Climatic Change (2008) 87:251–262 DOI 10.1007/s10584-007-9338-0

An integrated study of economic effects of and vulnerabilities to global warming on the Barents Sea cod

Arne Eide

Received: 6 July 2006 /Accepted: 3 October 2007 / Published online: 27 November 2007 # Springer Science + Business Media B.V. 2007

Abstract The Barents Sea area is characterised by a highly fluctuating physical environment causing substantial variations in the ecosystems and fisheries depending upon this. Simulations assuming different management regimes have been carried out to study how physical and biological effects of global warming influence the Barents Sea . A regional, high-resolution representation of the B2 world region (OECD90) scenario from the Intergovernmental Panel on Climate Change was used to calculate water temperatures and plankton biomasses by hydrodynamic modelling. These results were included in simulations performed by a multi-fleet, multi-species model, by which a fully integrated model linking to the global circulation model to the Barents Sea fisheries through a regional downscaling to the Barents Sea area is constructed. One factor of particular importance for the natural annual biological variations is the occasional inflow of young herring into the Barents Sea area. The herring inflow is difficult to predict and links to dynamical systems outside the Barents Sea area, complex recruitment mechanisms and oceanographic conditions. These processes are in the study represented by a stochastic representation of herring inflow based on historical observations. According to the performed simulations the fluctuations may slightly increase over the next 25 years, possibly caused by changes in temperature patterns. Six different management regimes have been included in the study and the results support earlier studies claiming that the choice of management regime potentially has a greater importance for biological and economic performance in the Barents Sea fisheries than impacts which derive from global warming over the next 25 years. A basic assumption for this conclusion is however that the Barents Sea ecosystem essentially preserves its structure and composition of today. Possible, unpredictable significant shifts in the ecosystem structure are not considered.

1 Introduction

Impacts of global warming on biological and economic systems are a major concern in the world today (Parry et al. 2007). However, it is a complex task to identify long term

A. Eide (*) Norwegian College of Science, University of Tromsø, Breivika, Tromsø 9037, Norway e-mail: [email protected] 252 Climatic Change (2008) 87:251–262 consequences of global warming on these systems. In particular local and regional effects are difficult to predict. Also normal variations in physical and biological systems, and in and between regions are substantial. Regional models, which have been developed to determine vulnerabilities of such systems, therefore include normal seasonal variations within and between years. Similarly to many other fisheries also the cod fisheries in the Barents Sea area closely links to seasonal variations. The modern history of reflects a development of adaptation techniques to varying environmental situations. The new concept of Harvest Control Rules (HCR) opens for a more dynamic adaptation compared with the traditional more static regulation procedures (Thompson 1999). This study is a part of the European BALANCE project, which aims to assess vulnerabilities due to climate changes in the Barents region. Within the BALANCE project integrated vulnerability studies are carried out, based on the Intergovernmental Panel on Climate Change (IPCC) Special Report on Emissions Scenarios (SRES; Nakícenovíc et al. 2000) scenario B2, world region OECD90. The B2 scenario assumes effective local solutions to economic, social, and environmental sustainability issues and intermediate economic development (the regional/environmental scenario). As the other scenarios also the B2 scenario therefore includes assumptions related to the magnitude and distribution of different types of economic activities throughout the simulation period and management decisions related to these activities. The precautionary principle in fisheries management fits the basic assumptions of the B2 scenario. Physical variables are brought into the integrated model structure in a more straight forward manner, as indicated in Fig. 1.

EconMult

Stock unit biomasses Catches Management EconSimp2000 AggMult

Temperatures Plankton biomasses SinMod

Physical variables REMO 5.1

Physical variables Intergovernmental Panel on Climate Change (IPCC) Special Report on Emissions Scenarios (SRES) B2 Physical Biological Economic Management environment environment activity decisions

Fig. 1 Flow chart of the integrated model structure covering the physical and biological environment, economic activity and management. EconSimp2000 is shown as the combined model of AggMult and EconMult. The B2 scenario covers all sectors, including management decisions, but only the physical variables are direct inputs to the integrated model, while the management integration (indicated by a dashed arrow) exists in terms of consistency. Up to the fishing activity the model integration has only one direction, while feedback mechanisms are included in the combined model (EconSimp2000) Climatic Change (2008) 87:251–262 253

Physical conditions are downscaled to the Barents region by applying the REMO5.1 model (Jacob et al. 2001). REMO5.1 is a regional climate model with ½ degree resolution, driven by the global atmospheric general circulation model ECHAM4 (Roeckner et al. 1996). The REMO5.1 climate simulation data is used in calculating values on physical variables in the 3D ocean circulation model SinMod (for more details see Pedersen et al. 2003; Slagstad and McClimans 2005). In addition to the Barents Sea hydrodynamics, SinMod also covers primary and secondary biological production in Atlantic and Arctic waters of the Barents Sea. Temperatures obtained by SinMod simulations indicate a slight increase in average sea temperature in the 50 meter upper layer of the Atlantic water the next 25 years (Fig. 2). The quarterly trend graphs in Fig. 2 show rather moderate increases, about half a Celsius degree within each quarter, much less than the normal quarterly variation over a year. The SinMod results lay however well within the probability range from −2 to + 3 Celsius degrees (compared with current average level) assumed in previous studies of global warming impacts on the Barents Sea fisheries (Eide and Heen 2002; Eide 2007). The uncertainty reflected by the rather wide probability range relates mainly to the two counterworking effects of warmer Atlantic water flowing into the Barents Sea basin and the slight reduction in Atlantic water inflow. The B2/REMO5.1/SinMod simulations propose that the two effects almost level out each other.

January - March April - June 2005 2010 2015 2020 2025 2030 2005 2010 2015 2020 2025 2030 8 8

C) 7 C) 7 ° ° 6 6

5 5 4 4 Temperature ( Temperature ( Temperature 3 3 2005 2010 2015 2020 2025 2030 2005 2010 2015 2020 2025 2030 Year Year

July - September October - December 2005 2010 2015 2020 2025 2030 2005 2010 2015 2020 2025 2030 8 8

7 C) 7 C) ° ° 6 6

5 5 4 4 Temperature ( Temperature Temperature ( Temperature 3 3 2005 2010 2015 2020 2025 2030 2005 2010 2015 2020 2025 2030 Year Year

Fig. 2 SinMod calculated average quarterly temperatures of the Atlantic water upper 50 m layer in the Barents Sea 2005–2030, based on simulations by the regional model Remo5.1 (points). The solid lines show the trends (linear regression model) while the dotted lines gives the corresponding 95% confidence intervals 254 Climatic Change (2008) 87:251–262

2 Materials and methods

The SinMod results have been implemented in the EconSimp2000 model (Eide 2007), the combined model of AggMult (Tjelmeland and Bogstad 1998) and EconMult (Eide and Flaaten 1998). AggMult covers the essential parts of the Barents Sea ecosystem which relates to ongoing commercial fisheries, cod and capelin being the most important fish species. Herring also plays a role in the Barents Sea ecosystem, as described below. Individual growth and predation is modelled by the feeding level principle (Ursin 1967), length based maturation and Beverton and Holt recruitment at age 0 (see Tjelmeland and Bogstad 1998 for details). The time resolution in AggMult is a quarter of a year, similar to the time step in the fleet model, EconMult. In addition to the major fish populations, cod, capelin and herring, the Barents Sea ecosystem is represented in AggMult by three plankton biomasses, small plankton in the southern Barents Sea (food for capelin and herring), small plankton in the northern Barents Sea (food for capelin) and large plankton organisms (food for cod, herring and capelin). In this study, the plankton growth rates are calculated within the SinMod model. While assuming the essential ecosystem structure of today to prevail, the system is influenced by global warming through water temperature changes and growth changes in plankton biomasses. The latter also acts as a proxy of oceanographic changes in the combined model, reflecting changes in spatial distribution and production, influencing fish production and harvest. Essentially EconSimp2000 is a deterministic model. The AggMult module has however the possibility of including some stochastic processes, particularly related to the complex interaction between the North Sea and the Barents Sea ecosystems, represented by occasional inflow from the south of young herring into the Barents Sea basin (Huse et al. 2002). The ecosystem of the Barents Sea is strongly dominated by the cod-capelin interaction where also the spring spawning herring plays an important role. The historical variation in herring stock recruitment performance and Barents Sea inflow is represented stochastically and Monte Carlo simulations of 100 runs has been carried out for a period of 25 years (2005–2030) for each management scenario. Cod cannibalism also represents a powerful buffering mechanism responding on environ- mental changes, including changes in the biological environment and a varying age composition within the cod population (Wikan and Eide 2004) and is built into the AggMult model. The fleet model (EconMult) is parameterised on the basis of Norwegian national statistics (Anon 2000) and fleet segments (Table 1) as in Eide (2007). 14 fleet segments are included in the cod fisheries as seen in Table 1. In addition to the cod fisheries 4 fleet segments operates in the capelin fishery which is automated to follow the present Harvest Control Rule (HCR) as an integrated part of the cod management. Six different management regimes are studied in the cod fisheries. The management regimes (Table 2) include two levels of constant total allowable catch (TAC), two regimes based on precautionary approach (PA), one limited entry regime (LE) and the pure open access (OA). In all cases except LE, the fleet composition is assumed to adapt dynamically to the economics of each fleet segment, assuming an annual entry rate of 5% and exit rate of 3%. Standard PA management is used as reference management regime. A new PA management regime, the 3 year rule PA management (3PA), was introduced to the fishery from 2005. The PA management rules of the Barents Sea cod fishery are the following: & Total Allowable Catch (TAC) is calculated on the basis of the precautionary fishing mortality rate (predefined to be 0.4, see for example Åsnes 2005). Climatic Change (2008) 87:251–262 255

Table 1 Vessels included in EconSimp2000

EconSimp Stat. Vessel categories in length, Targeted Fishing gears Home region group no. group weight or volume species

1 001 8.0–12.9 m Cod Gill nets/hand lines Northern Norway 2 002 13.0–20.9 m Cod Gill nets/hand lines Northern Norway 3 003 8.0–12.9 m Cod Danish seine Northern Norway 4 004 13.0–20.9 m Cod Danish seine Northern Norway 5 005 8.0–12.9 m Cod Long line Northern Norway 6 006 13.0 – 20.9 m Cod Long line Northern Norway 7 007 8.0–12.9 m Cod Sundry gears Southern Norway 8 008 13.0–20.9 m Cod Sundry gears Southern Norway 9 009 21.0–27.9 m Cod Danish seine Northern Norway 10 010 21.0–27.9 m Cod Sundry gears Norway 11 011 ≥28.0 m Cod Long line Norway 12 012 ≥28.0 m Cod Sundry gears Norway 13 013 ≥250BRT/500TE Cod Wet fish/freezers trawlers Norway 14 014 ≥250BRT/500TE Cod Industrial trawlers Norway 15 023 All Capelin Trawl Norway 16 027 <8,000 hl Capelin Purse seine Norway 17 028 ≥8,00 hl Capelin Purse seine Norway 18 029 All Capelin Purse seine Norway

The reference numbers (stat. group) refer to statistical categories used in the annually published surveys of the by the Norwegian Fisheries Directorate (Anon 2000)

& If the cod stock spawning biomass is estimated to be below the predefined precautionary approach level of 460 thousand tonnes, initially TAC setting is reduced correspondingly. If the cod stock spawning biomass is below the critical level of 220 thousand tonnes, TAC is set to 0.

The additional 3 year rules implemented in 2005 are the following: & TAC is calculated as above for three consecutive years based on stock prognosis and a TAC value is set equal the mean value of the three calculated PA-values of TAC. The prognostic model (PROST) makes use of a population dynamic model similar to what is assumed in virtual population analysis runs (VPA or XSA, see Åsnes 2005).

Table 2 Management regimes for the Barents Sea cod fisheries applied in the EconMult2000 simulations

No. Code Quota Fixed Management regimes fleet

1 PA Yes No Precautionary approach (0 regime) 2 3PA Yes No Precautionary approach and a the 3 year rule 3 LE No Yes Limited entry. 4 TAC1 Yes No Constant catch quota equal 242.5 thousand tonnes which is the actual Norwegian catch quota of 2005. 5 TAC2 Yes No Constant catch quota equal 400 thousand tonnes 6 OA no No Open access 256 Climatic Change (2008) 87:251–262

& If the above calculated average value of TAC exceeds + /− 10 per cent compared with TAC of the last year, the new TAC-value is adjusted until it equals 90% or 110% of the last year TAC-value. & If the spawning biomass is below the precautionary approach level, the 10% rule above does not apply. A necessary additional rule for making long term simulations has been added in this study: & The 10% rule is omitted if previous year’s spawning stock level was below the precautionary approach level and hence last year’s TAC value was reduced accordingly. The last TAC without more that 10% reduction of previous is then obtained as new TAC reference value. The new prognostic model for management purposes (PROST, see Åsnes 2005) is linked to EconSimp2000 and employed whenever PA management regimes are included. The fleet size of the limited entry management regime is calculated on the basis of the 2005 TAC value. Assuming a full utilising of the year 2000 fleet, 25% of the total Norwegian fleet is sufficient to obtain the TAC allocated to Norway. Hence in the LE management regime the dynamic fleet is replaced by a fixed fleet 25% of the year 2000 fleet, equally reduced in all fleet segments. Negative contribution margins are assumed to switch off the fishing activity of the fleet segment within the corresponding quarters of a year.

3 Results

100 simulations were performed for each of the six management regimes while varying the herring inflow randomly, applying the same 100 seed random processes in each management case. Each simulation covered a time period of 25 years, starting the simulations in year 2005. The variation caused by herring inflow over the simulation period is reflected in the error bars shown in Fig. 3 for the cod stock biomasses of different management regimes. The average values in the same figures may not have any significant interpretation except the expectancy of a rather stable level or trend throughout the simulation period. The first year of the simulation period will however be heavily influenced by the initial value of the simulations, which dampen out over time. The variance in the Monte Carlo simulations indicated by the error bars in Fig. 3 show significant differences between management regimes while such differences are not obvious over time. As global warming is increasing throughout the simulation period, the two five year periods in the beginning and in the end of the period were compared in order to investigate changes over time. The first five year period was set to 2007–2011 in order to reduce the influence from the initial condition of 2005, while the last five year period is the last five years of the simulation. Quantile plots of stock biomass and wage paying ability (profit before cost of labour) are shown in Figs. 4 and 5 for the two five year periods. The characteristics of the datasets of the two quantile plots and the additional datasets of Norwegian catches and average wage paying ability per vessel in the Norwegian fleet, standardised by number of fishers at each vessel, are shown in Fig. 6. Figures 5 and 6 introduce the concept of Wage Paying Ability (WPA), a way of measuring fleet profits adopted by the Norwegian Fisheries Directorate in their publications Climatic Change (2008) 87:251–262 257

Fig. 3 Error plots of cod stock PA 3PA biomass estimates (in mill. 4 4 tonnes) under six different man- 3 3 agement regimes, as presented in Table 1. 100 stochastic simula- 2 2 tions are performed of each man- agement regime, reproducing the 1 1 same seed random processes in all management cases. Average 2005 2015 2025 2005 2015 2025 biomass estimate of each year are LE TAC1 indicated as dots and the 4 4 corresponding standard deviation as solid vertical lines 3 3 2 2

1 1

2005 2015 2025 2005 2015 2025 TAC2 OA 4 4

3 3

2 2

1 1

2005 2015 2025 2005 2015 2025

on vessel economics (Anon 2000). WPA reflects the traditional way of regarding the fishing activity as a risk sharing operation between fishers, more similar to share holders than to employees and owner. The labour is then paid off by sharing the profits rather than earning a salary.

4 Discussion

The water temperature change in the Barents Sea the next 25 years (Fig. 2) caused by global warming is moderate compared with the natural variation within and between years. The temperature change develops over time and the simulations presented indicate the change on average to be about half a degree Celsius over the whole period. Possible impact from the changes in temperatures and plankton biomasses is seen in Fig. 4. The figure shows an increased fluctuation tendency over time, as low biomasses is lower and higher biomasses larger from the start to the end of the periods in all management regimes, as the quantile plots crosses the 45 degree lines upward when the biomass increases. It seems to be a rather robust conclusion based on Fig. 4 that the fluctuation behaviour of the cod stock biomass increases over the simulation period, independent from management. In particular this is valid for the high biomass levels, which tend to be even higher after increased global warming, while the lower biomass levels seem to be even lower in most cases. One important exception is however seen in the PA case, clearly differing from both 3PA and all other management regimes. Fig. 4 also show surprisingly stable fluctuation ranges on low biomass levels for TAC2, while both TAC1 and TAC2 represents the extremes in the upper biomass fluctuation rages. A similar pattern 258 Climatic Change (2008) 87:251–262

Fig. 4 Quantile plots of estimated stock biomass (in 1000 tonnes) the first (horizontal axes) and last (vertical axes) 5 years of the simulation period for each of the six investigated management regimes described in Table 2 is not shown in the case of OA, which span out the widest range of biomass fluctuation. But also here the overall picture of an increased tendency of biomass fluctuations over time is confirmed. A corresponding picture is obtained by the Box plots of the data sets (Fig. 6). The fleet dynamics contribute in replacing less economic efficient vessels by more efficient vessels from other fleet segments, sufficient to compensate the increased variation. The entry and exit rates seem not to be critical values, being far below the corresponding annual rates of biomass changes and changes in age composition with the cod stock. In the case of limited entry (LE) the entry and exit rates are however set equal zero, as the fleet composition and size is fixed. The LE quantile plot in Fig. 5 therefore reflects the same pattern as seen in Fig. 4. In the case of open access (OA) the effect of increased fluctuations is taken advantage of by increasing the fleet and the fleet activity beyond the limits of the other management regimes, which places the OA quantile plot above the 45 degree line in the positive area of wage paying ability. The quantile plots shown in Fig. 5 reflect also how the fleet properties convert the different biomass situations into values (WPA), given the management constraints. Without management (OA) the fleet is able to increase all positive WPA, but suffers correspondingly even bigger losses when profits turn negative. The most interesting observation from Fig. 5 is however that the pattern seen in PA and 3PA is significantly different from all other regimes, including OA. In PA and 3PA the increased biomass fluctuations is converted to reduced WPA fluctuations, as the previous low WPA- levels are replaced by higher and previous high WPA-levels are replaced by lower. The PA and 3PA regimes prove to stabilise the fleet performance over time, not only reducing the occurrence of low WPAs, but also the high earnings. It is a close linkage between occurrence of low stock biomass levels and the extreme highs. This may partly be caused by the fishing dynamics, as very low stock biomass levels Climatic Change (2008) 87:251–262 259

Fig. 5 Quantile plots of estimated total wage paying ability (bill. NOK) the first (horizontal axes) and last (vertical axes) 5 years of the simulation period for each of the six investigated management regimes described in Table 2 in an open fleet dynamics leads to fleet reductions and lower fleet capacity during the first period of stock recovery, which eventually brings the stock up to higher biomass levels than without such capacity reduction the years before. As long as the growth rate in the fleet is lower than the growth rate of the total biomass, the fleet will not be able to immediately adjust to the new stock situation. The stock dynamics, which is speeded up in the case of open access (OA), therefore unavoidably leads to a situation where a substantial resource rent is harvested in an unregulated fishery. Correspondingly negative profits are obtained when the stock collapses, but the peaks more than compensates the severe losses. Over the simulation period the open access therefore sum up the highest total resource rent, even though periods of huge losses exist. The same findings is described in Eide (2007), where the open access regime turned out the give the highest total resource rent summed up over the simulation period, as the richness of the extreme peaks more than compensate the loss of the extreme lows. The cost of stabilising the biomass development relates to the lack of extremely high biomasses, as seen in the cases of the PA-regimes. The PA harvest control rules works according to the intentions, adapting to new environmental conditions in a precautionary manner, levelling out the most pronounced biomass fluctuations. While all management regimes but the limited entry (LE) regime manage to keep or increase the average wage paying ability (Fig. 6), the LE regime shows a clear reduction although it clearly keeps the highest average wages (per fisher). Being the only management regime with restrictions on effort, LE distributes resource rent to the holders of fishing rights. The resource rent reduction over time reflects however reduced relative cost efficiency in the fleet composition as this is fixed initially. The fleet dynamics of the other management regimes provides the total fleet with a possibility to adapt to changes in the stock situation. 260 Climatic Change (2008) 87:251–262

4

3 onnes)

2

1 Total biomass (mill. t 0 PA 3PA LE TAC1 TAC2 OA

800

600

400

200 Norwegian catch (1000 tonnes) 0 PA 3PA LE TAC1 TAC2 OA

8

6

4

2

0

Wage paying-2 ability (bill. NOK )

PA 3PA LE TAC1 TAC2 OA

250

Average WPA (1000 NOK ) 0

PA 3PA LE TAC1 TAC2 OA

Fig. 6 Box-whisker-plots of cod stock biomasses (on top), Norwegian catches (mid graph) and Norwegian fleet wage paying abilities (on bottom) over periods of five years. Open boxes represents the first period of 5 years (2007–2011) and grey boxes (in front) the last 5 years in the simulation period (2025–2030). The six management regimes refer to the management in Table 2. Each box represents 50% of the dataset, the dotted line the median value. Total range indicated represents 95% of the data set Climatic Change (2008) 87:251–262 261

5 Conclusions

This study confirms the conclusions from other studies finding the impacts on the Barents Sea cod fisheries from global warming through water temperatures and other oceanographic changes to be insignificant compared with the normal environmental fluctuations experienced in this area, although increased fluctuations in stock biomass and stock age composition are found. Choice of management principles as reflected in the different management regimes studied, prove however strongly to influence the resulting biological and economic indicators. The principle of precautionary approach has the highest adaptive ability, while open access aims to take advantage of the biomass fluctuations as a consequence of no management. The adaptive capacity of the precautionary approach management systems have not been fully utilised in the simulations, as the current harvest control indicators constrain the adaptation system. Increased environmental fluctuations indicates that these indicators (spawning stock biomass and fishing mortality rate) may not be the optimal ones for adaptive management when the aim is to reduce vulnerability and at the same time improve the economic performance of the management system. To achieve this, indicators adapted to environmental fluctuations need to be developed. This work has yet not been carried out. In the open access case the economic benefits during periods of high stock biomasses more than outreach losses during periods of critical low biomasses. On average this pulse fishery produce profits exceeding what is obtained by managed systems. To some extent it depends on the input/output dynamics of the fishing fleet, but any reasonable assumptions regarding fleet dynamics point in the same direction; increased fluctuations benefits the economy of an open access fishery. The risk of collapses without later stock biomass recovery, points in the opposite direction. This risk may increase with increased fluctuations and possible occurrence of ecosystem threshold values, whereby the system is altered into other energy flow patterns. Existence of such threshold values may dramatically increase the overall risk of reduced future biomass peaks compared to previous biomass levels. Based on such reasoning, system vulnerabilities are expected to decrease by implementing adaptive precautionary approach management systems, compared to open access fishery. Successful adaptive management systems may also be able to deliver better economic results than what was found in the precautionary approach systems focused in this study. A particular challenge is to find adaptive system procedures whereby the economic benefits caused by natural fluctuations, is harvested without increased risk of stock collapses and ecosystem shifts. The basic assumption of this study, as stated in the introductory part, is that the ecosystem structure essentially remains as today and no significant ecosystem shift occurs during the warming period. The risk of significant changes due to global warming, also depends on how the natural resources are exploited, particularly the choice of management system. Increased vulnerability, which may be associated with increased fluctuations, could be outreached by implementing proper adaptive management systems based on precautionary principles. Precautionary approach management proves in the study to stabilise the fishery by utilising some adaptive management techniques. The cost of the reduced risk of depletion and the increased market stability is that highly productive periods fail to appear. From the managers point of view this tradeoffs seems to be even more pronounced in the situation of increased global warming, and the importance of finding more useful indicators in harvest control rules increases.

Acknowledgements The development of the EconMult model and the global warming impact analyses is part of the EU-funded project BALANCE EVK2-2002-00169. The BALANCE project has involved a 262 Climatic Change (2008) 87:251–262 multidisciplinary collaboration which has made it possible to parameterise a fully integrated modelling tool dealing with the management of the Barents Sea fish resources. The contribution from the BALANCE community, work packages leader Knut Heen and valuable comments and suggestions from Rik Leemans and two anonymous referees are greatly appreciated.

References

Anon (2000) Lønnsomhetsundersøkelser for helårsdrevne fiskefartøy 8 meter største lengde og over. (In Norwegian). The Norwegian Fisheries Directorate. Bergen, Norway Åsnes M (2005) Prost users guide, internal note, Norwegian Institute of Marine Research (IMR). Bergen, Norway Eide A (2007) Economic impacts of global warming – the case of the Barents Sea fisheries. Natural Resource Modeling 20(2):199–221 Eide A, Flaaten O (1998) Bioeconomic multispecies models of the Barents Sea fisheries. In: Rødseth T (ed.) models for multispecies management. Physica-Verlag, Heidelberg, New York, pp 141–17, ISBN 3-7908- 1001-0 Eide A, Heen K (2002) Economic impacts of global warming. A study of the fishing industry in North Norway. Fish Res 56:261–274 Huse G, Railsback S, Fernø A (2002) Modelling changes in migration pattern of herring: collective behaviour and numerical domination. J Fish Biol 60:571 Jacob D, Andrae U, Elgered G, Fortelius C, Graham LP, Jackson SD, Karstens U, Koepken C, Lindau R, Podzun R, Rockel B, Rubel F, Sass HB, Smith RND, van den Hurk BJJM, Yang X (2001) A comprehensive model intercomparison study investigating the water budget during the BALTEX- PIDCAP Period. Meteorol Atmos Phys 77:19–43 Nakícenovíc N, Alcamo J, Davis G, de Vries B, Fenhann J, Gaffin S, Gregory K, Grübler A, Jung TY, Kram T, Emilio la Rovere E, Michaelis L, Mori S, Morita T, Pepper W, Pitcher H, Price L, Riahi K, Roehrl A, Rogner H-H, Sankovski A, Schlesinger ME, Shukla PR, Smith S, Swart RJ, van Rooyen S, Victor N, Dadi Z (2000) Special report on emissions scenarios. Cambridge University Press, Cambridge, pp 599 Parry MP, Canziani OF, Palutikof J (eds) (2007) Climate Change 2007. Impacts, adaptation and vulnerability. Cambridge University Press, Cambridge, (in press). Pedersen OP, Slagstad D, Tande K (2003) Hydrodynamic model forecasts as a guide for process studies on plankton and larval fish. Fish Oceanogr 12(4/5):369–380 Roeckner E, Arpe K, Bengtsson L, Christoph M, Claussen M, Dümenil L, Esch M, Giorgetta M, Schlese U, Schulzweida U (1996) The atmospheric general circulation model ECHAM4: Model description and simulation of present-day climate. Max-Planck-Institut für Meteorologie, Hamburg, Report No. 218. Slagstad D, McClimans TA (2005) Modeling the ecosystem dynamics of the Barents Sea including the marginal ice zone: I. Physical and chemical oceanography. J Mar Syst 58(25):1–18 Thompson G (1999) Optimizing harvest control rules. In: presence of natural variability and parameter uncertainty. Proceedings, 5th NMFS NSAW. NOAA Tech. Memo. NMFSF/SPO-40. Tjelmeland S, Bogstad B (1998) Biological modelling. In: Models for multispecies management. Physica- Verlag, Heidelberg, New York, pp 69–92 ISBN 3-7908-1001-0 Ursin E (1967) A mathematical model of some aspects of fish growth, respiration, and mortality. J Fish Res Board Can 24(11):2355–2453 Wikan A, Eide A (2004) An analysis of a nonlinear stage-structured cannibalism model with application on the Northeast Arctic cod stock. Bull Math Biol 66:1685–1704