholar Crossref Search ADS WorldCat Gislason A. 2005 . Seasonal and spatial variability in egg production and biomass of Calanus finmarchicus around Iceland . Marine Ecology Progress Series , 286 : 177 – 192 . Google Scholar Crossref Search ADS WorldCat Hanssen H. 1997 . Mesozooplankton of the Laptev Sea and the adjacent eastern Nansen Basin—distribution and community structure in late summer . Berichte zur Polarforschung , 229 : 1 – 131 . Google Scholar OpenURL Placeholder Text WorldCat Hassel A. 1986 . Seasonal changes in zooplankton composition in the Barents Sea, with special attention to Calanus spp. (Copepoda) . Journal of Plankton Research , 8 : 329 – 339 . Google Scholar Crossref Search ADS WorldCat Hassel A. , Skjoldal H. R., Gjøsæter H., Loeng H., Omli L. 1991 . Impact of grazing from capelin (Mallotus villosus) on zooplankton: a case study in the northern Barents Sea in August 1985 . Polar Research , 10 : 371 – 388 . Google Scholar Crossref Search ADS WorldCat Hays G. C. , Richardson A. J., Robinson C. 2005 . Climate change and marine plankton . Trends in Ecology and Evolution , 20 : 337 – 344 . Google Scholar Crossref Search ADS PubMed WorldCat Heath M. R. , Jónasdóttir S. H. 1999 . Distribution and abundance of overwintering Calanus finmarchicus in the Faroe–Shetland Channel . Fisheries Oceanography , 8 : 40 – 60 . Google Scholar Crossref Search ADS WorldCat Helaouët P. , Beaugrand G. 2007 . Macroecology of Calanus finmarchicus and C. helgolandicus in the North Atlantic Ocean and adjacent seas . Marine Ecology Progress Series , 345 : 147 – 165 . Google Scholar Crossref Search ADS WorldCat Helaouët P. , Beaugrand G., Reid P. C. 2011 . Macrophysiology of Calanus finmarchicus in the North Atlantic Ocean . Progress in Oceanography , 91 : 217 – 228 . Google Scholar Crossref Search ADS WorldCat Helsel D. R. , Hirsch R. M. 1992 . Statistical Methods in Water Resources. Studies in Environmental Science . Elsevier , Amsterdam . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Hirche H. 1987 . Temperature and plankton . Marine Biology , 94 : 347 – 356 . Google Scholar Crossref Search ADS WorldCat Hirche H. J. 1991 . Distribution of dominant calanoid copepod species in the Greenland Sea during late fall . Polar Biology , 11 : 351 – 362 . Google Scholar Crossref Search ADS WorldCat Hirche H. J. 1997 . Life cycle of the copepod Calanus hyperboreus in the Greenland Sea . Marine Biology , 128 : 607 – 618 . Google Scholar Crossref Search ADS WorldCat Hirche H. J , Kattner G. 1993 . Egg production and lipid content of Calanus glacialis in spring: indication of a food-dependent and food-independent reproductive mode . Marine Biology , 117 : 615 – 622 . Google Scholar Crossref Search ADS WorldCat Hirche H. J , Kosobokova K. 2003 . Early reproduction and development of dominant calanoid copepods in the sea ice zone of the Barents Sea—need for a change of paradigms? Marine Biology , 143 : 769 – 781 . Google Scholar Crossref Search ADS WorldCat Hirche H. J. , Kosobokova K. 2007 . Distribution of Calanus finmarchicus in the northern North Atlantic and Arctic Ocean—Expatriation and potential colonization . Deep Sea Research Part II: Topical Studies in Oceanography , 54 : 2729 – 2747 . Google Scholar Crossref Search ADS WorldCat Hirche H. H , Kwasniewski S. 1997 . Distribution, reproduction and development of Calanus species in the Northeast Water in relation to environmental conditions . Journal of Marine Systems , 10 : 299 – 317 . Google Scholar Crossref Search ADS WorldCat Hirche H. J. , Mumm N. 1992 . Distribution of dominant copepods in the Nansen Basin, Arctic Ocean, in summer . Deep Sea Research Part A. Oceanographic Research Papers , 39 : 485 – 505 . Google Scholar Crossref Search ADS WorldCat Hirche H. , Baumann M. E. M., Kattner G., Gradinger R. 1991 . Plankton distribution and the impact of copepod grazing on primary production in Fram Strait, Greenland Sea . Journal of Marine Systems , 2 : 477 – 494 . Google Scholar Crossref Search ADS WorldCat Hirche H. , Brey T., Niehoff B. 2001 . A high-frequency time series at Ocean Weather Ship Station M (Norwegian Sea): population dynamics of Calanus finmarchicus . Marine Ecology Progress Series , 219 : 205 – 219 . Google Scholar Crossref Search ADS WorldCat Hirche H. , Hagen W., Mumm N., Richter C. 1994 . The Northeast Water polynya, Greenland Sea III. Meso- and macrozooplankton distribution and production of dominant herbivorous copepods during spring . Polar Biology , 14 : 491 – 503 . Google Scholar Crossref Search ADS WorldCat Hjøllo S. S. , Huse G., Skogen M. D., Melle W. 2012 . Modelling secondary production in the Norwegian Sea with a fully coupled physical/primary production/individual-based Calanus finmarchicus model system . Marine Biology Research , 8 : 508 – 526 . Google Scholar Crossref Search ADS WorldCat ICES . 2017 . Report of the Working Group on the Integrated Assessments of the Barents Sea (WGIBAR), 16-18 March 2017. ICES CM 2017/SSGIEA: 04. Ikeda T. , Skjoldal H. R. 1989 . Metabolism and elemental composition of zooplankton from the Barents Sea during early Arctic summer . Marine Biology , 100 : 173 – 183 . Google Scholar Crossref Search ADS WorldCat Jaschnov W. A. 1970 . Distribution of Calanus species in the seas of the Northern Hemisphere . Internationale Revue der Gesamten Hydrobiologie und Hydrographie , 55 : 197 – 212 . Google Scholar Crossref Search ADS WorldCat Ji R. , Ashjian C. J., Campbell R. G., Chen C., Gao G., Davis C. S., Cowles G. W., Beardsley R. C. 2012 . Life history and biogeography of Calanus copepods in the Arctic Ocean: an individual-based modeling study . Progress in Oceanography , 96 : 40 – 56 . Google Scholar Crossref Search ADS WorldCat Jónasdóttir S. H. 1999 . Lipid content of Calanus finmarchicus during overwintering in the Faroe–Shetland Channel . Fisheries Oceanography , 8 : 61 – 72 . Google Scholar Crossref Search ADS WorldCat Karnovsky N. , Kwasniewski S., Weslawski J., Walkusz W., Beszczynska-Möller A. 2003 . Foraging behavior of little auks in a heterogeneous environment . Marine Ecology Progress Series , 253 : 289 – 303 . Google Scholar Crossref Search ADS WorldCat Kjellerup S. , Dunweber M., Swalethorp R., Nielsen T. G., Møller E. F., Markager S., Hansen B. W. 2012 . Effects of a future warmer ocean on the coexisting copepods Calanus finmarchicus and C. glacialis in Disko Bay, western Greenland . Marine Ecology Progress Series , 447 : 87 – 108 . Google Scholar Crossref Search ADS WorldCat Kosobokova K. N. 1999 . The reproductive cycle and life history of the Arctic copepod Calanus glacialis in the White Sea . Polar Biology , 22 : 254 – 263 . Google Scholar Crossref Search ADS WorldCat Knutsen T. , Dalpadado P. 2009 . Økosystem barentshavet. In Havets ressurser og miljø 2009, pp 185 . Ed. by Gjøsæter H.. Institute of Marine Research , Bergen .http://hdl.handle.net/11250/116992. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Kvile K. O. , Fiksen O., Prokopchuk I., Opdal A. F. 2017 . Coupling survey data with drift model results suggests that local spawning is important for Calanus finmarchicus production in the Barents Sea . Journal of Marine Systems , 165 : 69 – 76 . Google Scholar Crossref Search ADS WorldCat Lee R. F. 1975 . Lipids of Arctic zooplankton . Comparative Biochemistry and Physiology Part B: Biochemistry and Molecular Biology , 51 : 263 – 266 . Google Scholar Crossref Search ADS WorldCat Lind S. , Ingvaldsen R. B. 2012 . Variability and impacts of Atlantic Water entering the Barents Sea from the north . Deep-Sea Research Part I: Oceanographic Research Papers , 62 : 70 – 88 . Google Scholar Crossref Search ADS WorldCat Lind S. , Ingvaldsen R. B., Furevik T. 2016 . Arctic layer salinity controls heat loss from deep Atlantic layer in seasonally ice-covered areas of the Barents Sea . Geophysical Research Letters , 43 : 5233 – 5242 . Google Scholar Crossref Search ADS WorldCat Madsen S. D. , Nielsen T. G., Hansen B. W. 2001 . Annual population development and production by Calanus finmarchicus, C. glacialis and C. hyperboreus in Disko Bay, western Greenland . Marine Biology , 139 : 75 – 93 . Google Scholar Crossref Search ADS WorldCat Melle W. , Skjoldal H. R. 1998 . Reproduction and development of Calanus finmarchicus, C. glacialis and C. hyperboreus in the Barents Sea . Marine Ecology Progress Series , 169 : 211 – 228 . Google Scholar Crossref Search ADS WorldCat Melle W. , Ellertsen B., Skjoldal H. R. 2004 . Zooplankton: the link to higher trophic levels. In The Norwegian Sea Ecosystem, pp. 137–202 . Ed. by Skjoldal H. R.. Tapir Academic Press , Trondheim . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Melle W. , Runge J., Head E., Plourde S., Castellani C., Licandro P., Pierson J. et al. 2014 . The North Atlantic Ocean as habitat for Calanus finmarchicus: Environmental factors and life history traits . Progress in Oceanography 129 : 244 – 284 . Google Scholar Crossref Search ADS WorldCat Møller E. F. , Nielsen T. G., Richardson K. 2006 . The zooplankton community in the Greenland Sea: Composition and role in carbon turnover . Deep Sea Research Part I: Oceanographic Research Papers , 53 : 76 – 93 . Google Scholar Crossref Search ADS WorldCat Orlova E. , Dalpadado P., Knutsen T., Nesterova V. N., Prokopchuk I. P. 2011 . Zooplankton. In The Barents Sea - Ecosystem, Resources, Management. Half a Century of Russian-Norwegian Cooperation, pp. 91–119 . Ed. by Jakobsen T., Ožigin V. K.. Tapir Academic Press , Trondheim . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Parent G. J. , Plourde S., Turgeon J. 2011 . Overlapping size ranges of Calanus spp. off the Canadian Arctic and Atlantic Coasts: impact on species’ abundances . Journal of Plankton Research , 33 : 1654 – 1665 . Google Scholar Crossref Search ADS WorldCat Parent G. J. , Plourde S., Turgeon J. 2012 . Natural hybridization between Calanus finmarchicus and C. glacialis (Copepoda) in the Arctic and Northwest Atlantic . Limnology and Oceanography , 57 : 1057 – 1066 . Google Scholar Crossref Search ADS WorldCat R Core Team . 2016 . R: A Language and Environment for Statistical Computing . R Foundation for Statistical Computing , Vienna, Austria . https://www.R-project.org/. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Reygondeau G. , Beaugrand G. 2011 . Future climate-driven shifts in distribution of Calanus finmarchicus . Global Change Biology , 17 : 756 – 766 . Google Scholar Crossref Search ADS WorldCat Runge J. A. , Plourde S., Joly P., Niehoff B., Durbin E. 2006 . Characteristics of egg production of the planktonic copepod, Calanus finmarchicus, on Georges Bank: 1994–1999 . Deep Sea Research Part II: Topical Studies in Oceanography , 53 : 2618 – 2631 . Google Scholar Crossref Search ADS WorldCat Sakshaug E. , Johnsen G., Kovacs K. 2009 . Ecosystem Barents Sea . Tapir Academic Press , Trondheim . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Sameoto D. , Wiebe P. H., Runge J., Postel L., Dunn J., Miller C., Coombs S. 2000 . Collecting zooplankton. In ICES Zooplankton Methodology Manual, pp. 55–82 . Ed. by Harris R., Wiebe P. H., Lenz J., Skjoldal H. R., Huntley M.. Academic Press , San Diego . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Scott C. L. , Kwasniewski S., Falk-Petersen S., Sargent J. R. 2000 . Lipids and life strategies of Calanus finmarchicus, Calanus glacialis and Calanus hyperboreus in late autumn, Kongsfjorden, Svalbard . Polar Biology , 23 : 510 – 516 . Google Scholar Crossref Search ADS WorldCat Skaret G. , Dalpadado P., Hjøllo S. S., Skogen M. D., Strand E. 2014 . Calanus finmarchicus abundance, production and population dynamics in the Barents Sea in a future climate . Progress in Oceanography , 125 : 26 – 39 . Google Scholar Crossref Search ADS WorldCat Skjoldal H. R. , Gjøsæter H., Loeng H. 1992 . The Barents Sea ecosystem in the 1980s: ocean climate, plankton, and capelin growth . ICES Marine Science Symposium , 195 : 278 – 290 . Google Scholar OpenURL Placeholder Text WorldCat Skjoldal H. R. , Wiebe P. H., Postel L., Knutsen T., Kaartvedt S., Sameoto D. D. 2013 . Intercomparison of zooplankton (net) sampling systems: Results from the ICES/GLOBEC sea-going workshop . Progress in Oceanography , 108 : 1 – 42 . Google Scholar Crossref Search ADS WorldCat Slagstad D. , Ellingsen I. H., Wassmann P. 2011 . Evaluating primary and secondary production in an Arctic Ocean void of summer sea ice: An experimental simulation approach . Progress in Oceanography , 90 : 117 – 131 . Google Scholar Crossref Search ADS WorldCat Smedsrud L. H. , Esau I., Ingvaldsen R. B., Eldevik T., Haugan P. M., Li C., Lien V. S., Olsen A., Omar A. M., Otterå O. H., et al. 2013 . The role of the Barents Sea in the Arctic climate system . Reviews of Geophysics , 51 : 415 – 449 . Google Scholar Crossref Search ADS WorldCat Sokal R. R. , Rohlf F. J. 2012 . Biometry - The Principles and Practice of Statistics in Biological Research . W. H. Freeman and Company , New York . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Søreide J. E. , Falk-Petersen S., Hegseth E. N., Hop H., Carroll M. L., Hobson K. A., Blachowiak-Samolyk K. 2008 . Seasonal feeding strategies of Calanus in the high-Arctic Svalbard region . Deep-Sea Research Part II: Topical Studies in Oceanography , 55 : 2225 – 2244 . Google Scholar Crossref Search ADS WorldCat Søreide J. E. , Leu E., Berge J., Graeve M., Falk-Petersen S. 2010 . Timing of blooms, algal food quality and Calanus glacialis reproduction and growth in a changing Arctic . Global Change Biology , 16 : 3154 – 3163 . Google Scholar OpenURL Placeholder Text WorldCat Stige L. C. , Dalpadado P., Orlova E., Boulay A. C., Durant J. M., Ottersen G., Stenseth N. C. 2014 . Spatiotemporal statistical analyses reveal predator-driven zooplankton fluctuations in the Barents Sea . Progress in Oceanography , 120 : 243 – 253 . Google Scholar Crossref Search ADS WorldCat Tande K. S. 1982 . Ecological investigations on the zooplankton community of Balsfjorden, northern Norway: Generation cycles, and variations in body weight and body content of carbon and nitrogen related to overwintering and reproduction in the copepod Calanus finmarchicus (Gunnerus) . Journal of Experimental Marine Biology and Ecology , 62 : 129 – 142 . Google Scholar Crossref Search ADS WorldCat Tande K. S. 1991 . Calanus in the North Norwegian fjords and in the Barents Sea . Polar Research , 10 : 389 – 407 . Google Scholar Crossref Search ADS WorldCat Tande K. , Hassel A., Slagstad D. 1985 . Gonad maturation and possible life cycle strategies in Calanus finmarchicus and Calanus glacialis in the northwestern part of the Barents Sea. In Marine Biology of Polar Regions and Effects of Stress on Marine Organisms, pp. 141–155 . Ed. by Gray J. S., Christiansen M. E.. Wiley and Sons , New York . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Tande K. S. , Slagstad D. 1992 . Regional and interannual variations in biomass and productivity of the marine copepod, Calanus finmarchicus, in sub-arctic environments . Oceanologica Acta , 15 : 309 – 321 . Google Scholar OpenURL Placeholder Text WorldCat Timofeev S. 2000 . Ecology of the Marine Zooplankton . Murmansk State Pedagogical Institute Press , Murmansk . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Torgersen T. , Huse G. 2005 . Variability in retention of Calanus finmarchicus in the Nordic Seas . ICES Journal of Marine Science , 62 : 1301 – 1309 . Google Scholar Crossref Search ADS WorldCat Tourangeau S. , Runge J. A. 1991 . Reproduction of Calanus glacialis under ice in spring in southeastern Hudson Bay, Canada . Marine Biology , 108 : 227 – 233 . Google Scholar Crossref Search ADS WorldCat Unstad K. H. , Tande K. S. 1991 . Depth distribution of Calanus finmarchicus and C. glacialis in relation to environmental conditions in the Barents Sea . Polar Research , 10 : 409 – 420 . Google Scholar Crossref Search ADS WorldCat Wassmann P. , Reigstad M., Haug T., Rudels B., Carroll M. L., Hop H., Gabrielsen G. W. et al. 2006 . Food webs and carbon flux in the Barents Sea . Progress in Oceanography , 71 : 232 – 287 . Google Scholar Crossref Search ADS WorldCat Weydmann A. , Carstensen J., Goszczko I., Dmoch K., Olszewska A., Kwasniewski S. 2014 . Shift towards the dominance of boreal species in the Arctic: inter-annual and spatial zooplankton variability in the West Spitsbergen Current . Marine Ecology Progress Series , 501 : 41 – 52 . Google Scholar Crossref Search ADS WorldCat Wiebe P. H. , Benfield M. C. 2003 . From the Hensen net toward four-dimensional biological oceanography . Progress in Oceanography , 56 : 7 – 136 . Google Scholar Crossref Search ADS WorldCat Wood S. N. 2017 . mgcv: Mixed GAM Computation Vehicle with GCV/AIC/REML Smoothness Estimation. https://cran.r-project.org/web/packages/mgcv/index.html. © International Council for the Exploration of the Sea 2017. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. © International Council for the Exploration of the Sea 2017.

Skogen, Morten, D;Hjøllo, Solfrid, S;Sandø, Anne, Britt;Tjiputra,, Jerry

doi: 10.1093/icesjms/fsy088pmid: N/A

Abstract The biogeochemistry from a global climate model (Norwegian Earth System Model) has been compared with results from a regional model (NORWECOM.E2E), where the regional model is forced by downscaled physics from the global model. The study should both be regarded as a direct comparison between a regional and its driving global model to investigate at what extent a global climate model can be used for regional studies, and a study of the future climate change in the Nordic and Barents Seas. The study concludes that the global and regional model compare well on trends, but many details are lost when a coarse resolution global model is used to assess climate impact on regional scale. The main difference between the two models is the timing of the spring bloom, and a non-exhaustive nutrient consumption in the global model in summer. The global model has a cold (in summer) and saline bias compared with climatology. This is both due to poorly resolved physical processes and oversimplified ecosystem parameterization. Through the downscaling the regional model is to some extent able to alleviate the bias in the physical fields, and the timing of the spring bloom is close to observations. The summer nutrient minimum is one month early. There is no trend in future primary production in any of the models, and the trends in modelled pH and ΩAr are also the same in both models. The largest discrepancy in the future projection is in the development of the CO2 uptake, where the regional suggests a slightly reduced uptake in the future. Introduction Human influence on the climate system is clear, and recent climate changes have had widespread impacts on human and natural systems. Without action, the world’s mean surface temperature is projected to rise over the 21st century (IPCC, 2014). Global coupled climate models (GCM) are generally capable of reproducing the observed trends in, e.g. the globally averaged atmospheric temperature. However, the global models do not have the horizontal resolution, which is needed in order to properly resolve the relevant features on regional scales (Melsom et al., 2009). As the spatial scales represented by the GCM may not be as fine as the end-use application requires, the GCM outputs will contain biases relative to the observational data, which preclude its direct use. Therefore, dynamical downscaling using so-called Regional Climate Models (RCM) is necessary to translate coarse global scale information into fine regional and local grids in order to obtain climate information on scales that are relevant to society. Over the past years, the emerging impacts of climate change on the Northeast Atlantic ecosystem have caused serious concerns (e.g. Fossheim et al., 2015; Eriksen et al., 2015). In the Polar regions, and in particular in the Arctic, there is significant spatial variability among GCMs (Overland and Wang, 2007; Steinacher et al., 2009). The Arctic region will continue to warm more rapidly than the global mean, and this, together with the systems limited adaptive capacity, makes the Arctic more vulnerable to climate change (IPCC, 2014). Therefore, any attempt to evaluate future ecosystem consequences of CO2 emissions will need predictions of ocean biogeochemistry changes at regional scale, along with detailed knowledge of biological responses. Some previous studies have focused on regional biogeochemistry in the Atlantic section of the Arctic Ocean. Using downscaled regional ocean models Bellerby et al. (2012) modelled both present and future climate ocean acidification in the Arctic with a focus on the Spitsbergen region, and Skogen et al. (2014) did a similar study of changes in ocean acidification for the whole Barents and Nordic seas. Skaret et al. (2014) and Slagstad et al. (2015) on the other hand, focused on changes in primary and secondary production in their studies with focus on the Barents Sea and the Arctic Ocean, respectively. For proper interpretation of climate change projections from RCMs, it is important to first assess the signal of climate change from GCMs to RCMs. This has been done for the atmosphere in a number of publications (e.g. Saini et al., 2015). Langehaug et al. (2018) used a high 0.25° and a medium 1.0° version of the global ocean-sea ice component of the Norwegian Earth System Model (NorESM1-ME) to assess the impact of increased ocean resolution, but to our knowledge there has not been any comparison of results from a GCM and a RCM. The goal of the present paper is to quantify whether an ocean RCM produce different projections than its driving GCM, based on climate change projections for the Nordic and Barents Seas. The performance and projected changes of the RCM are examined and compared with those of the driving GCM, thus the present study should be regarded as both a direct comparison between an RCM and its GCM and a study of the impact of climate change scenarios in the area. To achieve this a GCM (NorESM1-ME) has been downscaled and used to force a biogeochemical RCM (NORWECOM.E2E) for the Barents and Nordic seas under the RCP4.5 emission scenario. The focus has been on comparing model outputs on ocean carbon chemistry and primary production to investigate how the global and regional model compare for some key parameters. Such a comparison could provide new insights both on how well a GCM projection changes on a regional scale, and which future ecosystem questions can sufficiently be answered by a GCM with enough confidence. Material and methods NorESM1-ME The NorESM1-ME is a fully coupled climate carbon cycle model developed in Norway in collaboration with researchers from the National Center for Atmospheric Research at the United States. As such, some of its components are adopted from the Community Climate System Model (CCSM4), i.e. the atmospheric general circulation (Community Atmospheric Model, CAM4), land (Community Land Model, CLM4), and sea-ice (Community, CICE4) components (Gent et al., 2011). The ocean physical circulation is based on the Miami Isopycnic Coordinate Ocean Model (MICOM), coupled with the Hamburg Oceanic Carbon Cycle (HAMOCC5) model (Maier-Reimer et al., 2005; Tjiputra et al., 2013). In the upper ocean, the HAMOCC5 model consists of an NPZD-type ecosystem module, where the primary production is formulated as a function of phytoplankton growth and nutrient concentration within the top 100 m. In addition to multi-nutrients (i.e. nitrate, phosphate, and dissolved iron) co-limitation, the phytoplankton growth is limited by light availability and temperature. Simpler than the NORWECOM.E2E, HAMOCC5 model simulates a generic single class phytoplankton and zooplankton compartments. Below the mixed layer, particulate materials (particulate organic carbon, biogenic opal and calcium carbonate) sink at constant velocities and are remineralized at depth. Silicate (SI) concentration does not limit phytoplankton growth but is used to determine the portion of biogenic opal and calcium carbonate export production. Non-remineralized particles reaching the bottom layer, undergo chemical reactions with sediment porewaters, bioturbation and vertical advection within the sediment module. The formulations of the ecosystem dynamics are described in Six and Maier-Reimer (1996) and Maier-Reimer et al. (2005). In contrast to the NORWECOM.E2E, the ecosystem parameters in NorESM1-ME (HAMOCC5) are tuned toward the global marine ecosystem. The parameters in the HAMOCC5 model are tuned to optimize the large scale spatial distribution of surface primary production and biogeochemical tracers in the interior in order to reproduce the observed climatology features as well as the global ocean carbon sources and sinks. More details of the evaluation and performance of the ocean biogeochemistry in the NorESM1-ME model is available in Tjiputra et al. (2013). In this study, we apply the standard coupled RCP4.5 projection, following the standard CMIP5 (Coupled Model Intercomparison Project phase 5) protocol (Taylor et al., 2012). The RCP4.5 represents a future scenario where the global mean atmospheric radiative forcing approach 4.5 W m−2 by year 2100. Under this scenario, the atmospheric CO2 concentration pathway reaches 538 ppm at 2100. For the comparison with the regional model, we focus on analysing results from the 2006 to 2070 periods. ROMS downscaling To downscale the NorESM1-ME model (producing the physical forcing for the regional biogeochemical model), the Regional Ocean Modeling System (ROMS, Shchepetkin and McWilliams, 2005) is used. The ROMS model set-up is initialized from the NorESM1-ME model, and outputs from NorESM1-ME are also used at the open boundaries and as atmospheric forcing. A weak relaxation towards NorESM1-ME sea surface salinity with a time scale of 360 days was also applied. The model domain for the ROMS downscaling covers the North Atlantic, the Nordic and Barents Seas, and the Arctic Ocean from 30°N to the Bering Strait, with a horizontal model resolution of ∼10 × 10 km. In the vertical 35 generalized σ-coordinate (s) levels, stretched to increase vertical resolution near the surface and bottom, were used. The time step was 100 s. The ROMS model on this grid has previously been evaluated in a hindcast study in Melsom et al. (2009) and Sandø et al. (2014). Here, it was shown that downscaling reduced the biases in the Barents Sea projected by the global model, and that the downscaled results generally were closer to observations. For the present study the regional ROMS model was run for the period 2006–2070. More details on the set-up and performance of the present downscaling can be found in Sandø et al. (2018). NORWECOM.E2E The NORWegian ECOlogical Model system End-To-End (NORWECOM.E2E), a coupled physical, chemical, biological model system (Aksnes et al., 1995; Skogen et al., 1995; Skogen and Søiland, 1998), was developed to study primary production, nutrient budgets and dispersion of particles such as fish larvae and pollution. The model has been validated by comparison with field data in the Nordic and Barents seas (Skogen et al., 2007; Hjøllo et al., 2012; Skaret et al., 2014). Recently, it has been extended with a module to project ocean acidification (Skogen et al., 2014), and with Individual Based Models (IBMs) for Calanus finmarchicus (Hjøllo et al., 2012) and pelagic fish (Utne et al., 2012). In the present study, the model is run in offline mode. Physical ocean fields (velocities, salinity, temperature, water level and sea ice) from the ROMS downscaling (Section 2.2) has been interpolated from 5-daily means and used as physical forcing together with daily atmospheric (wind and short wave radiation) fields from the NorESM1-ME (Section 2.1) simulation. The horizontal grid used (Figure 1) is identical to a subdomain of the original ROMS grid, while in the vertical 20 sigma layers are used. The time step is 3600 s. Figure 1. View largeDownload slide Model domain for NORWECOM.E2E with bathymetry and boxes indicating the three seas used for statistics (Barents Sea, Greenland Sea, and Norwegian Sea). Shading denote water depth in meters. Figure 1. View largeDownload slide Model domain for NORWECOM.E2E with bathymetry and boxes indicating the three seas used for statistics (Barents Sea, Greenland Sea, and Norwegian Sea). Shading denote water depth in meters. The biochemical model is coupled to the physical model through the light, the hydrography and the horizontal and vertical movements of the water masses. The prognostic variables are dissolved inorganic nitrogen (DIN), phosphorous (PHO), and SI, two different types of phytoplankton (diatoms and flagellates), two detritus (dead organic matter) pools (N and P), diatom skeletal (biogenic silica) and oxygen. Two types of zooplankton (meso- and micro-zooplankton) are included based on a module taken from the ECOHAM4 model (Moll and Stegert, 2007; Stegert et al., 2009; Pätsch et al., 2009). The processes included are primary and secondary production, grazing by zooplankton on phytoplankton and detritus, respiration, algae death, remineralization of inorganic nutrients from dead organic matter, self shading, turbidity, sedimentation, resuspension, sedimental burial, and denitrification. The material produced by mortality is partly regenerated through the detritus pool, but a fraction of 10% is instantly regenerated as DIN (in nature as ammonia) and 25% as PHO available for uptake by phytoplankton (Garber, 1984; Bode et al., 2004). Ocean acidification is modelled using a submodule (Blackford and Gilbert, 2007; Skogen et al., 2014) for the carbonate system. The module is an implementation of the Haltafall speciation code (Ingri et al., 1967). The module calculates the carbonate system at any given point in space and time, using constants from Mehrbach et al. (1973) refitted by Dickson and Millero (1987). The inputs are temperature, salinity, dissolved inorganic carbon (DIC), total alkalinity (TA), and depth (pressure), whereas the outputs are pH, partial pressure of CO2 in seawater, carbonate and bicarbonate ion concentrations, and calcite and aragonite calcification states. In addition, the module calculates the air sea exchange of CO2 taking into account wind speed and atmospheric pCO2. The latter one uses the (Nightingale et al., 2000) parameterization for gas transfer velocity. TA is not a prognostic variable in the model. For oceanic regimes there is generally a well constrained relationship between salinity and TA as TA is conservative, and in the model an expression for the Nordic Seas and North Atlantic from Bellerby et al. (2005) (TA = 66.96 × S – 36.803) have been used for the calculation of TA. To allow the integration of the carbon system three state variables are added [detritus C pool, dissolved organic carbon (DOC) and DIC]. The carbon fluxes are following the nitrogen fluxes using the Redfield ratio [C:N = 5.68 (weight)], except for the remineralization rate of detritus C to DOC (10% day−1) taken from ECOHAM4 and the degradation rate of DOC (6% day−1 at 8 degrees, and Q10 = 2.6), which is taken from Lønborg et al. (2009). The new state variables have no impact on parameters in the original biogeochemical model. Future scenario of the atmospheric CO2 concentration based on the RCP4.5 of the fifth assessment report of the IPCC (2013) (Table AII.4.1), is used for the atmospheric boundary condition. Remineralization takes place both in the water column and in the sediments. Particulate matter has a sinking speed relative to the water and may accumulate on the bottom if the bottom stress is below a certain threshold value. Likewise, resuspension takes place if the bottom stress is above a limit. Parameterization of the biochemical processes is taken from literature based on experiments in laboratories and mesocosms, or deduced from field measurements (Pohlmann and Puls, 1994; Aksnes et al., 1995; Mayer, 1995; Gehlen et al., 1995; Lohse et al., 1995, 1996). A comparison between some key characteristics and parameters between NorESM1-ME and NORWECOM.E2E is given in Table 1. Table 1. Comparison of some model characteristics for NorESM1-ME and NORWECOM.E2E. NorESM1-ME NORWECOM.E2E Nutrients Nitrate, phosphate, silicate, iron Inorganic nitrogen, phosphate, silicate Phytoplankton One bulk Diatoms and flagellates Zooplankton One bulk Micro and meso P:N:C stoichiometry 1:16:122 1:16:112 POC remineralization rate (day–1) 0.03 0.0005, 0.0007, 0.0002 for N, P, and Si DOC remineralization rate (day–1) 0.004 0.028 (transient part) at 0°C Max phyto growth rate at 0° (day–1) 0.60 1.32 (dia), 0.88 (fla) Max zoo growth rate (day–1) 1.0 0.27 (meso), 0.33 (micro) at 0°C NorESM1-ME NORWECOM.E2E Nutrients Nitrate, phosphate, silicate, iron Inorganic nitrogen, phosphate, silicate Phytoplankton One bulk Diatoms and flagellates Zooplankton One bulk Micro and meso P:N:C stoichiometry 1:16:122 1:16:112 POC remineralization rate (day–1) 0.03 0.0005, 0.0007, 0.0002 for N, P, and Si DOC remineralization rate (day–1) 0.004 0.028 (transient part) at 0°C Max phyto growth rate at 0° (day–1) 0.60 1.32 (dia), 0.88 (fla) Max zoo growth rate (day–1) 1.0 0.27 (meso), 0.33 (micro) at 0°C Table 1. Comparison of some model characteristics for NorESM1-ME and NORWECOM.E2E. NorESM1-ME NORWECOM.E2E Nutrients Nitrate, phosphate, silicate, iron Inorganic nitrogen, phosphate, silicate Phytoplankton One bulk Diatoms and flagellates Zooplankton One bulk Micro and meso P:N:C stoichiometry 1:16:122 1:16:112 POC remineralization rate (day–1) 0.03 0.0005, 0.0007, 0.0002 for N, P, and Si DOC remineralization rate (day–1) 0.004 0.028 (transient part) at 0°C Max phyto growth rate at 0° (day–1) 0.60 1.32 (dia), 0.88 (fla) Max zoo growth rate (day–1) 1.0 0.27 (meso), 0.33 (micro) at 0°C NorESM1-ME NORWECOM.E2E Nutrients Nitrate, phosphate, silicate, iron Inorganic nitrogen, phosphate, silicate Phytoplankton One bulk Diatoms and flagellates Zooplankton One bulk Micro and meso P:N:C stoichiometry 1:16:122 1:16:112 POC remineralization rate (day–1) 0.03 0.0005, 0.0007, 0.0002 for N, P, and Si DOC remineralization rate (day–1) 0.004 0.028 (transient part) at 0°C Max phyto growth rate at 0° (day–1) 0.60 1.32 (dia), 0.88 (fla) Max zoo growth rate (day–1) 1.0 0.27 (meso), 0.33 (micro) at 0°C The incident irradiation used in the biochemical model is modelled using a formulation based on Skartveit and Olseth (1986, 1987) using short wave radiation outputs of the NorESM1-ME model, and corrected linearly at the sea surface using the modelled ice concentration. Initial fields for nutrients and DIC were interpolated from annual means of the NorESM1-ME simulation for the years 2001–2005, except for SI that has a large offset in the NorESM1-ME simulation with surface values close to 20 μM in the area of interest caused by advection of water with high SI from the Bering Sea. For SI typical winter values of Atlantic Water in the Norwegian Sea (SI = 5.5 μM, F. Rey, pers. comm.) were therefore used, together with some small initial amounts of algae (0.10 mg N m−3) for both diatoms and flagellates. For DOC only the transient part is considered, and the initial value is therefore set to zero. These values are also used at the open boundaries. Inorganic nitrogen is added to the system from the atmosphere, while there are no river inputs of nutrients and carbon. To absorb inconsistencies between the forced boundary conditions and the model results, a 7 gridcell “Flow Relaxation Scheme” (FRS) zone (Martinsen and Engedahl, 1987) is used around the open boundaries. The simulation started on 1 January 2006. After a 12 year spin-up (running the first year 12 times) the full model period (2006–2070) was run sequentially. Results Mean surface temperature and salinity for January 2006 are shown in Figure 2. As ocean physics is an input to the NORWECOM.E2E system, these results are from the ROMS downscaling (Sandø et al., 2018), and since the ROMS downscaling was initiated from the global model 1 January 2006, the figures are close to the initial field and similar in both models. The temperature field clearly shows how the warm water is transported with the Norwegian Coastal Current northwards into the Barents Sea and west of Spitsbergen. Surface salinity is well above 35 in most of the Barents and Norwegian seas, while water below 34 is only found in the Arctic Ocean, in the Greenland fjords and in the southeastern Barents Sea. Figure 2. View largeDownload slide Mean temperature (left) and salinity (right) for January 2006 from the ROMS model. Figure 2. View largeDownload slide Mean temperature (left) and salinity (right) for January 2006 from the ROMS model. Time series of annual mean sea surface temperature (SST) and salinity (SSS) for the Barents, Greenland and Norwegian seas (see Figure 1 for area definitions) for both the regional and global models are given in Figure 3. There is a general agreement on the magnitude between the two models in the Norwegian and Greenland seas SST, while the Barents Sea temperature is lower in the global (NorESM1-ME) than in the regional model. Using the Fitting Generalised Linear Models routine in R (glm), the trend in annual mean SST for the whole period has been computed. Except for the regional Greenland Sea, there is an increase in SST with the trend close to 0.02°C year−1, with a slightly stronger increase in the global model. There is a positive SST correlation between regional and global models in the Barents and Norwegian seas (r ≈ 0.5, p < 0.01), while SST is negative correlated in the Greenland Sea. The regional model is initialized from the global one, and there is an adjustment in SSS over the first decade of the simulation. In the Barents Sea the SSS increases rapidly before a similar decrease is seen. In the Greenland Sea SSS increases from 34.6 to 35.0, while in the Norwegian Sea there is similar decrease from 35.4 to 35.0. After this initial adjustment there is a steady decrease in the regional model (∼0.007 year−1) in all seas, while there are no trend in SSS in the global model. Figure 3. View largeDownload slide Annual mean sea surface temperature (SST, left) and salinity (SSS, right) for Barents Sea (black), the Greenland Sea (red), and the Norwegian Sea (green), for NorESM1-ME (solid line) and ROMS/NORWECOM.E2E (dashed line). Color refer to online version. Figure 3. View largeDownload slide Annual mean sea surface temperature (SST, left) and salinity (SSS, right) for Barents Sea (black), the Greenland Sea (red), and the Norwegian Sea (green), for NorESM1-ME (solid line) and ROMS/NORWECOM.E2E (dashed line). Color refer to online version. The annual cycle of SST is shown in Figure 4, and statistics where the annual cycles of the first 10 years are compared with that from the World Ocean Atlas (WOA) are given in Table 2. It is clear that the temperature increases from the first to the last 10 year periods. Comparing the global and regional models, the spring warming starts earlier in the regional model than in the global one. Summer maximum occurs in both models in August, and the summer maximum is higher in the regional model. The figure also show that the seasonal amplitude is larger in the regional model than in the global one, and that this amplitude is higher in the last decade than in the first. This increase in amplitude is up to 0.87°C for the regional model in the Greenland Sea. Table 2. Mean value, root mean square error (RMSE), and model bias of the seasonal cycle (N = 12) for the first 10 years (2006–2015) of sea surface temperature (Figure 4) and inorganic nitrate (Figure 6) for the two models compared with the observations from the World Ocean Atlas. Sea surface temperature Inorganic nitrate Mean RMSE Bias Mean RMSE Bias BSEA-WOA 2.9 3.4 BSEA-NORWECOM 2.8 0.9 0.2 7.0 4.3 −3.6 BSEA-NorESM 1.5 1.8 1.5 10.8 7.4 −7.4 GSEA-WOA 2.5 7.1 GSEA-NORWECOM 1.1 1.5 1.4 7.6 2.5 −0.4 GSEA-NorESM 1.1 1.8 1.3 12.3 5.7 −5.1 NSEA-WOA 7.1 7.1 NSEA-NORWECOM 5.6 1.5 1.5 6.1 2.2 1.0 NSEA-NorESM 5.8 1.3 1.2 11.4 4.8 −4.3 Sea surface temperature Inorganic nitrate Mean RMSE Bias Mean RMSE Bias BSEA-WOA 2.9 3.4 BSEA-NORWECOM 2.8 0.9 0.2 7.0 4.3 −3.6 BSEA-NorESM 1.5 1.8 1.5 10.8 7.4 −7.4 GSEA-WOA 2.5 7.1 GSEA-NORWECOM 1.1 1.5 1.4 7.6 2.5 −0.4 GSEA-NorESM 1.1 1.8 1.3 12.3 5.7 −5.1 NSEA-WOA 7.1 7.1 NSEA-NORWECOM 5.6 1.5 1.5 6.1 2.2 1.0 NSEA-NorESM 5.8 1.3 1.2 11.4 4.8 −4.3 Table 2. Mean value, root mean square error (RMSE), and model bias of the seasonal cycle (N = 12) for the first 10 years (2006–2015) of sea surface temperature (Figure 4) and inorganic nitrate (Figure 6) for the two models compared with the observations from the World Ocean Atlas. Sea surface temperature Inorganic nitrate Mean RMSE Bias Mean RMSE Bias BSEA-WOA 2.9 3.4 BSEA-NORWECOM 2.8 0.9 0.2 7.0 4.3 −3.6 BSEA-NorESM 1.5 1.8 1.5 10.8 7.4 −7.4 GSEA-WOA 2.5 7.1 GSEA-NORWECOM 1.1 1.5 1.4 7.6 2.5 −0.4 GSEA-NorESM 1.1 1.8 1.3 12.3 5.7 −5.1 NSEA-WOA 7.1 7.1 NSEA-NORWECOM 5.6 1.5 1.5 6.1 2.2 1.0 NSEA-NorESM 5.8 1.3 1.2 11.4 4.8 −4.3 Sea surface temperature Inorganic nitrate Mean RMSE Bias Mean RMSE Bias BSEA-WOA 2.9 3.4 BSEA-NORWECOM 2.8 0.9 0.2 7.0 4.3 −3.6 BSEA-NorESM 1.5 1.8 1.5 10.8 7.4 −7.4 GSEA-WOA 2.5 7.1 GSEA-NORWECOM 1.1 1.5 1.4 7.6 2.5 −0.4 GSEA-NorESM 1.1 1.8 1.3 12.3 5.7 −5.1 NSEA-WOA 7.1 7.1 NSEA-NORWECOM 5.6 1.5 1.5 6.1 2.2 1.0 NSEA-NorESM 5.8 1.3 1.2 11.4 4.8 −4.3 Figure 4. View largeDownload slide Annual cycle of sea surface temperature (SST) for the first (thin line) and last decades (thick line) for Barents Sea (black—left panel), the Greenland Sea (red—mid panel), and the Norwegian Sea (green—right panel), for NorESM1-ME (solid line) and ROMS/NORWECOM.E2E (dashed line). The X-es are the monthly means from the World Ocean Atlas for the period 2005–2012. Color refer to online version. Figure 4. View largeDownload slide Annual cycle of sea surface temperature (SST) for the first (thin line) and last decades (thick line) for Barents Sea (black—left panel), the Greenland Sea (red—mid panel), and the Norwegian Sea (green—right panel), for NorESM1-ME (solid line) and ROMS/NORWECOM.E2E (dashed line). The X-es are the monthly means from the World Ocean Atlas for the period 2005–2012. Color refer to online version. Time series of the annual mean upper 10 meters nutrients are shown in Figure 5. It is clearly seen that the values are much higher in the global than in the regional model, with a factor of 2 for inorganic nitrogen and PHO and a factor of four for SI (not shown). There is a small negative trend in the NorESM1-ME model value, and a positive correlation (r ≈ 0.4, p < 0.01) between PHO and SI in the Greenland Sea as well as the Norwegian Sea. The explanation for the large offset between the annual mean values is seen in Figure 6 where the seasonal cycle for inorganic nitrogen is shown. Some statistics where these seasonal cycles of the first 10 years are compared with that from the WOA are also given in Table 2. The figure shows that the winter values are close between the models, while there are large differences in the levels the rest of the year. The regional model utilizes all available nutrients in the upper 10 m, while there are large amounts of excess nutrients in the global model, consistent with the lower GPP. The seasonal cycles confirm the negative trend in annual means for the global model. Similar to the maximum in SST, the summer minimum in surface nutrients occur at the same time in NORWECOM.E2E and NorESM1-ME, and the decline in spring starts earlier in the regional model. Figure 5. View largeDownload slide Long term annual mean (0–10 m) inorganic nitrogen (left panel) 1s and, phosphorous (right panel) for Barents Sea (black), the Greenland Sea (red), and the Norwegian Sea (green), for NorESM1-ME (solid line) and NORWECOM (dashed line). Color refer to online version. Figure 5. View largeDownload slide Long term annual mean (0–10 m) inorganic nitrogen (left panel) 1s and, phosphorous (right panel) for Barents Sea (black), the Greenland Sea (red), and the Norwegian Sea (green), for NorESM1-ME (solid line) and NORWECOM (dashed line). Color refer to online version. Figure 6. View largeDownload slide Annual cycle (0–10 m) of inorganic nitrogen for the first 10 years (thin line) and last 10 years (thick line) for Barents Sea (black—left panel), the Greenland Sea (red—mid panel), and the Norwegian Sea (green—right panel), for NorESM1-ME (solid line) and NORWECOM (dashed line). The X-es are the monthly means from World Ocean Atlas. Color refer to online version. Figure 6. View largeDownload slide Annual cycle (0–10 m) of inorganic nitrogen for the first 10 years (thin line) and last 10 years (thick line) for Barents Sea (black—left panel), the Greenland Sea (red—mid panel), and the Norwegian Sea (green—right panel), for NorESM1-ME (solid line) and NORWECOM (dashed line). The X-es are the monthly means from World Ocean Atlas. Color refer to online version. Maps of annual mean primary production for both models averaged over the first decade of the simulation (2006–2015) are given in Figure 7. The gross primary production (GPP) is highest in the core of the warm Atlantic water in the Atlantic Current with lower values to the north and to the west. Comparing the two models, the GPP are generally higher in NORWECOM.E2E than in NorESM1-ME. This is most evident along the Greenland coast and in the northern Barents Sea, where the production in the global model is close to zero due to higher sea ice concentration in this model than in the regional (zero production along the southern boundary and high values to the northeast in the regional model is a boundary effect). In the last panel of Figure 7, annual mean net primary production (NPP) for the regional model for the same period is shown. The patterns are very similar to that of GPP, but the magnitude is much lower. On average, GPP in the regional model is 2.5 times higher than for the NPP. For GPP the mean values are 73, 68, and 139 g Cm−2 y−1 for NorESM1-ME in the Barents, Greenland, and Norwegian Sea, respectively, whereas the corresponding values for NORWECOM.E2E is 144, 154, and 180 g Cm−2 y−1. For NORWECOM.E2E NPP, the values for the same three domains are 54, 57, and 68 g Cm−2 y−1. Time series of annual GPP is given in Figure 8. In the Barents and Greenland seas, GPP is a factor two higher in the regional than in the global model, while in the Norwegian Sea the models are closer. There is no significant correlation in GPP between the models, but within each model and between the seas the correlation is in the range r = 0.3–0.4 (p < 0.01) for all combinations of seas. No clear trend in any of the models or seas are found. Maximum trend (absolute level) is found for the regional model in the Greenland Sea with a long term decline of -0.08 g Cm−2 y−1. A comparison of the GPP seasonal cycle (Figure 9) is consistent with that of the temperature and nutrient ones. The spring bloom in the regional model is earlier than for the global model (Figure 9), whereas the end of the production season is in all regions and models the same (August/September). In NORWECOM.E2E the production maximum is in May, while in NorESM1-ME the maximum is in June. The difference in the timing and maximum GPP between the two models can be attributed by the difference in the phytoplankton growth rate parameterizations, which is calibrated differently. Figure 7. View largeDownload slide Annual mean primary production (gC m−2 year−1) for the first decade (2006–2015). NorESM1-ME gross primary production (left), NORWECOM.E2E gross primary production (centre), and NORWECOM.E2E net primary production (right). Figure 7. View largeDownload slide Annual mean primary production (gC m−2 year−1) for the first decade (2006–2015). NorESM1-ME gross primary production (left), NORWECOM.E2E gross primary production (centre), and NORWECOM.E2E net primary production (right). Figure 8. View largeDownload slide Annual mean gross primary production (GPP) for Barents Sea (black), the Greenland Sea (red), and the Norwegian Sea (green), for NorESM1-ME (solid line) and NORWECOM (dashed line). Color refer to online version. Figure 8. View largeDownload slide Annual mean gross primary production (GPP) for Barents Sea (black), the Greenland Sea (red), and the Norwegian Sea (green), for NorESM1-ME (solid line) and NORWECOM (dashed line). Color refer to online version. Figure 9. View largeDownload slide Annual cycle of gross primary production (GPP) for the first 10 years (thin line) and last 10 years (thick line) for Barents Sea (black—left panel), the Greenland Sea (red—mid panel), and the Norwegian Sea (green—right panel), for NorESM1-ME (solid line) and NORWECOM (dashed line). Color refer to online version. Figure 9. View largeDownload slide Annual cycle of gross primary production (GPP) for the first 10 years (thin line) and last 10 years (thick line) for Barents Sea (black—left panel), the Greenland Sea (red—mid panel), and the Norwegian Sea (green—right panel), for NorESM1-ME (solid line) and NORWECOM (dashed line). Color refer to online version. Maps of annual mean pH in the upper 10 m for both models and averaged over the first decade of the simulation (2006–2015) are given in Figure 10. The magnitude is generally higher in NorESM1-ME (mean value 8.13 compared with 8.08), while the regional differences is more pronounced for NORWECOM.E2E, especially with lower pH levels along the Norwegian and the Greenland coast. There is a steady decline in pH in both models (see Figure 11, left panel), with a negative trend between -0.0021 and -0.0025 year−1 for both models and all three seas. The pH is slightly higher in the global model than in the regional one by a mean of 0.03 in Barents and Greenland seas and 0.05 in the Norwegian Sea. The results are similar for the saturation level of aragonite, Ωar (see Figure 11, right panel). In the global model the levels are slightly higher (0.1, 0.2, and 0.3 in Barents, Greenland, and Norwegian seas, respectively), with negative trends for both models close to -0.007 year−1. Figure 10. View largeDownload slide Annual mean pH (0–10 m) for the first decade (2006–2015). NorESM1-ME (left) and NORWECOM (right). Figure 10. View largeDownload slide Annual mean pH (0–10 m) for the first decade (2006–2015). NorESM1-ME (left) and NORWECOM (right). Figure 11. View largeDownload slide Annual mean (0–10 m) pH (left panel) and Ωar (right panel) for Barents Sea (black), the Greenland Sea (red), and the Norwegian Sea (green), for NorESM1-ME (solid line) and NORWECOM (dashed line). Color refer to online version. Figure 11. View largeDownload slide Annual mean (0–10 m) pH (left panel) and Ωar (right panel) for Barents Sea (black), the Greenland Sea (red), and the Norwegian Sea (green), for NorESM1-ME (solid line) and NORWECOM (dashed line). Color refer to online version. Spatial maps for annual CO2 flux for the first decade (2006–2015) are shown in Figure 12. The flux is higher in NorESM1-ME (10.69 compared with 5.75 mmol m−2 day−1 averaged over the whole area), while the regional differences is larger in NORWECOM.E2E. Maximum uptake of CO2 is in both models >20 mmol m−2 day−1 in parts of the Greenland Sea. While the global model has a net uptake of CO2 in the whole area, the regional model act as a source of CO2 in the Arctic and along the Greenland and Norwegian coasts. Time series in annual CO2 flux for the three different seas are shown in Figure 13. Both models suggest that the uptake is largest in the Greenland Sea. While the fluxes are similar between the seas for the NorESM1-ME model, there are large regional differences in the regional model ranging from <5 mmol m−2 day−1 in the Norwegian Sea to almost 15 mmol m−2 day−1 in the Greenland Sea. There is a positive trend (increasing uptake) in the flux in the global model with a maximum of 0.06 mmol m−2 day−1 year−1 in the Greenland sea, whereas NORWECOM.E2E give a positive trend in the Greenland sea (0.04 mmol m−2 day−1 year−1), a negative trend in the Norwegian Sea (-0.03 mmol m−2 day−1 year−1) and only a slight decrease in the Barents Sea (-0.005 mmol m−2 day−1 year−1). Averaged over the whole area, there is an increase in the uptake of CO2 in NorESM1-ME of 9% and NORWECOM.E2E of 7% from 2006 to 2070. The most prominent features in the seasonal cycle (see Figure 14) is the large positive flux during winter in the Greenland Sea, and the shift in season for the regional model in the Barents and Norwegian seas with lower winter values in the last 10 years than in the first period. The flux is always positive (ocean uptake) for both models, with the lowest values in the Norwegian Sea in the last period for the NORWECOM.E2E. For both models the strongest seasonal cycle is found in the Greenland Sea. Figure 12. View largeDownload slide Annual mean air–sea CO2 flux (mmol m−2 day−1) for the first decade (2006–2015). NorESM1-ME (left) and NORWECOM (right). Positive values are uptake. Figure 12. View largeDownload slide Annual mean air–sea CO2 flux (mmol m−2 day−1) for the first decade (2006–2015). NorESM1-ME (left) and NORWECOM (right). Positive values are uptake. Figure 13. View largeDownload slide Annual mean CO2-flux for Barents Sea (black), the Greenland Sea (red), and the Norwegian Sea (green), for NorESM1-ME (solid line) and NORWECOM (dashed line). Color refer to online version. Figure 13. View largeDownload slide Annual mean CO2-flux for Barents Sea (black), the Greenland Sea (red), and the Norwegian Sea (green), for NorESM1-ME (solid line) and NORWECOM (dashed line). Color refer to online version. Figure 14. View largeDownload slide Annual cycle of CO2-flux for the first 10 years (thin line) and last 10 years (thick line) for Barents Sea (black—left panel), the Greenland Sea (red—mid panel) and the Norwegian Sea (green—right panel), for NorESM1-ME (solid line) and NORWECOM (dashed line). Color refer to online version. Figure 14. View largeDownload slide Annual cycle of CO2-flux for the first 10 years (thin line) and last 10 years (thick line) for Barents Sea (black—left panel), the Greenland Sea (red—mid panel) and the Norwegian Sea (green—right panel), for NorESM1-ME (solid line) and NORWECOM (dashed line). Color refer to online version. Discussion Validation of present day climate Using data from WOA (https://www.nodc.noaa.gov/OC5/woa13/woa13data.html) (2005–2012 values) and averaging over the same areas give annual means of SST 2.9, 2.5, and 7.1°C for the Barents, Greenland, and Norwegian Sea, respectively, Table 2). In the Barents Sea the global model is too cold, while the regional model adjusts to the correct level after a ∼5 years. In the Greenland and Norwegian Seas, the global model is ∼1°C too cold. Here the regional model, which is initialized from the global one, has the same cold bias throughout the whole model period, due to too much vertical mixing. Except for the Barents Sea where the RMSE and bias is much lower in the regional model, the statistics for the annual cycle of the first 10 years are generally comparable for the two models. Comparing climatology of SSS from the WOA (34.5, 34.6, 35.0 for the Barents, Greenland, and Norwegian Sea, respectively), the NorESM1-ME is obviously much too saline in the Norwegian Sea, while the two other seas also have a positive bias. The regional model uses the first 5–10 years as spin-up to adjust the initial field. After this period a decreasing trend is bringing the SSS closer to a more realistic level over the simulation period, but the model never seems to stabilize at correct levels. Focusing on the seasonal cycle in SST (see Figure 4), the modelled maximum temperature in August and minimum in March is in accordance with the observations. The annual temperature amplitude is largely underestimated in the global model, while there is a good correspondence between the regional model and the observations. As already stated, the SI values in NorESM1-ME is unrealistically high. Comparing with the WOA, this is also the case for inorganic nitrogen and PHO. Omitting some small regional differences, the modelled annual mean inorganic nitrogen from the global model is close to 11 μM, whereas PHO is around 0.8 μM. The corresponding observations from WOA are 3.4, 7.1, and 7.1 μM for inorganic nitrogen for the Barents, Greenland, and Norwegian Sea, respectively, Table 2), and 0.4, 0.6, and 0.5 μM for PHO. Except for an overestimate in inorganic nitrogen in the Greenland Sea, the regional model has better fit to the observed annual means, despite being initialized from NorESM1-ME and using these values (N and P) on the open boundaries. An explanation of the discrepancies between the global model and observations are seen in the seasonal cycle of inorganic nitrogen in Figure 6. Except for the Barents Sea, there is a good agreement between both the models and the observed winter values. The difference in the annual means is due to the fact that when the NORWECOM.E2E model is able to utilize all available nutrients and bringing summer minimum close to zero in accordance with observations, the minimum in NorESM1-ME is around 6 μM. The reason for non-exhaustive nutrient consumption in NorESM1-ME in summer is because of a balance between further use of nutrients from phytoplankton growth, and the sources of new nutrients through remineralization of phytoplankton and zooplankton, the decay of DOC and remineralization of detritus. Here, the additional constraint on phytoplankton growth by zooplankton grazing (Table 1) dominates over nutrient re-supply predominantly by remineralization of detritus. The timing of the decline in surface nutrients in spring corresponds between the regional model and the observations, while the global model is delayed. However, the decline is faster in the regional model than observed so that the summer minimum is reached in June/July compared with the observed minimum in August. These findings are also confirmed from the RMSE and bias in Table 2, as the results from NORWECOM is much better than those from NorESM1-ME in all three domains. The study areas cover a large area south and north of the Arctic Circle, on both sides of the Arctic Front. Therefore, phytoplankton is exposed to wide variations in physical forcing factors such as light, temperature, and nutrient supply, which combined control the growth rate. To estimate the annual primary production under such conditions is almost impossible mainly for logistical reasons that result in a scarcity of measurements. Nevertheless, through combining existing measurements in the Norwegian Sea, an annual NPP rate of ∼80 g Cm−2 y−1 has been estimated (Rey, 2004). Annual NPP in the Greenland sea are apparently comparable with those in the Norwegian Sea, and estimated to be ∼70 g Cm−2 y−1 in Rey et al. (2000) and 81 g Cm−2 y−1 in the open Greenland Sea by Richardson et al. (2005). In the Barents Sea, estimates of primary production varies a lot between the different water masses. Titov and Orlova (2011) give a mean value for GPP in the Barents Sea of 111 g Cm−2 y−1, while Slagstad et al. (2011), using the SINMOD model, give a value of 102 g Cm−2 y−1 (GPP) and 53 g Cm−2 y−1 (NPP). The NPP in NORWECOM.E2E is at comparable magnitude to observations in both the Greenland and Norwegian seas, whereas in the Barents Sea, NPP is in agreement with the SINMOD model, while GPP is somewhat high. For NorESM1-ME, where only GPP is available, the values are obviously too low in the Barents and Greenland seas while in the Norwegian Sea (assuming a similar NPP:GPP ratio as for SINMOD and NORWECOM.E2E), the values are close to observations. Recalling the large reservoir of available summer nutrients in the global model, this seems consistent with the low production in the Barents and Greenland seas, however using the same argument in the Norwegian Sea would result in an annual primary production far above other estimates. Time-series observations from Ocean Weather Station Mike (OWSM) at (⁠ 66°N, 2°E) in the Norwegian Sea, have shown that the pre-bloom starts early March (average on 2 March), and that the time at which the bloom reaches its peak can vary by as much as 5–6 weeks from year to year. The observations indicate that the average time of the peak spring bloom is May 21st (mean for 1991–2003) and that the production season lasts until October. Maximum observed chlorophylla concentration in spring is barely >3 mg Chla m−3 (Rey, 2004). In the Greenland Sea, Richardson et al. (2005) make a summary of several studies and conclude that the spring bloom starts in March, maximum production occurs in May/June and that there still is a significant production going on in August. On the basis of modelling and observations, Titov (1995) describes the seasonal dynamics of primary production in the Barents Sea. It starts already in March and develops first in Atlantic and coastal waters and peeks in May/June before it slows down in June/July. The zone of the spring bloom moves to the north and northeast along the ice edge. In late summer and fall a second bloom is seen in the western Barents Sea formed by inflow of nutrient-rich Atlantic water. Chlorophylla observations along the Fugløya–Bjørnøya transect at the Barents Sea opening confirm this general picture with the pre-bloom starting in March, maximum chlorophylla in June and a second peak in August (Dalpadado et al., 2014). Onset of the spring bloom the NORWECOM.E2E model is in all areas in March with peak primary production in May in agreement with observations. From September onwards the GPP is close to zero, thus the model gives a shorter season than the observations. In the NorESM1-ME model the production season is even shorter starting in May with peak production in June, while (similar to the regional model) GPP is close to zero from September. Reflecting the shorter season, maximum production rate is highest in the global model up to 60 gC m−2 month−1 in the Norwegian Sea in June. The timing of the spring phytoplankton bloom depends strongly on the physical conditions, especially the development of the upper mixed layer (e.g. Taylor and Ferrari, 2011). However, when comparing the physics between the models there are no variables that can explain the large differences in the onset of the spring bloom. Instead, this discrepancy between the models is due to the parameterization of the phytoplankton growth (see Table 1). The maximum production rate at 0°C is considerably larger in NORWECOM.E2E (1.32 day−1) than in NorESM1-ME (0.60 day−1), which allows the former model to reach earlier net positive growth rate and hence bloom period. Using observation from 2001 to 2006 (monthly or higher frequency) from OWSM (Skjelvan et al., 2008) mean pH and ΩAr in the upper 10 m is estimated to 8.11 and 2.25, respectively. This is close to more recent observations (2012–2015) from a buoy operating at the same site reporting on minimum values of pH and ΩAr in winter around 8.06 and 1.85, increasing to 8.2 and 3.0 in summer (Chierici et al., 2016). Lauvset et al. (2016) mapped pH from the GLODAPv2 (Olsen et al., 2016) data set on a global 1° × 1° grid using the DIVA software (Troupin et al., 2012). On the basis of this, the average upper 10 m pH in the Barents, Greenland, and Norwegian seas are 8.11, 8.19, and 8.14, respectively. Compared with this, the global model is a little high in the Barents Sea, while the regional model is a little low in the Greenland Sea. Both models suggest the Norwegian Sea to have the lowest surface pH among the areas discussed, which contradicts the GLODAPv2 data set that has the lowest values in the Barents Sea. Also, the ΩAr can be computed from the GLODAPv2 data set. Using all available data points in the three boxes, the average values are 2.14, 2.01, and 2.32 for the Barents, Greenland, and Norwegian Sea, respectively. This suggests that ΩAr in the NorESM1-ME model is too low in the Barents Sea, while the NORWECOM.E2E has a low bias in all regions. Nevertheless, even if this is the most extensive data set available, the number of observations is limited. Between 1981 and 2013 the data base consists of 175 data points in the Barents Sea, 250 in the Greenland Sea, and 800 in the Norwegian Sea. Their seasonal and spatial coverage is limited and does not necessarily represent a full annual average in a regional sea. Mapping the observed pH and ΩAr at OWSM with the modelled mean from NORWECOM.E2E for the first five years (2006–2010) show that the model has an almost perfect match with pH = 8.12 and ΩAr = 2.18. The uptake of CO2 (Figure 13) is in accordance with Chen and Borges (2009) who suggest that in general high-latitude continental shelf seas tend to be net annual sinks of atmospheric CO2. Several estimates of the annual air–sea CO2 flux for the Barents Sea exist based on different data sets and approaches, ranging from 3.5 to 12 mmol m−2 day−1 (Bates and Mathis, 2009; Lauvset et al., 2013). Manizza et al. (2013) estimated the sink of CO2 in the Barents and Greenland seas to be 4 and 2.3 mmol m−2 day−1 in the period 1996–2007 using a regional physical–biogeochemical model for the Arctic, while Skogen et al. (2014) report on present day (20C3M) CO2 flux in the Barents Sea of 5.3 mmol m−2 day−1. Using Self Organised Maps, Yasunaka et al. (2016) estimated monthly gridded (⁠ 1° ×  1° ⁠) CO2 flux in the whole Arctic for the period 1997–2003. In the Barents Sea the mean net sink was estimated to be 10 mmol m−2 day−1, and in the Greenland and Norwegian Sea 11 mmol m−2 day−1. The seasonal cycle shows a maximum sink in winter (February/March) of 12 and 15 mmol m−2 day−1, respectively, and a minimum in summer (June/July/August) of ∼5 mmol m−2 day−1. The strong winter uptake in the Norwegian Sea in NorESM1-ME has been shown to be inconsistent with the data and likely as a result of the anomalously strong MLD bias rather than the biological processes (Gharamti et al., 2017). In the Barents Sea, NorESM1-ME is in the high end of previous estimates, while NORWECOM is well within. Both models have a less pronounced seasonal cycle than reported by Yasunaka et al. (2016). Nevertheless, there are large spatial differences (Figure 12) in both models. Recalculating both the annual mean and seasonal cycle in the NORWECOM.E2E model using the exact same boundaries as Yasunaka et al. (2016), gives a different picture. During the first decade the annual means for the Barents and Greenland/Norwegian (considered as one area) seas are 7 and 8 mmol m−2 day−1, respectively. There is also a clear seasonal cycle with winter maxima of 10 and 13 mmol m−2 day−1 and summer minima of 5 and 2 mmol m−2 day−1. Future climate changes Future changes in the downscaled physics in the Barents Sea has been studied in Sandø et al. (2018). There, to study a possible realization of the climate 50 years from now, the model mean values for the decade 2010–2019 was subtracted from the last decade 2060–2069. The downscaled model results show an increase in temperatures of ∼0.5– 1°C in most parts of the Barents Sea, somewhat more along the Polar Front in the Hopen Trench and in the northeastern parts of the Barents Sea in March (winter/spring), and somewhat less in September (summer/autumn), with a slight cooling in the southeastern parts. Reductions in sea ice extent reflects the increased temperatures and are most prominent in the northern Barents Sea, specifically along the northwestern coast of Novaya Zemlya and in the Barents Sea Exit, but also along the coast in the southeastern Barents Sea where sea ice is present during March. The inter-annual and decadal variability is quite substantial and bigger than the overall trend during the simulation period. There are no changes in future nutrient levels, and neither the global nor the regional model indicate any changes in the primary production at regional level. In general terms high-latitude spring-bloom ecosystems should benefit from increases in temperature giving increased regenerated production, but other factors like changes in mixed-layer-depth may alter this. The present study adds to other modelling studies that reports on the effect of climate change on primary production in the area, without any general agreement on how this will be effected. Steinacher et al. (2010) suggest a significant increase in primary production in 2100 under the A2 emission scenario based on an ensemble of 4 global models. Using a slightly different set-up of the NORWECOM model, Skaret et al. (2014) predicts a strong increase in primary production in the Barents Sea in 2065 under A1B using a regional downscaling of the GISS-AOM global climate model, while Slagstad et al. (2015) predicts a general decrease in primary production except for areas where ice retreats in 2100 under A1B using the SINMOD model and climate forcing from MPI-ECHAM5. Using POLCOMS-ERSEM forced by IPSL-CM4 under A1B, Barange et al. (2014) predicts a strong increase in phytoplankton biomass in the Greenland and Jan Mayen Economical exclusive economic zones (EEZs) and a slight decline in the Norwegian EEZ in 2050. The average pH of the surface waters of the global oceans has decreased from ∼8.2 before the onset of the industrial revolution to a present average of ∼8.1 (Caldeira and Wickett, 2003; Orr et al., 2005). Over the last quarter century the decrease has been by a rate of ∼0.0018 yr−1 at several open-ocean time-series sites (Feely et al., 2009). Lauvset et al. (2015) report on a decrease in the surface pH in the North Atlantic subpolar seasonally (NA-SPSS) biome of -0.0020 ± 0.0004 yr−1 between 1991 and 2011 using data from SOCAT collection (www.socat.info). Olafsson et al. (2009) report an even higher rate of -0.0024 yr−1 in the Iceland Sea for the period 1985–2008, thus the decline in pH in both models (on average -0.0024 yr−1) is within these observations. At the end of the century (2080–2100), the IPCC reports the global mean surface pH to decrease to 7.97 under RCP4.5 (Figure 2.5 in IPCC, 2014). This represents a decline of 0.14–0.15 compared with the level in 1986–2005. Using the modelled rate over a 75 year period (1995–2070) the models suggest a decrease in surface pH of 0.18. As the increase in atmospheric CO2 is low after 2070 under RCP4.5, and the decline in pH is believed to be even stronger in the Arctic, the modelled rate of future change in surface pH is in accordance with this prediction. A large proportion of marine life forms incorporate calcium carbonates in body armour, and ocean acidification leads to less favourable conditions for the formation of these minerals. Current surface seawaters are generally supersaturated with respect to calcium carbonates (Ω > 1), but saturation state decreases when more CO2 is dissolved in the water (reduced pH). A decrease in the carbonate concentration will thereby affect the survival of calcifying organisms, and when the carbonate concentration reaches a critical level the seawater will become corrosive for the calcifying organisms (Roleda et al., 2012). Changes in saturation state with respect to these minerals are therefore important for understanding how ocean acidification might impact future ecosystems. In 2002 the observed saturation horizon of aragonite (Ωar = 1) in the Norwegian and Greenland Seas was ∼2000 m (Børsheim and Golmen, 2010 based on Olsen et al., 2006). Compared with observations from 1981, there had been a shoaling of 150 m (or 7 m year−1). In the Iceland Sea Olafsson et al. (2009) report on a shoaling rate of 4 m year−1 in the period 1985–2008. Initially, undersaturation of Ωar in the Norwegian and Greenland seas appears below 3600 m in the regional model (not shown). At the end of the simulation the saturation horizon in these areas was ∼2600 m. This gives a shoaling of the saturation horizon of ∼1000 m in 64 years (16 m year−1). From an ensemble of climate models, Orr et al. (2005) report on the shoaling of the saturation horizon of aragonite in future climate. Following the IS92a emission scenario (723 ppm in 2100, which is slightly more pessimistic than RCP4.5 until 2060 when RCP4.5 stabilizes while IS92a continue to increase steadily) the annual average aragonite saturation shoaling in the North Atlantic (north of 50°N) during the 21st century, is 25 m year−1. The present shoaling from the regional downscaling is thereby well within the mean of the observed and predicted rates. The exchange of CO2 between atmosphere and seawater is relatively rapid with an equilibrium time scale of a year (Broecker and Peng, 1974), so that CO2 in surface waters in most ocean regions increase from year to year in proportion to the increased CO2 concentration in the atmosphere. However, there are several climate feedbacks related to the oceanic uptake of CO2. First higher temperature will increase the partial pressure of CO2 in the surface water and thereby reduce the ocean uptake. Second, climate change will have an effect on convective mixing and density stratification, which also will have an effect on the transport of CO2 into the ocean interior, and finally climate change will alter the natural carbon cycling through changes in biological production (Matear and Hirst, 1999). Averaged over the whole area, both models report on increase in ocean CO2 uptake (7% for the regional model and 9% for the global one). The increased SST will decrease the ocean uptake of CO2. As the biological production is approximately unchanged in the model both with respect to timing and magnitude, the increased uptake must be due to a positive contribution from changes in convective mixing and density stratification. For the regional model the increased uptake is strongest in the winter in the Greenland Sea (Figure 14). For the Barents Sea the regional model suggests a slight decrease in CO2 uptake. This is in disagreement with Skogen et al. (2014) who estimated an increase in the Barents Sea CO2 uptake from 5.3 to 8.5 mmol m−2 day−1 between 2000 and 2065 under emission scenario A1B. In their study, the main driver for this change was a change in DIC, which mainly contributed from a strong increase in the modelled primary and secondary production in the future climate, an increase that is not found in the present study. Concluding remarks The biogeochemistry from a global climate model (NorESM1-ME) has been compared with results from a regional model (NORWECOM.E2E) forced by a downscaling of the same climate simulation using the ROMS model. The focus has been to validate and determine the long term changes at regional scales, as the regional model obviously resolves local details that are not seen in the global model. The study concludes that the global model is able to reproduce several of today’s observations on a regional scale, but that there are many spatial details that are lost when a coarse resolution global model is used. The global model has a cold (in summer) and saline bias compared with climatology in the areas discussed, a bias that the regional model is able to alleviate to some extent. SI is unrealistically high in the global model, while winter values for inorganic nitrogen and PHO are close to the observations. On the other hand, the summer nutrient minimum is too high, as the global model is not able to utilize all nutrients in the upper layers, in contrast to the regional model. This results in a primary production in the Barents and Greenland seas below previous observational-based estimates, and a delayed onset of the spring bloom. The regional model is more reliable at projecting production level and timing, but the spring bloom develops too fast. Both models are comparable to the observations for pH and ΩAr, while NorESM1-ME is in the high end of CO2 flux estimates. There is a general agreement between the two models on future predictions, except for the development in SSS. There is no trend in future NPP in any of the models, while the trends in modelled pH and ΩAr are the same in both models. The largest discrepancy is in the development of the uptake of CO2, where the regional model suggests a slight reduced uptake in the future. Overall, when assessing present day climate impact on marine biogeochemistry and ecosystem, we demonstrate that a spatially coarse IPCC-class Earth System Model underperforms the regional model. One can argue that since the two applied biogeochemical models are different in their structure and parameterization, it is not only differences in scales (model resolution) and physical patterns, but also the inherent properties of the model formulations that contribute to the simulated differences. This problem was investigated by Skogen and Moll (2005) following up a previous study comparing two biogeochemical models in the North Sea (Skogen and Moll, 2000). In the first study it was shown that the two models agreed on annual mean primary production, its variability and the timing and size of the peak production. On the other hand, there was a low (even negative dependent of area) correlation in the production in different years between the two models. In the second study, the experiment was repeated, but both biogeochemical models were forced by the same physical model. The results were that the correlation between years became positive (changing from r = –0.49 to r = 0.63 for the North Sea annual production), and it was concluded that the single most important factor for a reliable modelling of phytoplankton and nutrient distributions and transports was a proper physical model. However, in addition to bias associated with poorly resolved physical processes, bias in the simulated seasonal cycle in key biogeochemical state variables are also attributed to the oversimplified ecosystem parameterization in the NorESM1-ME model, with a relatively slow phytoplankton growth compared with the grazing efficiency from zooplankton (Table 1). Data assimilation with the same global biogeochemical model (HAMOCC) indeed suggests that considerably model-data misfit in the seasonal cycle can be alleviated with a regional-varying ecosystem parameterization (Tjiputra et al., 2007). Nevertheless, when interpreting the results, several limitations should be taken into account. Models can only produce results, which are already predetermined by the model equations. As an example, climate change can favour other plankton groups than those included in the model and thereby potentially shift the ocean’s ability to serve as sink of CO2. The regional model is forced by downscaled ocean physics, but using atmospheric forcing from the global climate model. This might lead to, e.g. cold biases due to the insulation by the ice cover in the global model since the downscaled ocean physics has less sea ice in better correspondence with the observed ice extent (Sandø et al., 2018). The present study is only using one future scenario (RCP4.5) and one realization of it through the NorESM1-ME climate model. This is a clear limitation and has to be taken into consideration when interpreting the results. Through the ENSEMBLES project (http://ensembles-eu.metoffice.com) it was recommended to use results based on two or more RCM that again are forced by at least two Global Climate Models for climate impact studies (ENSEMBLES, 2009). The present study should therefore be considered as one member of a future ensemble of studies on the consequences of climate change. Upcoming studies using other models will either strengthen or weaken the findings, and thereby form an evaluation of the realism in the present set-up. Analysis of the uncertainty of a future projection in the context of global models is well illustrated by the work of Hawkins and Sutton (2009). Uncertainty is built up of three aspects: scenario uncertainty (reflecting the unknown future socio-economic landscape), model uncertainty (reflecting inaccuracies in the model) and internal variability (reflecting the difficulty in detecting a clear climate change signal until this averages out). In their work it is demonstrated how model and internal variability uncertainty decrease with lead time, while scenario uncertainty increases. This is also the case in the present study, which shows an evolution of an initial value problem, where the initial field and model uncertainties dominate, to a boundary value problem, where the emission scenario and corresponding atmospheric CO2 have the largest impact on the results. It is also demonstrated (Hawkins and Sutton, 2009), how, by moving from a global to a regional scale, the model and internal variability uncertainty can substantially increase. Therefore, single model simulations as forecasts of future conditions is questionable and should be viewed with an appropriate degree of caution. Acknowledgements This study was supported by the Centre for Climate Dynamics (SKD) in Bergen, Norway through the BIGCHANGE, PARADIGM, and LOES projects, and the FRAM centre in Tromsø, Norway through the Ocean Acidification Flagship. JT acknowledges Research Council of Norway through EVA (229771). References Aksnes D. , Ulvestad K. , Baliño B. , Berntsen J. , Egge J. , Svendsen E. 1995 . Ecological modelling in coastal waters: towards predictive physical–chemical–biological simulation models . Ophelia , 41 : 5 – 36 . Google Scholar Crossref Search ADS Barange M. , Merino G. , Blanchard J. L. , Scholtens J. , Harle J. , Allison E. H. , Allen J. I. , et al. 2014 . Impacts of climate change on marine ecosystem production in societies dependent on fisheries . Nature Climate Change , 4 : 211 – 216 . Google Scholar Crossref Search ADS Bates N. , Mathis T. 2009 . The Arctic Ocean marine carbon cycle: evaluation of air–sea CO2 exchanges, ocean acidification impacts and potential feedbacks . Biogeosciences , 6 : 2433 – 2459 . Google Scholar Crossref Search ADS Bellerby R. , Olsen A. , Furevik T. , Anderson L. 2005 . Response of the surface ocean CO2 system in the NordicSeas and North Atlantic to climate change. In The Nordic Seas: An Integrated Perspective, Geophysical Monograph Series , vol. 158 , pp. 189 – 198 . Ed. by Drange H. , Dokken T. , Furevik T. , Gerdes R. , Berger W. . American Geophyiscal Union . Bellerby R. G. J. , Silyakova A. , Nondal G. , Slagstad D. , Czerny J. , de Lange T. , Ludwig A. 2012 . Marine carbonate system evolution during the epoca arctic pelagic ecosystem experiment in the context of simulated arctic ocean acidification . Biogeosciences Discussions , 9 : 15541 – 15565 . Google Scholar Crossref Search ADS Blackford J. , Gilbert F. 2007 . pH variability and CO2 induced acidification in the North Sea . Journal of Marine Systems , 64 : 229 – 241 . Google Scholar Crossref Search ADS Bode A. , Barquero S. , Gonzales N. , Alvarez-Ossorio M. , Varela M. 2004 . Contribution of heterotrophic plankton to nitrogen regeneration in the upwelling ecosystem of La Coruna (NW Spain) . Journal of Plankton Research , 26 : 11 – 28 . Google Scholar Crossref Search ADS Børsheim K. , Golmen L. 2010 . Forsuring av havet. Kunnskapsstatus for norske farvann. Tech. Rep. TA 2575/2010, KLIF , Oslo, Norway , 101 pp. Broecker W. , Peng T-H. 1974 . Gas exchange rates between air and sea . Tellus , 26 : 21 – 35 . Google Scholar Crossref Search ADS Caldeira K. , Wickett M. 2003 . Anthropogenic carbon and ocean ph . Nature , 425 : 365. Google Scholar Crossref Search ADS PubMed Chen C. , Borges A. 2009 . Reconciling opposing views on carbon cycling in the coastal ocean: continental shelves as sinks and near-shore ecosystems as sources of atmospheric CO2 . Deep-Sea Research II , 56 : 578 – 581 . Google Scholar Crossref Search ADS Chierici M. , Skjelvan I. , Norli M. , Børsheim K. , Lauvset S. , Lødemel H. , Sørensen K. , et al. 2016 . Overvåking av havforsuring i norske farvann i 2015. Tech. Rep. M-573, Miljødirektoratet , Oslo, Norway , 114 pp. Dalpadado P. , Arrigo A. , Hjøllo S. , Rey F. , Ingvaldsen R. , Sperfeld E. , Dijken G. , et al. 2014 . Productivity in the Barents Sea – response to recent climate variability . PLoS One , 9 : e95273. Google Scholar Crossref Search ADS PubMed Dickson A. , Millero F. 1987 . A comparison of the equilibrium constants for the dissociation of carbonic acid in seawater media . Deep-Sea Res , 34 : 1733 – 1743 . Google Scholar Crossref Search ADS ENSEMBLES , 2009 . RCM-specific weights based on their ability to simulate the present climate calibrated for the ERA40-based simulations. ENSEMBLES (http://ensembles-eu.metoffice.com (last accessed 5 July 2018)) Project report D3.2.2 to the EU. 66pp. + 2app. Eriksen E. , Ingvaldsen R. , Nedreaas K. , Prozorkevich D. 2015 . The effect of recent warming on polar cod and beaked redfish juveniles in the Barents Sea . Regional Studies in Marine Science , 2 : 105 – 112 . Google Scholar Crossref Search ADS Feely R. , Doney S. , Cooley S. R. 2009 . Ocean acidification: present conditions and future changes in a high-CO2 world. Ocean Acidification. Present conditions and future changes in a high CO2 world . Oceanography , 22 : 36 – 47 . Google Scholar Crossref Search ADS Fossheim M. , Primicerio R. , Johannesen E. , Ingvaldsen R. , Aschan M. , Dolgov A. 2015 . Recent warming leads to a rapid borealization of fish communities in the Arctic . Nature Climate Change , 5 : 673 – 678 . Google Scholar Crossref Search ADS Garber J. 1984 . Laboratory study of nitrogen and phosphorous remineralization during decomposition of coastal plankton and seston . Estuarine, Coastal and Shelf Science , 18 : 685 – 702 . Google Scholar Crossref Search ADS Gehlen M. , Malschaert H. , Raaphorst W. 1995 . Spatial and temporal variability of benthic silica fluxes in the southeastern North Sea . Continental Shelf Research , 15 : 1675 – 1696 . Google Scholar Crossref Search ADS Gent P. R. , Danabasoglu G. , Donner L. J. , Holland M. M. , Hunke E. C. , Jayne S. R. , Lawrence D. M. , et al. 2011 . The community climate system model version 4 . Journal of Climate , 24 : 4973 – 4991 . Google Scholar Crossref Search ADS Gharamti M. , Tjiputra J. , Bethke I. , Samuelsen A. , Skjelvan I. , Bentsen M. , Bertino L. 2017 . Ensemble data assimilation for ocean biogeochemical state and parameter estimation at different sites . Ocean Modelling , 112 : 65 – 89 . Google Scholar Crossref Search ADS Hawkins E. , Sutton R. 2009 . The potential to narrow uncertainty in regional climate predictions . Bulletin of the American Meteorological Society , 90 : 1095 – 1107 . Google Scholar Crossref Search ADS Hjøllo S. , Huse G. , Skogen M. D. , Melle W. 2012 . Modelling secondary production in the Norwegian Sea with a fully coupled physical/primary production/individual-based Calanus finmarchicus model system . Marine Biology Research , 8 : 508 – 526 . Google Scholar Crossref Search ADS Ingri N. , Kakolowicz W. , Sillen L. , Warnqvist B. 1967 . Highspeed computers as a supplement to graphical methods: v. HALTAFALL, a general program for calculating the composition of equilibrium mixtures . Talanta , 14 : 1261. Google Scholar Crossref Search ADS PubMed IPCC , 2013 . Annex II: climate system scenario tables. In Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the 5th Report of the Intergovernmental Panel on Climate Change , pp. 1395 – 1445 . Ed. by Prather M. , Flato G. , Friedlingstein P. , Jones C. , Lamarque J. . Cambridge University Press , Cambridge, United Kingdom/New York, NY, USA . IPCC , 2014 . Climate Change 2014: Synthesis Report. contribution of working groups i, ii and iii to the 5th assessment report of the Intergovernmental Panel on Climate Change. 151 pp. Langehaug H. , Sandø A. , Årthun M. , Ilicak M. 2018 . Variability along the Atlantic water pathway in the forced Norwegian Earth System Model coastal zones . Climate Dynamics , 10.1007/s00382-018-4184-5. Lauvset S. , Chierici M. , Counillon F. , Omar A. , Nondal G. , Johannessen T. , Olsen A. 2013 . Annual and seasonal fCO2 and air–sea CO2 fluxes in the Barents Sea . Journal of Marine Systems , 113–114 : 62 – 74 . Google Scholar Crossref Search ADS Lauvset S. , Gruber N. , Landschutzer P. , Olsen A. , Tjiputra J. 2015 . Trends and drivers in global surface ocean pH over the past 3 decades . Biogeosciences , 12 : 1285 – 1298 . Google Scholar Crossref Search ADS Lauvset S. K. , Key R. M. , Olsen A. , van Heuven S. , Velo A. , Lin X. , Schirnick C. , et al. 2016 . A new global interior ocean mapped climatology: the 1 × 1 glodap version 2 . Earth System Science Data , 8 : 325 – 340 . Google Scholar Crossref Search ADS Lohse L. , Kloosterhuis H. T. , van Raaphorst W. , Helder W. 1996 . Denitrification rates as measured by the isotope pairing method and by the acetylene inhiluition technique in continental shelf sediments of the North Sea . Marine Ecology Progress Series , 132 : 169 – 179 . Google Scholar Crossref Search ADS Lohse L. , Malschaert F. , Slomp C. , Helder W. , Raaphorst W. 1995 . Sediment-water fluxes of inorganic nitrogen compounds along the transport route of organic matter in the North Sea . Ophelia , 41 : 173 – 197 . Google Scholar Crossref Search ADS Lønborg C. , Davidson K. , Alvarez-Salgado X. , Miller A. 2009 . Bioavailability and bacterial degradation rates of dissolved organic matter in a temperate coastal area during an annual cycle . Marine Chemistry , 113 : 219 – 226 . Google Scholar Crossref Search ADS Maier-Reimer E. , Kriest I. , Segschneider J. , Wetzel P. 2005 . The Hamburg Oceanic Carbon Cycle Circulation Model hamocc5.1 – Technical Description Release 1.1. Tech. Rep. Reports on Earth System Science 14 , Max Planck Institute for Meteorology , Hamburg, Germany . Manizza M. , Follows M. , Dutkiewicz S. , Menemenlis D. , Hill C. , Key R. 2013 . Changes in the Arctic Ocean CO2 sink (1996–2007): a regional model analysis . Global Biogeochemical Cycles , 27 : 1108 – 1118 . Google Scholar Crossref Search ADS Martinsen E. , Engedahl H. 1987 . Implementation and testing of a lateral boundary scheme as an open boundary condition in a barotropic ocean model . Coastal Engineering , 11 : 603 – 627 . Google Scholar Crossref Search ADS Matear R. , Hirst A. 1999 . Climate change feedback on the future oceanic CO2 uptake . Tellus B: Chemical and Physical Meteorology , 51 : 722 – 733 . Google Scholar Crossref Search ADS Mayer B. 1995 . Ein dreidimensionales, numerisches schwebstoff-transportmodell mit anwendung auf die Deutsche Bucht . Tech. Rep. GKSS 95/E/59, GKSS-Forschungszentrum Geesthacht GmbH , ISSN:0344-9629, 104 pp. Mehrbach C. , Culberson C. H. , Hawley J. E. , Pytkowicx R. M. 1973 . Measurement of the apparent dissociation constants of carbonic acid in seawater at atmospheric pressure . Limnology and Oceanography , 18 : 897 – 907 . Google Scholar Crossref Search ADS Melsom A. , Lien V. , Budgell W. 2009 . Using the regional ocean modeling system (ROMS) to improve the ocean circulation from a GCM 20th century simulation . Ocean Dynamics , 59 : 969 – 981 . Google Scholar Crossref Search ADS Moll A. , Stegert C. 2007 . Modeling pseudocalanus elongatus population dynamics embedded in a water column ecosystem model for the northern North Sea . Journal of Marine Systems , 64 : 35 – 46 . Google Scholar Crossref Search ADS Nightingale P. , Malin G. , Law C. , Watson A. , Liss P. , Liddicoat M. , Boutin J. , et al. 2000 . In situ evaluation of air–sea gas exchange parameterizations using novel conservative and volatile tracers . Global Biogeochemical Cycles , 14 : 373 – 388 . Google Scholar Crossref Search ADS Olafsson J. , Olafsdottir S. , Benoit-Cattin A. , Danielsen M. , Arnarson T. , Takahashi T. 2009 . Rate of Iceland Sea acidification from time series measurements . Biogeosciences , 6 : 2661 – 2668 . Google Scholar Crossref Search ADS Olsen A. , Key R. , van Heuven S. , Lauvset S. , Velo A. , Lin X. , Schirnick C. , et al. 2016 . The global ocean data analysis project version 2 (GLODAPv2) – an internally consistent data product for the world ocean . Earth System Science Data , 8 : 297 – 323 . Google Scholar Crossref Search ADS Olsen A. , Omar A. , Bellerby R. , Johannessen T. , Ninnemann U. , Brown K. , Olsson K. , et al. 2006 . Magnitude and origin of the anthropogenic CO2 increase and 13C suess effect in the Nordic seas since 1981 . Global Biogeochemical Cycles , 20 : GB3027 , doi:10.1029/2005GB002669. Google Scholar Crossref Search ADS Orr J. , Fabry V. , Aumont O. , Bopp L. , Doney S. , Feely R. A. G. , Gruber N. , et al. 2005 . Anthropogenic ocean acidification over the twenty-first century and its impact on calcifying organisms . Nature , 437 : 681 – 686 . Google Scholar Crossref Search ADS PubMed Overland J. , Wang M. 2007 . Future regional Arctic sea ice declines . Geophysical Research Letters , 34 : L17705. Google Scholar Crossref Search ADS Pätsch J. , Kühn W. , Moll A. H. L. 2009 . ECOHAM4 user guide – ECOsystem model, HAMburg, version 4. Tech. Rep. 01-2009, Institut Für Meereskunde , Hamburg , Germany . Pohlmann T. , Puls W. 1994 . Currents and transport in water, In Circulation and Contaminant Fluxes in the North Sea , pp. 345 – 402 . Ed. by Sündermann J. . Springer Verlag , Berlin . Rey F. 2004 . Phytoplankton: the grass of the sea. In The Norwegian Sea Ecosystem , pp. 97 – 136 . Ed. by Skjoldal H. . Tapir Academic Press , Trondheim, Norway . Rey F. , Noji T. , Miller L. 2000 . Seasonal phytoplankton development and new production in the Central Greenland Sea . Sarsia , 85 : 329 – 344 . Google Scholar Crossref Search ADS Richardson K. , Markager S. , Buch E. , Lassen M. , Kristensen A. 2005 . Seasonal distribution of primary production, phytoplankton biomass and size distribution in the Greenland Sea . Deep-Sea Research I , 52 : 979 – 999 . Google Scholar Crossref Search ADS Roleda M. , Boyd P. , Hurd C. 2012 . Before ocean acidification: calcifier chemistry lessons . Journal of Phycology , 48 : 840 – 843 . Google Scholar Crossref Search ADS PubMed Saini R. , Wang G. , Yu M. , Kim J. 2015 . Comparison of RCM and GCM projections of boreal summer precipitation over Africa . Journal of Geophysical Research: Atmospheres , 120 : 3679 – 3699 . Google Scholar Crossref Search ADS Sandø A. , Melsom A. , Budgell P. 2014 . Downscaling IPCC control run and future scenario with focus on the Barents Sea . Ocean Dynamics , 64 : 927 – 949 . Google Scholar Crossref Search ADS Sandø A. , Mousing E. , Budgell W. , Hjøllo S. , Skogen M. , Ådlandsvik B. 2018 . Possible Barents Sea climate and ecosystem changes in a moderate future emission scenario (RCP4.5) . Manuscript in Preparation . Shchepetkin A. F. , McWilliams J. C. 2005 . The regional oceanic modeling system (ROMS): a split-explicit, free-surface, topography-following-coordinate oceanic model . Ocean Modelling , 9 : 347 – 404 . Google Scholar Crossref Search ADS Six K. D. , Maier-Reimer E. 1996 . Effects of plankton dynamics on seasonal carbon fluxes in an ocean general circulation model . Global Biogeochemical Cycle , 10 : 559 – 583 . Google Scholar Crossref Search ADS Skaret G. , Dalpadado P. , Hjøllo S. , Skogen M. , Strand E. 2014 . Calanus finmarchicus abundance, production and population dynamics in the Barents Sea in a future climate . Progress in Oceanography , 125 : 26 – 39 . Google Scholar Crossref Search ADS Skartveit A. , Olseth J. A. 1986 . Modelling slope irradiance at high lattitudes . Solar Energy , 36 : 333 – 344 . Google Scholar Crossref Search ADS Skartveit A. , Olseth J. A. 1987 . A model for the diffuse fraction of hourly global radiation . Solar Energy , 38 : 271 – 274 . Google Scholar Crossref Search ADS Skjelvan I. , Falck E. , Rey F. , Kringstad S. 2008 . Inorganic carbon time series at Ocean Weather Station M in the Norwegian Sea . Biogeosciences , 5 : 549 – 560 . Google Scholar Crossref Search ADS Skogen M. , Budgell W. , Rey F. 2007 . Interannual variability in Nordic Seas primary production . ICES Journal of Marine Science , 64 : 889 – 898 . Google Scholar Crossref Search ADS Skogen M. , Moll A. 2000 . Interannual variability of the North Sea primary production: comparison from two model studies . Continental Shelf Research , 20 : 129 – 151 . Google Scholar Crossref Search ADS Skogen M. , Moll A. 2005 . Importance of ocean circulation in ecological modelling: an example from the North Sea . Journal of Marine Systems , 57 : 289 – 300 . Google Scholar Crossref Search ADS Skogen M. , Olsen A. , Børsheim K. , Sandø A. , Skjelvan I. 2014 . Modelling ocean acidification in the Nordic and Barents seas in present and future climate . Journal of Marine Systems , 131 : 10 – 20 . Google Scholar Crossref Search ADS Skogen M. , Søiland H. 1998 . A user’s guide to NORWECOM v2.0. The NORWegian ECOlogical Model system . Tech. Rep. Fisken Og Havet 18/98, Institute of Marine Research, Pb.1870, NO-5024 Bergen , 42 pp. Skogen M. , Svendsen E. , Berntsen J. , Aksnes D. , Ulvestad K. 1995 . Modelling the primary production in the North Sea using a coupled 3 dimensional Physical Chemical Biological Ocean model . Estuarine, Coastal and Shelf Science , 41 : 545 – 565 . Google Scholar Crossref Search ADS Slagstad D. , Ellingsen I. , Wassmann P. 2011 . Evaluating primary and secondary production in an Arctic Ocean void of summer sea ice: an experimental simulation approach . Progress in Oceanography , 90 : 117 – 131 . arctic Marine Ecosystems in an Era of Rapid Climate Change. http://www.sciencedirect.com/science/article/pii/S0079661111000243 (last accessed 5 July 2018). Google Scholar Crossref Search ADS Slagstad D. , Wassmann P. , Ellingsen I. 2015 . Physical constrains and productivity in the future Arctic Ocean . Frontiers in Marine Science , 2 : 85. Google Scholar Crossref Search ADS Stegert C. , Moll A. , Kreus M. 2009 . Validation of the three-dimensional ECOHAM model in the German Bight for 2004 including population dynamics of pseudocalnus elongatus . Journal of Sea Research , 62 : 1 – 15 . Google Scholar Crossref Search ADS Steinacher M. , Joos F. , Frölicher T. , Bopp L. , Cadule P. , Cocco V. , Doney S. , et al. 2010 . Projected 21st century decrease in marine productivity: a multi-model analysis . Biogeosciences , 7 : 979 – 1005 . Google Scholar Crossref Search ADS Steinacher M. , Joos F. , Frölicher T. , Plattner G-K. , Doney S. 2009 . Imminent ocean acidification in the Arctic projected with NCAR global coupled carbon cycle climate model . Biogeosciences , 6 : 515 – 533 . Google Scholar Crossref Search ADS Taylor J. R. , Ferrari R. 2011 . Shutdown of turbulent convection as a new criterion for the onset of spring phytoplankton blooms . Limnology and Oceanography , 56 : 2293 – 2307 . Google Scholar Crossref Search ADS Taylor K. E. , Stouffer R. J. , Meehl G. A. 2012 . An overview of CMIP5 and the experiment design . Bulletin of the American Meteorological Society , 93 : 485 – 498 . Google Scholar Crossref Search ADS Titov O. 1995 . Seasonal dynamics of primary produciton in the Barents sea . ICES CM 1995/Mini: 16. Mini-Symposium on Arctic Oceanographic Processes . ICES (International Council for the Exploration of the Seas, Copenhagen, Denmark). Titov O. , Orlova E. 2011 . Lower trophic levels. In The Barents Sea. Ecosystem, Resources and Management. Half a Century of Russian-Norwegian Cooperation , pp. 77 – 119 . Ed. by Jakobsen T. , Ozhigin V. . Tapir Academic Press , Trondheim, Norway , 825 pp. Tjiputra J. , Polzin D. , Winguth A. 2007 . Assimilation of seasonal chlorophyll and nutrient data into an adjoint threedimensional ocean carbon cycle model: sensitivity analysis and ecosystem parameter optimization . Global Biogeochemical Cycles , 21 : 10.1029/2006GB002745. Tjiputra J. F. , Roelandt C. , Bentsen M. , Lawrence D. M. , Lorentzen T. , Schwinger J. , Seland Ø. , et al. 2013 . Evaluation of the carbon cycle components in the Norwegian Earth System Model (NorESM) . Geoscientific Model Development , 6 : 301 – 325 . Google Scholar Crossref Search ADS Troupin C. , Barth A. , Sirjacobs D. , Ouberdous M. , Brankart J-M. , Brasseur P. , Rixen M. , et al. 2012 . Generation of analysis and consistent error fields using the data interpolating variational analysis (diva) . Ocean Modelling , 52–53 : 90 – 101 . Google Scholar Crossref Search ADS Utne K. , Hjøllo S. , Huse G. , Skogen M. 2012 . Estimating consumption of Calanus finmarchicus by planktivorous fish in the norwegian sea using a fully coupled 3d model system . Marine Biology Research , 8 : 527 – 547 . Google Scholar Crossref Search ADS Yasunaka S. , Murata A. , Watanabe E. , Chierici M. , Fransson A. , van Heuven S. , Hoppema M. , et al. 2016 . Mapping of the airesea CO2 flux in the Arctic Ocean and its adjacent seas: basin-wide distribution and seasonal to interannual variability . Polar Science , 10 : 323 – 334 . Google Scholar Crossref Search ADS © International Council for the Exploration of the Sea 2018. All rights reserved. For permissions, please email: [email protected] This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Koenker, Brittany L; Copeman, Louise A; Laurel, Benjamin J

doi: 10.1093/icesjms/fsy052pmid: N/A

Abstract The Arctic marine environment is rapidly changing with rising sea surface temperatures, declining sea ice habitat and projected increases in boreal species invasions. The success of resident Arctic fish will depend on both their thermal tolerance and their ability to cope with changing trophic interactions. Larval fish energetic condition is closely associated with mortality rates and therefore provides an indicator of overall well-being or fitness. In this study, we experimentally determined larval morphometric and lipid-based condition in an Arctic gadid (Arctic cod, Boreogadus saida) and a boreal gadid (walleye pollock, Gadus chalcogrammus) in response to different temperatures and food rations. Our results suggest that larval condition is highly sensitive to both factors but varies in a species- and ontogenetic-dependent manner. Results indicated that condition metrics based on length–weight relationships were not as sensitive as those based on lipid storage. Further, condition metrics changed with ontogeny and were best used within a developmental stage rather than across developmental stages. As expected, larval condition in first-feeding Arctic cod was higher at colder temperatures (2–5°C) than in the boreal gadid (5–12°C). However, at more developed larval stages the peak condition for Arctic cod was at warmer temperatures (7°C), while walleye pollock had the same thermal optimum as during earlier stages. Arctic cod were more sensitive to food ration at first feeding than walleye pollock, however; at later larval stages both species had a negative condition response to low food ration, especially at elevated temperatures (5 vs. 7°C). The lower thermal tolerance of Arctic cod, coupled with a higher sensitivity to food availability indicates that Arctic cod are particularly vulnerable to on-going environmental change. Arctic cod is a lipid-rich keystone species and therefore a reduction in their energetic condition during summer has the potential to affect the health of higher trophic levels throughout the Alaskan Arctic. Introduction Rapid ecosystem-level change is occurring throughout the Arctic because of rising temperatures in combination with declining sea ice volume and extent (Hoegh-Guldberg and Bruno, 2010). This, coupled with accelerated local extinction and invasion by North Pacific species, may result in higher species turnover rates and reorganization of the Arctic community structure (Cheung et al., 2009; Fossheim et al., 2015). Furthermore, sea ice reduction will potentially alter the regional primary productivity regime to the extent that a mismatch occurs between high-quality food production and key Arctic grazers (Søreide et al., 2010; Leu et al., 2011). The success of key Arctic species throughout the lipid-rich Arctic marine food web may largely be dictated by changing trophic interactions resulting from this mismatch (Falk-Petersen et al., 2007; Søreide et al., 2010). The latitudinal range of marine ectotherms is mainly dependent upon their thermal tolerance. Arctic cod (Boreogadus saida) is an ecologically important species in the Arctic where it plays a critical role as a mid-trophic prey item to marine mammals, seabirds, and other fish (Bluhm and Gradinger, 2008; Logerwell et al., 2015). As an ice-obligate species, Arctic cod will likely be impacted by sea ice retraction throughout the region (Fossheim et al., 2015). Walleye pollock (Gadus chalcogrammus) is a sub-Arctic species, which occupies a similar role throughout the Bering Sea shelves and the Gulf of Alaska (Bacheler et al., 2010). Climate warming and sea ice loss could result in the poleward migration of non-ice-obligate North Pacific species, such as walleye pollock (Rand and Logerwell, 2011; Fossheim et al., 2015). However, apparent avoidance of the Bering Sea cold pool by adult pollock (Overland and Stabeno, 2004) and continued formation of winter ice in the North Bering and Chukchi Seas (Sigler et al., 2011) may limit the potential for walleye pollock to become established in the Arctic (Hollowed et al., 2013). The early larval stage represents a critical period for marine fish species, often characterized by highly variable growth and survival (Letcher et al., 1996; Houde, 2008). This is particularly true in Polar Regions characterized by low light, cold temperatures, and reduced prey availability where mortality risks are enhanced and recruitment is ultimately impacted by overwintering success (Hurst, 2007). Under these conditions, larval survival depends largely on maximization of growth prior to overwintering (Fortier et al., 2006). Furthermore, developmental limitations of larval fish make this stage particularly sensitive to environmental stress and limit their ability to seek out suitable habitat under changing conditions (Rijnsdorp et al., 2009). Currently, basic understanding of the larval physiology of Arctic cod is largely lacking due to limitations in the logistics of under-ice sampling during the ice-covered Arctic winter and spring (Graham and Hop, 1995). This absence of physiological and ecological information on the early life stages of Arctic cod severely limits our ability to determine its survival potential in the face of ongoing climate change (Christiansen et al., 2014). Improved understanding of Arctic cod sensitivity to different environmental scenarios could help elucidate the factors that affect larval survival, population success, and eventual recruitment with changing ocean conditions. The nutritional condition of larval fish can have major impacts on mortality through direct (e.g. starvation) and indirect (e.g. prolonged stage duration and vulnerability to predation) mechanisms (Ehrlich et al., 1976, Shepherd and Cushing, 1980; Folkvord et al., 1996). As such, nutritional condition is a useful metric that can serve as an indicator of overall well-being or fitness (Jones et al., 1999). Condition indices have traditionally been based on the analysis of morphometric data, typically relating the actual weight of an individual to some “expected weight” or analysing a length–weight relationship (Bolger and Connolly, 1989). With these indices, it is assumed that at a given length, heavier fish are in better condition (Jones et al., 1999). Despite their widespread use, the inherent assumptions and limitations of these methods have been widely recognized. A range of condition indices (e.g. morphometric, biochemical, histological, etc.) with varying sensitivity to environmental stress exist (Suthers et al., 1992). Though morphometric indices (e.g. Fulton’s condition factor, K) are common (e.g. Neilson et al., 1986; Brodeur et al., 2000) in marine fish studies, direct condition measures (lipids) have been shown to be more sensitive to physiological stress in certain cases (Copeman et al., 2008). One advantage of lipid condition indices is the fact that lipid classes, particularly triacylglycerols (TAGs), quickly adjust to changes in feeding and so they can indicate shorter-term change in nutritional status (Lochmann et al., 1995). Conversely, morphometric or histological alterations are observed over longer time periods. Lipids are a limiting nutrient in cold-water marine ecosystems (Litzow et al., 2006) and affect the growth and survival of fish during early life stages (Lochmann et al., 1995; Copeman et al., 2002; Park et al., 2006). The energy storage of gadids during their early ontogeny is impacted by a number of factors (e.g. temperature, prey availability/quality), which are directly and indirectly tied to changing environmental conditions (Siddon et al., 2013). High temperatures increase metabolic demands and reduce the physiological ability of fish to store lipids (Jobling, 1988), in addition to altering the availability of prey and, thus, influencing the energy density of juvenile gadids (Heintz et al., 2013). By examining lipid class composition, it is possible to quantitatively measure energy reserves in an individual. TAGs are an energy storage lipid class that serve as an indicator of physiological state as they are the first lipids mobilized by fish during environmental stress (Fraser, 1989). To account for size dependency, TAG content can be measured relative to sterol (ST) compounds, a structural lipid class that serves as an adequate proxy for body size or dry weight (DWT) (Lochmann et al., 1995). The resultant TAG:ST ratio provides an index, which has been successfully used to measure larval fish, bivalves and crustacean condition (Fraser, 1989; Copeman and Laurel, 2010). As with growth, lipid condition has been shown to decrease near the upper thermal limit in four species of North Pacific juvenile gadids (Copeman et al., 2017). An experimental investigation of the nutritional condition of larval Arctic cod under various temperature and productivity scenarios has not previously been completed. In this study, total lipid and relative lipid class metrics were used to assess larval Arctic cod and walleye pollock condition in relation to changes in temperature and food availability. These lipid metrics were compared with morphometric condition indices based on length–weight residuals and body depth(BD):length ratios to determine the sensitivity of each. Specific objectives of this study were to (i) assess species-specific condition of Arctic cod and walleye pollock at two larval stages across a range of temperatures (Arctic cod: −1 to 9°C; walleye pollock: 0–12°C), (ii) determine how the interaction of temperature and food availability act on the species-specific condition of gadid larvae, and (iii) compare the sensitivity of different larval condition indices to changes in temperature and food availability. Methods Information on the lipid allocation and nutritional condition of larval Arctic cod and walleye pollock was obtained from laboratory experiments conducted at the Alaska Fisheries Science Center’s (AFSC) cold-water facilities at the Hatfield Marine Science Center (HMSC) in Newport, OR, USA. The results of these experiments also contributed to temperature-dependent growth and survival information (Koenker et al., this issue). The experimental methodology for eggs sources, egg incubation, live food preparation, general experimental design and husbandry as well as growth metrics have been previously described in detail (Koenker et al., this issue). Experimental design Briefly, laboratory experiments utilized larvae from the AFSC gadid broodstock programme. Broodstock were sourced from live juvenile fish collections of Arctic cod (70–85-mm SL) and walleye pollock (30–50-mm SL) collected during the spring or summer of 2011–2013. Juveniles were reared for over three years in the laboratory until they became active spawners (age 3+ fish) (collection details as in Laurel et al., 2016). Arctic cod adult broodstock were strip spawned in March of 2015 (for later stage experiment) and 2016 (for first-feeding experiment) and eggs from a single female were fertilized with milt from three males. Eggs were incubated at 1°C in a 4-L mesh pan suspended in a water bath until reaching ∼75% hatch level, at which time all hatched and unhatched larvae were transferred to 400-L stock tanks held at 2–3°C. Walleye pollock adult broodstock holding temperatures were reduced from 9 to 5°C in the fall and spawned naturally from February to late April 2015 (for both experiments). Eggs were retained in an egg basket from which the highest quality eggs were transferred directly to 100-L stock tanks and incubated at 5–6°C. Growth experiments were carried out in 38-L glass aquaria supplied with flow-through, temperature-controlled seawater. First-feeding Arctic cod larvae (mean 5.9-mm SL) were slow-acclimated to their corresponding temperature treatment and gently transferred to twelve aquaria for high food ration treatments (−1, 2, 5, and 9°C; n = 3 replicate tanks/temperature) and six for low food ration treatments (2 and 5°C; n = 3 replicate tanks/temperature) stocked at a density of ∼12 larvae per litre. First-feeding walleye pollock larvae (mean 4.7-mm SL) were acclimated in the same manner to twelve aquaria for high food ration treatments (0, 2, 5, and 12°C; n = 3 replicate tanks/temperature) and six for low food ration treatments (2 and 5°C; n = 3 replicate tanks/temperature) stocked at a density of ∼12 larvae per litre. Later stage larvae of both species remained in stock tanks and fed enriched rotifers (Brachionus sp.) twice daily at a density of 5 prey ml−1, before being gradually transitioned onto a diet of enriched brine shrimp (Artemia sp.) at a prey density of 2 prey ml−1 (see below for enrichment protocols). Later stage Arctic cod (78 dph; mean 11.3-mm SL) were transferred to twelve aquaria for high food ration treatments (0, 2, 5, and 7°C; n = 3 replicate tanks/temperature) and six for low food ration treatments (2 and 5°C; n = 3 replicate tanks/temperature) stocked at a density of ∼3 larvae per litre. A fourth replicate tank at 0 and 7°C was set up after 1 week to account for particularly high mortality in one of the replicate tanks at each temperature. An additional 9°C trial was conducted for later stage Arctic cod to assess the upper thermal limit for survival (n = 1 tank due to high mortality). Later stage walleye pollock (84 dph; mean 8.6-mm SL) were transferred to fifteen aquaria for high food ration treatments (0, 2, 5, 9, and 12°C; n = 3 replicate tanks/temperature) and nine for low food ration treatments (2, 5, and 9°C; n = 3 replicate tanks/temperature) in July 2015. First-feeding larval experiments received enriched rotifers (Brachionus sp.), while later stage experiments received enriched brine shrimp (Artemia sp.). Rotifers were cultured at 26°C in a high-density rotifer culture system from Aquatic Eco-Systems. Rotifers were harvest and enriched twice daily with Algamac 3050 (0.3 g per million rotifers; Aquafauna, Hawthorne, CA, USA) and Roti Grow Plus (daytime: 0.5 ml per million rotifers, overnight: 1.0 ml per million rotifers; Reed Mariculture, Campbell, CA, USA). Algamac was selected as a suitable enrichment because it contains a high proportion of long-chained fatty acids which are essential for North Pacific larval fish (Copeman and Laurel, 2010). Decapsulated brine shrimp were hatched for 24 h in hatching cones at 26–27°C before being enriched with Selco S.Presso (7.5 g per 15 l seawater; INVE Aquaculture, Nonthaburi, Thailand) for an additional 24 h. Both rotifers and brine shrimp were counted daily for quality control and to determine accurate prey counts. Prior to each feeding, Nanno 3600 algae paste (Reed Mariculture, Campbell, CA, USA) diluted with 2°C seawater was added to each tank to provide “green water” as a means of improving larval prey ingestion (Naas et al., 1996). First-feeding larvae in high food ration treatments were fed twice daily at prey densities of 5 prey ml−1, while low food ration treatments received prey densities of 0.5 prey ml−1 twice daily. Later stage larvae in high food ration treatments were fed twice daily at prey densities of 2 prey ml−1 and low food ration treatments received prey densities of 0.5 prey ml−1 once daily. All treatments received green water, including low food ration tanks that were not receiving prey in the afternoon. Tanks were clear of all live prey after ∼2 h following each feeding, indicating that all prey were either consumed or flowed out of the tank between feedings. This ensured that larvae were feeding on newly enriched prey and that prey quality did not deteriorate over the duration of the experiments. Throughout larval experiments, tanks were held at a 12:12-h light:dark photoperiod, with light levels ranging from 1.4 to 2.7 µE/m2sec at the surface of the water in the centre of each tank. Maintenance of tank temperatures, aeration, and flow rates was completed daily. Temperature was recorded in the morning and adjusted to within 0.5°C of the target temperature to account for fluctuations in ambient water temperature. Aeration was monitored to maintain gentle bubbling beneath the outflow mesh in each tank and flow rates were adjusted to within 270–330 ml min−1 each day. Tanks were siphoned daily, at least 2 h after feeding, to remove any mortalities along with excess food and debris. Due to differential mortality, experimental duration varied slightly among tanks (3–5 weeks for first-feeding experiments, 2–3 weeks for later stage experiments) Morphometric analyses Larvae from each tank were randomly sampled from throughout the water column for morphometric measurements (i.e., DWT, standard length [SL], BD) at the start and end of each experiment (12–35 days) as described in Koenker et al. (this issue). The number of individual larvae sampled varied based on mortality among tanks and throughout experiments. For first-feeding Arctic cod, one to five samples per tank containing 1 individual were used for end of experiment measurements (except for one tank at 9°C which was sampled at 3 weeks, rather than 5 weeks, due to mortality and contained five individuals). For first-feeding walleye pollock, 1 sample per tank containing 5–10 individuals was used for final measurements. For later stage experiments, 2–5 samples per tank containing one individual were used. Later stage BD:SL ratio analyses (Figures 2 and 6) also contain morphometric measurements for individuals used in lipid analyses outlined below. SL was determined as the length (mm) from the tip of the snout to the end of the notochord. BD was the width (mm) of the larvae just posterior to the anus not including the fin-fold. Length–weight residuals were obtained from the relationship between the natural log-transformed post-treatment mean SL (in mm) and mean DWT (in mg) for each tank in a single experiment. These length–weight residuals, at the tank level of observation, served as a morphometric condition index. Additionally, the ratio of BD:SL was computed for all larvae and applied as a measure of condition. BD represents the muscular tissue stored energy which is consumed by larvae after exhaustion of lipid reserves (Diaz et al., 2013). As such, larvae with a higher BD:SL were considered to be in better condition. Lipid extraction and analysis Additional larvae were sampled at the end of the experiments and used for total lipid and lipid class analyses. The number of individual larvae sampled was adjusted to obtain sufficient mass for lipid analyses (mean ∼200 µg total lipids per sample). For first-feeding Arctic cod, two pools of ten individuals per tank were sampled. For first-feeding walleye pollock, lipids per individual were limiting so all remaining larvae in each tank were pooled to produce one sample per tank (n = 7–52 individuals). For later stage experiments of both species, two to three samples per tank of 1 individual were used. Lipid samples were stored at −25°C in 2 ml of chloroform under nitrogen and sealed with Teflon tape for <3 months prior to lipid extraction. Lipids were extracted in chloroform and methanol according to Parrish (1987) using a modified Folch procedure (Folch et al., 1957). Thin layer chromatography with flame ionization detection with a MARK V Iatroscan (Iatron Laboratories, Tokyo, Japan) was used to classify and quantify total lipids and lipid classes using a procedure described by Lu et al. (2008) and modified by Copeman et al. (2016). Silica-coated Chromarods were spotted in duplicate with lipid extracts and a three-stage development system was employed for lipid class separation. The first separation involved a 90-s development in a chloroform:methanol:choloform-extracted water solution (150:120:30) to move polar lipids off the origin. The second separation involved a 48-min development in a hexane:diethyl ether:formic acid solution (99:1:0.05) and the final separation consisted of a 38-min development in a hexane:diethyl ether:formic acid solution (80:20:0.1). Following each separation, the rods were dried for 5 min and conditioned at constant humidity (∼32%) for an additional 5 min. After the final development, the Chromarods were scanned using PeakSimple software (ver. 3.67, SRI Inc.) and the signal (detected in mV) was quantified using lipid standards as described in Copeman et al. (2016). The resulting chromatograms were integrated to quantify absolute amounts of four lipid classes: TAGs, free fatty acids (FFA), sterols, and polar lipids. Lipid class values were taken as the average of duplicate runs for each lipid sample. For statistical analyses, tank means were used (n = 1–3 samples per tank as described above). Data analysis Primary growth and condition analyses were performed using RStudio statistical software (ver. 0.99.491, RStudio, Inc., Boston, MA, USA) using a significance level of α = 0.05. Tukey’s pairwise comparisons were performed using Minitab 17 statistical software (Minitab, Inc., State College, PA, USA) using a significance level of α = 0.05. Later stage Arctic cod at 9°C and later stage walleye pollock at 0°C were removed from condition analyses to account for size-selective mortality in treatments that experienced > 80% mortality within the first week of experiments (see Koenker et al., (this issue) for mortality details). Statistical tests were performed to assess the impacts of species, temperature, and food ration on larval condition and lipid storage within each ontogenetic stage. The effects of ontogeny within each species were not statistically tested as nutritional condition has been shown to be highly dependent on developmental stage (Richard et al., 1991), and thus, analysis across stages is not recommended. Furthermore, morphological and physiological changes during the early ontogeny of fish vary substantially within and among species (Pepin, 1995). Morphological condition indices are closely tied to species-specific allometric growth patterns and, so, this study considers temperature and food ration effects on morphological condition within a single species only. Lipid condition indices are less constrained by allometry, but still prone to species-specific lipid accumulation. Therefore, considerable care should be used when interpreting differences derived from the comparison of lipid condition across species in this study. For high food ration treatments, one-way analysis of variance (ANOVA) was used to assess the effects of temperature on morphological condition (length–weight residuals and BD:SL) within each stage. Additionally, two-way ANOVAs were used to determine species and temperature effects on lipid condition (TAG:ST) and total lipid storage (per larval DWT) within each stage. At intermediate temperatures where food rations were manipulated, two-way ANOVAs were used to assess the effects of temperature and food ration on morphological condition. Three-way ANOVAs were used to examine the interactive effects of temperature, food ration, and species on lipid condition and storage at each larval stage. Data were examined for normality and homogeneity of variance to satisfy the assumptions of ANOVA. Posteriori tests (Tukey’s pairwise comparisons) were performed to identify significant differences in morphometric (BD:SL) and lipid condition (TAG:ST, total lipids) with response to temperature within each experiment. For all statistical analyses, individual tanks were used as the level of observation and morphometric and lipid measurements taken at the end of laboratory experiments were used. Results Temperature effects on larval condition Logarithmic transformation was performed to linearize the relationship between post-treatment SL and DWT of larval fish from each experiment (Figure 1a, c, e, and g). The relationships for each experiment were described by the linear regression equations for first-feeding Arctic cod, first-feeding walleye pollock, later stage Arctic cod, and later stage walleye pollock shown in Figure 1. At the first-feeding stage, one-way ANOVAs detected no significant effect of temperature on the morphometric condition (length–weight residuals) of Arctic cod (F1,9 = 1.519, p = 0.249; Figure 1b) or walleye pollock (F1,10 = 1.767, p = 0.213; Figure 1d). Similarly, at the later larval stage, temperature did not statistically affect the condition of Arctic cod (F1,12 = 2.347, p = 0.152; Figure 1f) or walleye pollock (F1,10 = 1.439, p = 0.258; Figure 1h). The standard error of length–weight residual values was high because of large variation between tank means, likely contributing to the non-significant ANOVA results. Figure 1. Open in new tabDownload slide Linear relationships between log-transformed length (SL, mm) and log-transformed DWT (mg) of (a) first-feeding Arctic cod, (c) first-feeding walleye pollock, (e) later stage Arctic cod, and (g) later stage walleye pollock larvae and the residuals of this relationship for (b) first-feeding Arctic cod, (d) first-feeding walleye pollock, (f) later stage Arctic cod, and (h) later stage walleye pollock larvae in high food ration treatments at the end of laboratory experiments. Linear regression was fitted to mean tank data. Length–weight residuals were computed for each tank and plotted as treatment means ± 1 SE (n = 2–4 replicate tanks/treatment). Figure 1. Open in new tabDownload slide Linear relationships between log-transformed length (SL, mm) and log-transformed DWT (mg) of (a) first-feeding Arctic cod, (c) first-feeding walleye pollock, (e) later stage Arctic cod, and (g) later stage walleye pollock larvae and the residuals of this relationship for (b) first-feeding Arctic cod, (d) first-feeding walleye pollock, (f) later stage Arctic cod, and (h) later stage walleye pollock larvae in high food ration treatments at the end of laboratory experiments. Linear regression was fitted to mean tank data. Length–weight residuals were computed for each tank and plotted as treatment means ± 1 SE (n = 2–4 replicate tanks/treatment). The BD:SL ratio of gadid larvae under high food ration conditions generally increased across the experimental temperature range. For first-feeding Arctic cod, the effect of temperature on the BD:SL (one-way ANOVA; F1,9 = 3.891, p = 0.080) was not statistically significant (Figure 2a). For first-feeding walleye pollock, the BD:SL increased with temperature (from 0 to 12°C) as evidenced by a statistically significant temperature effect (F1,10 = 289.38, p < 0.001; Figure 2b). Similarly, a significant positive relationship between temperature and the BD:SL of both later stage Arctic cod (F1,12 = 95.436, p < 0.001) and walleye pollock (F1,10 = 27.902, p < 0.001) was statistically supported (Figure 2c and d). Figure 2. Open in new tabDownload slide BD:SL ratio of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae in high food ration treatments at the end of laboratory experiments. Data are treatment means ± 1 s.e. (n = 2–4 replicate tanks per treatment). Different letters indicate significant differences according to Tukey’s pairwise comparisons. Figure 2. Open in new tabDownload slide BD:SL ratio of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae in high food ration treatments at the end of laboratory experiments. Data are treatment means ± 1 s.e. (n = 2–4 replicate tanks per treatment). Different letters indicate significant differences according to Tukey’s pairwise comparisons. Total lipid and proportional lipid class composition for larvae in all experiments are included in Table 1. Among high food ration treatments, the largest proportion of each sample was comprised of PL, with first-feeding larvae containing proportionally more PL (84.6%) than later stage larvae (75.8%). The opposite was true with the TAG proportion, which was lower in first-feeding larvae (2.9%) than later stage larvae (12.8%). Both FFA (first-feeding: 0.5%; later stage: 0.8%) and ST (first-feeding: 11.9%; later stage: 10.6%) proportions remained relatively constant between stages. Table 1. Total lipid content (µg lipid mg−1 DWT) and lipid class composition of first-feeding and later stage Arctic cod and walleye pollock larvae under different temperature-food ration scenarios. Ontogenetic stage . Species . Temp (°C) . Food Ration . n . TL (μg mg−1) . TAG (%) . FFA (%) . ST (%) . PL (%) . First-feeding Arctic Cod −1 High 3 60.2 ± 3.1 3.1 ± 0.5 0.5 ± 0.1 7.4 ± 0.7 89.0 ± 0.7 2 High 2 90.9 ± 0.7 4.4 ± 0.0 0.4 ± 0.0 9.8 ± 0.6 85.4 ± 0.6 2 Low 3 85.5 ± 9.0 1.4 ± 0.3 0.5 ± 0.1 10.6 ± 0.8 87.4 ± 0.8 5 High 3 94.9 ± 5.1 4.4 ± 0.8 0.3 ± 0.1 10.8 ± 0.6 84.2 ± 1.4 5 Low 2 101.4 ± 6.7 2.9 ± 0.3 0.3 ± 0.0 10.8 ± 0.3 86.0 ± 0.6 Walleye Pollock 0 High 3 59.8 ± 4.8 1.3 ± 0.0 0.7 ± 0.0 14.0 ± 0.4 84.0 ± 0.4 2 High 3 55.3 ± 3.3 1.3 ± 0.2 0.7 ± 0.1 14.5 ± 0.7 83.5 ± 0.8 2 Low 1 29.9 ± 0.0 0.0 ± 0.0 4.1 ± 0.0 27.0 ± 0.0 68.9 ± 0.0 5 High 3 92.9 ± 9.4 3.7 ± 1.8 0.8 ± 0.3 12.2 ± 0.4 83.3 ± 2.6 5 Low 3 54.9 ± 10.8 1.6 ± 0.3 0.5 ± 0.1 12.0 ± 0.4 86.0 ± 0.3 12 High 2 74.4 ± 4.4 1.8 ± 0.2 0.4 ± 0.0 14.9 ± 1.5 82.6 ± 0.9 Later stage Arctic Cod 0 High 4 102.6 ± 2.4 5.6 ± 0.3 0.7 ± 0.1 10.2 ± 0.3 83.5 ± 0.6 2 High 3 104.9 ± 5.4 11.5 ± 0.5 0.7 ± 0.1 9.8 ± 0.6 78.0 ± 0.8 2 Low 3 99.0 ± 7.1 6.3 ± 1.0 0.7 ± 0.1 12.3 ± 0.6 80.7 ± 0.7 5 High 3 136.3 ± 5.0 19.0 ± 0.4 0.5 ± 0.1 10.9 ± 0.3 69.6 ± 0.7 5 Low 3 108.7 ± 1.4 14.1 ± 1.1 0.6 ± 0.1 11.0 ± 0.2 74.3 ± 1.3 7 High 4 138.1 ± 9.5 21.8 ± 2.1 0.5 ± 0.1 10.1 ± 0.4 67.6 ± 1.9 Walleye pollock 0 High 2 75.7 ± 0.0 6.8 ± 2.6 1.4 ± 0.2 13.2 ± 1.2 78.3 ± 1.8 2 High 3 113.1 ± 14.6 8.6 ± 1.2 0.9 ± 0.1 11.0 ± 0.6 79.5 ± 1.4 2 Low 3 97.2 ± 4.7 5.6 ± 1.0 0.9 ± 0.2 11.2 ± 0.5 82.2 ± 0.9 5 High 3 128.9 ± 5.9 14.7 ± 0.4 0.9 ± 0.1 11.1 ± 0.3 73.3 ± 0.7 5 Low 3 88.6 ± 4.7 4.3 ± 0.7 1.0 ± 0.3 13.8 ± 0.8 80.9 ± 0.3 9 High 3 126.0 ± 6.0 15.7 ± 2.0 0.6 ± 0.1 9.3 ± 0.5 74.4 ± 2.3 9 Low 3 95.0 ± 5.6 4.3 ± 0.8 0.9 ± 0.2 12.6 ± 0.6 82.2 ± 0.8 12 High 3 104.8 ± 4.6 11.8 ± 2.8 0.7 ± 0.1 9.8 ± 0.3 77.7 ± 2.7 Ontogenetic stage . Species . Temp (°C) . Food Ration . n . TL (μg mg−1) . TAG (%) . FFA (%) . ST (%) . PL (%) . First-feeding Arctic Cod −1 High 3 60.2 ± 3.1 3.1 ± 0.5 0.5 ± 0.1 7.4 ± 0.7 89.0 ± 0.7 2 High 2 90.9 ± 0.7 4.4 ± 0.0 0.4 ± 0.0 9.8 ± 0.6 85.4 ± 0.6 2 Low 3 85.5 ± 9.0 1.4 ± 0.3 0.5 ± 0.1 10.6 ± 0.8 87.4 ± 0.8 5 High 3 94.9 ± 5.1 4.4 ± 0.8 0.3 ± 0.1 10.8 ± 0.6 84.2 ± 1.4 5 Low 2 101.4 ± 6.7 2.9 ± 0.3 0.3 ± 0.0 10.8 ± 0.3 86.0 ± 0.6 Walleye Pollock 0 High 3 59.8 ± 4.8 1.3 ± 0.0 0.7 ± 0.0 14.0 ± 0.4 84.0 ± 0.4 2 High 3 55.3 ± 3.3 1.3 ± 0.2 0.7 ± 0.1 14.5 ± 0.7 83.5 ± 0.8 2 Low 1 29.9 ± 0.0 0.0 ± 0.0 4.1 ± 0.0 27.0 ± 0.0 68.9 ± 0.0 5 High 3 92.9 ± 9.4 3.7 ± 1.8 0.8 ± 0.3 12.2 ± 0.4 83.3 ± 2.6 5 Low 3 54.9 ± 10.8 1.6 ± 0.3 0.5 ± 0.1 12.0 ± 0.4 86.0 ± 0.3 12 High 2 74.4 ± 4.4 1.8 ± 0.2 0.4 ± 0.0 14.9 ± 1.5 82.6 ± 0.9 Later stage Arctic Cod 0 High 4 102.6 ± 2.4 5.6 ± 0.3 0.7 ± 0.1 10.2 ± 0.3 83.5 ± 0.6 2 High 3 104.9 ± 5.4 11.5 ± 0.5 0.7 ± 0.1 9.8 ± 0.6 78.0 ± 0.8 2 Low 3 99.0 ± 7.1 6.3 ± 1.0 0.7 ± 0.1 12.3 ± 0.6 80.7 ± 0.7 5 High 3 136.3 ± 5.0 19.0 ± 0.4 0.5 ± 0.1 10.9 ± 0.3 69.6 ± 0.7 5 Low 3 108.7 ± 1.4 14.1 ± 1.1 0.6 ± 0.1 11.0 ± 0.2 74.3 ± 1.3 7 High 4 138.1 ± 9.5 21.8 ± 2.1 0.5 ± 0.1 10.1 ± 0.4 67.6 ± 1.9 Walleye pollock 0 High 2 75.7 ± 0.0 6.8 ± 2.6 1.4 ± 0.2 13.2 ± 1.2 78.3 ± 1.8 2 High 3 113.1 ± 14.6 8.6 ± 1.2 0.9 ± 0.1 11.0 ± 0.6 79.5 ± 1.4 2 Low 3 97.2 ± 4.7 5.6 ± 1.0 0.9 ± 0.2 11.2 ± 0.5 82.2 ± 0.9 5 High 3 128.9 ± 5.9 14.7 ± 0.4 0.9 ± 0.1 11.1 ± 0.3 73.3 ± 0.7 5 Low 3 88.6 ± 4.7 4.3 ± 0.7 1.0 ± 0.3 13.8 ± 0.8 80.9 ± 0.3 9 High 3 126.0 ± 6.0 15.7 ± 2.0 0.6 ± 0.1 9.3 ± 0.5 74.4 ± 2.3 9 Low 3 95.0 ± 5.6 4.3 ± 0.8 0.9 ± 0.2 12.6 ± 0.6 82.2 ± 0.8 12 High 3 104.8 ± 4.6 11.8 ± 2.8 0.7 ± 0.1 9.8 ± 0.3 77.7 ± 2.7 TL, total lipid; TAG, triacylglycerols; FFA, free fatty acids; ST, sterols, PL, polar lipids. Arctic cod at both ontogenetic stages at 9°C are not included due to high mortality in treatment tanks. Data are treatment means ± 1 s.e. (n = 1–4 replicate tanks, depending on mortality). Open in new tab Table 1. Total lipid content (µg lipid mg−1 DWT) and lipid class composition of first-feeding and later stage Arctic cod and walleye pollock larvae under different temperature-food ration scenarios. Ontogenetic stage . Species . Temp (°C) . Food Ration . n . TL (μg mg−1) . TAG (%) . FFA (%) . ST (%) . PL (%) . First-feeding Arctic Cod −1 High 3 60.2 ± 3.1 3.1 ± 0.5 0.5 ± 0.1 7.4 ± 0.7 89.0 ± 0.7 2 High 2 90.9 ± 0.7 4.4 ± 0.0 0.4 ± 0.0 9.8 ± 0.6 85.4 ± 0.6 2 Low 3 85.5 ± 9.0 1.4 ± 0.3 0.5 ± 0.1 10.6 ± 0.8 87.4 ± 0.8 5 High 3 94.9 ± 5.1 4.4 ± 0.8 0.3 ± 0.1 10.8 ± 0.6 84.2 ± 1.4 5 Low 2 101.4 ± 6.7 2.9 ± 0.3 0.3 ± 0.0 10.8 ± 0.3 86.0 ± 0.6 Walleye Pollock 0 High 3 59.8 ± 4.8 1.3 ± 0.0 0.7 ± 0.0 14.0 ± 0.4 84.0 ± 0.4 2 High 3 55.3 ± 3.3 1.3 ± 0.2 0.7 ± 0.1 14.5 ± 0.7 83.5 ± 0.8 2 Low 1 29.9 ± 0.0 0.0 ± 0.0 4.1 ± 0.0 27.0 ± 0.0 68.9 ± 0.0 5 High 3 92.9 ± 9.4 3.7 ± 1.8 0.8 ± 0.3 12.2 ± 0.4 83.3 ± 2.6 5 Low 3 54.9 ± 10.8 1.6 ± 0.3 0.5 ± 0.1 12.0 ± 0.4 86.0 ± 0.3 12 High 2 74.4 ± 4.4 1.8 ± 0.2 0.4 ± 0.0 14.9 ± 1.5 82.6 ± 0.9 Later stage Arctic Cod 0 High 4 102.6 ± 2.4 5.6 ± 0.3 0.7 ± 0.1 10.2 ± 0.3 83.5 ± 0.6 2 High 3 104.9 ± 5.4 11.5 ± 0.5 0.7 ± 0.1 9.8 ± 0.6 78.0 ± 0.8 2 Low 3 99.0 ± 7.1 6.3 ± 1.0 0.7 ± 0.1 12.3 ± 0.6 80.7 ± 0.7 5 High 3 136.3 ± 5.0 19.0 ± 0.4 0.5 ± 0.1 10.9 ± 0.3 69.6 ± 0.7 5 Low 3 108.7 ± 1.4 14.1 ± 1.1 0.6 ± 0.1 11.0 ± 0.2 74.3 ± 1.3 7 High 4 138.1 ± 9.5 21.8 ± 2.1 0.5 ± 0.1 10.1 ± 0.4 67.6 ± 1.9 Walleye pollock 0 High 2 75.7 ± 0.0 6.8 ± 2.6 1.4 ± 0.2 13.2 ± 1.2 78.3 ± 1.8 2 High 3 113.1 ± 14.6 8.6 ± 1.2 0.9 ± 0.1 11.0 ± 0.6 79.5 ± 1.4 2 Low 3 97.2 ± 4.7 5.6 ± 1.0 0.9 ± 0.2 11.2 ± 0.5 82.2 ± 0.9 5 High 3 128.9 ± 5.9 14.7 ± 0.4 0.9 ± 0.1 11.1 ± 0.3 73.3 ± 0.7 5 Low 3 88.6 ± 4.7 4.3 ± 0.7 1.0 ± 0.3 13.8 ± 0.8 80.9 ± 0.3 9 High 3 126.0 ± 6.0 15.7 ± 2.0 0.6 ± 0.1 9.3 ± 0.5 74.4 ± 2.3 9 Low 3 95.0 ± 5.6 4.3 ± 0.8 0.9 ± 0.2 12.6 ± 0.6 82.2 ± 0.8 12 High 3 104.8 ± 4.6 11.8 ± 2.8 0.7 ± 0.1 9.8 ± 0.3 77.7 ± 2.7 Ontogenetic stage . Species . Temp (°C) . Food Ration . n . TL (μg mg−1) . TAG (%) . FFA (%) . ST (%) . PL (%) . First-feeding Arctic Cod −1 High 3 60.2 ± 3.1 3.1 ± 0.5 0.5 ± 0.1 7.4 ± 0.7 89.0 ± 0.7 2 High 2 90.9 ± 0.7 4.4 ± 0.0 0.4 ± 0.0 9.8 ± 0.6 85.4 ± 0.6 2 Low 3 85.5 ± 9.0 1.4 ± 0.3 0.5 ± 0.1 10.6 ± 0.8 87.4 ± 0.8 5 High 3 94.9 ± 5.1 4.4 ± 0.8 0.3 ± 0.1 10.8 ± 0.6 84.2 ± 1.4 5 Low 2 101.4 ± 6.7 2.9 ± 0.3 0.3 ± 0.0 10.8 ± 0.3 86.0 ± 0.6 Walleye Pollock 0 High 3 59.8 ± 4.8 1.3 ± 0.0 0.7 ± 0.0 14.0 ± 0.4 84.0 ± 0.4 2 High 3 55.3 ± 3.3 1.3 ± 0.2 0.7 ± 0.1 14.5 ± 0.7 83.5 ± 0.8 2 Low 1 29.9 ± 0.0 0.0 ± 0.0 4.1 ± 0.0 27.0 ± 0.0 68.9 ± 0.0 5 High 3 92.9 ± 9.4 3.7 ± 1.8 0.8 ± 0.3 12.2 ± 0.4 83.3 ± 2.6 5 Low 3 54.9 ± 10.8 1.6 ± 0.3 0.5 ± 0.1 12.0 ± 0.4 86.0 ± 0.3 12 High 2 74.4 ± 4.4 1.8 ± 0.2 0.4 ± 0.0 14.9 ± 1.5 82.6 ± 0.9 Later stage Arctic Cod 0 High 4 102.6 ± 2.4 5.6 ± 0.3 0.7 ± 0.1 10.2 ± 0.3 83.5 ± 0.6 2 High 3 104.9 ± 5.4 11.5 ± 0.5 0.7 ± 0.1 9.8 ± 0.6 78.0 ± 0.8 2 Low 3 99.0 ± 7.1 6.3 ± 1.0 0.7 ± 0.1 12.3 ± 0.6 80.7 ± 0.7 5 High 3 136.3 ± 5.0 19.0 ± 0.4 0.5 ± 0.1 10.9 ± 0.3 69.6 ± 0.7 5 Low 3 108.7 ± 1.4 14.1 ± 1.1 0.6 ± 0.1 11.0 ± 0.2 74.3 ± 1.3 7 High 4 138.1 ± 9.5 21.8 ± 2.1 0.5 ± 0.1 10.1 ± 0.4 67.6 ± 1.9 Walleye pollock 0 High 2 75.7 ± 0.0 6.8 ± 2.6 1.4 ± 0.2 13.2 ± 1.2 78.3 ± 1.8 2 High 3 113.1 ± 14.6 8.6 ± 1.2 0.9 ± 0.1 11.0 ± 0.6 79.5 ± 1.4 2 Low 3 97.2 ± 4.7 5.6 ± 1.0 0.9 ± 0.2 11.2 ± 0.5 82.2 ± 0.9 5 High 3 128.9 ± 5.9 14.7 ± 0.4 0.9 ± 0.1 11.1 ± 0.3 73.3 ± 0.7 5 Low 3 88.6 ± 4.7 4.3 ± 0.7 1.0 ± 0.3 13.8 ± 0.8 80.9 ± 0.3 9 High 3 126.0 ± 6.0 15.7 ± 2.0 0.6 ± 0.1 9.3 ± 0.5 74.4 ± 2.3 9 Low 3 95.0 ± 5.6 4.3 ± 0.8 0.9 ± 0.2 12.6 ± 0.6 82.2 ± 0.8 12 High 3 104.8 ± 4.6 11.8 ± 2.8 0.7 ± 0.1 9.8 ± 0.3 77.7 ± 2.7 TL, total lipid; TAG, triacylglycerols; FFA, free fatty acids; ST, sterols, PL, polar lipids. Arctic cod at both ontogenetic stages at 9°C are not included due to high mortality in treatment tanks. Data are treatment means ± 1 s.e. (n = 1–4 replicate tanks, depending on mortality). Open in new tab First-feeding Arctic cod were in higher lipid-based condition than first-feeding walleye pollock as evidenced by the significant effect of species (two-way ANOVA; F1,15 = 20.256, p < 0.001) on the TAG:ST of first-feeding larvae in high food ration treatments (Figure 3a and b). At this stage, temperature did not significantly impact the TAG: ST independently or as an interaction. At the later larval stage, a significant species–temperature interaction existed (F1,22 = 17.548, p < 0.001) such that Arctic cod were in better condition than walleye pollock at a given temperature and were also more sensitive to changes in temperature (Figure 3c and d). Figure 3. Open in new tabDownload slide TAG:ST ratio of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae in high food ration treatments at the end of laboratory experiments. First-feeding Arctic cod lipid data at 9°C was not collected because not enough larvae remained at the end of experiments due to high mortality. Data are treatment means ± 1 s.e. (n = 2–4 replicate tanks per treatment). Different letters indicate significant differences according to Tukey’s pairwise comparisons. Figure 3. Open in new tabDownload slide TAG:ST ratio of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae in high food ration treatments at the end of laboratory experiments. First-feeding Arctic cod lipid data at 9°C was not collected because not enough larvae remained at the end of experiments due to high mortality. Data are treatment means ± 1 s.e. (n = 2–4 replicate tanks per treatment). Different letters indicate significant differences according to Tukey’s pairwise comparisons. Finally, the total body lipid storage (lipid per DWT, µg mg−1) generally increased with temperature up to 5°C for first-feeding larvae as evidenced by a significant temperature effect on first-feeding larval total lipids (two-way ANOVA; F1,15 = 7.785, p = 0.014; Figure 4a and b). Later stage Arctic cod lipid storage also increased with temperature, while walleye pollock lipid storage reached a maximum at ∼5°C and then declined slightly resulting in a significant species–temperature interaction term (F1,22 = 12.358, p = 0.002; Figure 4c and d). Figure 4. Open in new tabDownload slide Total lipid (μg lipid mg−1 DWT) of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae in high food ration treatments at the end of laboratory experiments. Data are treatment means ± 1 s.e. (n = 2–4 replicate tanks per treatment). Different letters indicate significant differences according to Tukey’s pairwise comparisons. Figure 4. Open in new tabDownload slide Total lipid (μg lipid mg−1 DWT) of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae in high food ration treatments at the end of laboratory experiments. Data are treatment means ± 1 s.e. (n = 2–4 replicate tanks per treatment). Different letters indicate significant differences according to Tukey’s pairwise comparisons. The temperatures of maximum condition observed for each experiment based on different morphometric and lipid indices are included in Table 2. According to these indices, condition was maximized at 2–5°C and 7°C for first-feeding and later stage Arctic cod, respectively. Walleye pollock larval condition was maximized between 5 and 12°C at both ontogenetic stages. Table 2. Temperatures of maximum condition observed for first-feeding and later stage Arctic cod and walleye pollock based on different morphometric and lipid condition indices. Ontogenetic stage . Species . Condition index . Experimental temperature (°C) . First-Feeding Arctic Cod Length–weight residuals 5 BD:SL 5 TAG:ST 2 Total lipid storage 5 Walleye Pollock Length–weight residuals 12 BD:SL 12 TAG:ST 5 Total lipid storage 5 Later Stage Arctic Cod Length–weight residuals 7 BD:SL 7 TAG:ST 7 Total lipid storage 7 Walleye Pollock Length–weight residuals 12 BD:SL 12 TAG:ST 9 Total lipid storage 5 Ontogenetic stage . Species . Condition index . Experimental temperature (°C) . First-Feeding Arctic Cod Length–weight residuals 5 BD:SL 5 TAG:ST 2 Total lipid storage 5 Walleye Pollock Length–weight residuals 12 BD:SL 12 TAG:ST 5 Total lipid storage 5 Later Stage Arctic Cod Length–weight residuals 7 BD:SL 7 TAG:ST 7 Total lipid storage 7 Walleye Pollock Length–weight residuals 12 BD:SL 12 TAG:ST 9 Total lipid storage 5 Open in new tab Table 2. Temperatures of maximum condition observed for first-feeding and later stage Arctic cod and walleye pollock based on different morphometric and lipid condition indices. Ontogenetic stage . Species . Condition index . Experimental temperature (°C) . First-Feeding Arctic Cod Length–weight residuals 5 BD:SL 5 TAG:ST 2 Total lipid storage 5 Walleye Pollock Length–weight residuals 12 BD:SL 12 TAG:ST 5 Total lipid storage 5 Later Stage Arctic Cod Length–weight residuals 7 BD:SL 7 TAG:ST 7 Total lipid storage 7 Walleye Pollock Length–weight residuals 12 BD:SL 12 TAG:ST 9 Total lipid storage 5 Ontogenetic stage . Species . Condition index . Experimental temperature (°C) . First-Feeding Arctic Cod Length–weight residuals 5 BD:SL 5 TAG:ST 2 Total lipid storage 5 Walleye Pollock Length–weight residuals 12 BD:SL 12 TAG:ST 5 Total lipid storage 5 Later Stage Arctic Cod Length–weight residuals 7 BD:SL 7 TAG:ST 7 Total lipid storage 7 Walleye Pollock Length–weight residuals 12 BD:SL 12 TAG:ST 9 Total lipid storage 5 Open in new tab The significant one-way ANOVA (temperature effects) and two-way ANOVA (temperature and food ration effects) results across high food ration treatments are summarized in Tables 3 and 4, respectively. Table 3. Summary of p-values from one-way ANOVAs assessing the effect of temperature on larval condition across high food ration treatments according to different morphometric condition indices. Ontogenetic stage . Species . Length–weight residuals . BD:SL . First-feeding Arctic cod 0.249 0.080 Walleye pollock 0.213 <0.001 Later stage Arctic cod 0.152 <0.001 Walleye pollock 0.258 <0.001 Ontogenetic stage . Species . Length–weight residuals . BD:SL . First-feeding Arctic cod 0.249 0.080 Walleye pollock 0.213 <0.001 Later stage Arctic cod 0.152 <0.001 Walleye pollock 0.258 <0.001 Significant results (α = 0.05) for each condition index are indicated in bold. Open in new tab Table 3. Summary of p-values from one-way ANOVAs assessing the effect of temperature on larval condition across high food ration treatments according to different morphometric condition indices. Ontogenetic stage . Species . Length–weight residuals . BD:SL . First-feeding Arctic cod 0.249 0.080 Walleye pollock 0.213 <0.001 Later stage Arctic cod 0.152 <0.001 Walleye pollock 0.258 <0.001 Ontogenetic stage . Species . Length–weight residuals . BD:SL . First-feeding Arctic cod 0.249 0.080 Walleye pollock 0.213 <0.001 Later stage Arctic cod 0.152 <0.001 Walleye pollock 0.258 <0.001 Significant results (α = 0.05) for each condition index are indicated in bold. Open in new tab Table 4. Summary of p-values from two-way ANOVAs assessing the effects of temperature and species on larval condition across high food ration treatments according to different lipid condition indices. Ontogenetic stage . Results . TAG:ST . Total lipid storage . First-feeding Species < 0.001 0.146 Temperature 0.733 0.014 Species*Temperature 0.577 0.128 Later stage Species 0.293 0.707 Temperature < 0.001 0.066 Species*Temperature < 0.001 0.002 Ontogenetic stage . Results . TAG:ST . Total lipid storage . First-feeding Species < 0.001 0.146 Temperature 0.733 0.014 Species*Temperature 0.577 0.128 Later stage Species 0.293 0.707 Temperature < 0.001 0.066 Species*Temperature < 0.001 0.002 Significant results (α = 0.05) for each condition index are indicated in bold. Open in new tab Table 4. Summary of p-values from two-way ANOVAs assessing the effects of temperature and species on larval condition across high food ration treatments according to different lipid condition indices. Ontogenetic stage . Results . TAG:ST . Total lipid storage . First-feeding Species < 0.001 0.146 Temperature 0.733 0.014 Species*Temperature 0.577 0.128 Later stage Species 0.293 0.707 Temperature < 0.001 0.066 Species*Temperature < 0.001 0.002 Ontogenetic stage . Results . TAG:ST . Total lipid storage . First-feeding Species < 0.001 0.146 Temperature 0.733 0.014 Species*Temperature 0.577 0.128 Later stage Species 0.293 0.707 Temperature < 0.001 0.066 Species*Temperature < 0.001 0.002 Significant results (α = 0.05) for each condition index are indicated in bold. Open in new tab Temperature–prey interactive effects The interactive effects of temperature and food ration were assessed at intermediate temperatures (2 and 5°C) where food rations were manipulated. The logarithmic–transformed linear relationships between SL and DWT of larval fish at intermediate temperatures (2 and 5°C) and high and low food rations (Figure 5a, c, e, and g) were described by linear regression equations for first-feeding Arctic cod, first-feeding walleye pollock, later stage Arctic cod, and later stage walleye pollock. Analysis of morphometric condition (length–weight residuals) of first-feeding Arctic cod revealed no statistically significant independent or interactive effects of temperature and food ration (two-way ANOVA; Figure 5b). Similarly, the effects of temperature and food ration were not significant for first-feeding walleye pollock (Figure 5d). Figure 5. Open in new tabDownload slide Linear relationships between log-transformed length (SL, mm) and log-transformed DWT (mg) of (a) first-feeding Arctic cod, (c) first-feeding walleye pollock, (e) later stage Arctic cod, and (g) later stage walleye pollock larvae and the residuals of this relationship for (b) first-feeding Arctic cod, (d) first-feeding walleye pollock, (f) later stage Arctic cod, and (h) later stage walleye pollock larvae in high and low food ration treatments at the end of laboratory experiments. Linear regressions were fitted to mean tank data. Length–weight residuals were computed for each tank and plotted as treatment means ± 1 s.e. (n = 3 replicate tanks/treatment). Figure 5. Open in new tabDownload slide Linear relationships between log-transformed length (SL, mm) and log-transformed DWT (mg) of (a) first-feeding Arctic cod, (c) first-feeding walleye pollock, (e) later stage Arctic cod, and (g) later stage walleye pollock larvae and the residuals of this relationship for (b) first-feeding Arctic cod, (d) first-feeding walleye pollock, (f) later stage Arctic cod, and (h) later stage walleye pollock larvae in high and low food ration treatments at the end of laboratory experiments. Linear regressions were fitted to mean tank data. Length–weight residuals were computed for each tank and plotted as treatment means ± 1 s.e. (n = 3 replicate tanks/treatment). For first-feeding larvae, a significant effect of temperature on the BD:SL ratio was detected for both Arctic cod (two-way ANOVA; F1,8 = 15.056, p = 0.005; Figure 6a) and walleye pollock (two-way ANOVA; F1,8 = 109.682, p < 0.001; Figure 6b). The BD:SL was higher at 5°C than at 2°C for both first-feeding species. The effect of food ration on the BD:SL was significant for first-feeding Arctic cod (F1,8 = 33.808, p < 0.001), but not for walleye pollock (F1,8 = 2.308, p = 0.167). At both temperatures, Arctic cod larvae receiving high food rations were in better condition than those receiving low food rations. At the later larval stage, an independent food ration effect was detected for both Arctic cod (F1,8 = 48.922, p < 0.001; Figure 6c) and walleye pollock (F1,8 = 22.205, p = 0.002; Figure 6d). The BD:SL for both later stage species was higher under high food ration treatments than low food ration treatments. Unlike with first-feeding experiments, temperature sensitivity among later stage larvae varied with species. Temperature significantly affected the BD:SL of later stage Arctic cod (F1,8 = 26.338, p < 0.001), but not walleye pollock (F1,8 = 1.189, p = 0.307). Figure 6. Open in new tabDownload slide BD:SL of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae receiving high and low food rations at the end of laboratory experiments. Data are treatment means ± 1 s.e. (n = 3 replicate tanks per treatment). Figure 6. Open in new tabDownload slide BD:SL of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae receiving high and low food rations at the end of laboratory experiments. Data are treatment means ± 1 s.e. (n = 3 replicate tanks per treatment). Total lipid and proportional lipid class composition for larvae in 2 and 5°C treatments (Table 1) followed the same patterns as those detailed above. The largest proportion of each sample was comprised of PL, with first-feeding larvae containing more PL (82.9%) than later stage larvae (77.3%) on average. Once again, the opposite was true with the TAG proportion, which was lower in first-feeding larvae (2.5%) than later larval stage (10.5%). Both FFA (first-feeding: 1.0%; later stage: 0.8%) and ST (first-feeding: 13.5%; later stage: 11.4%) proportions remained relatively constant between stages. For first-feeding larvae, significant independent effects of species (F1,12 = 9.492, p = 0.010), temperature (F1,12 = 5.010, p = 0.045), and food ration (F1,12 = 11.211, p = 0.006) acting on the TAG:ST condition index was detected (three-way ANOVA; Figure 7a and b). Under the same experimental conditions (temperature and food ration), Arctic cod were in higher condition than walleye pollock at this stage. Additionally, nutritional condition of first-feeding larvae increased with temperature and with food ration. At the later larval stage, a significant three-way interaction between species, temperature, and food ration (F1,16 = 13.178, p = 0.002) on the TAG:ST was detected (Figure 7c and d). To further investigate this interaction, separate two-way ANOVAs considering temperature and food ration were conducted for each species. For later stage Arctic cod, the independent effects of temperature (F1,8 = 59.370, p < 0.001) and food ration (F1,8 = 40.494, p < 0.001) were statistically significant, but an interaction was not supported (Figure 7c). Larvae were in higher condition when receiving high food rations and in higher temperature treatments. Conversely, an interaction between temperature and food ration (F1,8 = 17.243, p = 0.003) was detected for later stage walleye pollock such that food sensitivity was significantly higher at 5°C than at 2°C (Figure 7d). Figure 7. Open in new tabDownload slide TAG:ST of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae receiving high and low food rations at the end of laboratory experiments. Data are treatment means ± 1 s.e. (n = 2–3 replicate tanks per treatment, except for firstfeeding walleye pollock at 2°C and low food ration where n = 1). Figure 7. Open in new tabDownload slide TAG:ST of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae receiving high and low food rations at the end of laboratory experiments. Data are treatment means ± 1 s.e. (n = 2–3 replicate tanks per treatment, except for firstfeeding walleye pollock at 2°C and low food ration where n = 1). Furthermore, a three-way ANOVA revealed a significant interactive effect of species and food ration (F1,12 = 8.120, p = 0.015), in addition to an independent temperature effect (F1,12 = 7.943, p = 0.016), acting on the total body lipid storage of first-feeding larvae (Figure 8a and b). For both species, total lipid per DWT increased with temperature. Additionally, first-feeding walleye pollock lipid storage was more sensitive to changes in food ration than first-feeding Arctic cod. At the later larval stage, a temperature-food ration interaction (F1,16 = 5.340, p = 0.035) was demonstrated by increased food sensitivity at higher temperatures for both species (Figure 8c and d). Figure 8. Open in new tabDownload slide Total lipid (μg lipid mg−1 DWT) of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae under high and low food rations at the end of laboratory experiments. Data are treatment means ± 1 s.e. (n = 2–3 replicate tanks per treatment, except for first-feeding walleye pollock at 2°C receiving a low food ration where n = 1). Figure 8. Open in new tabDownload slide Total lipid (μg lipid mg−1 DWT) of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae under high and low food rations at the end of laboratory experiments. Data are treatment means ± 1 s.e. (n = 2–3 replicate tanks per treatment, except for first-feeding walleye pollock at 2°C receiving a low food ration where n = 1). The significant two-way ANOVA (temperature and food ration effects) and three-way ANOVA (temperature, food ration, and species effects) results at intermediate temperatures (2 and 5°C) are summarized in Tables 5 and 6, respectively. Table 5. Summary of p-values from two-way ANOVAs assessing the effects of temperature and food ration on larval condition across high and low food ration treatments at intermediate temperatures (2 and 5°C) according to different morphometric condition indices. Ontogenetic stage . Species . Results . Length–weight residuals . BD:SL . First-Feeding Arctic Cod Temperature 0.840 0.005 Food Ration 0.301 <0.001 Temperature*Food Ration 0.925 0.985 Walleye Pollock Temperature 0.366 <0.001 Food Ration 0.719 0.167 Temperature*Food Ration 0.172 0.068 Later Stage Arctic Cod Temperature 0.525 <0.001 Food Ration 0.418 <0.001 Temperature*Food Ration 0.567 0.395 Walleye Pollock Temperature 0.120 0.307 Food Ration 0.571 0.002 Temperature*Food Ration 0.062 0.254 Ontogenetic stage . Species . Results . Length–weight residuals . BD:SL . First-Feeding Arctic Cod Temperature 0.840 0.005 Food Ration 0.301 <0.001 Temperature*Food Ration 0.925 0.985 Walleye Pollock Temperature 0.366 <0.001 Food Ration 0.719 0.167 Temperature*Food Ration 0.172 0.068 Later Stage Arctic Cod Temperature 0.525 <0.001 Food Ration 0.418 <0.001 Temperature*Food Ration 0.567 0.395 Walleye Pollock Temperature 0.120 0.307 Food Ration 0.571 0.002 Temperature*Food Ration 0.062 0.254 Significant results (α = 0.05) for each condition index are indicated in bold. Open in new tab Table 5. Summary of p-values from two-way ANOVAs assessing the effects of temperature and food ration on larval condition across high and low food ration treatments at intermediate temperatures (2 and 5°C) according to different morphometric condition indices. Ontogenetic stage . Species . Results . Length–weight residuals . BD:SL . First-Feeding Arctic Cod Temperature 0.840 0.005 Food Ration 0.301 <0.001 Temperature*Food Ration 0.925 0.985 Walleye Pollock Temperature 0.366 <0.001 Food Ration 0.719 0.167 Temperature*Food Ration 0.172 0.068 Later Stage Arctic Cod Temperature 0.525 <0.001 Food Ration 0.418 <0.001 Temperature*Food Ration 0.567 0.395 Walleye Pollock Temperature 0.120 0.307 Food Ration 0.571 0.002 Temperature*Food Ration 0.062 0.254 Ontogenetic stage . Species . Results . Length–weight residuals . BD:SL . First-Feeding Arctic Cod Temperature 0.840 0.005 Food Ration 0.301 <0.001 Temperature*Food Ration 0.925 0.985 Walleye Pollock Temperature 0.366 <0.001 Food Ration 0.719 0.167 Temperature*Food Ration 0.172 0.068 Later Stage Arctic Cod Temperature 0.525 <0.001 Food Ration 0.418 <0.001 Temperature*Food Ration 0.567 0.395 Walleye Pollock Temperature 0.120 0.307 Food Ration 0.571 0.002 Temperature*Food Ration 0.062 0.254 Significant results (α = 0.05) for each condition index are indicated in bold. Open in new tab Table 6. Summary of p-values from three-way ANOVAs assessing the effects of species, temperature, and food ration on larval condition across high and low food ration treatments at intermediate temperatures (2 and 5°C) according to different lipid condition indices. Ontogenetic stage . Results . TAG:ST . Total lipid storage . First-Feeding Species 0.010 <0.001 Temperature 0.045 0.016 Food Ration 0.006 0.044 Species*Temperature 0.125 0.061 Species*Food Ration 0.396 0.015 Temperature*Food Ration 0.510 0.890 Species*Temperature*Food Ration 0.289 0.348 Later Stage Species < 0.001 0.310 Temperature <0.001 0.036 Food Ration <0.001 <0.001 Species*Temperature <0.001 0.081 Species*Food Ration 0.475 0.272 Temperature*Food Ration 0.026 0.035 Species*Temperature*Food Ration 0.002a 0.981 Ontogenetic stage . Results . TAG:ST . Total lipid storage . First-Feeding Species 0.010 <0.001 Temperature 0.045 0.016 Food Ration 0.006 0.044 Species*Temperature 0.125 0.061 Species*Food Ration 0.396 0.015 Temperature*Food Ration 0.510 0.890 Species*Temperature*Food Ration 0.289 0.348 Later Stage Species < 0.001 0.310 Temperature <0.001 0.036 Food Ration <0.001 <0.001 Species*Temperature <0.001 0.081 Species*Food Ration 0.475 0.272 Temperature*Food Ration 0.026 0.035 Species*Temperature*Food Ration 0.002a 0.981 Significant results (α = 0.05) for each condition index are indicated in bold. a Additional separate two-way ANOVAs (temp and food ration) for each later stage species revealed significant temperature (p < 0.001) and food ration (p < 0.001) effects on Arctic cod condition and a significant temperature*food ration interaction (p = 0.003) acting on walleye pollock condition. Open in new tab Table 6. Summary of p-values from three-way ANOVAs assessing the effects of species, temperature, and food ration on larval condition across high and low food ration treatments at intermediate temperatures (2 and 5°C) according to different lipid condition indices. Ontogenetic stage . Results . TAG:ST . Total lipid storage . First-Feeding Species 0.010 <0.001 Temperature 0.045 0.016 Food Ration 0.006 0.044 Species*Temperature 0.125 0.061 Species*Food Ration 0.396 0.015 Temperature*Food Ration 0.510 0.890 Species*Temperature*Food Ration 0.289 0.348 Later Stage Species < 0.001 0.310 Temperature <0.001 0.036 Food Ration <0.001 <0.001 Species*Temperature <0.001 0.081 Species*Food Ration 0.475 0.272 Temperature*Food Ration 0.026 0.035 Species*Temperature*Food Ration 0.002a 0.981 Ontogenetic stage . Results . TAG:ST . Total lipid storage . First-Feeding Species 0.010 <0.001 Temperature 0.045 0.016 Food Ration 0.006 0.044 Species*Temperature 0.125 0.061 Species*Food Ration 0.396 0.015 Temperature*Food Ration 0.510 0.890 Species*Temperature*Food Ration 0.289 0.348 Later Stage Species < 0.001 0.310 Temperature <0.001 0.036 Food Ration <0.001 <0.001 Species*Temperature <0.001 0.081 Species*Food Ration 0.475 0.272 Temperature*Food Ration 0.026 0.035 Species*Temperature*Food Ration 0.002a 0.981 Significant results (α = 0.05) for each condition index are indicated in bold. a Additional separate two-way ANOVAs (temp and food ration) for each later stage species revealed significant temperature (p < 0.001) and food ration (p < 0.001) effects on Arctic cod condition and a significant temperature*food ration interaction (p = 0.003) acting on walleye pollock condition. Open in new tab Discussion Our understanding of gadid condition in response to environmental factors largely stems from studies of Atlantic cod, Gadus morhua (e.g. Neilson et al., 1986; Suthers et al., 1992; Lochmann et al., 1995) making direct inferences to polar species unsuitable. Furthermore, while both morphological (e.g. Brodeur et al., 2000; Laurel et al., 2016) and lipid (e.g. Heintz et al., 2013; Copeman et al., 2017) indices have been used separately to assess the condition of early life stage gadids, this study is unique in that it employs the use of multiple indices simultaneously to assess the relative strength and sensitivity of each. This study provides a direct examination of the morphometric and lipid condition of larval Arctic cod and walleye pollock under different temperature-food ration scenarios. Analyses from this study indicate that: (i) temperature directly impacts the larval condition of Arctic cod and walleye pollock, (ii) effects on larval condition (i.e. BD:SL, TAG:ST, and total lipid storage) vary with both temperature and food availability in a species-dependent manner, and (iii) morphometric and lipid indices should be used in combination, when possible, to sufficiently reflect changes in larval condition across temperature-food ration scenarios. Temperature effects on larval condition Morphometric condition based on length–weight residuals was not statistically sensitive to temperature differences. This could be partially due to the high variability between replicate tanks, particularly at the more extreme temperatures. At these temperatures, variability between tanks may have exceeded variability between treatments to the extent that treatment differences were not detectable. The BD:SL condition index generally increased with temperature in all experiments, except for first-feeding Arctic cod where condition was compromised at high temperatures. It has been suggested that at early life stages, fish growth in length precedes an increase in mass (Farbridge and Leatherland, 1987; Ferron and Leggett, 1994). As such, the fish at higher temperatures in this study were growing faster and had therefore likely begun to transition from lengthwise growth to an increase in BD and mass. Collectively, study metrics from Koenker et al. (this issue) and this work indicate that the overall well-being (manifested in terms of survival, growth, and condition) of first-feeding Arctic cod is reduced at 9°C. First-feeding and later stage Arctic cod were in highest morphometric condition (length–weight residuals and BD:SL) at 5 and 7°C, respectively, which coincides with temperatures of maximum growth (Koenker et al., this issue). In summer months, Arctic cod are commonly associated with thermal-salinity fronts ranging from 2 to 9°C in the nearshore Beaufort Sea (Moulton and Tarbox, 1987). This study indicates that these regions may provide optimal thermal habitat for larval Arctic cod, allowing for the maximization of growth and nutritional condition when food is abundant. Though the TAG:ST condition of later larval stages of both species were highly sensitive to changes in temperature, it was surprising that first-feeding larval TAG: ST condition did not demonstrate a relationship with temperature despite a general increase in total lipid storage. This largely relates to the relative proportion of different lipid classes accumulated at different ontogenetic stages. The proportion of TAG in first-feeding larvae (2.9%) was substantially lower than in later stage larvae (12.8%). Instead, first-feeding larval lipid composition was made up of considerably more PL. The TAG:ST condition index is sensitive to changes in temperature at the later larval stage but it may be that it should not be employed with first-feeding larvae that are unable to accumulate sufficient TAG to contribute to the TAG:ST index. Although the BD:SL condition index demonstrated a general increase in condition with temperature, the total lipid storage and later stage TAG:ST reveal unique temperature responses between species. As has been demonstrated for juvenile Arctic cod, the lipid storage of larval Arctic cod in this study increased with temperature over the range where individuals survived. In this case, lipid storage was highest at 5–7°C, with elevated mortality inhibiting lipid analyses above these temperatures. Walleye pollock lipid storage demonstrated a dome-shaped response to temperature with highest measured lipid storage at intermediate temperatures (5–9°C). This trend has been shown with boreal gadid species at the juvenile stage (Copeman et al., 2017) highlighting the high sensitivity of lipid storage to environmental temperature, that was not evident from body mass measures alone. Furthermore, according to the TAG:ST condition index, Arctic cod were in higher condition than walleye pollock at comparable temperatures. It has been suggested that increased lipid storage by juvenile Arctic cod, relative to boreal gadids, may indicate a life history strategy for prolonged overwintering (Copeman et al., 2017) in environments where winter temperatures are <0°C (Bouchard and Fortier, 2011). The results of this study provide evidence that rapid lipid accumulation by Arctic cod can be observed in the early larval stages. Based on these collective indices, maximum condition for Arctic cod was observed at 2–5°C for first-feeding larvae and at 7°C for later stage larvae. Summer observations from the field indicate a high abundance of larval Arctic cod in the northeastern Chukchi Sea between 0.3 and 5.9°C, with a lower abundance found in the northern Bering Sea and southern Chuckchi Sea between 0.7 and 9.7°C (Kono et al., 2016). This larval distribution pattern is in line with the temperatures of maximum condition reported in this study. Interactive temperature-prey effects on condition The residuals from the regression between log-transformed SL and log-transformed DWT can be interpreted as deviations from an average length–weight condition. This study demonstrates that the residuals of this relationship do not provide an adequate condition measure for assessing the impacts of food availability on larval fish species. Length–weight residuals did not detect a significant food ration effect for either species at any larval stage. Relative to other morphometric variables, BD has been found to be particularly responsive to starvation (Ehrlich et al., 1976; Diaz et al., 2013). In addition to detecting temperature effects throughout most larval experiments, the BD:SL index detected food ration effects on first-feeding Arctic cod condition and both later stage species. First-feeding Arctic cod were more food sensitive than first-feeding walleye pollock, which was contrary to expectation. At hatch, Arctic cod are larger and contain higher yolk reserves that led to the expectation that Arctic cod would be less susceptible to starvation at the onset of feeding. However, it may be that the observed food effect was due to improved foraging and feeding ability of Arctic cod at hatch (relative to walleye pollock) which allowed those under high food ration treatments to take better advantage of the abundant food environment than their low food ration counterparts. In first-feeding larvae across all high food ration treatments, the TAG:ST condition index did not indicate a significant relationship with temperature, though it proved to be sensitive to changes in both temperature and food availability at 2 and 5°C. At both stages, Arctic cod were in higher condition than walleye pollock and nutritional condition increased with temperature and food ration. Higher sensitivity to changes in food quantity at 5°C than at 2°C was expected due to elevated metabolic rates at higher temperatures. This pattern was demonstrated by later stage Arctic cod and walleye pollock through one or more lipid condition metrics. Throughout the Arctic, declining sea ice is expected to alter the primary productivity regime, potentially resulting in a mismatch between key Arctic grazers and the production of high-quality food (Søreide et al., 2010; Leu et al., 2011). It has been hypothesized that inter-annual variability in ocean conditions and shifting circulation patterns on the Chukchi Sea shelf may alter the distribution of large energy-rich Arctic zooplankton species (e.g. Calanus hyperboreus, Calanus glacialis) and increase the contribution of Pacific copepods (e.g. Neocalanus sp.) to the diets of Arctic fish species (Pinchuk and Eisner, 2017). Under this scenario, warm temperatures combined with altered timing and availability of lipid-rich prey may interact to decrease the condition of the larval gadid assemblage. Considerations for larval fish condition indices A comparison of the morphometric and lipid condition indices used in this study, suggests that multiple indices should be utilized, if possible, to assess the sensitivity of larval cod species to environmental conditions. Previous work with juvenile fish (e.g. Gilliers et al., 2004; Walther et al., 2010; Stowell, 2016) has demonstrated the limitations of using a single index to fully explain the impacts of environmental conditions on fish condition. Morphometric indices offer a logistic advantage, as they are often easier to obtain. However, the larval stage is characterised by rapid morphological changes (e.g. transition to exogenous feeding) (Neilson et al., 1986) which can make it difficult to differentiate between developmental changes and those related to environmental conditions (Ferron and Leggett, 1994). Results from this study indicate that the BD:SL was a good morphometric indicator of condition as it effectively detected species-specific responses to temperature and food availability at both larval stages. It has been suggested that lipid indices are more responsive to short-term change (Lochmann et al., 1995) and to physiological stress in some cases (Copeman et al., 2008). In this study, lipid indices were highly sensitive to species-dependent responses to changes in temperature and food availability at the later larval stage. However, it was found that first-feeding larvae were not yet feeding at a sufficient level to accumulate TAG and, therefore, the TAG: ST condition index was not suitable. Though first-feeding larvae did not appear to accumulate TAG, total lipid storage at this stage was sensitive to changes in temperature and food availability. Conclusions This study suggests that the nutritional condition of larval gadids is highly sensitive to changes in temperature and food availability, and that larval condition is highly variable between species. In the Canadian High Arctic, it has been hypothesized that ongoing warming in the short term will increase recruitment of age-0 Arctic cod in the presence of adequate prey availability (Bouchard et al., 2017). However, in regions like the Chukchi Sea, where ocean warming is most pronounced, rising temperatures may interact with reduced availability of lipid-rich prey to decrease the condition of larval gadids. Species-specific impacts to the lipid storage and condition of larval gadids associated with climate change will influence the quality of the forage fish assemblage in the Arctic. These species play a critical role in influencing energy transfer to upper trophic levels and changes in their lipid storage and condition will likely be recognized throughout the lipid-rich Arctic marine food web. Acknowledgements We would like to thank the NOAA-AFSC staff at the Hatfield Marine Science Center for the use of facilities, logistical support, and guidance for this work. Thanks to Paul Iseri for assistance with lab construction, tank design, and maintenance of the experimental setup. Also, thanks to Scott Haines, Michele Ottmar, and Eric Hanneman for assistance in larval husbandry and live food production. This project was supported with funding from the North Pacific Research Board (NPRB) Grant no. 1403 and this is NPRB publication no. 665. The findings and conclusions in the paper are those of the authors and do not necessarily represent the views of the National Marine Fisheries Service, NOAA. References Bacheler N. M. , Ciannelli L., Bailey K. M., Duffy-Anderson J. T. 2010 . Spatial and temporal patterns of walleye pollock (Theragra chalcogramma) spawning in the eastern Bering Sea inferred from egg and larval distributions . Fisheries Oceanography , 19 : 107 – 120 . Google Scholar Crossref Search ADS WorldCat Bluhm B. A. , Gradinger R. 2008 . Regional variability in food availability for Arctic marine mammals . Ecological Applications , 18 : S77 – S96 . Google Scholar Crossref Search ADS PubMed WorldCat Bolger T. , Connolly P. L. 1989 . The selection of suitable indices for the measurement and analysis of fish condition . Journal of Fish Biology , 34 : 171 – 182 . Google Scholar Crossref Search ADS WorldCat Bouchard C. , Fortier L. 2011 . Circum-arctic comparison of the hatching season of polar cod Boreogadus saida: a test of the freshwater winter refuge hypothesis . Progress in Oceanography , 90 : 105 – 116 . Google Scholar Crossref Search ADS WorldCat Bouchard C. , Geoffroy M., LeBlanc M., Majewski A., Gauthier S., Walkusz W., Reist J. D. et al. 2017 . Climate warming enhances polar cod recruitment, at least transiently . Progress in Oceanography , 156 : 121 – 129 . Google Scholar Crossref Search ADS WorldCat Brodeur R. D. , Wilson M. T., Ciannelli L. 2000 . Spatial and temporal variability in feeding and condition of age-0 walleye pollock (Theragra chalcogramma) in frontal regions of the Bering Sea . ICES Journal of Marine Science , 57 : 256 – 264 . Google Scholar Crossref Search ADS WorldCat Cheung W. W. L. , Lam V. W. Y., Sarmiento J. L., Kearney K., Watson R., Pauly D. 2009 . Projecting global marine biodiversity impacts under climate change scenarios . Fish and Fisheries , 10 : 235 – 251 . Google Scholar Crossref Search ADS WorldCat Christiansen J. S. , Mecklenburg C. W., Karamushko O. V. 2014 . Arctic marine fishes and their fisheries in light of global change . Global Change Biology , 20 : 352 – 359 . Google Scholar Crossref Search ADS PubMed WorldCat Copeman L. A. , Parrish C. C., Brown J. A., Harel M. 2002 . Effects of docosahexaenoic, eicosapentaenoic, and arachidonic acids on the early growth, survival, lipid composition and pigmentation of yellowtail flounder (Limanda ferruginea): a live food enrichment experiment . Aquaculture , 210 : 285 – 304 . Google Scholar Crossref Search ADS WorldCat Copeman L. A. , Parrish C. C., Gregory R. S., Wells J. S. 2008 . Decreased lipid storage in juvenile Atlantic cod (Gadus morhua) during settlement in cold-water eelgrass habitat . Marine Biology , 154 : 823 – 832 . Google Scholar Crossref Search ADS WorldCat Copeman L. A. , Laurel B. J. 2010 . Experimental evidence of fatty acid limited growth and survival in Pacific cod larvae . Marine Ecology Progress Series , 412 : 259 – 272 . Google Scholar Crossref Search ADS WorldCat Copeman L. A. , Laurel B. J., Boswell K. M., Sremba A. L., Klinck K., Heintz R. A., Vollenweider J. J. et al. 2016 . Ontogenetic and spatial variability in trophic biomarkers of juvenile saffron cod (Eleginus gracilis) from the Beaufort, Chukchi and Bering Seas . Polar Biology , 39 : 1109 – 1126 . Google Scholar Crossref Search ADS WorldCat Copeman L. A. , Laurel B. J., Spencer M., Sremba A. 2017 . Temperature impacts on lipid allocation among juvenile gadid species at the Pacific Arctic-Boreal interface: an experimental laboratory approach . Marine Ecology Progress Series , 566 : 183 – 198 . Google Scholar Crossref Search ADS WorldCat Diaz M. V. , Arano M. F., Pájaro M., Aristizábal E. O., Macchi G. J., Diaz M. V., Arano M. F., et al. 2013 . The use of morphological and histological features as nutritional condition indices of Pagrus pagrus larvae . Neotropical Ichthyology , 11 : 649 – 660 . Google Scholar Crossref Search ADS WorldCat Ehrlich K. F. , Blaxter J. H. S., Pemberton R. 1976 . Morphological and histological changes during the growth and starvation of herring and plaice larvae . Marine Biology , 35 : 105 – 118 . Google Scholar Crossref Search ADS WorldCat Falk-Petersen S. , Pavlov V., Timofeev S., Sargent J. R. 2007 . Climate variability and possible effects on arctic food chains: The role of Calanus. In Arctic Alpine Ecosystems and People in a Changing Environment, Pp. 147–166. Ed. by Ørbæk D. J. B., Kallenborn D. R., Tombre D. I., Hegseth D. E. N., Falk-Petersen D. S., Hoel D. A. H. Springer , Berlin, Heidelberg . 434 pp. Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Farbridge K. J. , Leatherland J. F. 1987 . Lunar cycles of coho salmon, Oncorhynchus kisutch. I. Growth and feeding . Journal of Experimental Biology , 129 : 165 – 178 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Ferron A. , Leggett W. C. 1994 . An appraisal of condition measures for marine fish larvae. In Advances in Marine Biology , pp. 217 – 303 . Ed. by Blaxter J. H. S., Southward A. J.. Elsevier , London . 364 pp. Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Folch J. , Lees M., Sloane-Stanley G. H. 1957 . A simple method for the isolation and purification of total lipids from animal tissues . Journal of Biological Chemistry , 226 : 497 – 509 . Google Scholar PubMed OpenURL Placeholder Text WorldCat Folkvord A. , Ystanes L., Johannessen A., Moksness E. 1996 . RNA: dNA ratios and growth of herring (Clupea harengus) larvae reared in mesocosms . Marine Biology , 126 : 591 – 602 . Google Scholar Crossref Search ADS WorldCat Fortier L. , Sirois P., Michaud J., Barber D. 2006 . Survival of Arctic cod larvae (Boreogadus saida) in relation to sea ice and temperature in the Northeast Water Polynya (Greenland Sea) . Canadian Journal of Fisheries and Aquatic Sciences , 63 : 1608 – 1616 . Google Scholar Crossref Search ADS WorldCat Fossheim M. , Primicerio R., Johannesen E., Ingvaldsen R. B., Aschan M. M., Dolgov A. V. 2015 . Recent warming leads to a rapid borealization of fish communities in the Arctic . Nature Climate Change , 5 : 673 – 677 . Google Scholar Crossref Search ADS WorldCat Fraser A. J. 1989 . Triacylglycerol content as a condition index for fish, bivalve, and crustacean larvae . Canadian Journal of Fisheries and Aquatic Sciences , 46 : 1868 – 1873 . Google Scholar Crossref Search ADS WorldCat Gilliers C. , Amara R., Bergeron J.-P., Le Pape O. 2004 . Comparison of growth and condition indices of juvenile flatfish in different coastal nursery grounds . Environmental Biology of Fishes , 71 : 189 – 198 . Google Scholar Crossref Search ADS WorldCat Graham M. , Hop H. 1995 . Aspects of reproduction and larval biology of Arctic cod (Boreogadus saida) . Arctic , 48 : 130 – 135 . Google Scholar Crossref Search ADS WorldCat Heintz R. A. , Siddon E. C., Farley E. V., Jr, Napp J. M. 2013 . Correlation between recruitment and fall condition of age-0 pollock (Theragra chalcogramma) from the eastern Bering Sea under varying climate conditions . Deep Sea Research Part II: Topical Studies in Oceanography , 94 : 150 – 156 . Google Scholar Crossref Search ADS WorldCat Hoegh-Guldberg O. , Bruno J. F. 2010 . The impact of climate change on the world’s marine ecosystems . Science , 328 : 1523 – 1528 . Google Scholar Crossref Search ADS PubMed WorldCat Hollowed A. B. , Planque B., Loeng H. 2013 . Potential movement of fish and shellfish stocks from the sub-Arctic to the Arctic Ocean . Fisheries Oceanography , 22 : 355 – 370 . Google Scholar Crossref Search ADS WorldCat Houde E. D. 2008 . Emerging from Hjort’s shadow . Journal of Northwest Atlantic Fishery Science , 41 : 53 – 70 . Google Scholar Crossref Search ADS WorldCat Hurst T. P. 2007 . Causes and consequences of winter mortality in fishes . Journal of Fish Biology , 71 : 315 – 345 . Google Scholar Crossref Search ADS WorldCat Jobling M. 1988 . A review of the physiological and nutritional energetics of cod, Gadus morhua L., with particular reference to growth under farmed conditions . Aquaculture , 70 : 1 – 19 . Google Scholar Crossref Search ADS WorldCat Jones R. E. , Petrell R. J., Pauly D. 1999 . Using modified length–weight relationships to assess the condition of fish . Aquacultural Engineering , 20 : 261 – 276 . Google Scholar Crossref Search ADS WorldCat Kono Y. , Sasaki H., Kurihara Y., Fujiwara A., Yamamoto J., Sakurai Y. 2016 . Distribution pattern of Polar cod (Boreogadus saida) larvae and larval fish assemblages in relation to oceanographic parameters in the northern Bering Sea and Chukchi Sea . Polar Biology , 39 : 1039 – 1048 . Google Scholar Crossref Search ADS WorldCat Laurel B. J. , Spencer M., Iseri P., Copeman L. A. 2016 . Temperature-dependent growth and behavior of juvenile Arctic cod (Boreogadus saida) and co-occurring North Pacific gadids . Polar Biology , 39 : 1127 – 1135 . Google Scholar Crossref Search ADS WorldCat Letcher B. H. , Rice J. A., Crowder L. B., Rose K. A. 1996 . Variability in survival of larval fish: disentangling components with a generalized individual-based model . Canadian Journal of Fisheries and Aquatic Sciences , 53 : 787 – 801 . Google Scholar Crossref Search ADS WorldCat Leu E. , Søreide J. E., Hessen D. O., Falk-Petersen S., Berge J. 2011 . Consequences of changing sea-ice cover for primary and secondary producers in the European Arctic shelf seas: timing, quantity, and quality . Progress in Oceanography , 90 : 18 – 32 . Google Scholar Crossref Search ADS WorldCat Litzow M. A. , Bailey K. M., Prahl F. G., Heintz R. 2006 . Climate regime shifts and reorganization of fish communities: the essential fatty acid limitation hypothesis . Marine Ecology Progress Series , 315 : 1 – 11 . Google Scholar Crossref Search ADS WorldCat Lochmann S. E. , Maillet G. L., Frank K. T., Taggart C. T. 1995 . Lipid class composition as a measure of nutritional condition in individual larval Atlantic cod (Gadus morhua) . Canadian Journal of Fisheries and Aquatic Sciences , 52 : 1294 – 1306 . Google Scholar Crossref Search ADS WorldCat Logerwell E. , Busby M., Carothers C., Cotton S., Duffy-Anderson J., Farley E., Goddard P. et al. 2015 . Fish communities across a spectrum of habitats in the western Beaufort Sea and Chukchi Sea . Progress in Oceanography , 136 : 115 – 132 . Google Scholar Crossref Search ADS WorldCat Lu Y. , Ludsin S. A., Fanslow D. L., Pothoven S. A. 2008 . Comparison of three microquantity techniques for measuring total lipids in fish . Canadian Journal of Fisheries and Aquatic Sciences , 65 : 2233 – 2241 . Google Scholar Crossref Search ADS WorldCat Moulton L. L. , Tarbox K. E. 1987 . Analysis of Arctic cod movements in the Beaufort Sea nearshore region, 1978-79 . Arctic , 40 : 43 – 49 . Google Scholar Crossref Search ADS WorldCat Naas K. , Huse I., Iglesias J. 1996 . Illumination in first feeding tanks for marine fish larvae . Aquacultural Engineering , 15 : 291 – 300 . Google Scholar Crossref Search ADS WorldCat Neilson J. D. , Perry R. I., Valerio P., Waiwood K. G. 1986 . Condition of Atlantic cod Gadus Morhua larvae after the transition to exogenous feeding: morphometrics, buoyancy and predator avoidance . Marine Ecology Progress Series , 32 : 229 – 235 . Google Scholar Crossref Search ADS WorldCat Overland J. E. , Stabeno P. J. 2004 . Is the climate of Bering Sea warming and affecting the ecosystem? EOS , 85 : 309 – 316 . Google Scholar Crossref Search ADS WorldCat Park H. G. , Puvanendran V., Kellett A., Parrish C. C., Brown J. A. 2006 . Effect of enriched rotifers on growth, survival, and composition of larval Atlantic cod (Gadus morhua) . ICES Journal of Marine Science , 63 : 285 – 295 . Google Scholar Crossref Search ADS WorldCat Parrish C. C. 1987 . Separation of aquatic lipid classes by chromarod thin-layer chromatography with measurement by latroscan flame ionization detection . Canadian Journal of Fisheries and Aquatic Sciences , 44 : 722 – 731 . Google Scholar Crossref Search ADS WorldCat Pepin P. 1995 . An analysis of the length-weight relationship of larval fish: limitations of the general allometric model . Fishery Bulletin , 93 : 419 – 426 . Google Scholar OpenURL Placeholder Text WorldCat Pinchuk A. I. , Eisner L. B. 2017 . Spatial heterogeneity in zooplankton summer distribution in the eastern Chukchi Sea in 2012–2013 as a result of large-scale interactions of water masses . Deep Sea Research Part II: Topical Studies in Oceanography , 135 : 27 – 39 . Google Scholar Crossref Search ADS WorldCat Rand K. M. , Logerwell E. A. 2011 . The first demersal trawl survey of benthic fish and invertebrates in the Beaufort Sea since the late 1970s . Polar Biology , 34 : 475 – 488 . Google Scholar Crossref Search ADS WorldCat Richard P. , Bergeron J.-P., Boulhic M., Galois R., Person-Le Ruyet J. 1991 . Effect of starvation on RNA, DNA and protein-content of laboratory-reared larvae and juveniles of Solea solea . Marine Ecology Progress Series , 72 : 69 – 77 . Google Scholar Crossref Search ADS WorldCat Rijnsdorp A. D. , Peck M. A., Engelhard G. H., Möllmann C., Pinnegar J. K. 2009 . Resolving the effect of climate change on fish populations . ICES Journal of Marine Science , 66 : 1570 – 1583 . Google Scholar Crossref Search ADS WorldCat Shepherd J. G. , Cushing D. H. 1980 . A mechanism for density-dependent survival of larval fish as the basis of a stock-recruitment relationship . ICES Journal of Marine Science , 39 : 160 – 167 . Google Scholar Crossref Search ADS WorldCat Siddon E. C. , Heintz R. A., Mueter F. J. 2013 . Conceptual model of energy allocation in walleye pollock (Theragra chalcogramma) from age-0 to age-1 in the southeastern Bering Sea . Deep Sea Research Part II: Topical Studies in Oceanography , 94 : 140 – 149 . Google Scholar Crossref Search ADS WorldCat Sigler M. , Renner M., Danielson S., Eisner L., Lauth R., Kuletz K., Logerwell E. et al. 2011 . Fluxes, fins, and feathers: relationships among the Bering, Chukchi, and Beaufort Seas in a time of climate change . Oceanography , 24 : 250 – 265 . Google Scholar Crossref Search ADS WorldCat Søreide J. E. , Leu E., Berge J., Graeve M., Falk-Petersen S. 2010 . Timing of blooms, algal food quality and Calanus glacialis reproduction and growth in a changing Arctic . Global Change Biology , 16 : 3154 – 3163 . Google Scholar OpenURL Placeholder Text WorldCat Stowell M. 2016 . Effects of dissolved oxygen and temperature on juvenile English sole condition. MS thesis, Oregon State University, Corvallis, OR, USA. Suthers I. M. , Fraser A., Frank K. T. 1992 . Comparison of lipid, otolith and morphometric condition indices of pelagic juvenile cod Gadus morhua from the Canadian Atlantic . Marine Ecology Progress Series , 84 : 31 – 40 . Google Scholar Crossref Search ADS WorldCat Walther B. D. , Elsdon T. S., Gillanders B. M. 2010 . Interactive effects of food quality, temperature and rearing time on condition of juvenile black bream Acanthopagrus butcheri . Journal of Fish Biology , 76 : 2455 – 2468 . Google Scholar Crossref Search ADS PubMed WorldCat Published by International Council for the Exploration of the Sea 2018. This work is written by US Government employees and is in the public domain in the US. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) Published by International Council for the Exploration of the Sea 2018. This work is written by US Government employees and is in the public domain in the US.

Koenker, Brittany, L;Laurel, Benjamin, J;Copeman, Louise, A;Ciannelli,, Lorenzo

doi: 10.1093/icesjms/fsy062pmid: N/A

Abstract Arctic cod (Boreogadus saida) is an ecologically significant species that is uniquely adapted to occupy ice edges, but warming and loss of sea ice are hypothesized to favour more facultative gadids, such as walleye pollock (Gadus chalcogrammus). To test this hypothesis, we experimentally measured the growth and survival of Arctic cod and walleye pollock at two larval stages across a range of temperature and food conditions in the laboratory. Results indicated early and late-stage Arctic cod larvae have a competitive growth and survival advantage over walleye pollock at low temperatures. However, these advantages are lost under warmer, food-productive conditions where walleye pollock larvae survived and experienced accelerated growth rates. Growth models developed from this study emphasize the need to account for both species- and stage-specific differences in the thermal response of closely related marine fish larvae. More broadly, these new vital rate data provide a mechanistic framework to forecast spatial-temporal shifts of gadids at the Arctic-boreal interface resulting from climatic warming and altered productivity regimes. Introduction In polar regions, mean near-surface temperatures are predicted to warm at rates exceeding global climate change averages (Serreze and Barry, 2011; IPCC, 2013) resulting in drastic impacts to regional ocean conditions. In the Chukchi Sea, August sea surface temperatures are warming at a rate of about + 0.5°C per decade (Richter-Menge et al., 2016) in combination with declining seasonal and perennial sea ice cover (Comiso et al., 2008; Moore and Stabeno, 2015; Frey et al., 2015). The resultant rapid ecosystem-level change in the Arctic marine environment (Hoegh-Guldberg and Bruno, 2010) is predicted to result in accelerated rates of species turnover and severe changes in marine biodiversity (Cheung et al., 2009; Fossheim et al., 2015). Sea ice extent and seawater temperatures in the polar marine ecosystems are closely tied and together influence processes related to food web dynamics, species physiology, and biogeochemical cycling (Doney et al., 2012). By the mid- to late- twenty-first century, forecasted changes in sea ice thickness and extent may result in an ice-free Arctic in the summer months (Holland et al., 2010; Moore and Stabeno, 2015). Altered timing of sea ice break up has the potential to change the primary productivity regime in the region to the extent that a mismatch occurs between the production of high-quality food and key Arctic grazers (Søreide et al., 2010; Leu et al., 2011). The indirect effects of temperature through changes in food supply and timing have been shown to be important in determining Atlantic cod recruitment in the North Sea (Beaugrand and Kirby, 2010). Thus, it is likely that bottom-up changes in trophic interactions will influence the success of key Arctic species at all levels of the marine food web (Søreide et al., 2010). However, a mechanistic understanding of these changes remains challenging due to the complex mosaic of direct (i.e. temperature) and indirect (i.e. changes in prey phenology) impacts that result from a rapidly changing climate (Smetacek and Nicol, 2005). The Alaskan Arctic region includes the US waters of the Chukchi and Beaufort Seas north of the Bering Strait (NPFMC, 2009). Arctic cod (Boreogadus saida) is an ecologically important species throughout this region, where it plays a critical mid-trophic role in channelling energy from plankton to upper trophic levels such as marine mammals, seabirds, and other fish (Craig et al., 1982; Bluhm and Gradinger, 2008; Logerwell et al., 2015). Arctic cod often associate with ice edges which may offer both predator refuge and feeding habitat in the form of ice-associated phytoplankton blooms. However, forecasted shrinkage of sea ice habitat could facilitate invasions by more generalist boreal fish species, such as walleye pollock (Gadus chalcogrammus), from the North Pacific (Barber et al., 2009; Rand and Logerwell, 2011; Fossheim et al., 2015). Walleye pollock occupies a similar ecologically important role in the Bering Sea and Gulf of Alaska, but because of their high commercial value, their life history and biology is comparatively much better understood than Arctic cod. This is particularly evident during the early life history, as sea ice makes it logistically challenging to conduct field work (e.g. Bouchard and Fortier, 2008, 2011; Falardeau et al. 2014) and live-animal experiments in the laboratory (e.g. Sakurai et al., 1998, Graham and Hop, 1995; Laurel et al., 2018). In contrast, the early life history of walleye pollock has been the focus of multiple laboratory studies examining the effects of temperature on egg and larval development and survival (e.g. Bacheler et al. 2010; Duffy-Anderson et al., 2016). The absence of comparative biological data on early life stages makes it difficult to determine whether and when poleward shifts of marine fish species might result from climate change (Christiansen et al., 2014). In marine environments, relatively small changes in temperature and food availability can influence the growth, development, and survival of fish (Peck et al., 2004; Pörtner and Peck, 2010). As poikilotherms, a number of critical physiological processes in fish are regulated by the temperature of the surrounding water. Typically, high latitude fish are characterized by a narrower thermal tolerance range (i.e. stenothermic) than those inhabiting mid-latitudes where wider seasonal fluctuations are common (Pörtner and Peck, 2010). In polar environments, summers are short and subzero temperatures persist during the prolonged winter spawning season (Bouchard and Fortier, 2011). Furthermore, thermal sensitivity in terms of growth and food conversion can vary considerably with ontogenetic stage (Björnsson et al. 2001). This is particularly important for fish at critical early life stages when predation, starvation, advective loss, and other external stressors enhance mortality risks (Zhao et al., 2001; Houde, 2008). First-feeding and early life stage cod larvae are likely more sensitive to variable temperatures than later stage larvae and early juveniles due to developmental limitations and low energy storage capacity (Pörtner and Farrell, 2008; Fouzai et al., 2015). These variations in thermal tolerance can be significant at the individual, population, and ecosystem level as different thermal tolerance windows may result in changes in species distributions (Pörtner and Farrell, 2008). The “Stage-Duration” hypothesis postulates that rapid growth allows larvae to reach a larger size where risk of mortality is reduced and survival to recruitment is improved (Miller et al., 1988; Houde, 2008). However, efforts to assess growth and survival potential in fish rely upon a strong understanding of species-specific environmental tolerances and food requirements (Jobling, 1988). Prey availability and temperature are arguably the most important factors affecting larval growth and early size-at-age (Otterlei et al., 1999). Growth and mortality, in turn, influence recruitment levels in marine fish (Houde, 2008). Therefore, by assessing the sensitivity of a species to environmental conditions affecting growth, it is possible to better understand their likelihood of survival and, thus, the factors dictating population success under a changing climate. Currently, the growth and survival of larval Arctic cod under various temperature and productivity scenarios has not been investigated experimentally. In this study, we examine the effects of temperature and food availability on the growth and survival of larval Arctic cod and walleye pollock. Specifically, the objectives of this study were to (i) compare temperature-dependent growth and survival at two larval stages across a range of temperatures (−1 to 12°C), (ii) determine the interactive effects of temperature and food availability acting on the species- and stage-specific larval growth and survival, and (iii) develop temperature-dependent growth models for first-feeding and later stage Arctic cod and walleye pollock larvae. Methods Egg sources Laboratory experiments utilized the Alaska Fisheries Science Center’s gadid broodstock programme and facilities at the Hatfield Marine Science Center in Newport, OR, USA. Arctic cod broodstock were collected as juvenile fish [70- to 85-mm standard length (SL)] in early August of 2012 and 2013 using a fyke net in the nearshore of Prudhoe Bay, AK (Beaufort Sea, 70.383°N–148.552°W). Walleye pollock broodstock were sourced from juvenile walleye pollock (30- to 50-mm SL at capture) that were collected using light and lift nets in the nearshore of Puget Sound, WA (48.135°N–122.760°W) in late June of 2011, 2012, and 2013. Fish were transported alive to the laboratory where they were weaned onto formulated foods and held under a 12:12-h light:dark photoperiod to mimic field conditions. All broodstock were fed daily to satiation using a combination of thawed krill and a gelatinized combination of squid, krill, herring, commercial fish food, amino acid supplements and vitamins (“gel food”, recipe and lipid content as in the control diet details from Copeman et al., 2013). Juveniles were reared for over 3 years in the laboratory until they became active spawners (age-3+ fish). Additional details on collection and husbandry of Arctic cod can be found in Laurel et al. (2016). Experiments were conducted for first-feeding larvae and later stage pre-flexion larvae of both species. Laboratory experiments for both later stage species and for first-feeding walleye pollock took place in 2015, whereas the first-feeding Arctic cod experiment took place the following spring in 2016. Egg incubation A full detailed description of egg collection and incubation of Arctic cod and walleye pollock can be found in Laurel et al. (2018). Briefly, adult Arctic cod broodstock (n = 27) were held at 2°C and non-lethally strip spawned in March of each year to produce a series of egg batches. Each egg batch consisted of a single female fertilized with milt from three males. Eggs batches were incubated between 1 and 2°C in a 4-l mesh pan suspended in a water bath until reaching ∼75% hatch level, at which time all hatched and unhatched larvae were transferred to a series of 400-l holding tanks maintained between 2 and 3°C until needed for experimentation. First-feeding larval experiments were sourced from a holding tank comprised of two batches spawned on the same day. Later stage feeding experiments were sourced from a holding tank comprised of three separate batches spawned across 2 successive days. Walleye pollock eggs were collected from a tank of 32 adult broodstock held at 5°C from February to late April 2015. Unlike Arctic cod, pollock are difficult to strip spawn non-lethally, and were instead allowed to spawn naturally in the tank. Fertilized eggs were retained in an egg basket for 24 h to assess quality and then transferred across a series of 100-l stock tanks held at 5–6°C until they were needed for experiments. Although parentage was unknown, the high volume of eggs indicated that multiple females contributed to the egg batches used in both first-feeding and later stage feeding experiments. General husbandry Growth experiments were carried out in a series of 38-l glass aquaria covered externally with black plastic and supplied with flow-through, temperature-controlled seawater. Throughout larval experiments, tanks were held at a 12:12-h light:dark photoperiod. Light levels ranged from 1.4 to 2.7 µE m−2 s−1 at the surface of the water in the centre of each tank. Maintenance of tank temperatures, aeration, and flow rates was completed daily. Temperature was recorded in the morning and adjusted if necessary to account for fluctuations in ambient water temperature. Aeration was checked to ensure gentle bubbling beneath the outflow mesh in each tank. Flow rates were adjusted to 300 ml min−1 at the start of each experiment using a stopwatch and visually modified to within 270–330 ml min−1 each day. Tanks were siphoned daily to remove any mortalities along with excess food and debris. This process was conducted at least 2 h after feeding when visibility was sufficient and most of the live prey had already exited the tank (see below). Tanks were monitored daily and experiments continued for 3 weeks (for later stage experiments) or 5 weeks (for first-feeding experiments). However, if an individual tank approached 100% mortality, larvae were sampled to ensure that morphometric data was collected from each tank. Due to differential mortality, experimental duration varied slightly among tanks (3–5 weeks for first-feeding experiments, 2–3 weeks for later stage experiments). Live food preparation Prior to each feeding, Nanno 3600 algae paste (Reed Mariculture, Campbell, CA, USA) diluted with 2°C seawater was added to each tank to provide “green water”. The addition of green water has been shown to alter the visual feeding environment and change light conditions in a manner that improves larval prey ingestion (Naas et al., 1996). First-feeding larval experiments received enriched rotifers (Brachionus sp.), while later stage experiments received enriched brine shrimp (Artemia sp.). Rotifers were cultured at 26°C in a high-density rotifer culture system from Aquatic Eco-Systems. Rotifers were harvested and enriched twice daily in conical tanks to produce two batches for morning and afternoon larval fish feedings. The daytime and overnight batches were enriched for 5 and 16 h, respectively, with Algamac 3050 (0.3 g per million rotifers; Aquafauna, Hawthorne, CA, USA) and RotiGrow Plus (daytime: 0.5 ml per million rotifers, overnight: 1.0 mL per million rotifers; Reed Mariculture, Campbell, CA, USA). Algamac was chosen as a suitable enrichment because it contains a high proportion of long-chained fatty acids which are important for North Pacific larval fish (Copeman and Laurel, 2010). Enrichment tanks were drained overnight through a 53 µm sieve, rinsed with seawater, and resuspended in cooler (5°C) seawater prior to feeding. Enriched rotifers were counted daily for quality control and prey counts. First-feeding larvae in high food ration treatments were supplied enriched rotifers twice daily at prey densities of 5 prey ml−1, while low food ration treatments received prey densities of 0.5 prey ml−1 twice daily. Decapsulated brine shrimp were hatched for 24 h in hatching cones at 26–27°C before being enriched with Selco S.Presso (7.5 g per 15-l seawater; INVE Aquaculture, Nonthaburi, Thailand) for an additional 24 h. Enriched brine shrimp were drained through nylon mesh and rinsed before being resuspended in seawater. Harvested brine shrimp were counted to determine accurate prey counts for larvae and for quality control. Later stage larvae in high food ration treatments were supplied enriched brine shrimp twice daily at prey densities of 2 prey ml−1 and low food ration treatments received prey densities of 0.5 prey ml−1 once daily. Low food ration tanks still received green water 2× daily despite not receiving prey in the afternoon. Tanks were clear of all live prey after ∼2 h of each feeding, indicating all prey were either consumed and/or flowed out of the tank between feedings. This ensured that prey quality did not deteriorate over the course of the experiment and larvae were feeding on newly enriched prey. Experimental design A summary of experimental temperatures, food rations, and tank replications for all larval experiments is provided in Table 1. Table 1. Experimental design (treatment temperatures, food ration, and replicate tanks) at the start of first-feeding and later stage Arctic cod and walleye pollock larval growth experiments. Ontogenetic stage Species Temperature (°C) Food rationa(HF: high food; LF: low food) No. of replicate tanks First-feeding Arctic Cod −1 HF: Enriched Brachionus sp; 3 2 3 5 5 prey ml−1; twice daily 3 9 3 2 LF: Enriched Brachionus sp; 3 5 0.5 prey ml−1; twice daily 3 Walleye Pollock 0 HF: Enriched Brachionus sp; 3 2 3 5 5 prey ml−1; twice daily 3 12 3 2 LF: Enriched Brachionus sp; 3 5 0.5 prey ml−1; twice daily 3 Later stage Arctic Cod 0 HF: Enriched Artemia sp; 4 2 3 5 2 prey ml−1; twice daily 3 7 4 9 1 2 LF: Enriched Artemia sp; 3 5 0.5 prey ml−1; once daily 3 Walleye Pollock 0 HF: Enriched Artemia sp; 3 2 3 5 2 prey ml−1; twice daily 3 9 3 12 3 2 LF: Enriched Artemia sp; 3 5 0.5 prey ml−1; once daily 3 9 3 Ontogenetic stage Species Temperature (°C) Food rationa(HF: high food; LF: low food) No. of replicate tanks First-feeding Arctic Cod −1 HF: Enriched Brachionus sp; 3 2 3 5 5 prey ml−1; twice daily 3 9 3 2 LF: Enriched Brachionus sp; 3 5 0.5 prey ml−1; twice daily 3 Walleye Pollock 0 HF: Enriched Brachionus sp; 3 2 3 5 5 prey ml−1; twice daily 3 12 3 2 LF: Enriched Brachionus sp; 3 5 0.5 prey ml−1; twice daily 3 Later stage Arctic Cod 0 HF: Enriched Artemia sp; 4 2 3 5 2 prey ml−1; twice daily 3 7 4 9 1 2 LF: Enriched Artemia sp; 3 5 0.5 prey ml−1; once daily 3 Walleye Pollock 0 HF: Enriched Artemia sp; 3 2 3 5 2 prey ml−1; twice daily 3 9 3 12 3 2 LF: Enriched Artemia sp; 3 5 0.5 prey ml−1; once daily 3 9 3 a Food rations were manipulated only at intermediate temperatures, 2 and 5°C. Table 1. Experimental design (treatment temperatures, food ration, and replicate tanks) at the start of first-feeding and later stage Arctic cod and walleye pollock larval growth experiments. Ontogenetic stage Species Temperature (°C) Food rationa(HF: high food; LF: low food) No. of replicate tanks First-feeding Arctic Cod −1 HF: Enriched Brachionus sp; 3 2 3 5 5 prey ml−1; twice daily 3 9 3 2 LF: Enriched Brachionus sp; 3 5 0.5 prey ml−1; twice daily 3 Walleye Pollock 0 HF: Enriched Brachionus sp; 3 2 3 5 5 prey ml−1; twice daily 3 12 3 2 LF: Enriched Brachionus sp; 3 5 0.5 prey ml−1; twice daily 3 Later stage Arctic Cod 0 HF: Enriched Artemia sp; 4 2 3 5 2 prey ml−1; twice daily 3 7 4 9 1 2 LF: Enriched Artemia sp; 3 5 0.5 prey ml−1; once daily 3 Walleye Pollock 0 HF: Enriched Artemia sp; 3 2 3 5 2 prey ml−1; twice daily 3 9 3 12 3 2 LF: Enriched Artemia sp; 3 5 0.5 prey ml−1; once daily 3 9 3 Ontogenetic stage Species Temperature (°C) Food rationa(HF: high food; LF: low food) No. of replicate tanks First-feeding Arctic Cod −1 HF: Enriched Brachionus sp; 3 2 3 5 5 prey ml−1; twice daily 3 9 3 2 LF: Enriched Brachionus sp; 3 5 0.5 prey ml−1; twice daily 3 Walleye Pollock 0 HF: Enriched Brachionus sp; 3 2 3 5 5 prey ml−1; twice daily 3 12 3 2 LF: Enriched Brachionus sp; 3 5 0.5 prey ml−1; twice daily 3 Later stage Arctic Cod 0 HF: Enriched Artemia sp; 4 2 3 5 2 prey ml−1; twice daily 3 7 4 9 1 2 LF: Enriched Artemia sp; 3 5 0.5 prey ml−1; once daily 3 Walleye Pollock 0 HF: Enriched Artemia sp; 3 2 3 5 2 prey ml−1; twice daily 3 9 3 12 3 2 LF: Enriched Artemia sp; 3 5 0.5 prey ml−1; once daily 3 9 3 a Food rations were manipulated only at intermediate temperatures, 2 and 5°C. Arctic cod larval experiments For first-feeding experiments, yolk-sac Arctic cod larvae (mean 5.9-mm SL) were transferred into 1-l beakers to acclimate to each temperature treatment in April 2016 at ∼90% hatch level. Once larvae were within 0.5°C of their target temperature, larvae were gently poured into their corresponding aquaria. First-feeding Arctic cod experiments utilized 12 aquaria for high food ration treatments (−1, 2, 5, and 9°C; n = 3 replicate tanks/temperature) and 6 for low food ration treatments (2 and 5°C; n = 3 replicate tanks/temperature) with larvae stocked at a density of 450 individuals per tank. The first-feeding Arctic cod experiment continued for 35 days, with the exception of the 9°C tanks which were sampled at 21 days and reached 100% mortality prior to 5 weeks. For the later stage experiment, Arctic cod larvae remained in stock tanks where they were “pulse fed” enriched rotifers (Brachionus sp.) twice daily at a density of 5 prey ml−1. Several weeks prior to the start of later stage experiments, larvae began receiving enriched brine shrimp (Artemia sp.) at a prey density of 2 prey ml−1 in addition to rotifers. Before later stage experiments began, larval gut content was visually examined to ensure that Artemia sp. were being consumed. Later stage larvae were in a stage of pre-flexion prior to the experiment. The later stage Arctic cod experiment began in June 2015 (78 dph) when larvae from the 400-l holding tanks achieved a mean SL of 11.3 mm. Later stage experiments used twelve aquaria for high food ration treatments (0, 2, 5, and 7°C; n = 3 replicate tanks/temperature) and six for low food ration treatments (2 and 5°C; n = 3 replicate tanks/temperature) with larvae stocked at a density of ∼100 individuals per tank. A fourth replicate tank at 0 and 7°C was set up after 1 week to account for particularly high mortality in one of the replicate tanks at each temperature. An additional 9°C trial was conducted for later stage Arctic cod to assess the upper thermal limit for survival (n = 1 tank due to high mortality). The later stage Arctic cod experiment continued for 19–20 days, with the exception of the added tanks which were sampled at the same time (14–15 days total) and one 7°C tank which was sampled at 16 days at it approached 100% mortality. Walleye pollock larval experiments In April 2015, first-feeding walleye pollock larvae (mean 4.7-mm SL) were transferred and slowly acclimated (as described above for Arctic cod) to twelve aquaria for high food ration treatments (0, 2, 5, and 12°C; n = 3 replicate tanks/temperature) and six for low food ration treatments (2 and 5°C; n = 3 replicate tanks/temperature) with larvae stocked at a density of 450 individuals per tank. The first-feeding walleye pollock experiment continued for 35 days, with the exception of one 12°C tank which was sampled at 21 days and reached 100% mortality prior to 5 weeks. Walleye pollock larvae used in the later stage experiment remained in stock tanks until July 2015 (84 dph) when they reached a larger length (mean 8.6-mm SL) and were exclusively weaned onto Artemia sp., similar to later stage Arctic cod. Prior to experiments, later stage walleye pollock were in a stage of pre-flexion. Later stage walleye pollock larvae were transferred following the same methods to fifteen aquaria for high food ration treatments (0, 2, 5, 9, and 12°C; n = 3 replicate tanks/temperature) and nine for low food ration treatments (2, 5, and 9°C; n = 3 replicate tanks/temperature). The later stage walleye pollock experiment continued for 26 days, with the exception of the 0 and 12°C tanks which were sampled at 12–13 days and the 9°C high food ration tanks which were sampled at 13–19 days as they approached 100% mortality. Data collection and analysis Survival was estimated differently for first-feeding and later stage larval experiments. Due to the small size and rapid decay of first-feeding larvae, survival estimates based on daily mortality counts were only possible for later stage larvae. As such, later stage larval mortality was assessed daily to quantify cumulative percent mortality for each species under different treatments. From these measurements, a daily mortality schedule was produced for each species based on the mean daily cumulative percent mortality at each temperature. Additionally, for the purpose of statistical analyses, each tank was assigned a time to 50% cumulative mortality (D50) which was determined as the first day of the experiments where cumulative mortality was at or above 50%. Two of the Arctic cod tanks were assigned a D50 equal to the last day of the experiment despite ∼55% of the larvae surviving the entire experimental period. For first-feeding larvae, survival was estimated from counts of remaining larvae at the end of the experiment. For consistency, this method of estimating survival was also used for later stage larvae to complement cumulative mortality estimates. To account for slight differences in experimental duration among tanks, a survival fraction representing the fraction of larvae surviving on a daily basis was computed. In this study, the survival fraction, S, was calculated for each tank according to the following equation: S=e-M In this equation, M is the daily mortality rate (day−1) derived from the exponential mortality model and calculated as: M=(lnN0-lnNt)/t when N0 is the initial number of larvae stocked, Nt is the number of larvae that survived to the end of the experiment, and t is the experiment duration in days. In instances when no larvae survived (two tanks at 12°C), calculation of the survival fraction was based on a single larvae surviving the duration of the experiment. Larvae from each tank were randomly sampled from throughout the water column for morphometric measurements (i.e. dry mass, SL, and body depth) at the start, middle, and end of each experiment. Larvae were anaesthetized with MS-222 (50 g l−1) and individual images were taken under calibrated magnification using a digital camera attached to a stereo microscope. Measurements for each fish were obtained from digital images using ImagePro software (Media Cybernetics, Bethesda, MD, USA). SL was determined as the length (mm) from the tip of the snout to the end of the notochord. Body depth was the width (mm) of the larvae posterior to the anus not including the fin-fold. Fish were then rinsed with a 3% ammonium formate solution to remove excess salts and placed on pre-weighed aluminium foil squares. Foils squares were then folded securely and placed in labelled slots on a baking sheet in a drying oven. Samples were dried at 55°C for a minimum of 48 h before determination of dry mass with a microbalance (Sartorius R16OP) to the nearest 1.0 µg. Specific growth rate (SGR), (% mass day−1) was calculated under different temperature-food ration treatments according to the following equation: SGR=100(eg-1) In this equation, g is the instantaneous growth coefficient calculated as: g=(lnWt-lnW0)/t where Wt is the final mean dry mass, W0 is the initial mean dry mass, and t is the number of days between measurements. All growth and survival analyses were performed using RStudio statistical software (ver. 0.99.491, RStudio, Inc., Boston, MA, USA). Survival (either S or D50) for each high food ration tank was analysed using a series of two-way analysis of variances (ANOVAs) examining the effects of temperature by species or ontogenetic stage. To account for the interactive effects of temperature and food availability, a three-way ANOVA was used to test for statistical differences in survival between species, temperature, and food ration at 2 and 5°C. An additional three-way ANOVA was used to consider the effects of stage, temperature, and food ration on survival at these intermediate temperatures. Data were examined for normality and homogeneity of variance to satisfy the assumptions of the ANOVA. A significance level of α = 0.05 was used in all analyses. Statistical differences in growth under high food ration treatments were determined using a two-way ANOVA to examine the effects of species and temperature, and also to examine the effects of ontogenetic stage and temperature. Analysis was conducted on tank replicates (average SGR/tank). Three-way ANOVAs were used to test for statistical differences in growth between species, temperature (2 and 5°C only), and food ration within each ontogenetic stage. Additional three-way ANOVAs were also used to assess the effect of ontogenetic stage, temperature (2 and 5°C only), and food ration on growth within species. A temperature-dependent growth model based on mean SGR for tank replicates was developed for the high food ration treatments. This model followed the form below: SGR=β0+β1T+β2T2+Isβ3+Isβ4T+Isβ5T2 where the SGR is the response variable, temperature (⁠ T ⁠) and temperature-squared (⁠ T2 ⁠) are explanatory variables, and species or stage (⁠ Is ⁠) is an indicator variable. The Akaike Information Criterion (AIC) was used to determine whether species-specific models of temperature-dependent growth were justified at each ontogenetic stage. That is, species-specific models were used when AIC values were lower than the simplified model that excluded the Is indicator variable and therefore assumed no difference in growth between the two species. Similarly, stage-specific models were used when AIC values were lower than the simplified model excluding the Is indicator variable which assumed no difference in growth between first-feeding and later larval stages. Following AIC, best-fit regression models (up to two parameters) were used to describe temperature-dependent growth relationships for each species and/or ontogenetic stage. To minimize the impact of size-selective mortality, any treatments experiencing > 80% mortality within the first week of experiments were not used in growth analysis or development of these explanatory growth models. Results Temperature effects on survival and growth The size-at-age (based on SL) of first-feeding and later stage Arctic cod and walleye pollock in high food ration treatments over the course of laboratory experiments is shown in Figure 1. The survival fraction ranged from 0.80–0.97 for Arctic cod and 0.77–0.94 for walleye pollock in high food ration experiments. Although first-feeding Arctic cod were reared across a cooler temperature range (−1 to 9°C) than walleye pollock (0–12°C), survival was lowest in the high temperature treatment for both species (Figure 2a and b; two-way ANOVA; F1, 18 = 10.581, p = 0.004). Similar patterns were observed at later stages, with highest survival of both species observed at 2°C (Figure 2c and d). However, unlike first-feeding larvae, there was significantly higher survival of Arctic cod than walleye pollock at comparable temperatures during the later stage (two-way ANOVA, F1, 26 = 14.688, p < 0.001). Figure 1. View largeDownload slide Size-at-age based on SL (mm) of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae in high food ration treatments over the course of laboratory experiments. Data are treatment means ±1 s.e. (n = 1–4 replicate tanks/treatment). Figure 1. View largeDownload slide Size-at-age based on SL (mm) of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae in high food ration treatments over the course of laboratory experiments. Data are treatment means ±1 s.e. (n = 1–4 replicate tanks/treatment). Figure 2. View largeDownload slide The survival fraction, or the fraction of larvae surviving on a daily basis, for (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock receiving high food rations. Data represent treatment means ±1 s.e. (n = 3 replicate tanks/treatment, except for first-feeding Arctic cod at 2 and 9°C where n = 2 and later stage Arctic cod at 0, 7, and 9°C where n = 4, 4, and 1, respectively). Figure 2. View largeDownload slide The survival fraction, or the fraction of larvae surviving on a daily basis, for (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock receiving high food rations. Data represent treatment means ±1 s.e. (n = 3 replicate tanks/treatment, except for first-feeding Arctic cod at 2 and 9°C where n = 2 and later stage Arctic cod at 0, 7, and 9°C where n = 4, 4, and 1, respectively). The impacts of temperature and ontogenetic stage on the survival fraction of each species differed. In Arctic cod, higher temperatures negatively impacted survival for both ontogenetic stages (two-way ANOVA; F1, 21 = 13.989, p = 0.001). However, in walleye pollock the survival impacts varied with ontogenetic stage (F1, 23 = 12.589, p = 0.002), driven by lower observed survival at the later stage than the first-feeding stage (Figure 2). The effects of temperature on survival in walleye pollock was not statistically significant (p = 0.094), although there was very low survival of late-stage larvae observed in the lowest temperature treatment For reasons indicated in the “Methods: Section, cumulative percent mortality was only quantified for later stage species (Figure 3). However, time to 50% cumulative mortality (D50) did not vary between species (F1,26 = 0.055, p = 0.817) and temperature treatments (F1,26 = 0.006, p = 0.937) or in the interaction term of the model (F1,26 = 2.726, p = 0.111). Despite not being significant, there were some notable differences in temperature-dependent survival between the two species. Later stage Arctic cod larvae at 9°C rapidly reached 50% cumulative mortality after one day, compared with a mean D50 of ≥ 8 days for all other temperature treatments receiving high food (Figure 4a). Conversely, later stage walleye pollock reached 50% cumulative mortality at 0°C after 2 days, whereas D50 among the remaining high food ration treatments was ≥ 10 days (Figure 4b). Figure 3. View largeDownload slide Daily mortality schedule based on the cumulative percent mortality (% day−1) of (a) later stage Arctic cod and (b) later stage walleye pollock receiving high food rations over the duration of laboratory experiments. Data represent treatment mean cumulative mortality based on daily tank mortality counts (n = 3 replicate tanks/treatment except for Arctic cod at 0, 7, and 9°C where n = 4, 4, and 1, respectively). Figure 3. View largeDownload slide Daily mortality schedule based on the cumulative percent mortality (% day−1) of (a) later stage Arctic cod and (b) later stage walleye pollock receiving high food rations over the duration of laboratory experiments. Data represent treatment mean cumulative mortality based on daily tank mortality counts (n = 3 replicate tanks/treatment except for Arctic cod at 0, 7, and 9°C where n = 4, 4, and 1, respectively). Figure 4. View largeDownload slide Time to 50% mortality (D50) in days for (a) later stage Arctic cod and (b) later stage walleye pollock. Data represent treatment means ±1 s.e. based on D50 values derived from daily tank mortality counts (n = 3 replicate tanks/treatment except for Arctic cod at 0, 7, and 9°C where n = 4, 4, and 1, respectively). Figure 4. View largeDownload slide Time to 50% mortality (D50) in days for (a) later stage Arctic cod and (b) later stage walleye pollock. Data represent treatment means ±1 s.e. based on D50 values derived from daily tank mortality counts (n = 3 replicate tanks/treatment except for Arctic cod at 0, 7, and 9°C where n = 4, 4, and 1, respectively). The SGR of larvae generally increased with temperature for each species within each ontogenetic stage. However, first-feeding Arctic cod larvae achieved maximum growth at 5°C whereas first-feeding pollock grew progressively faster with temperature to 12°C (Figure 5a and b). This pattern was reflected statistically by way of a significant interaction in the “species*temperature” term of the model of first-feeding larvae (F1,19 = 21.634, p < 0.001). Conversely, a significant interaction was not detected for later stage larvae, although Arctic cod larvae grew faster than walleye pollock across overlapping temperature treatments (F1,22 = 6.153, p = 0.021; Figure 5c and d). Figure 5. View largeDownload slide SGRs (% mass day−1) of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae receiving high food rations. Data are treatment means ±1 s.e. (n = 3 replicate tanks/treatment, except for first-feeding Arctic cod at 9°C where n = 2 and later stage Arctic cod at 0 and 7°C where n = 4). Later stage Arctic cod 9°C data and later stage walleye pollock 0°C data were not used in growth analysis as these treatments experienced > 80% mortality within the first week of experiments. Figure 5. View largeDownload slide SGRs (% mass day−1) of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae receiving high food rations. Data are treatment means ±1 s.e. (n = 3 replicate tanks/treatment, except for first-feeding Arctic cod at 9°C where n = 2 and later stage Arctic cod at 0 and 7°C where n = 4). Later stage Arctic cod 9°C data and later stage walleye pollock 0°C data were not used in growth analysis as these treatments experienced > 80% mortality within the first week of experiments. Additional two-way ANOVAs were conducted by species to determine the effects of ontogenetic stage and temperature on growth. A significant interaction between stage and temperature was found for both Arctic cod (F1,21 = 35.513, p < 0.001) and walleye pollock (F1,20 = 4.802, p = 0.040). This interaction was driven by increased temperature-dependent growth in later stage larvae of both species. Interaction of temperature and food availability The interactive effects of temperature and food ration were assessed at intermediate temperatures (2 and 5°C) where food rations were manipulated. The size-at-age of first-feeding and later stage Arctic cod and walleye pollock in these treatments over the course of laboratory experiments is shown in Figure 6. Figure 6. View largeDownload slide Size-at-age based on SL (mm) of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock receiving high and low food rations at intermediate temperature (2 and 5°C) over the duration of laboratory experiments. Data are treatment means ± 1 s.e. (n = 3 replicate tanks/treatment). Figure 6. View largeDownload slide Size-at-age based on SL (mm) of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock receiving high and low food rations at intermediate temperature (2 and 5°C) over the duration of laboratory experiments. Data are treatment means ± 1 s.e. (n = 3 replicate tanks/treatment). For first-feeding larvae, there was a significant interaction between the effects of species, temperature, and food ration explaining survival (three-way ANOVA; F1,15 = 5.855, p = 0.029; Figure 7a and b). To better understand the nature of this interaction, species were analysed separately using two-way ANOVAs with temperature and food ration as independent variables. The effects of temperature and food ration on the survival fraction of first-feeding Arctic cod were not significant, although the interaction was close to the statistical alpha (F1,7 = 4.750, p = 0.066). The near-significant interaction term was due to a more positive effect of food on survival within the warmer of the two temperature treatments (Figure 7a). There was also no significant interaction between temperature and food ration for first-feeding walleye pollock, but there was higher survival among high food ration treatments than those receiving low food at both 2 and 5°C (F1,8 = 6.738, p = 0.032; Figure 7b). Figure 7. View largeDownload slide The survival fraction of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae. Data represent treatment means ±1 s.e. (n = 3 replicate tanks/treatment except for first-feeding Arctic cod receiving high food rations at 2°C where n = 2). Figure 7. View largeDownload slide The survival fraction of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae. Data represent treatment means ±1 s.e. (n = 3 replicate tanks/treatment except for first-feeding Arctic cod receiving high food rations at 2°C where n = 2). For later stage larvae, there was a significant interaction between temperature and food ration (F1,16 = 6.091, p = 0.025) as well as a significant species effect (F1,16 = 10.838, p = 0.005) (Figure 7c and d). These statistical effects were due to higher survival in later stage Arctic cod compared to later stage walleye pollock larvae, and higher survival in low food ration tanks than high food ration tanks at 5°C for both species. Additional three-way ANOVAs were conducted for each species to determine the effects of ontogenetic stage, temperature, and food ration on survival. A significant three-way interaction was detected for Arctic cod (F1,15 = 7.117, p = 0.018), driven by observations that first-feeding Arctic cod larvae underwent more mortality under low food, warm conditions (5°C) but were relatively insensitive to such changes as later stage larvae (Figure 7a and c). In contrast, the negative effects of low food availability were not exacerbated at higher temperature in walleye pollock larvae (Figure 7b and d). However, first-feeding walleye pollock larvae were more food sensitive than later stage larvae, indicated by a significant interaction between ontogenetic stage and food ration (three-way ANOVA; F1,16 = 12.014, p = 0.003). The cumulative percent mortality of larvae receiving high and low food rations at 2 and 5°C spanning the duration of later stage laboratory experiments is shown in Figure 8. The time to 50% mortality (D50) was not significantly different between species (F1,16 = 1.200, p = 0.290), among temperatures (F1,16 = 0.048, p = 0.829), or across food rations (F1,16 = 1.614, p = 0.222) by way of the (three-way ANOVA) (Figure 9). The lack of significance was likely due to high error resulting from variability between tanks. However, a graphical trend was present demonstrating that tanks receiving high food rations took longer to reach 50% mortality than tanks receiving low food rations in most instances (Figure 8). Figure 8. View largeDownload slide Daily mortality schedule based on the cumulative percent mortality (% day−1) of (a) later stage Arctic cod and (b) later stage walleye pollock receiving high and low food rations at 2 and 5°C over the duration of laboratory experiments. Data represent treatment mean cumulative mortality based on daily tank mortality counts (n = 3 replicate tanks/treatment). Figure 8. View largeDownload slide Daily mortality schedule based on the cumulative percent mortality (% day−1) of (a) later stage Arctic cod and (b) later stage walleye pollock receiving high and low food rations at 2 and 5°C over the duration of laboratory experiments. Data represent treatment mean cumulative mortality based on daily tank mortality counts (n = 3 replicate tanks/treatment). Figure 9. View largeDownload slide Time to 50% mortality (D50) in days for (a) later stage Arctic cod and (b) later stage walleye pollock receiving high and low food rations at 2 and 5°C. Data represent treatment means ±1 s.e. based on D50 values derived from daily tank mortality counts (n = 3 replicate tanks/treatment). Figure 9. View largeDownload slide Time to 50% mortality (D50) in days for (a) later stage Arctic cod and (b) later stage walleye pollock receiving high and low food rations at 2 and 5°C. Data represent treatment means ±1 s.e. based on D50 values derived from daily tank mortality counts (n = 3 replicate tanks/treatment). A significant interaction between species and temperature (F1, 16 = 11.426, p = 0.004) and species and food ration (F1, 16 = 24.518, p < 0.001) on the SGR of first-feeding larvae at intermediate temperatures was detected (three-way ANOVA) (Figure 10a and b). First-feeding Arctic cod were more sensitive to food ration, whereas first-feeding walleye pollock were more sensitive to temperature. Similarly, for later stage larvae, there was a significant interaction between species and temperature (F1, 16 = 12.300, p = 0.003) on the larval growth rate at 2 and 5°C (Figure 10c and d), but unlike first-feeding larvae, later stage Arctic cod were more sensitive to temperature than later stage walleye pollock. Growth in the high food ration tanks was also higher than low ration treatments as indicated by significant single term effect of ‘food ration’ in the model (F1, 16 = 9.530, p = 0.007). Figure 10. View largeDownload slide SGRs (% mass day−1) of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae receiving high and low food rations at 2 and 5°C. Data are treatment means ±1 s.e. (n = 3 replicate tanks/treatment). Figure 10. View largeDownload slide SGRs (% mass day−1) of (a) first-feeding Arctic cod, (b) first-feeding walleye pollock, (c) later stage Arctic cod, and (d) later stage walleye pollock larvae receiving high and low food rations at 2 and 5°C. Data are treatment means ±1 s.e. (n = 3 replicate tanks/treatment). Last, a three-way ANOVA between ontogenetic stage, temperature, and food ration revealed a significant interaction between stage and temperature (F1, 16 = 13.900, p = 0.002) on the SGR of Arctic cod larvae (Figure 10a and c). The effect of temperature on growth was stronger for later stage Arctic cod than for first-feeding larvae. Similarly, a stage-temperature interaction (F1, 16 = 9.233, p = 0.008) was statistically supported for walleye pollock, but unlike Arctic cod, the effect of temperature on growth was stronger for first-feeding walleye pollock than for later stage walleye pollock (Figure 10b and d). Additionally, food ration was found to be significant for Arctic cod (F1, 16 = 52.135, p < 0.001), and just above the statistical alpha for walleye pollock (F1, 16 = 4.343, p = 0.054). Temperature-dependent growth models Based on AIC criteria, temperature-dependent growth models were developed separately for each species at each ontogenetic stage. For first-feeding larvae AIC values improved from 93.79 to 4.92 by including a species term in the model. At later stages, AIC scores modestly improved from 112.16 to 105.40 by including the species. Similarly, the growth model for Arctic cod and walleye pollock improved with the inclusion of an ontogenetic term to a model that pooled data across ontogenetic stages (Arctic cod, AIC = 54.77 vs. 77.05; walleye pollock, AIC =96.99 vs. 119.54; See Table 2) Table 2. Parameter estimates for temperature-dependent growth models for Arctic cod (AC) and walleye pollock (WP) under high food ration treatments. Ontogenetic stage Species T (°C) SGR (% mass day−1) Model Parameter estimates Y0 mean ± s.e. α mean ± s.e. β mean ± s.e. First-feeding AC −0.1 2.294 SGR=Yo+αT+βT2 2.349 ± 0.120 0.939 ± 0.072 −0.090 ± 0.008 1.9 3.727 4.8 4.809 9.0 3.486 WP 0.6 −0.755 SGR=Yo+αT+βT2 −1.141 ± 0.317 0.599 ± 0.178 −0.007 ± 0.018 2.3 0.122 4.9 1.682 10.6 4.424 Later stage AC 0.0 1.133 SGR=Yo+αT+βT2 1.140 ± 0.775 1.450 ± 0.594 −0.083 ± 0.085 2.2 3.803 4.4 6.003 6.7 7.094 WP 2.2 2.909 SGR=ae(βT) na −1.352 ±0.586 0.194 ± 0.042 4.7 3.998 8.6 5.678 11.2 12.609 Ontogenetic stage Species T (°C) SGR (% mass day−1) Model Parameter estimates Y0 mean ± s.e. α mean ± s.e. β mean ± s.e. First-feeding AC −0.1 2.294 SGR=Yo+αT+βT2 2.349 ± 0.120 0.939 ± 0.072 −0.090 ± 0.008 1.9 3.727 4.8 4.809 9.0 3.486 WP 0.6 −0.755 SGR=Yo+αT+βT2 −1.141 ± 0.317 0.599 ± 0.178 −0.007 ± 0.018 2.3 0.122 4.9 1.682 10.6 4.424 Later stage AC 0.0 1.133 SGR=Yo+αT+βT2 1.140 ± 0.775 1.450 ± 0.594 −0.083 ± 0.085 2.2 3.803 4.4 6.003 6.7 7.094 WP 2.2 2.909 SGR=ae(βT) na −1.352 ±0.586 0.194 ± 0.042 4.7 3.998 8.6 5.678 11.2 12.609 Mean treatment SGR values are based on replicate tanks (n = 3 except for first-feeding Arctic cod at ∼9°C where n = 2 and later stage Arctic cod at 0 and 7°C where n = 4) for each species at each temperature. See “Methods” section for additional details on model selection criteria. Table 2. Parameter estimates for temperature-dependent growth models for Arctic cod (AC) and walleye pollock (WP) under high food ration treatments. Ontogenetic stage Species T (°C) SGR (% mass day−1) Model Parameter estimates Y0 mean ± s.e. α mean ± s.e. β mean ± s.e. First-feeding AC −0.1 2.294 SGR=Yo+αT+βT2 2.349 ± 0.120 0.939 ± 0.072 −0.090 ± 0.008 1.9 3.727 4.8 4.809 9.0 3.486 WP 0.6 −0.755 SGR=Yo+αT+βT2 −1.141 ± 0.317 0.599 ± 0.178 −0.007 ± 0.018 2.3 0.122 4.9 1.682 10.6 4.424 Later stage AC 0.0 1.133 SGR=Yo+αT+βT2 1.140 ± 0.775 1.450 ± 0.594 −0.083 ± 0.085 2.2 3.803 4.4 6.003 6.7 7.094 WP 2.2 2.909 SGR=ae(βT) na −1.352 ±0.586 0.194 ± 0.042 4.7 3.998 8.6 5.678 11.2 12.609 Ontogenetic stage Species T (°C) SGR (% mass day−1) Model Parameter estimates Y0 mean ± s.e. α mean ± s.e. β mean ± s.e. First-feeding AC −0.1 2.294 SGR=Yo+αT+βT2 2.349 ± 0.120 0.939 ± 0.072 −0.090 ± 0.008 1.9 3.727 4.8 4.809 9.0 3.486 WP 0.6 −0.755 SGR=Yo+αT+βT2 −1.141 ± 0.317 0.599 ± 0.178 −0.007 ± 0.018 2.3 0.122 4.9 1.682 10.6 4.424 Later stage AC 0.0 1.133 SGR=Yo+αT+βT2 1.140 ± 0.775 1.450 ± 0.594 −0.083 ± 0.085 2.2 3.803 4.4 6.003 6.7 7.094 WP 2.2 2.909 SGR=ae(βT) na −1.352 ±0.586 0.194 ± 0.042 4.7 3.998 8.6 5.678 11.2 12.609 Mean treatment SGR values are based on replicate tanks (n = 3 except for first-feeding Arctic cod at ∼9°C where n = 2 and later stage Arctic cod at 0 and 7°C where n = 4) for each species at each temperature. See “Methods” section for additional details on model selection criteria. The growth models for first-feeding larvae indicated significantly higher growth by Arctic cod at modelled temperatures < 8.8°C, relative to walleye pollock (Figure 11a). Although Arctic cod were not reared at temperatures > 9°C, model extrapolation indicated walleye pollock growth surpassed Arctic cod at temperatures > 8.8°C. Arctic cod had approximately three times more growth than walleye pollock at temperatures between 0 and 5°C. Furthermore, Arctic cod achieved maximum growth at 5.2°C, while walleye pollock achieved maximum growth at 10.6°C. Figure 11. View largeDownload slide Explanatory models of SGRs; % mass day−1) for each species and ontogenetic stage. Top panels (a) and (b) compare species by ontogenetic stages whereas bottom panels (c) and (d) compare ontogenetic stages by species. All growth rates are based on larvae receiving high food rations. Data are tank means (n = 3 replicate tanks/treatment, except for first-feeding Arctic cod at 9°C where n = 2 and later stage Arctic cod at 0 and 7°C where n = 4). Later stage Arctic cod at 9°C and later stage walleye pollock at 0°C were not used in developing growth models due to high mortality (>80%) within the first week of experiments. The dashed line in the later stage Arctic cod model represents extrapolation of the model beyond the experimental temperature range. See “Methods” section and Table 2 for additional details on model selection criteria and parameters. Figure 11. View largeDownload slide Explanatory models of SGRs; % mass day−1) for each species and ontogenetic stage. Top panels (a) and (b) compare species by ontogenetic stages whereas bottom panels (c) and (d) compare ontogenetic stages by species. All growth rates are based on larvae receiving high food rations. Data are tank means (n = 3 replicate tanks/treatment, except for first-feeding Arctic cod at 9°C where n = 2 and later stage Arctic cod at 0 and 7°C where n = 4). Later stage Arctic cod at 9°C and later stage walleye pollock at 0°C were not used in developing growth models due to high mortality (>80%) within the first week of experiments. The dashed line in the later stage Arctic cod model represents extrapolation of the model beyond the experimental temperature range. See “Methods” section and Table 2 for additional details on model selection criteria and parameters. Later stage Arctic cod had higher growth than later stage walleye pollock across experimental temperatures from 2.2 to 6.7°C (Figure 11b). Extrapolated model growth suggests that walleye pollock had a growth advantage over Arctic cod at temperatures > 9.0°C. Within the observed temperature treatments, Arctic cod and walleye pollock achieved maximum growth at 6.7 and 11.2°C, respectively. Ontogenetic differences in growth were also observed within each species. First-feeding Arctic cod had higher growth from 0 to 2.3°C, but were surpassed by later stage larvae at temperatures > 2.3°C (Figure 11c). First-feeding Arctic cod experienced maximum growth at 5.2°C, while later stage Arctic cod had highest growth at a slightly higher temperature, 6.7°C. Later stage walleye pollock maintained a growth advantage over first-feeding walleye pollock across all modelled temperatures (Figure 11d). Both first-feeding and later stage walleye pollock achieved maximum growth at the upper end of the modelled temperature range (10.6 and 11.2°C, respectively). However, first-feeding walleye pollock growth increased at a relatively linear rate across the temperature range, whereas later stage walleye pollock growth appeared to have an exponential increase in growth at towards higher temperatures. Discussion This study is the first laboratory investigation of larval growth for feeding stages of Arctic cod and provided new growth and survival data for walleye pollock over a broader thermal range. Results indicated there was: (i) variable growth and survival responses by Arctic cod and walleye pollock with temperature, and (ii) stage-specific and species-specific differences in larval sensitivity to temperatures and food availability. Together, these data suggest common environmental conditions will impact these species differently, but need to be considered regionally within the ontogenetic stage of each species separately. These results are contextualized with other field and lab studies to determine how Arctic cod and walleye pollock may differentially respond to environmental variability resulting from climate change. Temperature-dependent survival and growth Species-specific differences in temperature-dependent survival were evident, as Arctic cod generally had higher survival than walleye pollock at lower temperatures (<7°C). At later stages, both species appeared to have increased survival at slightly warmer temperatures (highest survival at 2°C), possibly reflecting late-spring conditions during or shortly after ice-melt. However, walleye pollock larvae were clearly more tolerant of warm temperatures than Arctic cod at both stages. In addition, Arctic cod mortality occurred rapidly at 9°C (>80% in 2 days), suggesting that Arctic cod are highly sensitive to even short term increases in temperature. Temperature-dependent growth rates were also different between species, and these growth differences further changed with ontogeny. Larval Arctic cod maintained a growth advantage over walleye pollock at low temperatures, but were surpassed at ∼ 9°C during both larval stages. Within this range, maximum growth was achieved by Arctic cod at 5–7°C and by walleye pollock at 12°C. This thermal range was similar to laboratory work on juvenile gadids, which report maximum growth at 7°C for age-1 Arctic cod and 13°C for age-0 walleye pollock (Laurel et al., 2016). Laboratory findings by Kunz et al. (2016) also reported highest growth rates for age-2 Arctic cod were observed at 6°C, with decreased survival growth observed at 8°C. However, temperature-dependent growth patterns within the larval period shifted between the first-feeding and later feeding stages. For both species, later stage larvae had higher growth at temperatures >2°C and were more growth sensitive to temperature overall than first-feeding larvae. Although the growth rates of fish typically decline with body size and age (Jobling, 1988; Björnsson et al., 2007), growth rates typically increase through the larval stage (Campana, 1990; Baumann et al., 2006; Otterlei et al., 1999). It has been suggested that this increase in growth with size may be due to improved foraging ability with development at the larval stage (Pepin, 1991). First-feeding larvae also begin exogenous feeding before exhaustion of the yolk sac (Hunter, 1981) when the digestive system is not fully functional and enzyme activity levels remain relatively low (Kolkovski, 2001). These differences in temperature-dependent survival and growth broadly reflect the current distributions of these species in Alaskan waters. Arctic cod are principally restricted to the Chukchi and Beaufort Seas where offshore bottom temperatures seldom exceed 4°C in summer months (Sigler et al., 2011) . Summer surface temperatures vary along an inshore-offshore gradient, ranging from 0 to 2°C in offshore habitats (Crawford et al., 2012) to >14°C in the nearshore (Craig et al., 1982). In the Beaufort Sea, Arctic cod are commonly found shoaling near thermal-salinity fronts separating these surface water masses (Moulton and Tarbox, 1987). Further offshore, Arctic cod are either found in the near-surface waters or deeper, Atlantic-sourced water layer where temperatures range from 0 to 6°C (De Robertis et al., 2017). Although juvenile Arctic cod avoid the coldest intermediate depth waters (<0°C) originating from the Pacific (Crawford et al., 2012), larvae are typically hatching and feeding under extremely cold, unstratified conditions in spring following ice break-up (−1.8 to 0°C), possibly as a means extending the duration of the first summer growth period to achieve sufficient prewinter size (Bouchard and Fortier, 2011). The results from this study confirm that Arctic cod larvae can indeed successfully grow and maintain high survival at 0°C well into the late larval stages. In contrast, walleye pollock occupy a more diverse range of thermal habitats along a very broad latitudinal range extending from the Puget Sound to the Northern Bering Sea (e.g. Bailey et al., 1999). Eggs and larvae can be exposed to spring temperatures <0°C in the Bering Sea (Blood, 2002), and eggs develop and hatch successfully at temperatures between −1.0 and 12°C without experiencing significant malformations or mortality (Blood, 2002; Laurel et al., 2018). Although persistent, cold bottom water (<2°C) on the Bering Sea continental shelf (i.e. “cold pool”) may restrict poleward shifts in adult distribution and spawning activity (Mueter and Litzow, 2008), some portion of eggs are likely advecting into the cold pool (Blood, 2002). Age-0 juvenile stages of pollock are occasionally found in the Chukchi Sea (Logerwell et al., 2015), and although the thermal histories of these walleye pollock are unknown, the observations suggest larvae can indeed successfully develop and feed under some Arctic conditions. However, our laboratory data suggest feeding stages of walleye pollock larvae would undergo high rates of temperature-dependent mortality if spring-early summer conditions remained <2°C into the later larval stages. In the Bering Sea, feeding stage larvae are most associated with higher summer temperatures, more so than feeding conditions and other environmental parameters (Smart et al., 2012). The same study also noted that the temperature-relationships are stage-specific for walleye pollock larvae, with early stages associated with lower temperature and late stages associated with higher temperatures (Smart et al., 2012). These patterns support the observed shift in survival and growth we observed in later stage walleye pollock exposed to higher temperatures, and further emphasize the importance of including ontogeny in predictions of environmental response of marine fish larvae. Interaction of temperature and food availability Complex interactions of sea ice decline are linked to variable productivity (Arrigo and van Dijken, 2011) and will likely impact larval fish survival, growth, feeding, and condition (Thanassekos and Fortier, 2012; Kristiansen et al., 2014). Although we did not fully parameterize growth and survival across a continuous range of temperature-feeding scenarios, this study provided an indication of the relative impact of “match-mismatch” scenarios (Cushing, 1990) for these two species under cold or warm conditions in the spring (first-feeding stages) and early summer (late-feeding stages). In terms of survival, first-feeding walleye pollock exhibited sensitivity to food availability at both 2 and 5°C, while Arctic cod were not significantly impacted. Survival responses to the food environment at this stage are likely the result of species differences in size-at-hatch and corresponding foraging capabilities. At hatch, Arctic cod are substantially larger than walleye pollock (∼6.2 vs. 4.5 mm, respectively) and have 3–6× more yolk reserves (Laurel et al., 2018). These characteristics likely lower the risk to prey mismatch and may contribute to the higher food sensitivity of walleye pollock compared with Arctic cod observed in this study. At the later larval stage, it was unexpected to find that both species experienced higher survival under low food conditions in the 5°C treatment. An examination of the mortality schedule for these treatments indicated that the larvae in the low food ration treatments experienced higher mortality than larvae in the high food ration treatments at the early onset of the experiment. This suggests there was an immediate negative impact of low food conditions on these larvae. The high larval mortality measured in the high food treatments was observed at the end of the experiment, possibly because these larvae were growing faster and approaching the onset of flexion. Ultimately, physiological and morphological changes during larval fish development improve feeding and growth efficiency, but metamorphic changes are considered to be energy-demanding processes which may interfere with growth and survival when feeding ability is compromised during developmental changes (Geffen et al., 2007). It has been shown that the onset of flexion and metamorphosis is generally size-dependent but can also be influenced by both environmental and nutritional factors (Falk-Petersen, 2005). Therefore, larvae in the high food ration tanks may have experienced a size- or growth-dependent critical period near the end of the experiment. Growth sensitivity to temperature and food also shifted with ontogeny. At the first-feeding larval stage, Arctic cod growth was more sensitive to food availability, while walleye pollock were more temperature sensitive. An assessment of growth rates across ontogenetic stages revealed that the temperature sensitivity of Arctic cod increased with age, while it decreased for walleye pollock. Other studies of larval Arctic cod have demonstrated the highly variable nature of the relative impacts of temperature and prey availability on growth. For example, prey density has been shown to limit the feeding success (and subsequent growth) of Arctic cod larvae in Hudson Bay (Fortier et al., 1996), yet contradictory findings from the Northeast Water polynya indicate that larval Arctic cod feeding success is largely determined by temperature with little to no impact of prey density on the feeding success of larvae of all sizes (Michaud et al., 1996). Impacts of climate change The results of this study provide a clear indication that temperature is a key factor determining growth and survival rates in larval Arctic cod and walleye pollock. Laboratory-derived growth rates are increasingly being used to understand the mechanisms impacting growth rates of fish in the wild (Folkvord, 2005), therefore we anticipate the growth models presented here will provide some information on how these species will respond to ocean warming. These data are also important in the development of Individual Based Models (IBMs) incorporating biophysical transport models (e.g. Petrik et al., 2016) and bioclimate envelope models (e.g. Pearson and Dawson, 2003) used to define current and future biogeographies, as well as, better understand the potential for acclimatization of these species to changing conditions (Pörtner and Farrell, 2008). It is important to recognize that the offspring from this study are sources from broodstock that were grown and matured in the laboratory under environmental conditions that differed from natural conditions. The influence of maternal experience on the thermal reaction norms of the embryos studied in this experiment are unknown. Therefore, the derived parameters on growth and survival derived from this study will remain uncertain in the absence of knowing maternal effects on offspring phenotypes. Our results clearly emphasize the need to consider both the species-specific and within species, ontogenetic thermal reaction norms of growth and survival, as marine species from different regions will be exposed to seasonally dynamic changes in ocean temperatures throughout their life history. Distinct temperature-dependent growth models have been used to describe juvenile gadid growth in laboratory experiments (Laurel et al., 2016), but it is clear from this study that stage-specific growth models will be required for larvae during the winter-spring transition in the Arctic. It is also notable that stage-specific models were required for both species across a relatively short developmental period from first-feeding to later larval stages (2–3 months). These shifts in temperature-dependent growth within and between species also change the thermal “tipping-points” where one species has a growth advantage over the other. For example, juvenile growth models indicate that Arctic cod have a growth advantage at low temperatures, but are surpassed by walleye pollock at temperature >2.5°C (Laurel et al., 2016). In contrast, at the larval stage, Arctic cod maintained a growth advantage over a wider range of temperatures, up to 6.7°C without notable changes in observed mortality. Ultimately, changes in extreme temperatures, rather than mean temperatures, may be critical for a species’ persistence in a region (Stachowicz et al., 2002). The low survival and reduced growth potential relative to walleye pollock at 9°C in the laboratory demonstrates that late hatching Arctic cod may experience reduced survival under extreme summer conditions in some areas. The thermal range under which larval Arctic cod can survive, despite being even narrower than the juvenile stage, may still be broad enough to survive or even benefit from further spring warming and early ice breakup in the Chukchi and Beaufort Sea. Long-term scenarios depend on whether ice coverage and cold temperatures in the North Bering and Chukchi Seas will continue to serve as a barrier preventing spawning activity of walleye pollock from advancing poleward. Conclusions In conclusion, Arctic cod were able to maximize growth and survival at lower temperatures than walleye pollock larvae. Rising temperatures and altered productivity regimes associated with climate change in the Arctic have the potential to constrain the habitat available to Arctic cod and, thus, dramatically decrease its competitive strength relative to North Pacific gadid species, like walleye pollock. Temperature-dependent growth models developed in this study emphasize the need to consider species- and stage-specific differences in the growth during the larval period. Furthermore, significant impacts to the growth and survival of Alaskan gadids from continued warming in the Arctic have implications for recruitment and population success of these species and those that prey upon them. Knowledge of the habitat requirements of these ecologically important species is essential for effective resource management, and is key to understanding the broad implications of global change. Acknowledgements We thank the NOAA-AFSC staff at the Hatfield Marine Science Center for the use of facilities, logistical support, and guidance. Thanks to Paul Iseri for assistance with lab construction, tank design, and maintenance of the experimental setup. Thanks also to Scott Haines, Michele Ottmar, and Mara Spencer for assistance in larval husbandry and live food production. Finally, we thank J. Napp, C. Ryer and C. Vestfals for reviewing earlier drafts of this article. This project was supported with funding from the North Pacific Research Board (NPRB) grant number: R1403. This study is NPRB contribution no. 673. The findings and conclusions in the paper are those of the authors and do not necessarily represent the views of the National Marine Fisheries Service, NOAA. References Arrigo K. R. , van Dijken G. L. 2011 . Secular trends in Arctic Ocean net primary production . Journal of Geophysical Research: Oceans , 116 : C09011. Bacheler N. M. , Ciannelli L. , Bailey K. M. , Duffy-Anderson J. T. 2010 . Spatial and temporal patterns of walleye pollock (Theragra chalcogramma) spawning in the eastern Bering Sea inferred from egg and larval distributions . Fisheries Oceanography , 19 : 107 – 120 . Google Scholar Crossref Search ADS Bailey K. M. , Quinn T. J. , Bentzen P. , Grant W. S. 1999 . Population structure and dynamics of walleye pollock, Theragra chalcogramma . Advances in Marine Biology , 37 : 179 – 255 . Google Scholar Crossref Search ADS Barber D. G. , Lukovich J. V. , Keogak J. , Baryluk S. , Fortier L. , Henry G. H. R. 2009 . The changing climate of the Arctic . Arctic , 61 : 7 – 26 . Google Scholar Crossref Search ADS Baumann H. , Grohsler T. , Kornilovs G. , Makarchouk A. , Feldmann V. , Temming A. 2006 . Temperature-induced regional and temporal growth differences in Baltic young-of-the-year sprat Sprattus sprattus . Marine Ecology Progress Series , 319 : 311 – 311 . Beaugrand G. , Kirby R. R. 2010 . Climate, plankton and cod . Global Change Biology , 16 : 1268 – 1280 . Google Scholar Crossref Search ADS Björnsson B. , Steinarsson A. , Oddgeirsson M. 2001 . Optimal temperature for growth and feed conversion of immature cod (Gadus morhua L.) . ICES Journal of Marine Science , 58 : 29 – 38 . Google Scholar Crossref Search ADS Björnsson B. , Steinarsson A. , Árnason T. 2007 . Growth model for Atlantic cod (Gadus morhua): effects of temperature and body weight on growth rate . Aquaculture , 271 : 216 – 226 . Google Scholar Crossref Search ADS Blood D. M. 2002 . Low-temperature incubation of walleye pollock (Theragra chalcogramma) eggs from the southeast Bering Sea shelf and Shelikof Strait, Gulf of Alaska . Deep-Sea Research II , 49 : 6095 – 6108 . Google Scholar Crossref Search ADS Bluhm B. A. , Gradinger R. 2008 . Regional variability in food availability for Arctic marine mammals . Ecological Applications , 18 : S77 – S96 . Google Scholar Crossref Search ADS PubMed Bouchard C. , Fortier L. 2008 . Effects of polynyas on the hatching season, early growth and survival of polar cod Boreogadus saida in the Laptev Sea . Marine Ecology Progress Series , 355 : 247 – 256 . Google Scholar Crossref Search ADS Bouchard C. , Fortier L. 2011 . Circum-arctic comparison of the hatching season of polar cod Boreogadus saida: a test of the freshwater winter refuge hypothesis . Progress in Oceanography , 90 : 105 – 116 . Google Scholar Crossref Search ADS Campana S. E. 1990 . How reliable are growth back-calculations based on otoliths? Canadian Journal of Fisheries Aquatic Sciences , 47 : 2219 – 2227 . Google Scholar Crossref Search ADS Cheung W. W. L. , Lam V. W. Y. , Sarmiento J. L. , Kearney K. , Watson R. , Pauly D. 2009 . Projecting global marine biodiversity impacts under climate change scenarios . Fish and Fisheries , 10 : 235 – 251 . Google Scholar Crossref Search ADS Christiansen J. S. , Mecklenburg C. W. , Karamushko O. V. 2014 . Arctic marine fishes and their fisheries in light of global change . Global Change Biology , 20 : 352 – 359 . Google Scholar Crossref Search ADS PubMed Comiso J. C. , Parkinson C. L. , Gersten R. , Stock L. 2008 . Accelerated decline in the Arctic sea ice cover . Geophysical Research Letters , 35 : L01703. Google Scholar Crossref Search ADS Copeman L. A. , Laurel B. J. 2010 . Experimental evidence of fatty acid limited growth and survival in Pacific cod larvae . Marine Ecology Progress Series , 412 : 259 – 272 . Google Scholar Crossref Search ADS Copeman L. A. , Laurel B. J. , Parrish C. C. 2013 . Effect of temperature and tissue type on fatty acid signatures of two species of North Pacific juvenile gadids: a laboratory feeding study . Journal of Experimental Marine Biology and Ecology , 448 : 188 – 196 . Google Scholar Crossref Search ADS Craig P. C. , Griffiths W. B. , Haldorson L. , McElderry H. 1982 . Ecological studies of Arctic cod (Boreogadus saida) in Beaufort Sea coastal waters, Alaska . Canadian Journal of Fisheries and Aquatic Sciences , 39 : 395 – 406 . Google Scholar Crossref Search ADS Crawford R. E. , Vagle S. , Carmack E. C. 2012 . Water mass and bathymetric characteristics of polar cod habitat along the continental shelf and slope of the Beaufort and Chukchi seas . Polar Biology , 35 : 179 – 190 . Google Scholar Crossref Search ADS Cushing D. H. 1990 . Plankton production and year-class strength in fish populations: an update of the match/mismatch hypothesis . Advances in Marine Biology , 26 : 249 – 293 . Google Scholar Crossref Search ADS De Robertis A. , Taylor K. , Wilson C. D. , Farley E. V. 2017 . Abundance and distribution of Arctic cod (Boreogadus saida) and other pelagic fishes over the U.S. Continental Shelf of the Northern Bering and Chukchi Seas . Deep Sea Research Part II: Topical Studies in Oceanography , 135 : 51 – 65 . Google Scholar Crossref Search ADS Doney S. C. , Ruckelshaus M. , Duffy J. E. , Barry J. P. , Chan F. , English C. A. , Galindo H. M. et al. 2012 . Climate change impacts on marine ecosystems . Annual Review of Marine Science , 4 : 11 – 37 . Google Scholar Crossref Search ADS PubMed Duffy-Anderson J. T. , Barbeaux S. J. , Farley E. , Heintz R. , Horne J. K. , Parker-Stetter S. L. , Petrik C. et al. 2016 . The critical first year of life of walleye pollock (Gadus chalcogrammus) in the eastern Bering Sea: implications for recruitment and future research . Deep-Sea Research Part II-Topical Studies in Oceanography , 134 : 283 – 301 . Google Scholar Crossref Search ADS Falk-Petersen I. B. 2005 . Comparative organ differentiation during early life stages of marine fish . Fish and Shellfish Immunology , 19 : 397 – 412 . Google Scholar Crossref Search ADS PubMed Falardeau M. , Robert D. , Fortier L. 2014 . Could the planktonic stages of polar cod and Pacific sand lance compete for food in the warming Beaufort Sea? ICES Journal of Marine Science , 71 : 1956 – 1965 . Google Scholar Crossref Search ADS Folkvord A. 2005 . Comparison of size-at-age of larval Atlantic cod (Gadus morhua) from different populations based on size- and temperature-dependent growth models . Canadian Journal of Fisheries and Aquatic Sciences , 62 : 1037 – 1052 . Google Scholar Crossref Search ADS Fortier L. , Gilbert M. , Ponton D. , Ingram R. G. , Robineau B. , Legendre L. 1996 . Impact of freshwater on a subarctic coastal ecosystem under seasonal sea ice (southeastern Hudson Bay, Canada). III. Feeding success of marine fish larvae . Journal of Marine Systems , 7 : 251 – 265 . Google Scholar Crossref Search ADS Fossheim M. , Primicerio R. , Johannesen E. , Ingvaldsen R. B. , Aschan M. M. , Dolgov A. V. 2015 . Recent warming leads to a rapid borealization of fish communities in the Arctic . Nature Climate Change , 5 : 673 – 677 . Google Scholar Crossref Search ADS Fouzai N. , Opdal A. F. , Jørgensen C. , Fiksen Ø. 2015 . Effects of temperature and food availability on larval cod survival: a model for behaviour in vertical gradients . Marine Ecology Progress Series , 529 : 199 – 212 . Google Scholar Crossref Search ADS Frey K. E. , Moore G. W. K. , Cooper L. W. , Grebmeier J. M. 2015 . Divergent patterns of recent sea ice cover across the Bering, Chukchi, and Beaufort Seas of the Pacific Arctic Region . Progress in Oceanography , 136 : 32 – 49 . Google Scholar Crossref Search ADS Geffen A. J. , van der Veer H. W. , Nash R. D. M. 2007 . The cost of metamorphosis in flatfishes . Journal of Sea Research , 58 : 35 – 45 . Google Scholar Crossref Search ADS Graham M. , Hop H. 1995 . Aspects of reproduction and larval biology of Arctic cod (Boreogadus saida) . Arctic , 48 : 130 – 135 . Google Scholar Crossref Search ADS Hoegh-Guldberg O. , Bruno J. F. 2010 . The impact of climate change on the world’s marine ecosystems . Science , 328 : 1523 – 1528 . Google Scholar Crossref Search ADS PubMed Holland M. M. , Serreze M. C. , Stroeve J. 2010 . The sea ice mass budget of the Arctic and its future change as simulated by coupled climate models . Climate Dynamics , 34 : 185 – 200 . Google Scholar Crossref Search ADS Houde E. D. 2008 . Emerging from Hjort’s shadow . Journal of Northwest Atlantic Fishery Science , 41 : 53 – 70 . Google Scholar Crossref Search ADS Hunter J. R. 1981 . Feeding ecology and predation of marine fish larvae. In Marine Fish Larvae: Morphology, Ecology, and Relation to Fisheries, pp. 34–77 . Ed. by Lasker R. . University of Washington Press , Seattle, WA, USA . 131 pp. IPCC . 2013 . Long-term Climate Change: Projections, Commitments and Irreversibility. In Climate Change 2013: The Physical Science Basis. IPCC Working Group I Contribution to AR5 , Cambridge University Press , Cambridge, UK and New York, NY, USA . 107 pp. Jobling M. 1988 . A review of the physiological and nutritional energetics of cod, Gadus morhua L., with particular reference to growth under farmed conditions . Aquaculture , 70 : 1 – 19 . Google Scholar Crossref Search ADS Kolkovski S. 2001 . Digestive enzymes in fish larvae and juveniles—implications and applications to formulated diets . Aquaculture , 200 : 181 – 201 . Google Scholar Crossref Search ADS Kristiansen T. , Stock C. , Drinkwater K. F. , Curchitser E. N. 2014 . Mechanistic insights into the effects of climate change on larval cod . Global Change Biology , 20 : 1559 – 1584 . Google Scholar Crossref Search ADS PubMed Kunz K. L. , Frickenhaus S. , Hardenberg S. , Johansen T. , Leo E. , Pörtner H.-O. , Schmidt M. et al. 2016 . New encounters in Arctic waters: a comparison of metabolism and performance of polar cod (Boreogadus saida) and Atlantic cod (Gadus morhua) under ocean acidification and warming . Polar Biology , 39 : 1137 – 1153 . Google Scholar Crossref Search ADS Laurel B. J. , Spencer M. , Iseri P. , Copeman L. A. 2016 . Temperature-dependent growth and behavior of juvenile Arctic cod (Boreogadus saida) and co-occurring North Pacific gadids . Polar Biology , 39 : 1127 – 1135 . Google Scholar Crossref Search ADS Laurel B. J. , Copeman L. A. , Spencer M. , Iseri P. 2018 . Comparative effects of temperature on rates of development and survival of eggs and yolk-sac larvae of Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus) . ICES Journal of Marine Science , 75 : 2403 – 2412 . Leu E. , Søreide J. E. , Hessen D. O. , Falk-Petersen S. , Berge J. 2011 . Consequences of changing sea-ice cover for primary and secondary producers in the European Arctic shelf seas: timing, quantity, and quality . Progress in Oceanography , 90 : 18 – 32 . Google Scholar Crossref Search ADS Logerwell E. , Busby M. , Carothers C. , Cotton S. , Duffy-Anderson J. , Farley E. , Goddard P. et al. 2015 . Fish communities across a spectrum of habitats in the western Beaufort Sea and Chukchi Sea . Progress in Oceanography , 136 : 115 – 132 . Google Scholar Crossref Search ADS Michaud J. , Fortier L. , Rowe P. , Ramseier R. 1996 . Feeding success and survivorship of Arctic cod larvae, Boreogadus saida, in the Northeast Water polynya (Greenland Sea) . Fisheries Oceanography , 5 : 120 – 135 . Google Scholar Crossref Search ADS Miller T. J. , Crowder L. B. , Rice J. A. , Marschall E. A. 1988 . Larval size and recruitment mechanisms in fishes: toward a conceptual framework . Canadian Journal of Fisheries and Aquatic Sciences , 45 : 1657 – 1670 . Google Scholar Crossref Search ADS Moore S. E. , Stabeno P. J. 2015 . Synthesis of Arctic Research (SOAR) in marine ecosystems of the Pacific Arctic . Progress in Oceanography , 136 : 1 – 11 . Google Scholar Crossref Search ADS Moulton L. L. , Tarbox K. E. 1987 . Analysis of Arctic cod movements in the Beaufort Sea nearshore region, 1978-79 . Arctic , 40 : 43 – 49 . Google Scholar Crossref Search ADS Mueter F. J. , Litzow M. A. 2008 . Sea ice retreat alters the biogeography of the Bering Sea continental shelf . Ecological Applications , 18 : 309 – 320 . Google Scholar Crossref Search ADS PubMed Naas K. , Huse I. , Iglesias J. 1996 . Illumination in first feeding tanks for marine fish larvae . Aquacultural Engineering , 15 : 291 – 300 . Google Scholar Crossref Search ADS NPFMC . 2009 . Fishery Management Plan for Fish Resources of the Arctic Management Area . Anchorage , AK . Otterlei E. , Nyhammer G. , Folkvord A. , Stefansson S. O. 1999 . Temperature- and size-dependent growth of larval and early juvenile Atlantic cod (Gadus morhua): a comparative study of Norwegian coastal cod and northeast Arctic cod . Canadian Journal of Fisheries and Aquatic Sciences , 56 : 2099 – 2111 . Google Scholar Crossref Search ADS Pearson R. G. , Dawson T. P. 2003 . Predicting the impacts of climate change on the distribution of species: are bioclimate envelope models useful? Global Ecology and Biogeography , 12 : 361 – 371 . Google Scholar Crossref Search ADS Peck L. S. , Webb K. E. , Bailey D. M. 2004 . Extreme sensitivity of biological function to temperature in Antarctic marine species . Functional Ecology , 18 : 625 – 630 . Google Scholar Crossref Search ADS Pepin P. 1991 . Effect of temperature and size on development, mortality, and survival rates of the pelagic early life history stages of marine fish . Canadian Journal of Fisheries and Aquatic Sciences , 48 : 503 – 518 . Google Scholar Crossref Search ADS Petrik C. M. , Duffy-Anderson J. T. , Castruccio F. , Curchitser E. N. , Danielson S. L. , Hedstrom K. , Mueter F. 2016 . Modelled connectivity between walleye pollock (Gadus chalcogrammus) spawning and age-0 nursery areas in warm and cold years with implications for juvenile survival . ICES Journal of Marine Science , 73 : 1890 – 1900 . Google Scholar Crossref Search ADS Pörtner H. O. , Farrell A. P. 2008 . Physiology and climate change . Science , 322 : 690 – 692 . Google Scholar Crossref Search ADS PubMed Pörtner H. O. , Peck M. A. 2010 . Climate change effects on fishes and fisheries: towards a cause-and-effect understanding . Journal of Fish Biology , 77 : 1745 – 1779 . Google Scholar Crossref Search ADS PubMed Rand K. M. , Logerwell E. A. 2011 . The first demersal trawl survey of benthic fish and invertebrates in the Beaufort Sea since the late 1970s . Polar Biology , 34 : 475 – 488 . Google Scholar Crossref Search ADS Richter-Menge J. , Overland J. E. , Mathis J. T. (Eds). 2016 . Arctic Report Card 2016. NOAA Arctic Program. www.arctic.noaa.gov/Report-Card Sakurai Y. , Ishii K. , Nakatani T. , Yamaguchi H. , Anma G. , Jin M. 1998 . 15 Reproductive characteristics and effects of temperature and salinity on the development and survival of eggs and larvae of Arctic cod (Boreogadus saida) . Memoirs of the Faulty of Fisheries Hokkaido University , 45 : 77 – 89 . Serreze M. C. , Barry R. G. 2011 . Processes and impacts of Arctic amplification: a research synthesis . Global and Planetary Change , 77 : 85 – 96 . Google Scholar Crossref Search ADS Sigler M. , Renner M. , Danielson S. , Eisner L. , Lauth R. , Kuletz K. , Logerwell E. et al. 2011 . Fluxes, fins, and feathers: relationships among the Bering, Chukchi, and Beaufort Seas in a time of climate change . Oceanography , 24 : 250 – 265 . Google Scholar Crossref Search ADS Smart T. I. , Duffy-Anderson J. T. , Horne J. K. , Farley E. V. , Wilson C. D. , Napp J. M. 2012 . Influence of environment on walleye pollock eggs, larvae, and juveniles in the southeastern Bering Sea . Deep-Sea Research Part II-Topical Studies in Oceanography , 65-70 : 196 – 207 . Google Scholar Crossref Search ADS Smetacek V. , Nicol S. 2005 . Polar ocean ecosystems in a changing world . Nature , 437 : 362 – 368 . Google Scholar Crossref Search ADS PubMed Søreide J. E. , Leu E. , Berge J. , Graeve M. , Falk-Petersen S. 2010 . Timing of blooms, algal food quality and Calanus glacialis reproduction and growth in a changing Arctic . Global Change Biology , 16 : 3154 – 3163 . Stachowicz J. J. , Terwin J. R. , Whitlatch R. B. , Osman R. W. 2002 . Linking climate change and biological invasions: ocean warming facilitates nonindigenous species invasions . Proceedings of the National Academy of Sciences of the United States of America , 99 : 15497 – 15500 . Google Scholar Crossref Search ADS PubMed Thanassekos S. , Fortier L. 2012 . An Individual Based Model of Arctic cod (Boreogadus saida) early life in Arctic polynyas: i. Simulated growth in relation to hatch date in the Northeast Water (Greenland Sea) and the North Water (Baffin Bay) . Journal of Marine Systems , 93 : 25 – 38 . Google Scholar Crossref Search ADS Zhao Y. , Chen Y. , Brown J. A. 2001 . Impacts of egg and larval size on survival and growth of Atlantic cod under different feeding conditions . Journal of Fish Biology , 59 : 569 – 581 . Google Scholar Crossref Search ADS Published by International Council for the Exploration of the Sea 2018. This work is written by US Government employees and is in the public domain in the US. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Laurel, Benjamin, J;Copeman, Louise, A;Spencer,, Mara;Iseri,, Paul

doi: 10.1093/icesjms/fsy042pmid: N/A

Abstract Changes in Arctic fish assemblages resulting from climate change will likely be determined by the differential thermal response of key species during their early life history. In this study, we incubated multiple batches of eggs and larvae of two ecologically important gadids co-occurring at the Pacific–Arctic interface, Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus). Fertilized egg batches (n = 11 Arctic cod; n = 6 walleye pollock) were collected in the late winter/early spring from laboratory broodstock held under simulated seasonal environmental conditions. Image and lipid analyses indicated that Arctic cod eggs and larvae were ∼25–35% larger than walleye pollock and had nearly 3–6× more energetic reserves. Two batches of eggs from each species were incubated in replicated containers (n = 3/batch/temperature) at −0.4, 1.2, 2.5, 3.8, 5.0, 9.0, and 12.0°C for Arctic cod and −0.8, 0.3, 2.2, 4.5, 9.0, and 12.0°C for walleye pollock. Both species had very similar low thermal tolerance, but Arctic cod were much more sensitive to higher thermal stress in terms of hatch success and size-at-hatch. For example, Arctic cod hatch success declined precipitously at temperatures above 3.5°C yet remained above 50% in walleye pollock at 9°C. Arctic cod also had significantly longer development times, such that embryos could survive for ∼4 months at temperatures <0°C from the time of spawning to first-feeding. Collectively, these results indicate Arctic cod have a much smaller thermal window for survival, but can survive for longer periods in the absence of food than walleye pollock at cold temperatures. These temperature-dependent rates will be useful in the development of population forecasts and biophysical transport models for these species in the northern Bering, Chukchi, and Beaufort seas. Introduction The early ontogeny of fish species is considered to be most temperature sensitive (Pörtner and Peck, 2010), which has important implications on biogeography by way of upper/lower thermal tolerance and dispersal potential (Bradbury et al., 2008). Regional thermal conditions also can drive mortality rates by way of regulating development and growth rate (“Stage-Duration” hypothesis; Houde, 2008) and yolk-reserve use that determine timing of first-feeding; for example, Match/Mismatch hypothesis (Cushing, 1995). These and other early life-history parameters can be incorporated into individual-based models (IBM) to quantify natural mortality rates (e.g. Cowan et al., 1996) that combined with physical circulation models provide a means of examining how the environment regulates populations and habitat connectivity (Petrik et al., 2015, 2016). Walleye pollock (Gadus chalcogrammus) comprise one of the world’s largest commercial fisheries and are ecologically important in the eastern and western North Pacific. In Alaska, young walleye pollock form a key component of the Bering Sea food web by acting as prey for a variety of seabirds, mammals, and piscivorous species (Aydin and Mueter, 2007). The direct and interactive role of temperature on recruitment processes has been an intensive area of study on walleye pollock since the early 1990s (Napp et al., 2000; Smart et al., 2012; Sigler et al., 2016), and walleye pollock has also been the focus of laboratory studies to examine the effects of temperature on egg and larval development and survival (see Duffy-Anderson et al., 2016 and references therein). Arctic cod (Boreogadus saida) is an ecologically significant species that is adapted to occupy ice-edges, but with declining sea-ice, is hypothesized to have increasing spatial over-lap with competing fish species moving poleward (Rand and Logerwell, 2011; Fossheim et al., 2015). Despite its important role in the Arctic marine food web, little biological information is available for early stages of Arctic cod (Altukhov, 1981; Sakurai et al., 1998), largely due to the difficulty of sampling under the ice during the winter–spring spawning period (Bouchard and Fortier, 2008). To date, there are no data on spawning and developmental rates of Arctic cod from the Alaskan Arctic, i.e. Eastern North Bering, Chukchi, and Beaufort Seas. The absence of early life stage information on this and other Arctic species is an identified knowledge gap, and has hampered modelling efforts forecasting biotic changes in the Arctic resulting from warming and loss of sea-ice (Christiansen et al., 2014). In this study, we compare how temperature impacts development rates in the eggs and larvae of Arctic cod and walleye pollock. Specifically, we compare egg characteristics shortly after fertilization (lipid content, egg size) and subsequently quantify the temperature-dependent effects on hatch timing, hatch success, size-at-hatch, and time-to-starvation. Using these endpoints, we tested the following hypotheses: H1—Arctic cod have a more stenothermic developmental response (narrower thermal range) than walleye pollock; H2—Arctic cod are more temperature sensitive than walleye pollock; that is, higher rates of change with temperature for each developmental metric. We discuss these results in the context of various climate scenarios during the spring larval period at the Pacific–Arctic interface. Methods Parental broodstock Eggs for these experiments were sourced from captive broodstock of Arctic cod and walleye pollock held at the AFSC laboratory in Newport, Oregon. Arctic cod broodstock were collected as age-1 juveniles (70–85 mm SL) by fyke net in late July and early August 2012 from the Beaufort Sea (Prudhoe Bay, AK, 70.383°N–148.552°W). Following shipment, 30 fish (5 males, 25 females) were acclimated to 5°C and maintained in a 6-m diameter round tank with flow-through seawater. Holding temperatures gradually adjusted to 1–2°C during late fall before finally adjusting to 5°C in the late spring of the following year. Walleye pollock broodstock were collected as age-0 juveniles (∼20–30 mm total length) in multiple years (2006–2015) from coastal waters of Puget Sound at Port Townsend, Washington (48.135°N–122.760°W), using a light and lift system. Walleye pollock were reared in the laboratory at 7–9°C during the summer and 2–6°C during the spawning season. At age-3, fish were transferred to a 6-m tank and maintained under the same seasonally adjusted temperatures. Walleye pollock broodstock consisted of a group of 40 age 3–6 individuals estimated at a 1:1 sex ratio. Spawning and rearing system In 2015, fish of both species showed signs of maturity in early March. Arctic cod were checked daily by gently squeezing the abdomen to determine if eggs were freely flowing, clear, and hydrated. Eggs from females meeting these criteria were dispensed into a dry stainless steel bowl nested on ice. Females were returned to their original holding tank alive and milt from three males was added to eggs. Eggs and milt were gently mixed for a 10–20 s period before 1 l of 2°C seawater was added to the gametes. The bowl containing eggs and milt was then transferred to 2°C water baths to maintain constant temperature for a 10 min period. Fertilized eggs were then transferred to meshed 4 l baskets floating in 2°C water baths until distribution in the exposure vessels (below). Water inflow to the baskets was provided through a small plastic tube positioned under the water’s surface. Eggs were collected and incubated in this manner from a total of 12 spawning events (batches) for these experiments. For walleye pollock, fish were allowed to spawn naturally in the tank. Floating eggs were collected daily from the surface outflow of the tank. Although specific parentage of collected eggs was not known, we selected spawning dates with high egg production (reflecting spawning by multiple females) in an attempt to maximize genetic diversity among incubated eggs. Eggs were inspected under a dissecting scope to determine viability before transfer to meshed 4 l baskets held in 2°C water. Eggs were collected and incubated in this manner from a total of six spawning events (batches) for these experiments. Experimental setup and design Two egg batches from each species were selected for use in an extended temperature incubation experiment. Batches were selected on the basis of high fertilization rates (>95%) and consistency in egg shape, size, and colour. From each egg batch of each species, 2 ml of eggs of Arctic cod (∼400 eggs) and 4 ml of eggs from walleye pollock (∼800 eggs) were incubated in a series of 1-L 220 µm mesh-bottom containers. Note that the estimated egg stocking densities for each beaker would be adjusted to actual counts based on daily mortality and hatch counts at the end of the experiment. The mesh-bottom containers were suspended in a series of temperature-controlled, flow-through water baths (66 × 46 × 38 cm) maintained at −0.4, 1.2, 2.5, 3.8, 5.0, 9.0, and 12.0°C for Arctic cod and −0.8, 0.3, 2.2, 4.5, 9.0, and 12.0°C for walleye pollock. Eggs were acclimated to each temperature treatment at rate of 1°C per hour on the day of spawning. Temperatures were maintained within 0.2° of their nominal treatment during the course of the experiment. A total of three replicate containers were used for each egg batch of each species at each temperature treatment for a total of 120 separate containers for egg incubation. Temperature-controlled seawater was supplied to each of the seawater baths at a rate of 2–3 l min−1. Water was exchanged daily by gently lifting and lowering containers in each seawater bath. Eggs from each replicate beaker were monitored daily thereafter to measure mortality and hatch. Eggs were considered dead when they were both opaque in colour and on the bottom of the meshed container. Dead eggs were counted and removed by pipette. As larvae hatched, a subset of individuals were counted and transferred by pipette to a separate corresponding 1 l meshed beaker held at the same temperature to follow larval daily mortality and time to starvation. The transfer of newly hatched larvae to the adjacent beaker was done daily and terminated after 25% of the eggs hatched in the original incubation beaker. Thereafter, newly hatched larvae were counted and transferred to larger feeding tanks (300 l) to develop husbandry protocols for future growth studies. At 50% hatch, a sub-sample of larvae was collected for length/weight measurement (n = 35/beaker/temperature/batch/species; see below). Replicate beakers of eggs and larvae continued to be checked daily until all embryos either hatched or died. Floating eggs that never hatched were counted as “dead” eggs occurring on the last day of observed hatch from the corresponding beaker. Egg/larval measurements Within the first 24 h, eggs from each batch were sampled in triplicate for lipid content (n = 50–100 pooled eggs/sample; see Lipid analyses) and dry mass (DM) (n = 5 pooled eggs/sample). An additional 20 eggs from each batch were imaged and measured under a dissecting scope to determine egg size (diameter) to the nearest 0.01 mm. After eggs had incubated in the experimental treatments, larvae were sampled (n = 35) from each replicate 1 l beaker when 50% of the eggs in the beaker had hatched. Sampled larvae from each replicate beaker were anesthetized using a 0.0005 ppm solution of tricane methanesulfonate (MS-222) prior to image and weight measurements of: (i) standard length (SL) in mm (n = 10 individuals/beaker), (ii) myotome height (body depth) at the anus (MH) in mm (n = 10 individuals/beaker), and (iii) triplicate measures of DM in milligrams (mg) using pooled larvae (n = 5/foil). All larvae were rinsed in 2 ml of 3% ammonium formate solution on a 24 mm stainless steel screen to rid excess salt and inorganic material. Rinsed eggs and larvae were then transferred to a 1 × 1 cm pre-weighed foil and placed into a drying oven set at 60°C. Sample and foils were reweighed 48 h later. Individual DM were calculated by subtracting the known foil weight and dividing by the number of individuals on the foil. All DM measurements were conducted on a microbalance (Sartorius R16OP) to the nearest µg. Lipid analyses Lipid analyses were conducted on both batches of each species used in the temperature incubation experiments. Additional lipid analyses were conducted on the remaining Arctic cod batches to determine if female size was linked to lipid content variance among batches. Lipids were extracted in chloroform/methanol according to Parrish (1998) using a modified Folch procedure (Folch et al., 1957). Larvae were individually pipetted onto a pre-combusted 47 mm glass fibre filter (Watman GF C) and rinsed with filtered seawater. The filter containing the larvae was then transferred to a lipid-clean vial, immersed in chloroform and frozen at –20°C until the end of the experiment after which they underwent lipid analysis, within 6 months of larval sampling. Total lipids were determined using thin-layer chromatography with flame-ionization detection (TLC/FID) with a MARK V Iatroscan (Iatron Laboratories, Tokyo, Japan) and modified from those methods described by Lu et al. (2008) as in Copeman et al. (2016). Total lipids were expressed in absolute amounts (lipid per egg and lipid per DM, µg g−1). Data analysis Statistical analyses were conducted in Systat (version 11.0, Systat Software, Inc.: www.systat.com). A general linear model (GLM) was used to statistically examine the effects of “species,” “temperature,” and “batch” on relationships of hatch success (%), time to 50% hatch (days), hatch duration (days), size-at-first hatch (mm SL), and time to starvation (days). Temperature was considered a categorical variable in the model and restricted to two nominal treatment conditions (-1.0 and 2.0°C) where batches of both species experienced sufficient survival to characterize the dependent variable. Interactions were explored between “species” and “temperature” whereas “batch” was nested within the “species” term of the model. Tukey's range tests were used to determine significant differences among means in each model. Data were checked for normality using the Kolmogorov–Smirnov test. Following statistical analyses, temperature relationships for hatch and survival characteristics for each species were described by a series of 2- and 3-parameter non-linear regression models over the full temperature range. For visual purposes in figures, batches were modelled separately for each species when significant differences were detected in the Tukey’s range tests. In instances where “batch” significantly interacted with the species term in the 3-way GLM separate 2-way GLMs were run for each species to determine if the batch effects remained significant within each species to justify fitting batch-specific non-linear regressions. However, in the model summary table, batches were pooled to model temperature relationships on species level regardless of statistical differences. Results Spawning and egg batch characteristics Spawning in the laboratory of Arctic cod occurred during a relatively discrete period (March 3–17) and was almost identical in the 2016 and 2017 spawning periods that were observed outside the egg incubation study. In contrast, walleye pollock spawning activity extended over a 2 month period beginning in early April, although no additional batches were collected after 20 April 2016 (Table 1). Females of both species showed signs of swelling and reduced interest in food 6–8 weeks prior to the spawning period. Arctic cod males repeatedly spawned within a season, but females released all their eggs in a single batch and did not spawn again during that year. Two of the Arctic cod females released their eggs naturally during the night, although these eggs were not included in the batches analyzed in this study. All the remaining females released their eggs during the day using strip spawning described in Methods. Walleye pollock released all of their eggs naturally at night and were collected from the egg collectors the following morning. Nearly all the Arctic cod females that spawned in 2015 spawned again in both 2016 and 2017, indicating iteroparity. Walleye pollock females also survived across multiple years of spawning activity in the lab, although actual parentage of egg batches within and across years was not determined. Table 1. Descriptive characteristics of egg batches of AC and WP collected from broodstock held in the laboratory. Species Date Egg batch Female size (cm TL) Egg diameter (mm) Egg dry mass (mg egg−1) Total Lipid (µg egg−1) Lipid density (µg mg dry mass−1) Arctic cod 3/3/15 1 29.0 1.65 ± 0.03 0.221 ± 0.011 33.37 ± 4.59 151.00 ± 14.09 3/3/15 2 32.0 1.62 ± 0.02 0.191 ± 0.001 26.03 ± 5.23 136.28 ± 15.42 3/5/15 3 30.5 1.66 ± 0.05 0.208 ± 0.007 31.34 ± 8.36 150.67 ± 17.32 3/6/15 4 26.5 1.65 ± 0.05 0.192 ± 0.005 26.40a 137.50 3/6/15 5 25.0 1.79  ± 0.05 0.240 ± 0.001 40.59 ± 3.35 169.13 ± 8.30 3/8/15 6 25.0 1.62  ± 0.04 0.202 ± 0.001 22.32 ± 2.26 110.50 ± 7.59 3/8/15 7 28.0 1.64 ± 0.06 0.189 ± 0.002 21.25a 112.43 3/9/15 8 30.5 1.69 ± 0.07 0.199 ± 0.005 na 3/9/15 9 30.5 1.64 ± 0.05 0.144 ± 0.005 na 3/11/15 10 32.0 1.70 ± 0.05 0.198 ± 0.004 20.65 ± 3.14 104.29 ± 7.21 3/17/15 11 32.0 1.73 ± 0.05 0.207 ± 0.001 24.93 ± 2.18 120.43 ± 6.89 Walleye pollock 4/3/15 1 na 1.28 ± 0.05 0.101 ± 0.001 6.65 ± 2.17 65.84 ± 5.12 4/13/15 2 na 1.32 ± 0.04 0.124 ± 0.002 7.35 ± 1.62 59.27 ± 4.69 4/14/15 3 na 1.31 ± 0.03 0.125 ± 0.002 na 4/15/15 4 na 1.28 ± 0.04 0.114 ± 0.002 na 4/17/15 5 na 1.29 ± 0.03 0.113 ± 0.003 na 4/20/15 6 na 1.32 ± 0.04 0.120 ± 0.002 na Species Date Egg batch Female size (cm TL) Egg diameter (mm) Egg dry mass (mg egg−1) Total Lipid (µg egg−1) Lipid density (µg mg dry mass−1) Arctic cod 3/3/15 1 29.0 1.65 ± 0.03 0.221 ± 0.011 33.37 ± 4.59 151.00 ± 14.09 3/3/15 2 32.0 1.62 ± 0.02 0.191 ± 0.001 26.03 ± 5.23 136.28 ± 15.42 3/5/15 3 30.5 1.66 ± 0.05 0.208 ± 0.007 31.34 ± 8.36 150.67 ± 17.32 3/6/15 4 26.5 1.65 ± 0.05 0.192 ± 0.005 26.40a 137.50 3/6/15 5 25.0 1.79  ± 0.05 0.240 ± 0.001 40.59 ± 3.35 169.13 ± 8.30 3/8/15 6 25.0 1.62  ± 0.04 0.202 ± 0.001 22.32 ± 2.26 110.50 ± 7.59 3/8/15 7 28.0 1.64 ± 0.06 0.189 ± 0.002 21.25a 112.43 3/9/15 8 30.5 1.69 ± 0.07 0.199 ± 0.005 na 3/9/15 9 30.5 1.64 ± 0.05 0.144 ± 0.005 na 3/11/15 10 32.0 1.70 ± 0.05 0.198 ± 0.004 20.65 ± 3.14 104.29 ± 7.21 3/17/15 11 32.0 1.73 ± 0.05 0.207 ± 0.001 24.93 ± 2.18 120.43 ± 6.89 Walleye pollock 4/3/15 1 na 1.28 ± 0.05 0.101 ± 0.001 6.65 ± 2.17 65.84 ± 5.12 4/13/15 2 na 1.32 ± 0.04 0.124 ± 0.002 7.35 ± 1.62 59.27 ± 4.69 4/14/15 3 na 1.31 ± 0.03 0.125 ± 0.002 na 4/15/15 4 na 1.28 ± 0.04 0.114 ± 0.002 na 4/17/15 5 na 1.29 ± 0.03 0.113 ± 0.003 na 4/20/15 6 na 1.32 ± 0.04 0.120 ± 0.002 na Batches used in temperature incubation experiments are highlighted in grey. Egg diameters are based on means (±1 SD) of 25–30 eggs per batch. Egg dry mass is based on means (±1 SD) of three pooled samples (n = 5 individuals foil−1) for each batch. Lipid content is based on means (±1 SD) of two pooled samples (n = 100 eggs sample−1) per batch. a On the basis of a single pooled sample of 100 eggs. Table 1. Descriptive characteristics of egg batches of AC and WP collected from broodstock held in the laboratory. Species Date Egg batch Female size (cm TL) Egg diameter (mm) Egg dry mass (mg egg−1) Total Lipid (µg egg−1) Lipid density (µg mg dry mass−1) Arctic cod 3/3/15 1 29.0 1.65 ± 0.03 0.221 ± 0.011 33.37 ± 4.59 151.00 ± 14.09 3/3/15 2 32.0 1.62 ± 0.02 0.191 ± 0.001 26.03 ± 5.23 136.28 ± 15.42 3/5/15 3 30.5 1.66 ± 0.05 0.208 ± 0.007 31.34 ± 8.36 150.67 ± 17.32 3/6/15 4 26.5 1.65 ± 0.05 0.192 ± 0.005 26.40a 137.50 3/6/15 5 25.0 1.79  ± 0.05 0.240 ± 0.001 40.59 ± 3.35 169.13 ± 8.30 3/8/15 6 25.0 1.62  ± 0.04 0.202 ± 0.001 22.32 ± 2.26 110.50 ± 7.59 3/8/15 7 28.0 1.64 ± 0.06 0.189 ± 0.002 21.25a 112.43 3/9/15 8 30.5 1.69 ± 0.07 0.199 ± 0.005 na 3/9/15 9 30.5 1.64 ± 0.05 0.144 ± 0.005 na 3/11/15 10 32.0 1.70 ± 0.05 0.198 ± 0.004 20.65 ± 3.14 104.29 ± 7.21 3/17/15 11 32.0 1.73 ± 0.05 0.207 ± 0.001 24.93 ± 2.18 120.43 ± 6.89 Walleye pollock 4/3/15 1 na 1.28 ± 0.05 0.101 ± 0.001 6.65 ± 2.17 65.84 ± 5.12 4/13/15 2 na 1.32 ± 0.04 0.124 ± 0.002 7.35 ± 1.62 59.27 ± 4.69 4/14/15 3 na 1.31 ± 0.03 0.125 ± 0.002 na 4/15/15 4 na 1.28 ± 0.04 0.114 ± 0.002 na 4/17/15 5 na 1.29 ± 0.03 0.113 ± 0.003 na 4/20/15 6 na 1.32 ± 0.04 0.120 ± 0.002 na Species Date Egg batch Female size (cm TL) Egg diameter (mm) Egg dry mass (mg egg−1) Total Lipid (µg egg−1) Lipid density (µg mg dry mass−1) Arctic cod 3/3/15 1 29.0 1.65 ± 0.03 0.221 ± 0.011 33.37 ± 4.59 151.00 ± 14.09 3/3/15 2 32.0 1.62 ± 0.02 0.191 ± 0.001 26.03 ± 5.23 136.28 ± 15.42 3/5/15 3 30.5 1.66 ± 0.05 0.208 ± 0.007 31.34 ± 8.36 150.67 ± 17.32 3/6/15 4 26.5 1.65 ± 0.05 0.192 ± 0.005 26.40a 137.50 3/6/15 5 25.0 1.79  ± 0.05 0.240 ± 0.001 40.59 ± 3.35 169.13 ± 8.30 3/8/15 6 25.0 1.62  ± 0.04 0.202 ± 0.001 22.32 ± 2.26 110.50 ± 7.59 3/8/15 7 28.0 1.64 ± 0.06 0.189 ± 0.002 21.25a 112.43 3/9/15 8 30.5 1.69 ± 0.07 0.199 ± 0.005 na 3/9/15 9 30.5 1.64 ± 0.05 0.144 ± 0.005 na 3/11/15 10 32.0 1.70 ± 0.05 0.198 ± 0.004 20.65 ± 3.14 104.29 ± 7.21 3/17/15 11 32.0 1.73 ± 0.05 0.207 ± 0.001 24.93 ± 2.18 120.43 ± 6.89 Walleye pollock 4/3/15 1 na 1.28 ± 0.05 0.101 ± 0.001 6.65 ± 2.17 65.84 ± 5.12 4/13/15 2 na 1.32 ± 0.04 0.124 ± 0.002 7.35 ± 1.62 59.27 ± 4.69 4/14/15 3 na 1.31 ± 0.03 0.125 ± 0.002 na 4/15/15 4 na 1.28 ± 0.04 0.114 ± 0.002 na 4/17/15 5 na 1.29 ± 0.03 0.113 ± 0.003 na 4/20/15 6 na 1.32 ± 0.04 0.120 ± 0.002 na Batches used in temperature incubation experiments are highlighted in grey. Egg diameters are based on means (±1 SD) of 25–30 eggs per batch. Egg dry mass is based on means (±1 SD) of three pooled samples (n = 5 individuals foil−1) for each batch. Lipid content is based on means (±1 SD) of two pooled samples (n = 100 eggs sample−1) per batch. a On the basis of a single pooled sample of 100 eggs. Egg diameter was larger in Arctic cod (1.62–1.79 mm) compared with walleye pollock (1.28–1.32 mm), and there was more variation in size among the batches of Arctic cod. Egg size variation among Arctic cod batches was not linked to female size (F1, 10 = 0.0597, p = 0.8124), but was weakly predictive (albeit statistically insignificant) of egg DM (F1, 10 = 4.7126, p = 0.0580). Lipid content (μg egg-1) was also 2–6 times higher than for walleye pollock, with high variation noted among Arctic cod batches. The lipid content variation among batches of Arctic cod was not linked to female size or egg size (F1, 8 = 2.8739, p = 0.1338), although the largest eggs (Batch 5) also had the highest lipid content among all batches analyzed (Table 1). Temperature effects on survival and hatch timing Time-to-hatch decreased with increasing temperatures in both species, but Arctic cod eggs developed at much slower rates than walleye pollock at comparative temperatures; that is, ∼20–30 days longer from the upper to lower temperature range (Figures 1 and 2; Table 2). Statistical analysis indicated a significant interaction in the species and temperature term of the model (F1, 18 = 449.961, p < 0.001) as well as a significant batch effect (F1, 18 = 19.061, p = 0.001). The species-specific GLM indicated that the batch interaction was driven by 2–3 day earlier hatching in low-lipid batch of Arctic cod (Tukey’s HSD = 1.833, p < 0.001) with no change in hatch timing between the two in walleye pollock batches (Tukey’s HSD = 0.083, p = 0.992). Table 2. Temperature-dependent regression models describing hatch characteristics of AC (Boreogadus saida) and WP (Gadus chalcogrammus) based on incubation experiments. Species Temperature function (T) Batch effect? Model r2 Arctic cod Time to 50% hatch (days) Y f=61.131-11.329T+0.982T2 0.98 Hatch success (%) N f=87.926+1.266T-3.658T2 0.98 Size-at-hatch (mm SL) N f=6.872-0.190T-0.040T2 0.92 Time to 50% starvation (days post-hatch) N f=47.214-6.390T+2.856T2-0.586T3 0.95 Walleye pollock Time to 50% hatch (days) N f=7.606+22.635e-0.319T 0.98 Hatch success (%) Y f=72.127+5.775T-0.801T2 0.80 Size-at-hatch (mm SL) N f=4.867+0.020T-0.008T2 0.85 Time to 50% starvation (days post-hatch) N f=33.943-2.808T+0.037T2 0.94 Species Temperature function (T) Batch effect? Model r2 Arctic cod Time to 50% hatch (days) Y f=61.131-11.329T+0.982T2 0.98 Hatch success (%) N f=87.926+1.266T-3.658T2 0.98 Size-at-hatch (mm SL) N f=6.872-0.190T-0.040T2 0.92 Time to 50% starvation (days post-hatch) N f=47.214-6.390T+2.856T2-0.586T3 0.95 Walleye pollock Time to 50% hatch (days) N f=7.606+22.635e-0.319T 0.98 Hatch success (%) Y f=72.127+5.775T-0.801T2 0.80 Size-at-hatch (mm SL) N f=4.867+0.020T-0.008T2 0.85 Time to 50% starvation (days post-hatch) N f=33.943-2.808T+0.037T2 0.94 Model fits are based on pooled batches of replicate beakers for each temperature (n = 6) from each species. Table 2. Temperature-dependent regression models describing hatch characteristics of AC (Boreogadus saida) and WP (Gadus chalcogrammus) based on incubation experiments. Species Temperature function (T) Batch effect? Model r2 Arctic cod Time to 50% hatch (days) Y f=61.131-11.329T+0.982T2 0.98 Hatch success (%) N f=87.926+1.266T-3.658T2 0.98 Size-at-hatch (mm SL) N f=6.872-0.190T-0.040T2 0.92 Time to 50% starvation (days post-hatch) N f=47.214-6.390T+2.856T2-0.586T3 0.95 Walleye pollock Time to 50% hatch (days) N f=7.606+22.635e-0.319T 0.98 Hatch success (%) Y f=72.127+5.775T-0.801T2 0.80 Size-at-hatch (mm SL) N f=4.867+0.020T-0.008T2 0.85 Time to 50% starvation (days post-hatch) N f=33.943-2.808T+0.037T2 0.94 Species Temperature function (T) Batch effect? Model r2 Arctic cod Time to 50% hatch (days) Y f=61.131-11.329T+0.982T2 0.98 Hatch success (%) N f=87.926+1.266T-3.658T2 0.98 Size-at-hatch (mm SL) N f=6.872-0.190T-0.040T2 0.92 Time to 50% starvation (days post-hatch) N f=47.214-6.390T+2.856T2-0.586T3 0.95 Walleye pollock Time to 50% hatch (days) N f=7.606+22.635e-0.319T 0.98 Hatch success (%) Y f=72.127+5.775T-0.801T2 0.80 Size-at-hatch (mm SL) N f=4.867+0.020T-0.008T2 0.85 Time to 50% starvation (days post-hatch) N f=33.943-2.808T+0.037T2 0.94 Model fits are based on pooled batches of replicate beakers for each temperature (n = 6) from each species. Figure 1. View largeDownload slide The effects of temperature on cumulative hatch (%) of eggs of Arctic cod (AC; Boreogadus saida) and walleye pollock (WP; Gadus chalcogrammus). Triangles represent higher lipid content batches (AC Batch 5 and WP Batch 1) and circles represent lower-lipid batches (AC Batch 6 and WP Batch 2). Data are means ±1 SD based on counts of hatched larvae in three replicate tanks at each temperature. Note, displayed data restricted to temperature treatments where successful hatching was observed among all batches (i.e. >−1 and <5°C). Figure 1. View largeDownload slide The effects of temperature on cumulative hatch (%) of eggs of Arctic cod (AC; Boreogadus saida) and walleye pollock (WP; Gadus chalcogrammus). Triangles represent higher lipid content batches (AC Batch 5 and WP Batch 1) and circles represent lower-lipid batches (AC Batch 6 and WP Batch 2). Data are means ±1 SD based on counts of hatched larvae in three replicate tanks at each temperature. Note, displayed data restricted to temperature treatments where successful hatching was observed among all batches (i.e. >−1 and <5°C). Figure 2. View largeDownload slide The effects of temperature on time to 50% hatch in eggs of Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus). Fitted lines are the results of 2-parameter regression models listed in Table 2. Solid and dashed lines represent model fits to separate batches within a species where significant differences were detected in the general linear model. Data represent observations from individual replicate 1 l beakers (n = 3/treatment). Figure 2. View largeDownload slide The effects of temperature on time to 50% hatch in eggs of Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus). Fitted lines are the results of 2-parameter regression models listed in Table 2. Solid and dashed lines represent model fits to separate batches within a species where significant differences were detected in the general linear model. Data represent observations from individual replicate 1 l beakers (n = 3/treatment). Overall, hatch success in Arctic cod was much more temperature-sensitive than walleye pollock, indicated by the narrower thermal range and steep decline in the proportion of eggs that hatched above 3°C (Figure 3; Table 2). This pattern was reflected statistically in the GLM by way of a significant interaction in the species and temperature term of the model (F1, 18 = 93.186, p < 0.001). The GLM also indicated a significant batch effect (F2, 18 = 27.977, p < 0.001), which was only apparent in walleye pollock, driven by higher hatch success overall in Batch 1 versus Batch 2 eggs (Tukey HSD = 16.005, p <0.001; Figure 3). Figure 3. View largeDownload slide The effects of temperature on hatch success (%) in Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus). Fitted lines are the results of 2-parameter regression models listed in Table 2. Solid and dashed lines represent model fits to separate batches within a species where significant differences were detected in the general linear model. Data represent observations from individual replicate 1 l beakers (n = 3/treatment). Figure 3. View largeDownload slide The effects of temperature on hatch success (%) in Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus). Fitted lines are the results of 2-parameter regression models listed in Table 2. Solid and dashed lines represent model fits to separate batches within a species where significant differences were detected in the general linear model. Data represent observations from individual replicate 1 l beakers (n = 3/treatment). Size-at-hatch (SL mm) decreased with temperature, but the pattern was more strongly apparent in Arctic cod compared with walleye pollock (Figure 4; Table 2). This pattern was reflected in the significant species * temperature interaction term of the GLM (F1, 18 = 68.071, p < 0.001) and there were no significant differences within batches of either species (F2, 18 = 3.138, p = 0.068). In terms of body depth (MH mm), Arctic cod were larger and temperature had a positive effect on MH on both species within the temperature range of the statistical model (p < 0.001). However, the non-linear model fits across the full range of temperatures indicated that a relationship was only apparent in Arctic cod; largest Arctic cod larvae were apparent between the 1.2°C and 2.5°C treatment (Figure 5; Table 2). There was also a batch effect detected in Arctic cod (Tukey HSD = 13.030, p < 0.001), with larger eggs (Batch 5) producing larger larvae (Figure 5). Figure 4. View largeDownload slide The effects of temperature on size-at-hatch (standard length [SL] mm) in Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus). Fitted lines are the results of 2-parameter regression models listed in Table 2. Solid and dashed lines represent model fits to separate batches within a species where significant differences were detected in the general linear model. Data represent observations from individual replicate 1 l beakers (n = 3/treatment). Figure 4. View largeDownload slide The effects of temperature on size-at-hatch (standard length [SL] mm) in Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus). Fitted lines are the results of 2-parameter regression models listed in Table 2. Solid and dashed lines represent model fits to separate batches within a species where significant differences were detected in the general linear model. Data represent observations from individual replicate 1 l beakers (n = 3/treatment). Figure 5. View largeDownload slide The effects of temperature on body depth (myotome height [MH] mm) in Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus). Fitted lines are the results of 2-parameter regression models listed in Table 2. Solid and dashed lines represent model fits to separate batches within a species where significant differences were detected in the general linear model. Data represent observations from individual replicate 1 l beakers (n = 3/treatment). Figure 5. View largeDownload slide The effects of temperature on body depth (myotome height [MH] mm) in Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus). Fitted lines are the results of 2-parameter regression models listed in Table 2. Solid and dashed lines represent model fits to separate batches within a species where significant differences were detected in the general linear model. Data represent observations from individual replicate 1 l beakers (n = 3/treatment). Following hatch, temperature negatively impacted the time in which both species could survive in the absence of food (F1, 18 = 77.397, p < 0.001), but Arctic cod yolk-sac larvae were able to survive 10–14 days longer than walleye pollock across the lower temperature range (−1 to 3.5°C; F1, 18 = 79, 763, p < 0.001). The precipitous decline in time to starvation for Arctic cod larvae in the 5°C treatment (Batch 2 only) suggests a temperature-species interaction, indicated by way of the non-linear model fits (see Figure 6; Table 2). However, this was not evident over the temperature range used in the statistical model, that is, −1.0 to 2.0°C (F2, 18 = 0.845, p = 0.135). Figure 6. View largeDownload slide The effects of temperature on days to 50% starvation (days post-hatch) in Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus). Fitted lines are the results of 3-parameter regression models listed in Table 2. Data represent observations from individual replicate 1 l beakers (n = 3/treatment). Figure 6. View largeDownload slide The effects of temperature on days to 50% starvation (days post-hatch) in Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus). Fitted lines are the results of 3-parameter regression models listed in Table 2. Data represent observations from individual replicate 1 l beakers (n = 3/treatment). Discussion Hatch success The species differences in temperature-dependent hatch successes were striking, and clearly demonstrate why Arctic cod are limited to higher latitudes where winter–spring temperatures are predictably <0°C. Hatch success in Arctic cod precipitously declined above 2°C, yet remained relatively high in walleye pollock across a broad range of temperatures up to 12°C. The actual upper thermal tolerance for walleye pollock eggs in Alaskan waters has not been examined (Blood et al., 1994; Blood, 2002), although eggs sourced from the Western Pacific also maintain high hatching rates up to temperatures of 10°C (Nakatani and Maeda, 1984). The relatively narrow thermal range over which Arctic cod eggs successfully hatched is similar to the upper thermal limits observed (3–3.5°C) in other studies; (Sakurai et al., 1998; Kent et al., 2016) and near the maximum temperature (3.3°C where hearts fail to keep up with further temperature increases, Drost et al., 2016a). These data suggest suitable spawning habitats are much more spatially and temporally restricted for Arctic cod than for walleye pollock. This is in distinct contrast to older juvenile stages, where laboratory studies indicate that Arctic cod have greater cardiovascular resilience to temperature acclimation (Drost et al., 2016b) and can successfully grow across a much broader range of thermal environments (−1 to 12°C; Laurel et al., 2017). The significant change in the ontogenetic thermal response of Arctic cod appears to be an important biological characteristic to consider in population or range shift models for this species. The Northern Bering Sea is considered to be the upper latitudinal range for walleye pollock, yet hatch success was relatively high in the -0.6°C treatment. Although eggs were not incubated at temperatures near freezing (−1.8°C), there is no evidence to suggest that low thermal tolerance of eggs is restricting spawning in the higher Arctic. More likely, indirect effects of temperature on egg distributions, by way of spawning habitat preference of adults and/or advective processes, limit northern range limits of walleye pollock at the Bering–Chukchi interface (Mueter et al., 2011; Hollowed et al., 2012). In addition to a dynamic thermal environment, Arctic cod eggs are exposed to a broad range of salinities resulting from ice melt, river input, and Bering Sea water through the Bering Strait (Danielson et al., 2017). Bouchard and Fortier (2008) found that increased freshwater input was associated with earlier hatch timing in Arctic cod. However, given Arctic cod eggs are highly temperature sensitive, this association is likely due to the warming influence of the freshwater rather than a salinity cue on hatch timing. In the Eastern Bering Sea, egg distributions of walleye pollock are more strongly linked to temperature than salinity (Smart et al., 2012). And despite the influence that salinity can have on hatch success in cod species, most species are generally tolerant to a broad range of salinities before there is notable impacts on survival (Nissling and Westin, 1991). For Arctic cod, Sakurai et al. (1998) reported a relatively broad range of salinities over which eggs can survive to hatch (12.9–51.6 psu), with highest (70%) survival occurring in 32.1 and 40.8 psu treatments compared with <20% survival at 22.5 and 51.6 psu. Thermal conditions are therefore more likely driving hatch characteristics and larval distributions of these species during the winter–spring transition. Time to 50% hatch Prior studies on the developmental rate of walleye pollock indicate that temperature may affect populations differently. Blood (2002) found that Shelikof Strait walleye pollock eggs developed more slowly at 2°C compared with eggs collected from the Bering Sea. Earlier work by Blood et al. (1994) also found significantly longer development times in western versus eastern North Pacific walleye pollock populations across a broader range of incubation temperatures. Development times for Arctic cod ranged from 64 to 67 days at −0.4°C to 31–35 days at 3.8°C. The temperature-dependent hatch-timing model from this study closely matched estimates and measurements reported from other regions. These include reports of 77–79 days at −1.5°C and 35 days at 1.5°C of eggs sourced from the White Sea (Aronovich et al., 1975; Altukhov, 1981), to 43–44 days at 2°C from broodstock collected around Resolute Bay, Canada (Graham and Hop, 1995) and 35–75 days across −1 to 3°C thermal range from broodstock collected from St. Lawrence Island (Sakurai et al., 1998). Given that spawning is occurring in winter months under the ice in most regions, the egg development times for Arctic cod at temperatures <0°C are predicted to be between 61 and 85 days based on the hatch timing model from this study. As with hatch success, the temperature effects on time to 50% hatch were profoundly different between the two gadid species. Although evidence suggests early life stages of Arctic cod are metabolically more active than walleye pollock at cold temperatures (Laurel et al., 2016), walleye pollock hatched earlier across all temperature treatments. The factors contributing to variable hatch timing both within and across fish species has been linked to egg size; that is, larger eggs require longer development times (Pauly and Pullin, 1988; Pepin, 1991). In our study, Arctic cod time to 50% hatch was about two times longer than that of walleye pollock across all temperatures where both species successfully developed. The larger Arctic cod egg batch also developed 2–3 days longer than the small egg batch at each incubation temperature. However, there was no statistical difference in hatch timing between the two batches of walleye pollock eggs. Walleye pollock egg size is reported to have an influence on hatch timing at colder temperature (<2°C; Blood, 2002), but the available data suggest mixed patterns rather than a general trend of egg-size influence on development rates. For example, Bering Sea walleye pollock mean egg sizes are larger (1.45–1.72) than those from Shelikof Strait (1.32–1.47), yet Shelikof Strait eggs take ∼10 days longer to hatch at 0°C (Blood, 2002). In another study, larger eggs from the Bering Sea (1.45–1.72 mm) develop more slowly than the smaller eggs collected from Shelikof Strait (Kendall, 2001). These mixed trends may reflect interactions in the temperature-dependent metabolic rates of certain populations of walleye pollock with the overall role of egg size on development rates. Size-at-hatch Size-at-hatch in fish larvae is highly variable both within and across related species (Pepin, 1991). Larger eggs generally produce larger larvae (e.g. Marteinsdottir and Steinarsson, 1998), so it was not surprising that Arctic cod were significantly larger than walleye pollock at hatch. Within-species variation is often attributed to population differences or the impact of hatch timing within a batch of eggs (Chambers et al., 1989; Laurel et al., 2008). The role of temperature on size-at-hatch is poorly understood, but is a well-known phenomenon in marine fish larvae across a range of taxa (Pauly and Pullin, 1988; Blaxter, 1991; Pepin et al., 1997). While previous studies have described the influence of temperature on size-at-hatch in walleye pollock, the variation has been relatively small. For example, Blood et al. (1994) reports a 0.4−mm change in the length across a 3.9°C temperature range. In our study, size-at-hatch of walleye pollock larvae varied by ∼0.6 mm SL over 12.8°C. In contrast, Arctic cod mean length at hatch decreased strongly with temperature by ∼1.5 mm SL over a 5.4°C temperature range. The ecological significance of variable size-at-hatch may be associated with foraging or swimming capabilities of individual larvae, although seldom has this been tested (Miller et al., 1988). Porter and Bailey (2007) found that late-hatching walleye pollock (larger) larvae were more responsive to predator signals than early-hatching larvae. In Pacific cod, larger larvae resulting from later hatch may have reduced yolk reserves and a shortened time-period to begin feeding on exogenous food sources (Laurel et al., 2008). However, in Arctic cod, there appears to be no trade-off in larval size and lipid reserves at hatch (Copeman and Laurel, unpublished data), suggesting that hatching at lower temperatures provides aa dual benefit for this species. Temperature-dependent post-hatch survival The upper thermal tolerance for post-hatch embryonic larvae was not fully explored in this study, as post-hatch treatments were limited to the thermal range of egg incubation. Early feeding stages of Arctic cod larvae acclimated to 7°C were able to grow for 4 weeks under laboratory conditions, but suffered higher mortality rates than larvae incubated at temperature <5°C (Koenker et al., in review). Unfed Arctic cod embryos were also able to survive longer in the absence of food than walleye pollock at cold temperatures; for example, 50% survival in Arctic cod yolk-sac larvae at 45 days post-hatch at 1.2°C. This is considerably longer than observations of Arctic cod yolk-sac larvae reported by Sakurai et al. (1998) of ∼30% survival in Arctic cod yolk-sac larvae at 30 days post-hatch at 1.5°C. Some of these differences may also be due to population or parental effects, although there was no statistical differences in post-hatch survival detected between the two Arctic cod batches in our study that had notable differences in lipid content. The observed differences may be attributable to the timing in which newly hatched larvae were transferred into the starvation tanks for each of the studies. Larvae from this study were immediately transferred at the beginning of the hatch cycle whereas Sakurai et al. (1998) initiated experiments when larvae were at 50% hatch with a higher likelihood of lower yolk reserves (sensuLaurel et al., 2008). From a species comparison perspective, Arctic cod larvae were able to survive 10–14 days longer than walleye pollock larvae in the absence of food across equivalent temperature environments. This suggests Arctic cod are physiologically more robust to match-mismatch scenarios than walleye pollock larvae under winter–spring conditions in the Arctic. Hatch timing of Arctic cod larvae can be extremely variable across their geographic range, sometimes occurring under the ice, and there is high selective pressure to achieve as much growth as possible in the first summer to survive the first over-wintering period (Bouchard and Fortier, 2011). Increased size and yolk reserves of larvae may therefore be important adaptations to maximize the duration of that first-year growth period. Although Arctic cod are generally exposed to consistent winter–spring temperatures <0°C (Bouchard and Fortier, 2008), food availability is spatially–temporally episodic by way of freshwater run-off events and summer ice-edge blooms (Wassmann, 2006). In general, it appears that Arctic cod embryos have physiology and adaptations well-suited to survive at a narrow range of low temperatures (“stenothermic”). Arctic cod juveniles are often ice-associated (Lønne and Gulliksen, 1989; Gradinger and Bluhm, 2004), and consequently are able to grow at sub-zero temperatures when in the ice. As summer waters become stratified and surface waters warm, Arctic cod juveniles (40–60 mm SL) are able to increase their thermal window and grow rapidly over a broader range of temperatures (−1 to 12°C, Laurel et al., 2017) before narrowing again towards maturity (age 1+ in deeper waters (0–9°C, Laurel et al., 2016). In contrast, walleye pollock eggs in the Bering Sea are found throughout the water column, and can be exposed to considerable temperature variation in the early spring. Mooring data in the eastern Bering Sea indicate temperatures during the spawning period can vary by as much as 3.5°C year-to-year, which translates to ∼2 weeks of variable incubation time (Blood et al., 1994). Reduced energetic reserves coupled with variable sea-ice and phytoplankton timing in the Bering Sea (Napp et al., 2000; Sigler et al., 2016) suggests that walleye pollock larvae may be more vulnerable to match/mismatch processes than Arctic cod in the current climate regime. Similar conclusions were reached by Pepin et al. (2014) based on observed weak relationships in feeding success and growth for Arctic cod larvae compared with more temperate species. However, it is likely that climate change will differentially impact the phenology and metabolic demands of these two species in ways that will destabilize established patterns of synchrony. Conclusions Understanding the thermal sensitivity of egg and larval stages in closely related or competing species provides a means of assessing their likelihood of survival with changing climate. The temperature-dependent hatch characteristics and early embryonic development of Arctic cod are distinctly different than the closely related walleye pollock. While both species have similarly low thermal tolerance during egg development, Arctic cod were much more sensitive to increases in temperature in terms of hatch success and size-at-hatch. The larger eggs and larger larvae with increased lipid reserves in Arctic cod corresponded with significantly longer development times, such that embryos could survive for ∼4 months from the time of spawning until they have to begin feeding at temperatures <0°C. This is ∼2 months longer than walleye pollock at equivalent temperatures based on the combined temperature-dependent hatch and time-to-starvation models. Such differences may contribute to the biogeographic separation of these species in the Pacific Arctic region. That is, the narrow thermal range of Arctic cod embryos may restrict their distribution to high latitudes, whereas the energetic reserves of walleye pollock embryos are insufficient to successfully extend their range into high Arctic environments. Undoubtedly, biogeographic ranges of these two marine species are driven by a host of mechanisms acting across the full life history of the individual. However, the distinct embryonic characteristics and thermal rates of these two closely related gadids appears to be an important component to consider in understanding current distributions and future range shifts in the wake of climate change in the Arctic. Acknowledgements We thank T. Hurst, J. Napp, and C. Ryer for reviewing earlier drafts of this manuscript. Thanks also to Bill Kopplin, Robert Fechhelm, Kyle McCain, Bill Streever, and the LGL Limited field crew for their assistance in the collection of Arctic cod in Prudhoe Bay as well as to Scott Haines, Michele Ottmar, and Eric Hanneman for their assistance in the fish transport and animal husbandry. We would like to thank Angie Sremba, Leah Feinberg, Kalyn Hubbard, and Kristina Mccan for technical assistance with lipid extractions and lipid class measurements. Funding This project was supported with funding from the North Pacific Research Board (NPRB) grant #R1403. This study is NPRB contribution #666. The findings and conclusions in the paper are those of the authors and do not necessarily represent the views of the National Marine Fisheries Service, NOAA. References Altukhov K. A. 1981 . The reproduction and development of the Arctic cod, Boreogadus saida, in the White Sea . Journal of Ichthyology/Voprosy Ikhtiologii , 19 : 93 – 21 . Aronovich T. , Doroshev S. , Spectorova L. , Makhotin V. 1975 . Egg incubation and larval rearing of navaga (Eleginus navaga Pall.), polar cod (Boreogadus saida Lepechin) and arctic flounder (Liopsetta glacialis Pall.) in the laboratory . Aquaculture , 6 : 233 – 242 . Google Scholar Crossref Search ADS Aydin K. , Mueter F. 2007 . The Bering Sea – a dynamic food web perspective . Deep-Sea Research Part II – Topical Studies in Oceanography , 54 : 2501 – 2525 . Google Scholar Crossref Search ADS Blaxter J. 1991 . The effect of temperature on larval fishes . Netherlands Journal of Zoology , 42 : 336 – 357 . Google Scholar Crossref Search ADS Blood D. M. 2002 . Low-temperature incubation of walleye pollock (Theragra chalcogramma) eggs from the southeast Bering Sea shelf and Shelikof Strait, Gulf of Alaska . Deep-Sea Research Part II – Topical Studies in Oceanography , 49 : 6095 – 6108 . Google Scholar Crossref Search ADS Blood D. M. , Matarese A. C. , Yoklavich M. M. 1994 . Embryonic-development of walleye pollock, Theragra chalcogramma, from Shelikof Strait, Gulf-of-Alaska . Fishery Bulletin , 92 : 207 – 222 . Bouchard C. , Fortier L. 2008 . Effects of polynyas on the hatching season, early growth and survival of polar cod Boreogadus saida in the Laptev Sea . Marine Ecology Progress Series , 355 : 247 – 256 . Google Scholar Crossref Search ADS Bouchard C. , Fortier L. 2011 . Circum-arctic comparison of the hatching season of polar cod Boreogadus saida: a test of the freshwater winter refuge hypothesis . Progress in Oceanography , 90 : 105 – 116 . Google Scholar Crossref Search ADS Bradbury I. , Laurel B. , Robichaud D. , Rose G. , Snelgrove P. , Gregory R. , Cote D. , et al. 2008 . Discrete spatial dynamics in a marine broadcast spawner: re-evaluating scales of connectivity and habitat associations in Atlantic cod (Gadus morhua L.) in coastal Newfoundland . Fisheries Research (Amsterdam) , 91 : 299 – 309 . Google Scholar Crossref Search ADS Chambers R. C. , Leggett W. C. , Brown J. A. 1989 . Egg size, female effects, and the correlations between early life-history traits of capelin, Mallotus villosus – an appraisal at the individual level . Fishery Bulletin , 87 : 515 – 523 . Christiansen J. S. , Mecklenburg C. W. , Karamushko O. V. 2014 . Arctic marine fishes and their fisheries in light of global change . Global Change Biology , 20 : 352 – 359 . Google Scholar Crossref Search ADS PubMed Copeman L. A. , Laurel B. J. , Boswell K. M. , Sremba A. L. , Klinck K. , Heintz R. A. , Vollenweider J. J. , et al. 2016 . Ontogenetic and spatial variability in trophic biomarkers of juvenile saffron cod (Eleginus gracilis) from the Beaufort, Chukchi and Bering Seas . Polar Biology , 39 : 1109 – 1126 . Google Scholar Crossref Search ADS Cowan J. H. , Houde E. D. , Rose K. A. 1996 . Size-dependent vulnerability of marine fish larvae to predation: an individual-based numerical experiment . ICES Journal of Marine Science , 53 : 23 – 37 . Google Scholar Crossref Search ADS Cushing D. 1995 . Recruitment. Population and Regulation in the Sea . Cambridge University Press , Cambridge . Danielson S. L. , Eisner L. , Ladd C. , Mordy C. , Sousa L. , Weingartner T. J. 2017 . A comparison between late summer 2012 and 2013 water masses, macronutrients, and phytoplankton standing crops in the northern Bering and Chukchi Seas . Deep-Sea Research Part II – Topical Studies in Oceanography , 135 : 7 – 26 . Google Scholar Crossref Search ADS Drost H. , Fisher J. , Randall F. , Kent D. , Carmack E. , Farrell A. 2016a . Upper thermal limits of the hearts of Arctic cod Boreogadus saida: adults compared with larvae . Journal of Fish Biology , 88 : 718 – 726 . Google Scholar Crossref Search ADS Drost H. , Lo M. , Carmack E. , Farrell A. 2016b . Acclimation potential of Arctic cod (Boreogadus saida) from the rapidly warming Arctic Ocean . Journal of Experimental Biology , 219 : 3114 – 3125 . Google Scholar Crossref Search ADS Duffy-Anderson J. T. , Barbeaux S. J. , Farley E. , Heintz R. , Horne J. K. , Parker-Stetter S. L. , Petrik C. , et al. 2016 . The critical first year of life of walleye pollock (Gadus chalcogrammus) in the eastern Bering Sea: implications for recruitment and future research . Deep-Sea Research Part II – Topical Studies in Oceanography , 134 : 283 – 301 . Google Scholar Crossref Search ADS Folch J. , Lees M. , Stanley G. H. S. 1957 . A simple method for the isolation and purification of total lipides from animal tissues . Journal of Biological Chemistry , 226 : 497 – 509 . Google Scholar PubMed Fossheim M. , Primicerio R. , Johannesen E. , Ingvaldsen R. B. , Aschan M. M. , Dolgov A. V. 2015 . Recent warming leads to a rapid borealization of fish communities in the Arctic . Nature Climate Change , 5 : 673 – 678 . Google Scholar Crossref Search ADS Gradinger R. , Bluhm B. 2004 . In-situ observations on the distribution and behavior of amphipods and Arctic cod (Boreogadus saida) under the sea ice of the High Arctic Canada Basin . Polar Biology , 27 : 595 – 603 . Google Scholar Crossref Search ADS Graham M. , Hop H. 1995 . Aspects of reproduction and larval biology of Arctic cod (Boreogadus saida) . Arctic , 48 : 130 – 135 . Google Scholar Crossref Search ADS Hollowed A. B. , Barbeaux S. J. , Cokelet E. D. , Farley E. , Kotwicki S. , Ressler P. H. , Spital C. , et al. 2012 . Effects of climate variations on pelagic ocean habitats and their role in structuring forage fish distributions in the Bering Sea . Deep-Sea Research Part II – Topical Studies in Oceanography , 65 : 230 – 250 . Google Scholar Crossref Search ADS Houde E. D. 2008 . Emerging from Hjort’s shadow . Journal of Northwest Atlantic Fishery Science , 41 : 53 – 70 . Google Scholar Crossref Search ADS Kendall A. W. 2001 . Specific gravity and vertical distribution of walleye pollock (Theragra chalcogramma) eggs. National Oceanic and Atmospheric Administration, National Marine Fisheries Service, Alaska Fisheries Science Center, Seattle, WA 98115-0070. Processed Report 2001, 88 pp. Kent D. , Drost H. , Fisher J. , Oyama T. , Farrell A. 2016 . Laboratory rearing of wild Arctic cod Boreogadus saida from egg to adulthood . Journal of Fish Biology , 88 : 1241 – 1248 . Google Scholar Crossref Search ADS PubMed Koenker B. , Laurel B. J. , Copeman L. A. , Ciannelli L. in review. Effects of temperature and food availability on the survival and growth of larval Arctic cod (Boreogadus saida) and walleye pollock (Gadus chalcogrammus) . ICES Journal of Marine Science . Laurel B. J. , Copeman L. A. , Spencer M. , Iseri P. 2017 . Temperature-dependent growth as a function of size and age in juvenile Arctic cod (Boreogadus saida) . ICES Journal of Marine Science , 74 : 1614 – 1621 . Google Scholar Crossref Search ADS Laurel B. J. , Hurst T. P. , Copeman L. A. , Davis M. W. 2008 . The role of temperature on the growth and survival of early and late hatching Pacific cod larvae (Gadus macrocephalus) . Journal of Plankton Research , 30 : 1051 – 1060 . Google Scholar Crossref Search ADS Laurel B. J. , Spencer M. , Iseri P. , Copeman L. A. 2016 . Temperature-dependent growth and behavior of juvenile Arctic cod (Boreogadus saida) and co-occurring North Pacific gadids . Polar Biology , 39 : 1127 – 1135 . Google Scholar Crossref Search ADS Lønne O. J. , Gulliksen B. 1989 . Size, age and diet of Polar cod, Boreogadus saida (Lepechin 1773), in Ice Covered Waters . Polar Biology , 9 : 187 – 191 . Google Scholar Crossref Search ADS Lu Y. H. , Ludsin S. A. , Fanslow D. L. , Pothoven S. A. 2008 . Comparison of three microquantity techniques for measuring total lipids in fish . Canadian Journal of Fisheries and Aquatic Sciences , 65 : 2233 – 2241 . Google Scholar Crossref Search ADS Marteinsdottir G. , Steinarsson A. 1998 . Maternal influence on the size and viability of Iceland cod Gadus morhua eggs and larvae . Journal of Fish Biology , 52 : 1241 – 1258 . Miller T. J. , Crowder L. B. , Rice J. A. , Marschall E. A. 1988 . Larval size and recruitment mechanisms in fishes: toward a conceptual framework . Canadian Journal of Fisheries and Aquatic Sciences , 45 : 1657 – 1670 . Google Scholar Crossref Search ADS Mueter F. J. , Bond N. A. , Ianelli J. N. , Hollowed A. B. 2011 . Expected declines in recruitment of walleye pollock (Theragra chalcogramma) in the eastern Bering Sea under future climate change . ICES Journal of Marine Science , 68 : 1284 – 1296 . Google Scholar Crossref Search ADS Nakatani T. , Maeda T. 1984 . Thermal effect on the development of walleye pollock eggs and their upward speed to the surface . Bulletin of the Japanese Society of Scientific Fisheries , 50 : 937 – 942 . Google Scholar Crossref Search ADS Napp J. M. , Kendall A. W. , Schumacher J. D. 2000 . A synthesis of biological and physical processes affecting the feeding environment of larval walleye pollock (Theragra chalcogramma) in the eastern Bering Sea . Fisheries Oceanography , 9 : 147 – 162 . Google Scholar Crossref Search ADS Nissling A. , Westin L. 1991 . Egg mortality and hatching rate of Baltic cod (Gadus morhua) in different salinities . Marine Biology , 111 : 29 – 32 . Google Scholar Crossref Search ADS Parrish C. C. 1998 . Lipid biogeochemistry of plankton, settling matter and sediments in Trinity Bay, Newfoundland. I. Lipid classes . Organic Geochemistry , 29 : 1531 – 1545 . Google Scholar Crossref Search ADS Pauly D. , Pullin R. S. V. 1988 . Hatching time in spherical, pelagic, marine fish eggs in response to temperature and egg size . Environmental Biology of Fishes , 22 : 261 – 271 . Google Scholar Crossref Search ADS Pepin P. 1991 . Effect of temperature and size on development, mortality, and survival rates of the pelagic early life-history stages of marine fish . Canadian Journal of Fisheries and Aquatic Sciences , 48 : 503 – 518 . Google Scholar Crossref Search ADS Pepin P. , Orr D. , Anderson J. 1997 . Time to hatch and larval size in relation to temperature and egg size in Atlantic cod (Gadus morhua) . Canadian Journal of Fisheries and Aquatic Sciences , 54 : 2 – 10 . Google Scholar Crossref Search ADS Pepin P. , Robert D. , Bouchard C. , Dower J. F. , Falardeau M. , Fortier L. , Jenkins G. P. , et al. 2014 . Once upon a larva: revisiting the relationship between feeding success and growth in fish larvae . ICES Journal of Marine Science , 72 : 359 – 373 . Google Scholar Crossref Search ADS Petrik C. M. , Duffy-Anderson J. T. , Castruccio F. , Curchitser E. N. , Danielson S. L. , Hedstrom K. , Mueter F. 2016 . Modelled connectivity between walleye pollock (Gadus chalcogrammus) spawning and age-0 nursery areas in warm and cold years with implications for juvenile survival . ICES Journal of Marine Science , 73 : 1890 – 1900 . Google Scholar Crossref Search ADS Petrik C. M. , Duffy-Anderson J. T. , Mueter F. , Hedstrom K. , Curchitser E. N. 2015 . Biophysical transport model suggests climate variability determines distribution of Walleye Pollock early life stages in the eastern Bering Sea through effects on spawning . Progress in Oceanography , 138 : 459 – 474 . Google Scholar Crossref Search ADS Porter S. M. , Bailey K. M. 2007 . The effect of early and late hatching on the escape response of walleye pollock (Theragra chalcogramma) larvae . Journal of Plankton Research , 29 : 291 – 300 . Google Scholar Crossref Search ADS Pörtner H-O. , Peck M. 2010 . Climate change effects on fishes and fisheries: towards a cause‐and‐effect understanding . Journal of Fish Biology , 77 : 1745 – 1779 . Google Scholar Crossref Search ADS PubMed Rand K. M. , Logerwell E. A. 2011 . The first demersal trawl survey of benthic fish and invertebrates in the Beaufort Sea since the late 1970s . Polar Biology , 34 : 475 – 488 . Google Scholar Crossref Search ADS Sakurai Y. , Ishii K. , Nakatani T. , Yamaguchi H. , Anma G. , Jin M. 1998 . Reproductive characteristics and effects of temperature and salinity on the development and survival of eggs and larvae of Arctic cod (Boreogadus saida) . Memoirs of the Faculty of Fisheries Hokkaido University , 45 : 77 – 89 . Sigler M. F. , Napp J. M. , Stabeno P. J. , Heintz R. A. , Lomas M. W. , Hunt G. L. 2016 . Variation in annual production of copepods, euphausiids, and juvenile walleye pollock in the southeastern Bering Sea . Deep Sea Research Part II – Topical Studies in Oceanography , 134 : 223 – 234 . Google Scholar Crossref Search ADS Smart T. I. , Duffy-Anderson J. T. , Horne J. K. , Farley E. V. , Wilson C. D. , Napp J. M. 2012 . Influence of environment on walleye pollock eggs, larvae, and juveniles in the southeastern Bering Sea . Deep-Sea Research Part II – Topical Studies in Oceanography , 65–70 : 196 – 207 . Google Scholar Crossref Search ADS Wassmann P. 2006 . Structure and function of contemporary food webs on Arctic shelves: an introduction . Progress in Oceanography , 71 : 123 – 128 . Google Scholar Crossref Search ADS Published by International Council for the Exploration of the Sea 2018. This work is written by US Government employees and is in the public domain in the US. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)