MARINE ECOLOGY PROGRESS SERIESMar Ecol Prog Ser

Vol. 339: 283299, 2007 Published June 6

INTRODUCTION

Although fish eggs and larvae may suffer mortalitythrough oil spills, there are few reported cases inwhich oil spills have conclusively had a significant im-pact on fish stocks (IPIECA 1997). However, this doesnot mean that fish stocks cannot be significantly af-fected by oil spills. Most data on large oil spills derivesfrom temperate and subtropical environments, where

biological productivity and oil degradation rates arehigh, ecosystems relatively complex, and fish stocksoften spawn over a longer period of the year or evenyear-round. A significant proportion of the unexploitedoil reserves (approx. 25%; see USGS 2000) are, how-ever, found in the polar regions, and oil companies arepresently turning towards these regions to developnew oil fields. In these areas physical and biological oildegradation is likely to be slower than in more temper-

Inter-Research 2007 www.int-res.com*Corresponding author. Email: n.c.stenseth@bio.uio.no

REVIEW

Fish and oil in the LofotenBarents Sea system:synoptic review of the effect of oil spills on fish

populations

Dag . Hjermann1, Arne Melsom2, Gjert E. Dingsr1, Jol M. Durant1, Anne Maria Eikeset1, Lars Petter Red2, 3, Geir Ottersen4, 5, Geir Storvik1, 6,

Nils Chr. Stenseth1, 7,*

1Centre for Ecological and Evolutionary Synthesis (CEES), Department of Biology, University of Oslo, PO Box 1066 Blindern,0316 Oslo, Norway

2Norwegian Meteorological Institute, PO Box 43 Blindern, 0313 Oslo, Norway3Department of Geosciences, Section Meteorology and Oceanography, University of Oslo, PO Box 1022 Blindern,

0315 Oslo, Norway4Institute of Marine Research, Gaustadallen 21, 0349 Oslo, Norway

5Bjerknes Centre for Climate Research/GEOS, University of Bergen, Allgaten 55, 5007 Bergen, Norway6Department of Mathematics, University of Oslo, PO Box 1053 Blindern, 0316 Oslo, Norway

7Institute of Marine Research, Fldevigen Marine Research Station, 4817 His, Norway

ABSTRACT: The Lofoten-Barents Sea area, which contains some of the most valuable fish stocks ofthe Atlantic Ocean, is being considered for offshore oil production. We review the effects of ahypothetical oil spill on fishes in this area, with a focus on effects on the egg and larval stage of the3 dominating fish stocks: NE Arctic cod Gadus morhua, Barents Sea capelin Mallotus villosus, andNorwegian spring-spawning herring Clupea harengus. In particular, we emphasise that the long-term population impact of an oil spill depends on ecological and oceanographic factors, some ofwhich have been poorly explored. Among these are (1) effects of the physical state of the ocean, espe-cially mesoscale circulation features, on the advection of oil and fish larvae, (2) effects of the spatialdistribution of spawners, (3) effects of harvesting on stock structure and length of the spawning sea-son, (4) effects of natural mortality and species interactions subsequent to an oil spill, and (5) chronicsublethal effects from persistent oil residues.

KEY WORDS: Advection Fishes Pollution Petroleum Vulnerability

Resale or republication not permitted without written consent of the publisher

Hadi Abaza

Mar Ecol Prog Ser 339: 283299, 2007

ate regions, due to low temperatures(Garrett et al. 2003, Brakstad et al.2004). Moreover, polar ecosystems arefairly simple (Hamre 1994, Hillebrand2004) and thus possibly more vulnera-ble to changes in the abundance of thefew key species (Paine 1980). One ex-ample is that the collapse of the BarentsSea capelin stock in the 1980s signifi-cantly affected several trophic levels in-cluding the capelins prey, zooplankton(Dalpadado et al. 2001), predators suchas cod (Hjermann et al. 2004a) and harpseals (Haug & Nilssen 1995), and alter-native prey of its predators such asshrimp (Worm & Myers 2003). Perhapsmost importantly, fish stocks such ascod and herring are close to their cli-matic limits in these habitats, and dueto these environmental constraints theytend to have short, intensive spawningseasons and localised spawning areas.As a result, their eggs and larvaethelife stages most susceptible to oil expo-suremay be relatively concentratedalong a narrow advection route. Thesestocks are among the economically andecologically most important fish stocksof the Norwegian and Barents Seas. For more than amillennium, this ecosystem (Fig. 1) has been one of themain food chambers for Europe and together withfarmed salmon is the main reason why Norway is theworlds third largest exporter of fishes, measured byexport value. Norway is also the worlds third largestexporter of oil. Exploitation of oil and gas will shortlybegin in the LofotenBarents Sea area (later denotedLBS; see also Fig. 1), with a gas project (Snhvit) start-ing production in 2007, and a nearby oil field (Goliat)to be developed soon. In 2006, it was decided thatsome of the most controversial areas from a fisheriesviewpoint (Aglen et al. 2005) will not be opened for oilexploration until 2010 (Anonymous 2006). However,these areas are also those that are most likely to con-tain large oil and gas reserves (Anonymous 2006) andtheir opening will continue to be advocated by the oilindustry. In addition, a substantial number of tankerssail along the Norwegian coast, carrying Siberian oilfrom Russian oil terminals. Risk analyses indicate thatRussian export traffic constitutes a higher risk for oilaccidents than LBS oil production (Blom-Jensen &Dervo 2003), especially as new oil terminals are beingbuilt on the coast of NW Russia. It has been esti-mated that this traffic may increase strongly, from12 million metric t of oil in 2004 (almost 1 tanker d1) to50 to 150 million metric t in the next decade (Bam-

bulyak & Frantzen 2005). Although the issues of thisreview are also relevant for tanker spills, we have con-centrated on spills from oil installations such as drillingplatforms. However, much of the data we review stemfrom ship accidents.

Herein, we review available information on the im-pact of possible oil spills in the LBS on 3 fish stocks ofmajor economic importance: the North-east Arctic(NEA) cod, the Barents Sea (BS) capelin, and the Nor-wegian spring-spawning (NSS) herring (Hamre 1994).The most important spawning grounds of the former 2stocks are inside the LBS; the latter spawns mainlysouth of LBS but its larvae are advected into the LBSwith the Norwegian Coastal Current (NCC; Fig. 1). Inaddition, the 3 stocks interact strongly with each otherthrough predatorprey relationships (Bogstad et al.2000, Hamre 2003, Hjermann et al. 2004a,b, 2007).

In Norway, an environmental impact assessment(EIA) has to be performed before a new area is openedto oil explorations. Part of such an EIA is an assessmentof the potential impact of an oil spill. The EIA for oildevelopment in the LBS region was prepared in 2003(Anonymous 2003), and included a report on possibleeffects of accidental oil spills on life in the watercolumn (Johansen et al. 2003). With this report as astarting point, we review some factors that affect theimpact of an oil spill on fishes in the LBS area. Rather

284

Fig. 1. LofotenBarents Sea ecological system with spawning locations andadvection routes of eggs and larvae of 3 fish stocks; red: North-east Arctic cod;purple: Norwegian spring-spawning herring; green: capelin (Aglen et al. 2005).Dotted line: maximum extension of the 3 species in the Barents Sea in the firstsummer following spawning. Light blue: continental shelf (

Hjermann et al.: Fish and oil in the LofotenBarents Sea system

than providing a comprehensive review of the vastliterature on the toxicological effects of oil, we wish toemphasise some oceanographic and ecological aspects(Fig. 2) that are often considered only superficially inEIAs of oil spills, and which need to be better under-stood in order to obtain a clear picture of the conse-quences of oil exploitation on the marine resources inthis region (Fig. 3).

ENVIRONMENTAL IMPACT ASSESSMENT (EIA)

The oil spill section of the EIAs conducted beforeopening oil exploration fields includes the develop-ment of possible oil trajectories and concentrations of

oil in the water column which, combined with distribu-tion maps of sensitive organisms, is used as a basis forassessing likely ecological impact scenarios. Regionsthat may be affected by oil pollution are determined bysimulated trajectories and weathering of the oil, andusually defined as regions where the probability of animpact from an accidental spill exceeds 5%. Tradition-ally, drift trajectories have been computed using anempirical relation between surface winds and theocean current with which the oil spill is advected(Martinsen 1982). In more recent work, the drift trajec-tories have been computed from the flow fields ofocean circulation models (Martinsen et al. 1994), tak-ing the full 3-dimensionality of the flow field andhydrology into account (Wettre et al. 2001). Due to the

steady expansion of available com-puter resources, such simulations arenow performed with a grid mesh thatresolves features such as filaments andeddies. In Norwegian waters, this re-quires a grid mesh with a resolution of5 km. Such a simulation was usedfor production of drift trajectories inthe most extensive examination yet ofpotential oil spills on the ecology of theLBS region (Johansen et al. 2003).

The drift trajectories are used to sim-ulate drift of both hydrocarbons andfish eggs/larvae, and thereby calculatethe exposure of eggs or larvae to hydro-carbons (Johansen et al. 2003). This canthen be combined with results from lab-oratory experiments on the toxicologi-cal effects of hydrocarbons to calculatethe proportion of the stock that is sub-ject to exposure above some set level.However, several aspects are often dis-regarded. Firstly, the ecosystem is oftenviewed in a static way; for instance, it isfrequently assumed that the proportionof fish spawning at each spawning loca-tion is the same between years, while inreality, this varies substantially. Thevariation between spawning locationsmay be caused both by climate (pre-vailing oceanographic conditions) aswell as by the age/size structure of thespawning stock. In cod and herring,which spawn in the coastal current,spawning results in a train of larvae,whose length depends on the spawningtime (which varies for biological rea-sons) and whose route depends onweather conditions. Secondly, it is as-sumed that all larvae are of equal

285

Movementof oil

Movement ofeggs/larvae

Egg/larvaeexposure

Oil toxicity

Spawnerpopulationstructure

Oil weatheringand mixing

Egg/larvaemortality

Wheremortalityoccurs

Survival ofremainingpopulation

Otherspecies

Populationsize andstructure

Spawninglocation and

period

Fisheries

Climate

OIL SPILL

Fisheries

Days YearsTime after oil spill

Fig. 3. Oil spill sites used in the simulations of the ULB 7-c spill (Johansen et al. 2003)

Fig. 2. Factors that affect the long-term impact of oil spills. Theexisting Environmental Impact Assessment (EIA) focuses oncomputing movement of oil and eggs/larvae, calculating theexposure of eggs/larvae to oil, and estimating the consequentmortality. However, the consequences of an oil spill are also pro-foundly affected by other factors (see What affects the impact

of an oil spill on a fish population?)

Hadi Abaza

Hadi Abaza

Mar Ecol Prog Ser 339: 283299, 2007

worth, i.e. that, in the absence of an oil spill, all larvaehave an equal chance of survival to fishable size ormaturity. However, natural mortality after the larvalstage varies substantially both spatially and amongyears (Ciannelli et al. 2007). Spatial variation is impor-tant since an oil spill kills larvae in a specific area.Thus, the impact of a spill on a population will vary,depending on whether the larvae killed were thosethat would otherwise have had the best chance ofreaching fishable size, or whether they were destinedto have been transported to an area in which theywould have a very low chance of survival. Year-to-yearvariation in natural survivalwhich may depend onthe state of the ecosystem (abundance of prey, preda-tors, and competitors)will affect the populationslong-term response to oil spill mortality. Finally, wewill briefly consider the possibilities of chronic effectsof those oil components with long persistence times.Such chronic effects are largely disregarded in thepresent EIA.

THE LOFOTENBARENTS SEA PELAGICECOSYSTEM

Oceanography

The LBS region consists of the Barents Sea, an openarcto-boreal shelf-sea with an average depth of about230 m (Zenkevitch 1963), and the narrow continentalshelf along the Norwegian coast downto Lofoten (Fig. 1). The ocean circula-tion is dominated by the NCC andthe Norwegian Atlantic Current (NAC)(Helland-Hansen & Nansen 1909,Stre & Mork 1981, Orvik & Niiler2002). The NCC is associated withlow-saline coastal water masses. Thesewater masses originate in the Skager-rak and are the result of mixing withfreshwaterprimarily from rivers dis-charging into the Baltic Sea, but alsointo the North Sea from the UK, conti-nental Europe and Scandinavia (Al-bretsen & Red 2006). The NCC passesthe southern tip of Norway as it exitsfrom the Skagerrak, and continuesalong the Norwegian coastal shelf tothe Barents Sea. The NAC on the otherhand is characterised by water massesof Atlantic origin that flow into theNordic Seas across the IcelandFaeroeRidge and through the FaeroeShet-land Channel (sterhus & Hansen2000). This inflow is easily recognisable

in satellite images (Fig. 4) as a warm-water mass in-truding into the Nordic Seas as far north and east asthe Barents Sea. As shown in Fig. 4, the inflow is sepa-rated by 2 fronts, an easterly front separating it fromthe NCC, and a westerly front separating it from thepolar water masses on the broad Greenland shelf andeastwards. At the entrance to the Barents Sea theNAC divides into 2 branches (Fig. 4), one entering theBarents Sea, and one continuing northwards towardsSpitsbergen (Svalbard). Other major characteristics ofthe ocean circulation in the area are the presence ofstrong tides and (not least) many mesoscale structuressuch as filaments, meanders and eddies. Eddies maydiverge from the front and influence circulation andhence residence time. Such mesoscale features onscales of 5 to 10 km are prominent in the Barents Sea(Fig. 5). Note, for instance, the strong current filamentor jet-like structure along the Norwegian coast presentin both the satellite image (Fig. 5a) and the model fore-cast in Fig. 5 and how, in general, the dark areas in thesatellite image correspond to the areas of strong cur-rents in the model forecast. This reflects the fact thatareas of strong currents are areas of strong verticalmixing which tends to reduce the backscatter signal tothe satellite. Due to the relatively warm Atlantic watermasses of the NAC, sea temperatures in the LBS areaare higher than in other regions at similar latitudes. Asa result, the southern part of the Barents Sea staysice-free even in the most severe winters. Year-to-yearvariability in temperature south of the oceanic Polar

286

Fig. 4. Sea surface temperature (SST) and sea-ice extent in the northern NorthAtlantic and Nordic Seas Ocean and Sea Ice Satellite Application (OSI-SAF)product (http://saf.met.no); 7 day satellite imagery composite, centered on April11, 2004. White: sea ice; gray: cloud masked (contaminated) areas. Latitude-longitude grid on land masses at 10 intervals. NAC: warm-water tongue of

Norwegian Atlantic Current

Hjermann et al.: Fish and oil in the LofotenBarents Sea system

front is strongly influenced by the inflow of Atlanticwater (Loeng 1991, Ingvaldsen et al. 2003). This vari-ability appears to be mainly wind-driven and hencelinked to the North Atlantic Oscillation (NAO) (Hurrellet al. 2003). The correlation between the NAO and theinflow and temperature of the Barents Sea variesgreatly with time, being most pronounced early in thelast century and over the last 3 to 4 decades (Dicksonet al. 2000, Ottersen et al. 2001).

Ecology of the focal fish stocks: NEA cod, NSSherring and BS capelin

These fish stocks all spawn along the Norwegiancoast in winter to spring (Fig. 1). The NSS herring havethe southernmost spawning distribution, with spawn-ing sites spread from around 58 to 69 N, but espe-cially between 62 and 64 N (Aglen et al. 2005). Theyspawn on the bottom, and their larvae drift towards theBarents Sea with the NCC current. The herring stay inthe Barents Sea until they have grown to a size of20 cm (typically at the age of 3 yr) and then migrate tothe Norwegian Sea, where they remain for the rest oftheir lifespan. Spawning NEA cod are distributed fur-ther north, with spawning sites at 62 to 71 N, but withthe spawning sites at Lofoten and the Rst bank(around 68 N) being the most important (Aglen et al.2005). Cod spawn pelagically, and their eggs hatchinto larvae as they drift northwards. The cod juvenilesreach the Barents Sea in summer, and then shift from a

pelagic to a demersal life style in September to Octo-ber; they remain in the Barents Sea for the rest of theirlife cycle except for spawning migrations after matura-tion at an age of 6 to 8 yr. Of the 3 species, BS capelinspawn furthest north (north of 70 N), close to the coaston the southern fringe of the Barents Sea, at depths of10 to 60 m. They spawn on sandy/gravel bottoms,attaching their eggs to gravel on the bottom. The eggshatch after 3 to 8 wk and the larvae migrate upwards tothe upper water layers, in which they are advectednorthward into the Barents Sea during summer andautumn (Gjster 1998, Aglen et al. 2005).

Economic and ecological importance of the fishstocks

Economically, the NEA cod is by far the most impor-tant of the 3 stocks. Dried cod originating from thecods spawning grounds at Lofoten has been one ofNorways largest export items for about 1000 yr. Atpresent, most of the other large stocks of Atlantic codhave collapsed, with the NEA stock the largest remain-ing, accounting for about 50% of the total annual codcatch (2004: ca. 600000 metric t of NEA cod includingunreported landings; ICES 2005). Second in economicimportance is the worlds largest herring stock, theNSS herring, which supports an annual fishery of up to1.2 million metric t (one of the largest fisheries in theNorth Atlantic). The BS capelin stock is also the worldslargest stock of this species (catch up to 2.9 million

287

Russia

Bare

nts S

ea

Bear Island

NC

72N

30E

25Eb.

NAC

Fig. 5. (a) Barents Sea as seen by the Moderate Resolution Imaging Spectroradiometer (MODIS), captured by the Aqua satellite on July 19,2003 (spatial resolution: 250 m). The bright turquoise areas are blooms of phytoplankton, probably the coccolithophore Emiliania huxleyi.NC: North Cape. (b) An ocean model forecast showing the sea surface current at practically the same time (July 24, 2003), using the forecastsystem of the Norwegian Meteorological Institute. Arrow length is proportional to current strength, which is also indicated by the colors fromlight blue (weak) to yellow and red (strong). The latitudelongitude grid has a resolution of 1. NAC: the Norwegian Atlantic current

Mar Ecol Prog Ser 339: 283299, 2007

metric t yr1. However, its economic importance ispartly indirect in that: (1) it comprises the most impor-tant prey item for the cod (Kjesbu et al. 1998, Marshallet al. 1999); (2) both capelin and herring are importantsources of fishmeal and fish oil, and thereby contribute(as fish food) to the huge Norwegian aquaculture sec-tor (Norwegian farmed Atlantic salmon, 570 000 t yr1,accounts for almost half of the worlds Atlantic salmonproduction).

In the context of biodiversity, these 3 fish stocks arevitally important as prey for large populations of sea-birds (20 million in the summer) and large populationsof seals and whales. They also play an important role inenergy and nutrient flow. For instance, capelin is theonly species able to effectively exploit the rich plank-ton bloom along the ice edge (Gjster & Loeng 1987,Hassel et al. 1991, Gjster et al. 2002). It effectivelytransports a substantial amount of energy from theremote central and northern Barents Sea to coastalareas, where it becomes easily available to piscivorousfishes, seabirds, mammals and fisheries restricted tothe southern parts of the Barents Sea. In contrast,much of the biomass accumulated by the herringmoves out of the Barents Sea when the 3-yr old herringreturns to the Norwegian Sea.

This review does not deal with other fish stocks inthe LBS, such as the coastal cod stocks, haddockMelanogrammus aeglefinus, saithe Pollachius virens,2 species of redfish Sebastes spp., and Greenlandhalibut Reinhardtius hippoglossoides, although theyare (or have been) important both economically andecologically.

EFFECTS OF OIL SPILLS ON FISHES

When assesing the effects of oil exploration, the EIAsconsider issues such as the amount of stranded oil, thelength of the affected coastline, and the regional prob-ability of an impact in the case of an accidental oil spill.In impact studies for the LBA area, various approacheshave been applied to examine transport of oil after anaccidental oil spill and its impact on fishes. The oceancurrents that advect oil were initially derived frommonthly climatologies and empirical relations betweencurrents and winds for the 35 yr period 1955 to 1990(Rudberg 2003a).

Results from 3-dimensional hydrodynamic ocean cir-culation models were used in the EIA on fish stocks(Johansen et al. 2003). Ocean circulation simulationswere performed with 2 different models. The durationof both simulations was limited to a single model yearbecause of the intensive use of computer resourcesrequired. A grid mesh with a horizontal resolution of4 km was applied in the simulations, and tides were

included in both cases. At this resolution, the resultsreveal the presence of ocean circulation features suchas meanders and ocean eddies in addition to largerscale circulation structures. The simulation of egg andlarvae advection was based on model results for cur-rents for the relevant year. Oil was assumed to bespilled from 1 of 6 possible locations in the area (Fig. 3).From the simulations, Johansen et al. (2003) calculatedhow large a fraction of the population would beexposed to 90 ppb of the water-soluble fraction of oil(or 8000 ppb hours; see same section below).

An oil spill from the surface from either a platform ora ship will form a plume on the surface downstreamfrom the source. Although oil (density 0.79 to 1.00 103 kg m3) is lighter than seawater (density 1.03 103 kg m3) some of it will enter the water column be-low the slick by dispersion through wave action and byvertical mixing and chemical dissolution. The extent ofdispersion depends on the composition of the oil aswell as on the weather (wave energy). Dissolution is aless important pathway, since the most soluble sub-stances are light aromatics (e.g. benzene, toluene),which are the first to be lost through evaporation(ITOPF 2002). Oil persistence depends on many fac-tors, especially density (ITOPF 2002). The oil of theLBS area is of medium density, 0.86 to 0.91 103 kg m3

or American Petroleum Institute (API) gravity 24 to 33(Rudberg 2003b), i.e. with a persistence in the order ofmonths (National Research Council 1985). The oil insome areas has a high wax content (up to 13%) and isthus fairly viscous (1.968 Pa s) (Rudberg 2003b). Atlow temperatures, natural breakdown of such oil isslow and cannot be effectively combated with disper-sants (ITOPF 2002). The oil on board Russian exporttankers varies in quality, but the Russian Export BlendCrude Oil (REBCO, or Urals) typically has a density ofaround 0.87 103kg m3.

We focus on the long-term population consequencesof oil impacting early stages of the fishes, i.e. eggs andlarvae (Fig. 2) for 2 reasons: (1) In laboratory studies,adult fishes were able to detect petroleum at very lowconcentrations (Hellstrm & Dving 1983, Dauble et al.1985, Beitinger 1990, Farr et al. 1995); (2) large num-bers of dead fishes have seldom been reported after oilspills. Thus, juvenile and adult fish appear to be capa-ble of avoiding water with high hydrocarbon concen-trations. One notable exception was the oil spill follow-ing the Amoco Cadiz ship accident, which killedlarge numbers of fishes, including a locally high pro-portion of 1 yr-old fishes of the commercially importantplaice and sole. This was probably due to massiveamounts of emulsified oil in the shallow waters wherethe spill occurred (IPIECA 1997). Oil at very low con-centrations (ppb levels) can also taint adult fishes (i.e.impart unpleasant odours and flavour to their flesh).

288

Hjermann et al.: Fish and oil in the LofotenBarents Sea system

Fishes can become tainted directly from the water orfrom sediments (via absorption through the gills andskin), or through contaminated prey species (IPIECA1997). The slightest suspicion of tainting or that fishesmay constitute a human health hazard can renderthem unmarketable for extended periods (Birtwell &McAllister 2002). For this reason, fishing is sometimesbanned in the area of an oil spill. Also, oil-pollutedsediments may negatively affect burrowing fishes suchas flatfishes or sandlance in several ways (see Birtwell& McAllister 2002 and references therein). However,the focal species of the present review are not closelylinked to the seabed.

In contrast to juveniles and adults, fish eggs and lar-vae are planktonic and thus unable to escape from pol-luted water; they are exposed to any toxic compoundsthe water may contain. The toxicity of oil for fish eggsand larvae has been reviewed, for example, by Clark(2001). Toxicological mechanisms are complex, ascrude oil is a mixture of different hydrocarbons andother organic and inorganic substances, and its compo-sition varies greatly between oil types. In addition, theperseverance and toxicity of oil are affected by severalother factors such as weathering, dispersion and emul-sification, which again depend on external factors suchas weather conditions. Toxicological laboratory studieson animals typically use oil itself (fresh or weathered),the light aromatics benzene, toluene, ethylbenzene,and xylene (the BTEX components, which make up 80to 90% of the water-soluble fraction), or the polycyclicaromatic hydrocarbons (PAHs). The toxicity of fresh oilis correlated to its PAH content, which therefore hasbeen thought to be the primary toxin of oil. However,toxicity and PAH are not correlated in weathered oils(Barron et al. 1999). In addition, toxicity test proce-dures typically use 25C water and warm-water organ-isms, which may be more robust than cold-waterorganisms (Perkins et al. 2005). Relatively few studieshave been performed on eggs and larvae of the organ-isms reviewed in the present paper. Booman et al.(1995) tested the susceptibility of NEA cod larvae toBTEX, and found them to be relatively susceptiblecompared to those laboratory organisms more com-monly used in toxicity tests. For instance, yolk-sac lar-vae of NEA cod showed reduced oxygen uptake whenexposed for 24 h to 2080 ppb of the water-solublefraction of oil. However, Booman et al. (1995) found noeffects on growth or first feeding success. Toxicologicalstudies have also traditionally been conducted in theabsence of UV light, but the toxicity of PAH andweathered oil has been found to increase with a factorof 2 to >1000 in the presence of sunlight (Barron et al.2003). The use of chemical dispersants to avoid stran-ded oil on the beaches will in general increase thetoxic effects of oil on fish larvae, at least temporarily

(Couillard et al. 2005). In the EIA for the LBS, Johansenet al. (2003) found the lowest LD50 (concentration lead-ing to 50% mortality) to be 900 ppb among 93 studiesof planktonic organisms (the studies used differenttypes of oil products and either invertebrates, fishes oralgae). Using a security factor of 10, they set thepredicted no-effect concentration (PNEC) at 90 ppb(g l1) hydrocarbons. Since most studies used a 96 hexposure, this is equivalent to approximately 8000(90 96) ppb h.

A critical issue that has arisen in recent years, in par-ticular following the Exxon Valdez accident, is theeffect of sublethal damage to eggs and larvae. Al-though acute lethality tests are useful for generatingguidelines to protect against physiological death (i.e.mortality), they ignore ecological death, i.e. pollutiondamage that renders fishes unable to function in anecological context, even if they are not visibly harmed(Scott & Sloman 2004). PAHs cause a range of abnor-malities such as morphological deformities and cyto-genetic abnormalities (Hose et al. 1996) and embryoniccardiac dysfunction (Incardona et al. 2004). Moresubtle, but possibly serious effects include lastingdisruption of complex behaviour, such as predatoravoidance, reproductive and social behaviour (Scott &Sloman 2004). For fish eggs that develop in intertidalsediments, such as Pacific herring and pink salmon,such effects can result from PAH concentrations as lowas 0.4 ppb (Carls et al. 1999). It has also been shownthat weathered crude oil can cause immunosuppres-sion and expression of a viral disease, viral hemor-rhagic septicemia virus (VHSV). The 1989 Pacific her-ring year-class, which was exposed to Exxon Valdezoil at the egg stage, displayed a high incidence ofVHSV as well as an extremely low survival until theage of spawning. This caused a dramatic collapse ofthe stock in 1993. Carls et al. (2002) argued that theExxon Valdez oil spill might, at least partly, havecaused this collapse, although this could not be con-clusively demonstrated.

WHAT AFFECTS THE IMPACT OF AN OIL SPILLON A FISH POPULATION?

The effect on a fish population of an oil spill in anarea with fish larvae depends to a great extent onoceanographic and ecological conditions. The extent ofthe spill, the weather conditions at the moment ofimpact, the time of year and several ecological aspectsall influence the extent of the impact on the year-classaffected. Below, we discuss some of these aspects.While the current EIA is in general scientifically sound,there are some topics which it deals with fairly super-ficially.

289

Hadi Abaza

Hadi Abaza

Hadi Abaza

Hadi Abaza

Hadi Abaza

Hadi Abaza

Hadi Abaza

Mar Ecol Prog Ser 339: 283299, 2007

Physical state of the ocean, especiallymesoscale circulation

The oceanic conditions at the time ofthe spill and immediately afterwardsaffect the advection of both oil andspawning products. The amount of oilthat is dissolved and dispersed fromthe spill site into the water columndepends on the weather, ocean condi-tions, and location and depth of thespill (spills at the sea bottom give riseto different hydrocarbon concentra-tions than surface spills). Almost allthe existing oil pollution impact stud-ies for Norwegian waters have beenbased on empirical relations betweenhistorical winds and ocean currents.The wind strength in the LBS area canbe characterised by the North AtlanticOscillation (NAO) index (Ottersen etal. 2001, Stenseth et al. 2002, Hurrellet al. 2003, Stenseth et al. 2003). Thisindex is a measure of the strength ofthe North Atlantic westerlies, whichhave a significant impact on thestrength and position of the NAC (Or-vik et al. 2001). However, the weatherin this region is not solely determinedby the NAO, which is why an ap-proach using historical winds has beenconsidered preferable to simply de-ducing the impact from the NAO. Withthis approach, ocean circulation fea-tures will be on the same scale as theatmospheric scale, with possible modi-fications for coastal areas. However,because of the different properties ofair and water, oceanic scales aresmaller than atmospheric scales by afactor of 1/10 to 1/100 (Charney &Flierl 1981).

Smaller scale oceanic features com-prise eddies, filaments and meanders(mesoscale features) (Fig. 5). Thesemesoscale features influence how oilspills affect biological activity in theocean. Generally, the water mass of aneddy differs from its surroundings. Aneddy traps particles and dissolved ma-terial (e.g. fish eggs and hydrocarbons,if present) until the eddy dissipates oris dispersed. In order to include meso-scale effects, the results of hydrody-namic ocean-circulation models have

290

a

b

MainlandMainland

Coastalcurrent

spawning(April)

larvae(June)

Age 1(Februaryfollowing year)0-group

(September)

Widespreadspawning

Long spawningperiod

Naturalmortalityhigh inoceanicareas

c

MMainland

Coastalcurrent

spawning(April)

Mostspawning in

oceanic areas

Short spawningperiod

No spatialvariation in

naturalmortality

larvae(June)

Age 1(Februaryfollowing year)

0-group(September)

Fig. 6. Biological and oceanographic factors may determine how an oil spill af-fects mortality in cod larvae, and thus the North-east Arctic (NEA) cod stock.(a) Red: spawning areas; green: position of the eggs and larvae in April andMay; blue arrow: movement of fish eggs and larvae with currents (NCC andNAC). (b,c) Location of a year-class of NEA cod at 4 stages: eggs at the mainspawning grounds of Lofoten, larvae drifting along the coast in the NorwegianCoastal Current (NCC), 0-group fish in the Barents Sea, and 1 yr olds at roughlythe same location as the 0-group. An oil spill in the oceanic part of the NCC killspart of the larval population (shaded area), with small impact in (b) and large im-pact in (c). (b) Cod spawning occurs throughout coastal and oceanic sites over anextended period, resulting in a long train of cod larvae; the oil spill only killslarvae that would have had a small chance of surviving the first winter. (c) Codspawn mostly in the oceanic areas and for a short period, resulting in a moreconcentrated distribution of larvae, and the larvae killed by oil would have hadan average (or higher) chance of future survival. Oceanography determines thedegree of mixing (i.e. spatial distribution from eggs to larvae to the 0-group stage)

Hjermann et al.: Fish and oil in the LofotenBarents Sea system

recently been used in impact studies. As mentioned inan earlier subsection, results from 2 such simulationswere used in the EIA for the LBS area (Johansen et al.2003). A horizontal resolution of 4 km allows formesoscale features to be represented in such sim-ulations (exemplified in Fig. 5). Nevertheless, the useof such results for calculation of mesoscale statistics isquestionable for 2 reasons: (1) the use of a singlemodel year in each simulation means that this doesnot reflect year-to-year variability in the larger-scaleocean circulation associated with (e.g.) the NAOindex, and its impact on mesoscale structures. Large-scale ocean circulation provides the energy formesoscale motion since energy arising from large-scale stratification and momentum is converted intomesoscale energy (e.g. Fossum 2006, Fossum & Red2006). (2) A single realization for each period meansthat the small-scale variability associated with thestochastic nature of flow instability is not included. Inorder to describe the statistics associated with meso-scale variability, a proper ensemble of simulations (i.e.a set of realizations of the system) must be computed,

similar to the manner in which medium-range weatherforecasts are routinely produced.

Distribution of spawning sites

The only way in which fishes can decide where tobe at the egg and larval stages is by the parents' choiceof spawning time and site (Fig. 7), and by the densityof the eggs (which determine their average depth).For each of the 3 species, the proportion of individualsspawning at each spawning location varies substan-tially between years. Herring spawning sites canchange dramatically both from one year to the nextand on multidecadal time scales. The spawning pat-tern of cod is more stable, but there is still considerablevariation in the distribution of spawning cod amongthe main areas of Lofoten (Vestfjorden and Yttersida;Fig. 7, Table 1). Spawning capelin have an easterly dis-tribution (28 to 33 E) in some years and a westerlydistribution (18 to 31 E) in others (Gjster et al.1998). Changes in temperature, stock size or mean age

of the spawning stock, and changes in eggpredation from herring are possible explana-tions for the observed patterns. However,there is no simple relationship between (e.g.)stock size and climatic conditions.

Year-to-year variability in the spawningarea may be correlated with the severity ofthe impact from an oil spill, since certainocean circulation regimes (correlated withweather patterns such as NAO) may be asso-ciated with both anomalous hydrocarbon con-

291

2003

Yttersida

Vestf

jorde

n

70

69

68

67

10 12 14 16 18

2002 2004

Fig. 7. Gadus morhua. Spatial distribution of spawning North-east Arctic (NEA) cod in 2002 to 2004 in the coastal area Vest-fjorden and the more oceanic area of Yttersida (indicated on the left map: Salthaug 2002, Mehl & Nedreaas 2003, Mehl 2004).Shading indicates the strength of acoustic registrations of cod (darker shading = high density of cod spawners). See Table 1

for additional data

Cod stock Cod standing Cod first-time Capelinin Vestfjorden stock biomass spawners biomass

(%) (103 metric t) (%) (103 metric t)

2002 36 158 57 35002003 21 263 83 21002004 4 286 38 700

Table 1. Gadus morhua and Mallotus villosus. Stock data for the spawning periods shown in Fig. 7

Mar Ecol Prog Ser 339: 283299, 2007

tents from an oil spill, and with a shift in the preferen-tial spawning site. A change in spawning area or timeof spawning will thus affect the drift pattern of eggsand larvae which, in turn, will affect the probabilityof contact with oil from an oil spill.

Length of the spawning season as an ecological andevolutionary consequence of harvesting

To fully understand external influences such asfishing pressure and perturbations such as oil spills,it is important to study their impact on the life-history traits of the relevant population. Decrease ofgenetic variation may lead to a population that ismore susceptible to environmental disturbances suchas oil spills by reducing its ability to adapt andrespond to stress. A growing body of evidence sug-gests that high fishing pressure and size-selectiveharvesting may result in fisheries-induced evolutionof some life-history traits (Heino & God 2002),including age and size at maturation (Heino et al.2002a,b, Barot et al. 2004, Ernande et al. 2004, Olsenet al. 2004). In the case of NEA cod, the decrease inage and size at maturation observed since 1950 hasbeen suggested to be not only a plastic but also agenetic response to a high fishing pressure, as wellas changes in the fisheries size selectivity (Heino etal. 2002b, Eikeset et al. 2005). Other life-history traitsmay also be affected by fisheries-induced changes;

e.g. individual growth rate (Conover & Munch 2002)and reproductive investment (Reznick et al. 1996,Reznick & Ghalambor 2005, Rijnsdorp et al. 2005).Therefore, it is important to take fishery-inducedevolution into account when assessing the sensitivityof fish populations to oil spills. In the case of cod,it has also been shown that compared to first-timespawners, older individuals that have spawned be-fore (repeat spawners) arrive earlier on the spawn-ing grounds, spawn over a longer time span, produc-ing several times more eggs during the spawningseason and over a wider range of vertical distribution(Kjesbu et al. 1992, 1996, Marshall et al. 1998). As aresult, eggs and larvae of repeat spawners spreadout much more horizontally, vertically and tempo-rally, the plume of eggs and larvae downstreamfrom the spawning site is longer, and therefore thepopulation will be less susceptible to oil spills(Fig. 6). Since the 1950s, the proportion of older,larger cod has decreased dramatically. While 90% ofthe spawning stock biomass originated from fish ofage 10 yr and older in 1947, this value was 2.5% in2002 (Ottersen et al. 2006). Therefore, cod eggs andlarvae have become increasingly concentrated inspace and time, and thus the consequences of an oilaccident may be more severe. Physical conditionsalso influence the distribution and advection of eggsand larvae (e.g. the less wind, the more larvae willremain high in the water column, where concentra-tions of hydrocarbons are highest).

Natural mortality and species inter-actions subsequent to oil spill mortality

In the absence of human interference,there is enormous natural mortality ofmany fish species from the larval stage tothe time when they recruit to the fish-eries. For instance, for NEA cod, survivalis in the order of 106 from eggs to age 3yr (i.e. ~1 out of 1000 000 eggs survives).Still, this mortality appears to vary a gooddeal between years (in the order of ~10for NEA cod) and between areas. Thesurvival of NEA cod from age 5 mo to age11 mo appears to vary substantially spa-tially as well, being higher in the easternBarents Sea (Ciannelli et al. 2007; seealso present Fig. 8). One possible reasonis that small cod may be protected fromcannibalism in colder waters (Ciannelli etal. 2007). Since an oil spill kills larvae in alimited area, this will influence the de-gree to which the population is affected

292

Fig. 8. Gadus morhua. Spatial patterns of North-east Arctic cod survival fromage 5 mo (September) to age 11 mo (February), during the period from 1980 to2004, as estimated by an additive GAM model including geographic co-ordinates (latitude and longitude), bottom depth, fish length and wintertemperature as covariates. Circles are proportional to the relative survival (i.e.larger circles show better survival). Circles of different sizes within the samegrid location reflect the interannual variability of the survival metric (from

Ciannelli et al. 2007)

Hadi Abaza

Hadi Abaza

Hadi Abaza

Hjermann et al.: Fish and oil in the LofotenBarents Sea system

by the oil spill. For instance, assuming that an oil spillkills 20% of the fish larvae, and that these 20% werelarvae that would have been advected to ecologicallyunfavourable areas and thus have suffered 100% mor-tality, then the real effect of the oil spill would be closeto zero when the cohort reaches fishable age (e.g. age 3yr). On the other hand, if the 20% larvae killed werethose that would have been advected to favourable ar-eas and thus have had the best chance of survival, thenthe final effect would be more than 20%. Therefore, anunderstanding of the spatial variability in survival andits causes is crucial (Fig. 6). The fact that an oil spilldoes not affect all eggs or larvae equally can affect theconclusions about the effects of an oil spill. For instance,according to Johansen et al. (2003: their Fig. 7.8), an oilspill from the Troms I petroleum region will tend toaffect/kill cod larvae in the eastern part of their distrib-ution area. Assuming that these will also have aneasterly position at the 0-group stage, these larvae arethose with the best chance of survival from the 0-groupstage to age 1 yr (Ciannelli et al. 2007; Fig. 8).

Even if the primary effect of an oil spill was to reducea single year-class of a single species, ecological inter-actions would lead to secondary effects arising frompredation, competition and cannibalism (Fig. 9). Sucheffects could be either intraspecific (e.g. changedgrowth and life-time among the surviving portion ofthe year-class as well as other year-classes) or interspe-cific (i.e. changes in temporal distribution, growth andsurvival among the prey, predators and competitors ofthe species affected by oil). Intraspecific secondaryeffects can reduce or increase the primary effect,(compensation or depensation, respectively). Compen-sation is expected if the stock is bottom-up controlled(i.e. if prey is the limiting factor). For instance, if 15%

of the cod larvae were killed by oil (pri-mary effect), reduced competitionamong the survivors might partly com-pensate for the mortality, reducing theyear-class by 15% at age 3 yr as aresult of the oil spill. Whether compen-sation or depensation occurs may welldepend on the ecological and/or cli-matic state of the ecosystem. In the caseof cod, a mechanism for such shiftscould be cannibalism, which is moreprevalent when the capelin stock is

small. Thus, the early survival of cod may be bottom-up controlled when capelin is abundant and top-downcontrolled when capelin is scarce.

Interspecific secondary effects are highly likely in theLBS system since the 3 species are strongly interlinked(e.g. Dingsr et al. 2007, Hjermann et al. 2007): (1)Capelin is a key food source for cod, and thus highcapelin biomass has a significant positive effect on cod,reducing cannibalism, and increasing reproductiveoutput of the latter (Marshall et al. 1999). Likewise, codpredation has a significantly negative effect on capelinsurvival (Hjermann et al. 2004c). (2) Capelin recruit-ment is heavily affected by predation by juvenileherring on capelin larvae; the capelin population cancollapse (decrease by >95%) following years withabundant juvenile herring (Gjster & Bogstad 1998).Because of their intimate interdependence, all 3 stocksare likely to be affected should only one of them be di-rectly exposed to pollution. The extent and duration ofsuch cascading effects depends on the structure of theecosystem and the strength of interactions, which tosome degree is known for at least the Barents Seaecosystem (Tjelmeland & Bogstad 1998, Bogstad etal. 2000). Also, if an oil spill directly disrupts thetemporal distribution of only one trophic level, othertrophic levels may become temporally mismatched,thus weakening trophic links the essence of thematch-mismatch hypothesis (Hjort 1914, Cushing 1990,Durant et al. 2007).

Chronic sublethal effects of persistent oil residues

The immediate effects of oil spills are reflected inacute mortality of sea birds and oceanic mammals

293

Fig. 9. Gadus morhua, Mallotus villosus and Clupea harengus. Trophicrelationships between the main components of the food web, including cod

cannibalism

Hadi Abaza

Hadi Abaza

Mar Ecol Prog Ser 339: 283299, 2007

through oil slicks. However, the Exxon Valdez oilspill has shown that the mortality immediately follow-ing an oil spill can be less damaging than long-term(e.g. a decade) chronic exposure to persistent oilresidues in the environment (Peterson et al. 2003).Many of the populations that suffered heavy short-term mortality in the 1989 Exxon Valdez spill haverecovered, but not all some populations are stillrecovering slowly and some not at all. 15 years afterthe accident, significant amounts of Exxon Valdez oilremain in intertidal and subtidal sediments and belowmussel beds. Thus, species such as sea otters andharlequin ducks that prey on benthic organisms, stillexhibit increased levels of the detoxification enzymeCYP1A, and suffer higher mortality in the areaimpacted by the spill (Peterson et al. 2003). The ele-vated level of toxicity also impacts those fishes whoseeggs develop in intertidal habitats. One example isthat among pink salmon exposed to

Hjermann et al.: Fish and oil in the LofotenBarents Sea system

may be gained by direct observation, some (in manycases considerable) uncertainties will always remain.While such uncertainties can be integrated in theMonte Carlo procedure, this source of uncertainty istypically ignored.

KNOWLEDGE GAPS AND RESEARCH NEEDS

The biological effects of oil spills are generally ex-tremely hard to predict. Their impact is affected by somany parameters that the effects of no 2 oil spills arethe same. The amount of oil spilled is hardly correlatedwith the ecological impact of the spill. Measured in theamount of oil spilled, the Exxon Valdez oil spill ranksonly as no. 40 among oil spills that occurred between1967 and 1994, but cost Exxon 3.5 billion dollars indirect expenses (ExxonMobil 2004). In comparison,15 times as much oil was spilled in the IXTOC I blow-out in the Mexican Gulf in 1979, and 6500 times asmuch in the Persian Gulf in 1991 (Paine et al. 1996). Anoil slick that was barely visible killed 10 000 seabirds ina fiord close to Vard, Northern Norway in 1979 (Bam-bulyak & Frantzen 2005). Thus, ecological impact isobviously determined by other factors in addition tospill size, such as location (closeness to shoreline), timeof year, atmospheric as well as oceanic conditions atthe time of the spill, and type of oil. However, as wehave attempted to show herein, the effects of an oilspill also depend in more subtle ways on its inter-actions with prevailing oceanographic and ecologicalconditions. Thus, assessing the potential effects ofoil spills demands careful consideration of both thephysical and the biological processes involved.

With regard to physical processes, drift trajectories,and hence the region of influence of a potential acci-dental oil spill, are traditionally computed using anempirical relationship between surface winds and theocean current with which the oil spill is advected.Thus, based on wind data from an atmospheric circula-tion analysis, drift trajectories are completely deter-ministic. In one sense, the ocean is subservient to theatmosphere: In a hypothetical case of 2 years withidentical atmospheric forcing, the ocean circulation,and drift trajectories, becomes identical. However, asdemonstrated by the results of recent ocean circulationstudies using an ensemble of realizations, the relation-ship between ocean circulation and atmospheric forc-ing is not deterministic (e.g. Metzger & Hurlburt 2001,Melsom et al. 2003). In a study with a mesh configura-tion (i.e. grid size and map projection) similar to thesimulations of Johansen et al. (2003), Melsom (2005)found that in frontal regions, about 13 of the variabilitycan be attributed to the non-deterministic variabilityassociated with flow instabilities resulting in meso-

scale features such as eddies, filaments and meanders(e.g. Fossum 2006, Fossum & Red 2006). This is in factan underestimate, since the variance in the probabilitydistributions for (e.g.) salinity and temperature wasunderestimated in Melsoms (2005) study.

In order to fill the gaps on physical aspects in ourknowledge of how oceanic transport of oil will affect itsimpact on marine life, we must address the 2 issues notyet appropriately analyzed in the existing impactstudies of the LBS area: (1) How does large-scale vari-ability affect mesoscale circulation and thereby thevariability of transport pathways, and (2) what is theprobability distribution of transport pathways asso-ciated with non-deterministic variability? In order toresolve question (1), a hydrodynamic model must beapplied to a time period that spans various atmosphericforcing regimes in a manner that reproduces meso-scale features. Question (2) is resolved by an ensemblesimulation, whereby the results have a probability dis-tribution that is similar to the observed distribution.

Both approaches are computationally demanding.The fact that the hydrodynamic model of Johansen etal. (2003) was limited to two 1 yr simulations was prob-ably due to the limits of the computer resources avail-able at that time. This situation is changing. Moreover,an ocean circulation ensemble that exhibits a properprobability distribution in the context of mesoscalevariability is yet to be defined, and constitutes a scien-tific challenge, since the mesoscale circulation is notresolved by synoptic observations in space and time.We suggest that inclusion of continuously perturbedatmospheric forcing fields will yield more realisticprobability distributions than the approach using ini-tial state perturbations of decadal simulations thatwere performed by Melsom (2005).

With regard to ecological aspects, we have pointedto some unknowns for these 3 species. This may seemparadoxical, since cod and herring have been inthe focus of Norwegian fishery science for more than100 years. However, even for NEA cod, one of the best-studied fish stocks in the world, there are gaps in ourknowledge that significantly hamper our ability topredict the effects of an oil spill. This is illustrated bythe Exxon Valdez spill, for which it has proved diffi-cult to assess the oil spill damage and the recoverytime even in the case of populations that had been verywell-studied prior to the accident, such as the killerwhale Orcinus orca and pink salmon Oncorhynchusgorbuscha.

We believe, however, that existing predictions forpossible oil spill impacts can be improved by usingavailable data. Typically, the expected impact of an oilspill is given as a function of approximate location (e.g.distance from the coast) and time of year (month).Oceanographic models should be able to provide more

295

Mar Ecol Prog Ser 339: 283299, 2007

detailed guidance on the risks associated with spills atdifferent locations. Also, as detailed above, the impactwill depend on the ecosystems ecological state (abun-dance of key species, such as capelin in the case ofLBS) as well as its climatic state (e.g. high or low NAOyears). These variables can affect both the location/distribution of eggs/larvae, the drift patterns of botheggs/larvae and oil, as well as the consequences tofish populations (including effects of ecosystem inter-actions) of oil-caused mortality at the larval stage. Withregard to possible cascade effects of oil spills causinglarval mortality, oil spills comprise but one kind ofpulse perturbation (e.g. Chan et al. 2003), and thusother kinds of perturbations (i.e. climatic and eco-logical conditions causing mortality in specific lifestages) can provide insight into such cascade effects.Ecological/genetic models can predict the combinedeffects of fisheries and oil on fish dynamics, i.e. howfisheries affect susceptibility of fish stocks to oil spillsand vice versa.

Even in a pristine state, the Barents SeaNorwegianSea ecosystem may not be characterised by balanceand stability. For instance, recruitment (i.e. larval mor-tality) of the herring, a key species in this system, isextremely variable from year to year, in part because ofvariations in the oceanic climate (Fiksen & Slotte 2002).Hence, the ecosystem is inherently stochastic: Evenwith a perfect knowledge of the system and the lethaland non-lethal effects of an oil spill, we could not pre-dict its effects on the ecosystem for more than a fewyears ahead. Further, our knowledge on ecosystemprocesses in this system is imperfect. Since no largedegree of certainty can be achieved in any conclusionsabout long-term effects, at best we can attempt, bymodelling, to attain a quantitative indication of thepossible outcomes of oil spills in the ecosystem context.Qualitatively, we can assess at which places and timesan oil spill may be expected to have the most signifi-cant long-term effects, and under which climatic re-gimes the effect of spills will be most harmful.

Acknowledgements. We thank the following NorwegianResearch Council programmes: Havet og Kysten for sup-porting the LEO (Long-term Effects of Oil accidents) project,and Polar Climate Research for supporting the ECOBE(Effects of the North Atlantic Variability on the Barents SeaEcosystem) project. We also thank VISTA (www.vista.no) forfunding D..H, and Hydro for financial support. Commentsprovided by 4 reviewers on an earlier version of the paper aregreatly appreciated.

LITERATURE CITED

Aglen A, Gjster H, Holst JC, Klungsoyr J, Olsen E (2005)Verdifulle omrder for torsk, hyse, sild og lodde i omrdetLofotenBarentshavet. Institute of Marine Research, Ber-gen, Norway

Albretsen J, Red LP (2006) Sensitivity of Skagerrak dy-namics to freshwater discharges: insight from a numericalmodel. In: Dahlin H, Ryder P (eds) European operationaloceanography: present and future. Proc 4th EuroGOOSConf, June 69, 2005, Brest, France. Elsevier, Amsterdam

Anonymous (2003) Utredning av konsekvenser av helrigpetroleumsvirksomhet i omrdet LofotenBarentshavet.Ministry of Petroleum and Energy, Oslo

Anonymous (2006) New monitoring of the marine environ-ment in the north. Press release from the Ministry of Envi-ronment, Norway. Oslo, March 31, 2006

Bambulyak A, Frantzen B (2005) Oil transport from the Russ-ian part of the Barents Region, Svanhovd EnvironmentalCentre, Svanhovd, Norway

Barot S, Heino M, OBrien L, Dieckmann U (2004) Long-termtrend in the maturation reaction norm of two cod stocks.Ecol Appl 14:12571271

Barron MG, Podrabsky T, Ogle S, Ricker RW (1999) Are aro-matic hydrocarbons the primary determinant of petroleumtoxicity to aquatic organisms? Aquat Toxicol 46:253268

Barron MG, Carls MG, Short JW, Rice SD (2003) Photo-enhanced toxicity of aqueous phase and chemically dis-persed weathered Alaska North Slope crude oil to Pacificherring eggs and larvae. Environ Toxicol Chem 22:650660

Beitinger TL (1990) Behavioral reactions for the assessmentof stress in fishes. J Gt Lakes Res 16:495528

Birtwell IK, McAllister CD (2002) Hydrocarbons and theireffects on aquatic organisms in relation to offshore oil andgas exploration and oil well blowout scenarios in BritishColumbia,1985. Can Tech Rep Fish Aquat Sci 2391 Fish-eries and Oceans Canada, West Vancouver

Blom-Jensen B, Dervo HJ (2003) Sannsynlighet for hendelsermed store oljeutslipp (ULB 7e). Scandpower, Kjeller

Bogstad B, Haug T, Mehl S (2000) Who eats whom in theBarents Sea? In: Vkingsson GA (ed) Minke whales, harpand hooded seals: major predators in the North Atlanticecosystem. North Atlantic Marine Mammal Commision,Troms

Booman C, Midty F, Smith AT, Westrheim K, Fyn L (1995)Effekter av olje p marine organismersrlig p fiske-larvens frste nringsopptak. Fisken Havet 1995:9, Insti-tute of Marine Research, Bergen

Brakstad OG, Bonaunet K, Nordtug T, Johansen O (2004)Biotransformation and dissolution of petroleum hydro-carbons in natural flowing seawater at low temperature.Biodegradation 15:337346

Carls MG, Rice SD, Hose JE (1999) Sensitivity of fish embryosto weathered crude oil. Part I. Low-level exposure duringincubation causes malformations, genetic damage, andmortality in larval Pacific herring (Clupea pallasi). EnvironToxicol Chem 18:481493

Carls MG, Marty GD, Hose JE (2002) Synthesis of the toxico-logical impacts of the Exxon Valdez oil spill on Pacificherring (Clupea pallasi) in Prince William Sound, Alaska,USA. Can J Fish Aquat Sci 59:153172

Chan KS, Stenseth NC, Lekve K, Gjster J (2003) Modelingpulse disturbance impact on cod population dynamics: the1988 algal bloom of Skagerrak, Norway. Ecol Monogr 73:151171

Charney JG, Flierl GR (1981) Oceanic analogues of large-scale atmospheric motions. In: Warren BA, Wunsch C(eds) The evolution of physical oceanography. MIT Press,Cambridge, MA

Ciannelli L, Dingsr GE, Bogstad B, Ottersen G, Chan KS,Gjster H, Stiansen JE, Stenseth NC (2007) Spatialanatomy of species survival: effects of predation and cli-

296

Hjermann et al.: Fish and oil in the LofotenBarents Sea system

mate-driven environmental variability. Ecology 88:635646Clark RB (2001) Marine Pollution, Oxford University Press,

OxfordConover DO, Munch SB (2002) Sustaining fisheries yields

over evolutionary time scales. Science 297:9496Couillard CM, Lee K, Legare B, King TL (2005) Effect of dis-

persant on the composition of the water-accommodatedfraction of crude oil and its toxicity to larval marine fish.Environ Toxicol Chem 24:14961504

Cushing DH (1990) Plankton production and year-classstrength in fish populationsan update of the match mis-match hypothesis. Adv Mar Biol 26:249293

Dalpadado P, Borkner N, Bogstad B, Mehl S (2001) Distribu-tion of Themisto (Amphipoda) spp. in the Barents Sea andpredator-prey interactions. ICES J Mar Sci 58:876895

Dauble DD, Gray RH, Skalski JR, Lusty EW, Simmons MA(1985) Avoidance of a water-soluble fraction of coal liquidby fathead minnows. Trans Am Fish Soc 114:754760

Dickson RR, Osborn TJ, Hurrell JW, Meincke J and 5 others(2000) The Arctic Ocean response to the North Atlanticoscillation. J Clim 13:26712696

Dingsr GE, Ciannelli L, Chan KS, Ottersen G, Stenseth NC(2007) Density dependence and density independenceduring the early life stages of 4 marine fish stocks. Ecology88:625634

Durant JM, Hjermann D, Ottersen G, Stenseth NC (2007)Climate and the match or mismatch between predatorrequirements and resource availability. Clim Res 33:271283

Eikeset AM, Dunlop E, Dieckmann U (2005) Fisheries-induced evolution in Northeast Arctic cod. ICES CM2005/AA:16

Ernande B, Dieckmann U, Heino M (2004) Adaptive changesin harvested populations: plasticity and evolution of ageand size at maturation. Proc R Soc Lond Ser B 271:415423

ExxonMobil (2004) ExxonMobil sets Valdez record straight.News Release October 6, 2004. (Retrieved from www.exxonmobil.com/ on March 23, 2007)

Farr AJ, Chabot CC, Taylor DH (1995) Behavioral avoidanceof fluoranthene by fathead minnows (Pimephales prome-las). Neurotoxicol Teratol 17:265271

Fiksen O, Slotte A (2002) Stock-environment recruitmentmodels for Norwegian spring spawning herring (Clupeaharengus). Can J Fish Aquat Sci 59:211217

Fossum I (2006) Analysis of instabilities and mesoscalemotion off southern Norway. J Geophys Res 111:C08006,doi: 10.1029/2005JC003228

Fossum I, Red LP (2006) Analysis of instabilities and meso-scale motion in continuously stratified ocean models.J Mar Res 64:319353

Garrett RM, Rothenburger SJ, Prince RC (2003) Biodegrada-tion of fuel oil under laboratory and Arctic marine condi-tions. Spill Science Technol Bull 8:297302

Gjster H (1998) The population biology and exploitation ofcapelin (Mallotus villosus) in the Barents Sea. Sarsia 83:453496

Gjster H, Bogstad B (1998) Effects of the presence of her-ring (Clupea harengus) on the stock-recruitment relation-ship of Barents Sea capelin (Mallotus villosus). Fish Res38:5771

Gjster H, Loeng H (1987) Growth of the Barents Seacapelin, Mallotus Villosus, in relation to climate. EnvironBiol Fish 20:293300

Gjster H, Dommasnes A, Rottingen B (1998) The BarentsSea capelin stock 19721997. A synthesis of results fromacoustic surveys. Sarsia 83:497510

Gjster H, Dalpadado P, Hassel A (2002) Growth of Barents

Sea capelin (Mallotus villosus) in relation to zooplanktonabundance. ICES J Mar Sci 59:959967

Hamre J (1994) Biodiversity and exploitation of the main fishstocks in the NorwegianBarents Sea ecosystem. Bio-divers Conserv 3:473492

Hamre J (2003) Capelin and herring as key species for theyield of north-east Arctic cod. Results from multispeciesmodel runs. Scient Mar 67 (Suppl. 1):315323

Hassel A, Skjoldal HR, Gjster H, Loeng H, Omli L (1991)Impact of grazing from Capelin (Mallotus villosus) on zoo-planktona case-study in the northern Barents Sea inAugust 1985. Polar Res 10:371388

Haug T, Nilssen KT (1995) Ecological implications of harpseal Phoca groenlandica invasions in northern Norway. In:Blix AS, Walle L, Ulltang O (eds) Whales, seals, fish andman, Vol 4. Elsevier, Amsterdam, p 545556

Heino M, God OR (2002) Fisheries-induced selection pres-sures in the context of sustainable fisheries. Bull Mar Sci70:639656

Heino M, Dieckmann U, God OR (2002a) Measuring prob-abilistic reaction norms for age and size at maturation.Evolution 56:669678

Heino M, Dieckmann U, God OR (2002b) Reaction normanalysis of fishery-induced adaptive change and the caseof the Northeast Arctic cod. ICES CM Y:14

Heintz RA, Rice SD, Wertheimer AC, Bradshaw RF, Thrower FP,Joyce JE, Short JW (2000) Delayed effects on growth andmarine survival of pink salmon Oncorhynchus gorbuschaafter exposure to crude oil during embryonic development.Mar Ecol Prog Ser 208:205216

Helland-Hansen B, Nansen F (1909) The Norwegian Sea:Its physical oceanography based upon the Norwegian re-search 19001904, Part 1, No. 2. Fiskeridir Skr Ser Hav-unders 3

Hellstrm T, Dving KB (1983) Perception of diesel oil by cod(Gadus morhua L.). Aquat Toxicol 4:303315

Hillebrand H (2004) Strength, slope and variability of marinelatitudinal gradients. Mar Ecol Prog Ser 273:251267

Hjermann D, Stenseth NC, Ottersen G (2004a) The popula-tion dynamics of northeast Arctic cod through two de-cades: an analysis based on survey data. Can J Fish AquatSci 61:17471755

Hjermann D, Stenseth NC, Ottersen G (2004b) Indirect cli-matic forcing of the Barents Sea capelin: a cohort-effect.Mar Ecol Prog Ser 273:229238

Hjermann D, Ottersen G, Stenseth NC (2004c) Competitionamong fishermen and fish causes the collapse of BarentsSea capelin. Proc Natl Acad Sci USA 101:1167911684

Hjermann D, Bogstad B, Eikeset AM, Ottersen G, GjsterH, Stenseth NC (2007) Food web dynamics affect North-East Arctic cod recruitment. Proc R Soc Lond Ser B 274:661669

Hjort J (1914) Fluctuations in the great fisheries of northernEurope viewed in the light of biological research. RappP-V Run Cons Int Explor Mer 20:1228

Hose JE, McGurk D, Marty GD, Hinton DE, Brown ED, BakerTT (1996) Sublethal effects of the Exxon Valdez oil spill onherring embryos and larvae: Morphological, cytogenetic,and histopathological assessments, 19891991. Can J FishAquat Sci 53:23552365

Hurrell JW, Kushnir Y, Ottersen G, Visbeck M (eds) (2003)The North Atlantic Oscillation: climate significance andenvironmental impact, Vol 134. American GeophysicalUnion, Washington, DC

ICES (2005) Report of the Arctic Fisheries Working Group.ICES CM 2005/ACFM:20

Incardona JP, Collier TK, Scholz NL (2004) Defects in cardiac

297

Mar Ecol Prog Ser 339: 283299, 2007

function precede morphological abnormalities in fishembryos exposed to polycyclic aromatic hydrocarbons.Toxicol Appl Pharmacol 196:191205

Ingvaldsen R, Loeng H, Ottersen G, dlandsvik B (2003)Climate variability in the Barents Sea during the 20th cen-tury with a focus on the 1990s. ICES Mar Sci Symp 219:160168

IPIECA (1997) Biological impacts of oil pollution: fisheries.IPIECA Rep Ser 8, International Petroleum Industry Envi-ronmental Conversation Association, London

ITOPF (2002) Fate of marine oil spills. Tech Info Pap, Inter-national Tanker Owners Pollution Federation, London

Johansen , Skognes K, Aspholm O, stby C, Moe KA,Fossum P (2003) Uhellsutslipp av oljekonsekvenser ivannsylen (ULB 7-c). Foundation for Scientific and Indus-trial Research at the Norwegian Institute of Technology(SINTEF), Trondheim

Kjesbu OS, Kryvi H, Sundby S, Solemdal P (1992) Buoyancyvariations in eggs of Atlantic cod (Gadus morhua L.) inrelation to chorion thickness and egg sizetheory andobservations. J Fish Biol 41:581599

Kjesbu OS, Solemdal P, Bratland P, Fonn M (1996) Varia-tion in annual egg production in individual captiveAtlantic cod (Gadus morhua). Can J Fish Aquat Sci 53:610620

Kjesbu OS, Witthames PR, Solemdal P, Walker MG (1998)Temporal variations in the fecundity of Arcto-Norwegiancod (Gadus morhua) in response to natural changes infood and temperature. J Sea Res 40:303321

Loeng H (1991) Features of the physical oceanographic condi-tions of the Barents Sea. Polar Res 10:518

Marshall CT, Kjesbu OS, Yaragina NA, Solemdal P, Ulltang O(1998) Is spawner biomass a sensitive measure of thereproductive and recruitment potential of NortheastArctic cod? Can J Fish Aquat Sci 55:17661783

Marshall CT, Yaragina NA, Lambert Y, Kjesbu OS (1999)Total lipid energy as a proxy for total egg production byfish stocks. Nature 402:288290

Martinsen EA (1982) Operational oil drift model for Nor-wegian waters and surrounding oceans. Tech Rep 59,Norwegian Meteorological Institute, Oslo (in Norwegian)

Martinsen EA, Melsom A, Sveen V, Grong E, Reistad M, Hal-vorsen N, Johansen , Skognes K (1994) The operationaloil drift system at DNMI. Tech Rep 125, Norwegian Mete-orological Institute, Oslo

Mehl S (2004) Abundance of spawning Northeast Arctic codspring 2004. Toktrapport Nr. 92004, Institute of MarineResearch, Bergen, Norway

Mehl S, Nedreaas K (2003) Abundance of spawning North-east Arctic cod spring 2003, Toktrapport Nr. 132003,Institute of Marine Research, Bergen, Norway

Melsom A (2005) Mesoscale activity in the North Sea as seenin ensemble simulations. Ocean Dyn 55:338350

Melsom A, Metzger EJ, Hurlburt HE (2003) Impact of remoteoceanic forcing on Gulf of Alaska sea levels and mesoscalecirculation. J Geophys Res Oceans 108:C113346, doi:10.1029/2002JC001742

Metzger EJ, Hurlburt HE (2001) The nondeterministic natureof Kuroshio penetration and eddy shedding in the SouthChina Sea. J Phys Oceanogr 31:17121732

National Research Council (1985) Oil in the sea. III: Inputs,fates and effects. National Academies Press, Washington,DC

Olsen EM, Heino M, Lilly GR, Morgan MJ, Brattey J, ErnandeB, Dieckmann U (2004) Maturation trends indicative ofrapid evolution preceded the collapse of northern cod.Nature 428:932935

Orvik KA, Niiler P (2002) Major pathways of Atlantic waterin the northern North Atlantic and Nordic Seas towardArctic. Geophys Res Lett 29:1896

Orvik KA, Skagseth O, Mork M (2001) Atlantic inflow to theNordic Seas: current structure and volume fluxes frommoored current meters, VM-ADCP and SeaSoar-CTDobservations, 19951999. Deep-Sea Res I 48:937957

sterhus S, Hansen B (2000) North AtlanticNorwegian Seaexchanges. Progr Oceanogr 45:109208

Ottersen G, Planque B, Belgrano A, Post E, Reid PC, StensethNC (2001) Ecological effects of the North Atlantic Oscilla-tion. Oecologia 128:114

Ottersen G, Hjermann D, Stenseth NC (2006) Changes inspawning stock structure strengthens the link betweenclimate and recruitment in a heavily fished cod stock. FishOceanogr 15:230243

Paine RT (1980) Food webs linkage, interaction strengthand community infrastructure. The 3rd Tansley Lecture.J Anim Ecol 49:667685

Paine RT, Ruesink JL, Sun A, Soulanille EL, Wonham MJ,Harley CDG, Brumbaugh DR, Secord DL (1996) Troubleon oiled waters: lessons from the Exxon Valdez oil spill.Annu Rev Ecol Systs 27:197235

Perkins RA, Rhoton S, Behr-Andres C (2005) Comparativemarine toxicity testing: a cold-water species and standardwarm-water test species exposed to crude oil and disper-sant. Cold Regions Sci Technol 42:226236

Peterson CH, Rice SD, Short JW, Esler D, Bodkin JL, Balla-chey BE, Irons DB (2003) Long-term ecosystem responseto the Exxon Valdez oil spill. Science 302:20822086

Reznick DN, Ghalambor CK (2005) Can commercial fishingcause evolution? Answers from guppies. Can J Fish AquatSci 62:791801

Reznick DN, Rodd FH, Cardenas M (1996) Life-history evolu-tion in guppies (Poecilia reticulata: Poeciliidae).4. Paral-lelism in life-history phenotypes. Am Nat 147:319338

Rijnsdorp AD, Grift RE, Kraak SBM (2005) Fisheries-inducedadaptive change in reproductive investment in North Seaplaice (Pleuronectes platessa)? Can J Fish Aquat Sci 62:833843

Rudberg A (2003a) Oljedriftsmodellering i Lofoten og Bar-entshavet. Spredning av olje ved akutte utslipp. ULBdelutredning 7-a. DNV rapport nr. 20030385, Det NorskeVeritas, Hvik

Rudberg A (2003b) Oljedriftsmodellering i Lofoten og Bar-entshavet; spredning av olje ved akutte utslipp til sj. ULBDelutredning 7-a, Det Norske Veritas, Hvik

Stre R, Mork M (1981) The Norwegian Coastal Current, Vol1 & 2. Proc Norwegian Coastal Current Symp, Universityof Bergen, Bergen, Norway

Salthaug A (2002) Revidert rapport fra tokt med F/F G.O.Sars, Lofoten 16.03.02.04.02. Toktrapport Nr. 342002,Institute of Marine Research, Bergen

Scott GR, Sloman KA (2004) The effects of environmentalpollutants on complex fish behaviour: integrating be-havioural and physiological indicators of toxicity. AquatToxicol 68:369392

Stenseth NC, Mysterud A, Ottersen G, Hurrell JW, Chan KS,Lima M (2002) Ecological effects of climate fluctuations.Science 297:12921296

Stenseth NC, Ottersen G, Hurrell JW, Mysterud A, Lima M,Chan KS, Yoccoz NG, Adlandsvik B (2003) Studyingclimate effects on ecology through the use of climateindices: the North Atlantic Oscillation, El Nio-SouthernOscillation and beyond. Proc R Soc Lond Ser B 270:20872096

Tjelmeland S, Bogstad B (1998) MULTSPECa review of a

298

Hjermann et al.: Fish and oil in the LofotenBarents Sea system

multispecies modelling project for the Barents Sea. FishRes 37:127142

USGS (2000) World petroleum assessment 2000Descriptionand results, US Geological Survey, Reston, VA

Webster L, Angus L, Topping G, Dalgarno EJ, Moffat CF(1997) Long-term monitoring of polycyclic aromatic hydro-carbons in mussels (Mytilus edulis) following the Braer oilspill. Analyst 122:14911495

Webster L, McIntosh AD, Moffat CF, Dalgarno EJ, BrownNA, Fryer RJ (2000) Analysis of sediments from ShetlandIsland voes for polycyclic aromatic hydrocarbons, steranesand triterpanes. J Environ Monitor 2:2938

Webster L, Fryer RJ, Dalgarno EJ, Megginson C, Moffat CF(2001) The polycyclic aromatic hydrocarbon and geo-

chemical biomarker composition of sediments from voesand coastal areas in the Shetland and Orkney Islands.J Environ Monitor 3:591601

Wettre C, Johansen , Skognes K (2001) Development of a3-dimensional oil drift model at DNMI. Res Rep No. 133,Norwegian Meteorological Institute, Oslo

Wolfe DA, Hameedi MJ, Galt JA, Watabayashi G and 8 others(1994) The fate of the oil spilled from the Exxon Valdez.Environ Sci Technol 28:A560A568

Worm B, Myers RA (2003) Meta-analysis of cod-shrimp inter-actions reveals topdown control in oceanic food webs.Ecology 84:162173

Zenkevitch LA (1963) Biology of the seas of the USSR. Allanand Unwin, London

299

Editorial responsibility: Howard Browman (Associate Editor-in-Chief), Storeb, Norway

Submitted: June 11, 2006; Accepted: October 10, 2006Proofs received from author(s): May 3, 2007