An ecosystem modelling framework for incorporating climate regime shifts into fisheries management

Progress in Oceanography 115 (2013) 53–64

Contents lists available at SciVerse ScienceDirect

Progress in Oceanography

journal homepage: www.elsevier .com/ locate /pocean

An ecosystem modelling framework for incorporating climate regime shiftsinto fisheries management

Caihong Fu a,⇑, R. Ian Perry a, Yunne-Jai Shin c,d, Jake Schweigert a, Huizhu Liu b

a Pacific Biological Station, Fisheries and Oceans Canada, 3190 Hammond Bay Road, Nanaimo, BC, Canada V9T 6N7b Vancouver Island University, Nanaimo, BC, Canada V9T 5S5c IRD Institut de Recherche pour le Développement CRH Centre de Recherche Halieutique Méditerranéenne et Tropicale UMR 212 EME, Avenue Jean Monnet, BP 171, 34203 Sètecedex, Franced UCT University of Cape Town, Marine Research Institute (MA-RE) and Department of Zoology, Rondebosch 7701, Cape Town, South Africa

a r t i c l e i n f o

Article history:Available online 23 March 2013

0079-6611/$ - see front matter Crown Copyright � 2http://dx.doi.org/10.1016/j.pocean.2013.03.003

⇑ Corresponding author. Tel.: +1 250 729 8373; faxE-mail address: [email protected] (C. Fu)

a b s t r a c t

Ecosystem-based approaches to fisheries management (EBM) attempt to account for fishing, climate var-iability and species interactions when formulating fisheries management advice. Ecosystem models thatinvestigate the combined effects of ecological processes are vital to support the implementation of EBMby assessing the effectiveness of management strategies in an ecosystem context. In this study, an indi-vidual-based ecosystem model was used to demonstrate how species at different trophic levels and ofdifferent life histories responded to climate regimes and how well different single- or various multi-spe-cies fisheries at different intensities perform in terms of human benefits (yield) and trade-offs (fisheryclosures) as well as their impacts on the ecosystem. In addition, other performance indicators were alsoused to evaluate management strategies. The simulations indicated that under no fishing, each speciesresponded to the regimes differently due to different life history traits and different trophic interactions.Fishing at the level of natural mortality (F = M) produced the highest yields within each fishery, however,an F adjusted for the current productivity conditions (regime) resulted in much fewer fishery closurescompared with F = M, indicating the advantage of implementing a policy of a regime-specific F fromthe stand point of conservation and fishery stability. Furthermore, a regime-specific F strategy generallyresulted in higher yield and fewer fishery closures compared with F = 0.5M. Other performance indicatorsalso pointed to the advantage of using a regime-specific F strategy in terms of the stability of both eco-system and fishery production. As a specific example, fishing the predators of Pacific herring under allmulti-species fisheries scenarios increased the yield of Pacific herring and reduced the number of herringfishery closures. This supports the conclusion that an exploitation strategy which is balanced across alltrophic levels produces better outcomes, as advocated by other researchers.

Crown Copyright � 2013 Published by Elsevier Ltd. All rights reserved.

1. Introduction

Ecosystem-based approaches to fisheries management (EBM)attempt to account for ecosystem processes, including fishingand/or climate variability in conjunction with species interactions,when formulating fisheries management advice (Sissenwine andMurawski, 2004). Numerous studies have illustrated that changesin climate-ocean regimes are associated with fluctuations in theabundance of fish populations and in species composition (e.g.Mantua et al., 1997; McFarlane et al., 2000; Lees et al., 2006). In re-sponse to the call for incorporating climatic-ocean changes into themanagement of marine living resources, a new set of modellingtools, ranging from population, to community, to ecosystems, havebeen developed (e.g., Barange et al., 2010; Stock et al., 2011). Most

013 Published by Elsevier Ltd. All r

: +1 250 756 7053..

studies in this area have focused on single-species-based popula-tion models (e.g., Parma and Deriso, 1990; King and McFarlane,2006a; A’mar et al., 2009; Haltuch and Punt, 2011; Punt, 2011).However, the dynamics of fish populations within an ecosystemare more complicated that what single-species models can depict.In addition, they are not always coherent with physical forcing butrather exhibit different spatial and temporal patterns dependingon the different life history patterns, species interactions and vary-ing fishery exploitation of the species involved (Polovina, 2005;Lees et al., 2006; deYoung et al., 2008; Barange et al., 2010). Lim-ited knowledge of the physical drivers, physical-biological link-ages, and causative agents for the dynamics of ecosystemsremains an impediment to developing management strategieswhich incorporate changing environments (deYoung et al., 2008).In order to assess the effectiveness of management strategies inan ecosystem context, it has been advocated to develop compre-hensive, robust and highly mechanistic end-to-end ecosystemmodels that investigate the combined effects of ecological pro-

ights reserved.

http://dx.doi.org/10.1016/j.pocean.2013.03.003

mailto:[email protected]

http://dx.doi.org/10.1016/j.pocean.2013.03.003

http://www.sciencedirect.com/science/journal/00796611

http://www.elsevier.com/locate/pocean

54 C. Fu et al. / Progress in Oceanography 115 (2013) 53–64

cesses through trophic interactions, environmental disturbances,and fishing (Link, 2002; Pikitch et al., 2004; Polovina, 2005; Pla-gányi, 2007; Barange et al., 2010; Stock et al., 2011).

End-to-end ecosystem models attempt to represent the entireecological system along with the associated abiotic environmentextending to climate impacts (Fulton, 2010). Most common exam-ples of end-to-end models include Ecopath with Ecosim (EwE,Christensen and Walters, 2004), Atlantis (Fulton et al., 2004), andOSMOSE (Shin and Cury, 2001, 2004). In this study, we applied anewer version of the OSMOSE model (Fu et al., 2012) to explore cli-mate regime shifts and fishing under different types of fisheries(single-species fisheries as well as multi-species fisheries) in theStrait of Georgia (SoG) in British Columbia, Canada (Fig. 1). OS-MOSE is a multi-species individual-based model which explicitlyincludes the dynamics of modelled species and their interactions,has greater detail (e.g. spatial overlap with prey) at the species le-vel, and can be used to explore hypotheses regarding the mecha-nisms underlying the sustainability of ecosystem properties.

For western North America and the adjoining North Pacific,Gedalof and Smith (2001) identified 12 climate regimes since1650, with an average duration of 23 years (range 16–83 years).The SoG is a semi-enclosed sea covering an area of approximately6900 km2 (Thomson, 1981). Direct and indirect connections withthe Northeast Pacific (e.g. Li et al., 2013; Mackas et al., 2013; Perry,2013) drive the SoG to undergo similar decadal scale climate re-gime shifts to those that have occurred in the North Pacific (Ged-alof and Smith, 2001). Observations available since 1970 revealedthat the strait has been warming at the same rate as the uppermost150 m of the open ocean (Masson and Cummins, 2007). Climate re-gime shifts affect key physical processes: enrichment, initiation,concentration and retention (Perry et al., 2012), which inducechanges in phytoplankton and zooplankton biomass whose avail-ability to fish as food are fundamental to successful recruitmentin many marine fish populations (Lees et al., 2006). In the SoG,changes of phytoplankton species assemblages have been observedas a result of changes in temperature of only a few degrees (Hob-son and McQuoid, 2001); large copepod zooplankton biomass hasdecreased since 2001 possibly due to climate-induced mismatchesof seasonal timing of phytoplankton (Johannessen and Macdonald,

Fig. 1. Map of the Strait of Georgia, Canada.

2009). The studies by Li et al. (2013) and Mackas et al. (2013) sug-gest that recent (past 20 years) changes in zooplankton composi-tion in the SoG are likely related to climate regime shifts in theNorth Pacific.

In response to climate regime shifts observed in the North Paci-fic (Hare and Mantua, 2000) and also in response to fishery prac-tices (Beamish et al., 2004), the structures of the fish communityin the SoG have changed. The SoG ecosystem supports a numberof species of current and past commercial interest, including Pacificherring (Clupea pallasi), Pacific hake (Merluccius productus), spinydogfish (Squalus acanthias), walleye pollock (Theragra chalcogram-ma), Pacific cod (Gadus macrocephalus) and lingcod (Ophiodon elon-gates), and important non-harvested species such as harbour seal(Phoca vitulina). Typical changes in the structure of the SoG fishcommunity include the steady increase of harbour seals since the1970s (Olesiuk, 1993), the drastic declines of a number of ground-fish species such as Pacific cod and lingcod since the 1970s (Loganet al., 2005), and the decline in size-at-age for several of the pre-dominant fish species including Pacific hake (King and McFarlane,2006b) and herring (Schweigert et al., 2009).

While the impacts of climate regime shifts on fish populationscan manifest themselves in physiology, geographic range and phe-nology at population, species, community and ecosystem levels(Hughes, 2000; McCarty, 2001; Vasseur and McCann, 2005; Che-ung et al., 2011), we examined only regime shift impacts that wereexpressed as changes in the production of phytoplankton and zoo-plankton, and which then affected the dynamics of fish populationsthrough bottom-up forcing. In addition, fishery exploitation is an-other key factor affecting the dynamics of fish species in the SoG(Perry et al., 2012) through top-down forcing. A mechanistic Nutri-ent–Phytoplankton–Zooplankton–Detritus (NPZD) model is cur-rently being developed for this region, although to date theresults are available for only a few years (Angelica Peña, Institueof Ocean Sciences, Sidney, B.C., Canada, personal communication).As a consequence, and due to a lack of time series observations ofphytoplankton biomass (Johannessen and Macdonald, 2009), wefocused our simulations on hypothetical ‘what-if’ scenarios.Through the simulation of various climate regimes and fishery sce-narios, the objectives of this paper are to (1) demonstrate how spe-cies at different trophic levels and with different life historiesrespond to productivity regimes; (2) show how well different sin-gle- or multi-species fisheries at different fishing levels perform interms of human benefits (yield) and trade-offs (fishery closures) aswell as their impacts on the ecosystem.

2. Materials and methods

2.1. The SoG OSMOSE model

To examine how climate change and fishing may affect speciesin the SoG ecosystem that are connected through the food web, wesimulated the dynamics of 10 major taxa (as in Fu et al. (2012))that vary in life history and exploitation intensity and amongwhich there are strong predator–prey interactions. The 10 taxaare euphausiids, Pandalid shrimp (Pandalus spp.), Pacific herring,Pacific hake, walleye pollock, Pacific cod, lingcod, spotted ratfish(Hydrolagus colliei), spiny dogfish, and harbour seal. Euphausiidsserve as major food sources for Pacific herring (Stout et al.,2001), Pacific hake (McFarlane and Beamish, 1985), walleye pol-lock (Shaw and McFarlane, 1986), spiny dogfish (Ware and McFar-lane, 1995), and spotted ratfish (unpublished data from thegroundfish survey in the SoG, Pacific Biological Station, Nanaimo,2012). Pandalid shrimps are important prey for lingcod (Casset al., 1990), Pacific cod and walleye pollock (Yang, 1993). Pacificherring is commercially important (the most dominant species in

C. Fu et al. / Progress in Oceanography 115 (2013) 53–64 55

commercial catches, Fu et al., 2012) and ecologically significant byserving as a major food source for many species including Pacifichake (McFarlane and Beamish, 1985), Pacific cod (Westrheim andFoucher, 1987), lingcod (Cass et al., 1990), spiny dogfish (Wareand McFarlane, 1995), and harbour seal (Olesiuk, 1993). Harbourseal is included in the model as a major predator, consuming an an-nual average of 27,324 tons of prey since 1999 (Peter Olesiuk, Pa-cific Biological Station, Nanaimo, B.C., Canada, personalcommunication). The diet of harbour seal is dominated (75%) byPacific herring and hake (Olesiuk, 1993).

The life cycle of each species included in the OSMOSE model ismodelled, starting with eggs (for fish species) or newborn (formammals), at a time step of 3 months. At the first time step wheneggs or newborn are produced, eggs or newborn are split into anumber of super-individuals called schools, which are distributedspatially according to input distribution maps specified for age 0.Distribution maps specified for each age group were obtainedbased on geo-referenced occurrence data from both commercialfisheries and research surveys. Within each map, the SoG was di-vided into 1300 spatial grid cells each with dimensions of4 � 4 km2. At each time step, OSMOSE simulates the biological pro-cesses for these schools, including growth, predation, starvation,additional mortality D due to other uncounted causes, fishing,reproduction, and spatial movement (including migration). Basicbiological parameter values for each modelled species are givenin Table 1.

The average growth of each school follows the von Bertalanffygrowth model, but the growth for each individual school is ad-justed based on the quantity of food ingested during a time step(Shin and Cury, 2004). Species interactions are fulfilled throughthe process of size-based predation. Predation occurs under theconditions of size suitability (within a minimum and a maximumpredator to prey size ratio) and spatio-temporal co-occurrence be-tween a predator and its prey. As in Fu et al. (2012), the opportu-nistic predation process is superimposed with a diet suitabilitymatrix (Table 2) to allow deliberate selection/exclusion of certainfood items. Except for euphausiids, Pandalid shrimp, spiny dogfish,and harbour seal, all other species are divided into juvenile andmature categories based on pre-defined sizes (Pacific herring:15 cm, Pacific hake and walleye pollock: 30 cm, Pacific cod:60 cm, lingcod: 65 cm, spotted ratfish: 30 cm) for constructingthe diet suitability matrix. Pacific cod and lingcod juveniles are fur-ther separated into two groups because their young-of-the-year(<25 cm) are in different locations from other age classes (e.g., ineel grass beds, Cass et al., 1990). The dividing size of spiny dogfishis set at 60 cm corresponding to 15 years old to reflect the fact thatspiny dogfish younger than 15 form pelagic groups (Beamish andSweeting, 2009). Schools of all modelled species except for harbour

Table 1Basic biological parameter values for each of the species included in the OSMOSE model. Gparameters a and b for the weight-at-length function; relative fecundity u is the number ofeach female; Amat , Amax, Arec are the ages at maturity, maximum longevity, and recruitmentconsideration; biomasses are the initial values of year 2005.

Species Growth Re

L1 (cm) k (year �1) t0 (year) a (g cm�3) b u

Euphausiids 1.84 1.68 �0.20 0.009 2.920 24Pandalid shrimp 2.37 0.92 0.23 0.001 3.069 34Pacific herring 26.3 0.36 �0.03 0.007 3.000 20Pacific hake 44.50 0.46 �0.17 0.001 2.997 10Walleye pollock 44.50 0.92 0.57 0.005 2.4406 93Pacific cod 87.50 0.39 �0.01 0.007 3.096 56Lingcod 90.50 0.22 �1.15 0.000 3.217 26Spotted ratfish 79.0 0.20 0.00 0.007 3.013 2Spiny dogfish 105.52 0.06 �4.28 0.003 3.060 7.Harbour seal 150.22 0.63 �1.52 0.011 3.116 1

seal are subject to predation. In addition, each species is providedwith a certain biomass of unspecified prey to account for the factthat each modelled species consumes prey species other thanthose included explicitly in the model. Phytoplankton (diatoms),small zooplankton (ciliates), and copepods are added to each gridcell to serve as prey. Diatoms and ciliates serve as food for eup-hausiids, and copepods provide food for euphausiids, juveniles ofseveral fish species as well as Pacific herring adults (Table 2). Whileit is possible that hungry prey is behaviourally more exposed andvulnerable to predation, to avoid increased model complexity, weassume that within each grid cell, a predator school uniformly eatsthe available modelled prey species up to its maximum ingestionrate (fixed at 3.5 g of prey per gram body weight of predator peryear, Shin and Cury, 2001). If the modelled prey species are de-pleted before the predator school reaches its maximum ingestionrate, then the predator switches to unspecified prey.

A school is subject to starvation mortality if the food ration istoo low to provide the basic fish maintenance requirements (Shinand Cury, 2004). Fishing mortality is assumed to be knife-edged,i.e., all schools become vulnerable to fishing when they reach theage of recruitment to the fishery (Arec). At the end of each time step,OSMOSE models the reproduction process, which involves addingeggs or newborn to the modelled system. The amount of eggs ornewborn released depends on the season, the age of maturity(AMat in years), the spawning biomass, the sex ratio (set to 1:1 forall species) and the fecundity parameter (/, number of eggsspawned per gram of mature female or number of young producedby each female). Thus, recruitment emerges from the annual sur-vival of eggs and juveniles. At each of the subsequent time steps,the spatial distribution is updated in a random fashion but withinthe grid cells defined by the distribution maps. All the modelledspecies are year round residents in the SoG except for Pacific her-ring, which migrates out of the system in summer and fall. In themodel, Pacific herring are removed from the SoG during these sea-sons and subject to average growth and natural mortality (M) only.

The SoG OSMOSE model was initialized at the biomass levels of2005 (Li et al., 2010; Table 1) for each of the ten modelled speciesas well as diatoms, ciliates, and copepods in the same way as Fuet al. (2012). The SoG OSMOSE model was run for 200 years underno fishing and no trophic interactions (all the values in the dietsuitability matrix were changed to 0), thus each modelled speciesfed solely on its prey. The model was tuned by adjusting theamount of prey and larval mortality for each species until each spe-cies reached the equilibrium condition of 2005. This exercise as-sumes food availability is not a limiting factor and therefore thatthe dynamics of each modelled species depends on larval mortal-ity. Then the species trophic interactions were activated by usingthe diet suitability values shown in Table 2, and the model was

rowth parameter values for the von Bertalanffy curve include L1 , k, and t0, as well aseggs spawned per gram of mature female per year, or number of young produced per

to fishery; M is natural mortality when species trophic interactions are not taken into

production Survivalship Biomass

(eggs g�1) Amat (year) Amax (year) Arec (year) M (year�1) (t km�2)

,469 1 2 1 1.151 20.05 2 4 2 0.576 1.00 3 9 2 0.435 16.5978 3 11 2 0.356 4.3268 3 8 3 0.489 1.1834 3 8 3 0.489 0.070

6 17 5 0.186 0.556(per $) 5 15 3 0.261 2.02 (per $) 35 90 15 0.043 5.710(per $) 4 25 1 0.156 0.287

Tabl

e2

Die

tsu

itab

ility

mat

rix

wit

hpr

edat

ors

onth

efi

rst

colu

mn;

The

Rom

annu

mer

als

assi

gned

toea

chpr

eysp

ecie

sfr

omIt

oX

corr

espo

ndto

the

pred

ator

spec

ies

onth

efi

rst

colu

mn

from

euph

ausi

ids

toha

rbou

rse

alre

spec

tive

ly.F

ish

size

sar

ein

cm.

Pred

ator

sPr

ey

III

III<

15II

I>15

IV<

30IV

>30

V<

40V

>40

VI<

25V

I<60

VI>

60V

II<

25V

II<

65V

II>

65V

III<

30V

III>

30IX

<60

IX>

60X

Cop

epod

s

I:Eu

phau

siid

s0

00

00

00

00

00

00

00

00

00

1II

:Pa

nda

lid

shri

mp

00

00

00

00

00

00

00

00

00

00

III:

Paci

fic

her

rin

g<

151

00

00

00

00

00

00

00

00

00

1II

I:Pa

cifi

ch

erri

ng

>15

10

00

00

00

00

00

00

00

00

01

IV:

Paci

fic

hak

e<

301

01

00

00

00

00

00

00

00

00

1IV

:Pa

cifi

ch

ake

>30

10

11

10

10

00

00

00

00

00

00

V:

Wal

leye

poll

ock

<40

10

11

10

00

00

00

00

00

00

01

V:

Wal

leye

poll

ock

>40

10

11

11

10

00

00

00

00

00

00

VI:

Paci

fic

cod

<25

01

11

00

00

00

00

00

00

00

01

VI:

Paci

fic

cod

<60

01

11

00

11

11

00

00

01

00

00

VI:

Paci

fic

cod

>60

01

11

00

11

11

00

00

01

00

00

VII

:Li

ngc

od<

250

11

10

00

00

00

00

00

00

00

1V

II:

Lin

gcod

<65

01

11

11

10

11

00

00

01

00

00

VII

:Li

ngc

od>

650

11

11

11

11

10

00

00

11

00

0V

III:

Spot

ted

ratfi

sh<

301

10

00

00

00

00

00

00

00

00

1V

III:

Spot

ted

ratfi

sh>

301

10

01

01

00

00

00

00

00

00

0IX

:Sp

iny

dogfi

sh<

601

01

11

01

10

00

00

00

00

00

1IX

:Sp

iny

dogfi

sh>

601

11

11

11

11

10

00

00

11

00

1X

:H

arbo

ur

seal

00

11

11

11

11

10

00

01

11

00


tuned again to reach the same equilibrium condition by reducingthe initial M (now termed as D, such that, M = D + predation mor-tality), and by reducing the amount of prey to account for the factthat some food is provided by the modelled species.

2.2. Productivity regimes and fishing scenarios

Fishing directly affects the dynamics of the targeted species, butalso indirectly affects other species via trophic interactions. Eachscenario was run 20 times for 200 years with only the last 60 yearsbeing used for manipulating climate regimes through planktondynamics as well as fishing strategies. The 60-year period was di-vided into three equal-length productivity regimes. During the 1st20-year productivity regime, plankton (diatom, ciliate, and cope-pod) biomasses were kept at the levels of 2005. During the 2nd20-year productivity regime, the biomasses of these planktongroups were reduced to a quarter of the 2005 levels, and duringthe 3rd regime, plankton biomasses were increased and main-tained at half of the 2005 levels. We selected plankton biomasseslower than 2005 for the last two regimes for the purpose of obtain-ing more conservative simulation results. Six fisheries were simu-lated during the last 60 years including one single-species fishery(herring), three 2-species fisheries (herring along with hake, pol-lock, and dogfish, respectively), one 3-species fishery (her-ring + hake + pollock), and one 4-species fishery(herring + hake + pollock + dogfish). For each fishery, five differentfishing scenarios were simulated following King and McFarlane(2006a) with fishing mortality (F) set at 25% of natural mortality(0.25M), 50% of M (0.5M), M, regime-specific F (F = M for years 1–20, F = 0.25M for years 21–40, and F = 0.5M for years 41–60), or re-gime-specific F with a 3-year lag (F = M for years 1–23, F = 0.25Mfor years 24–43, and F = 0.5M for years 44–60) to account for thefact that regime detection is often delayed (Table 3). Each fishingscenario was subjected to the harvest control rule which requiredthe fishery to be closed for each year and species if the species bio-mass was below a cutoff level of 25% of the initial biomass. Fishingat F = M represents a typical harvest rate proposed by fisheriesmanagement, and regime-specific F scenarios (F = M, F = 0.25Mand F = 0.5M for regimes 1, 2 and 3, respectively) were used to re-flect the relative levels of productivity in each respective regime(i.e. high, low and moderate). By implementing various F levelsin the simulations, ranging from the typical level of F = M to themuch lower level of F = 0.25M, we aim to identify an optimal F levelfor a certain environmental (regime) condition.

2.3. Performance indicators

Similar to King and McFarlane (2006a), each fishery strategy(single- or multi-species fishery at a different fishing level) wasevaluated based on yield as the human benefit and number of fish-ery closures as the trade-off. While other economical indicatorssuch as landed value can be used as performance measures, wechose yield because it provides a more direct reflection of fisheryimpacts on an ecosystem. In addition, we explored properties re-lated to biomass (both species-specific and system biomass), yield,and their ratio particularly in terms of their changes as measures offishery performance and fishery impacts on the ecosystem. Systembiomass (B) is generally observed to be a more stable quantity thanspecies biomass (Bs) and is used as a measure of resource potential,referring to the production capacity and the potential contributionof the ecosystem as an exploitable marine resource (Shin et al.,2010a,b). Relating B to yield, the performance indicator ‘‘1/(yield/biomass)’’ denoted as B/Y, measures the inverse level of exploita-tion or total fishing pressure on an ecosystem with its valuedecreasing under increasing fishing pressure (Shin et al.,2010a,b). We propose here a similar performance indicator ‘‘1-

Table 3Scenarios of fishing mortality F (year�1) on each species. Numbers inside parentheses represent number of years under that level of F.

Scenarios Species

Pacific herring Pacific hake Walleye pollock Spiny dogfish

1 0.109 0 0 02 0.218 0 0 03 0.435 0 0 04 0.435(20), 0.109(20), 0.218(20) 0 0 05 0.435(23), 0.109(20), 0.218(17) 0 0 06 0.109 0.089 0 07 0.218 0.178 0 08 0.435 0.356 0 09 0.435(20), 0.109(20), 0.218(20) 0.356(20), 0.089(20), 0.178(20) 0 0

10 0.435(23), 0.109(20), 0.218(17) 0.356(23), 0.089(20), 0.178(17) 0 011 0.109 0 0.122 012 0.218 0 0.245 013 0.435 0 0.489 014 0.435(20), 0.109(20), 0.218(20) 0 0.489(20), 0.122(20), 0.245(20) 015 0.435(23), 0.109(20), 0.218(17) 0 0.489(23), 0.122(20), 0.245(17) 016 0.109 0 0 0.01117 0.218 0 0 0.02218 0.435 0 0 0.04319 0.435(20), 0.109(20), 0.218(20) 0 0 0.043(20), 0.011(20), 0.022(20)20 0.435(23), 0.109(20), 0.218(17) 0 0 0.043(23), 0.011(20), 0.022(17)21 0.109 0.089 0.122 022 0.218 0.178 0.245 023 0.435 0.356 0.489 024 0.435(20), 0.109(20), 0.218(20) 0.356(20), 0.089(20), 0.178(20) 0.489(20), 0.122(20), 0.245(20) 025 0.435(23), 0.109(20), 0.218(17) 0.356(23), 0.089(20), 0.178(17) 0.489(23), 0.122(20), 0.245(17) 026 0.109 0.089 0.122 0.01127 0.218 0.178 0.245 0.02228 0.435 0.356 0.489 0.04329 0.435(20), 0.109(20), 0.218(20) 0.356(20), 0.089(20), 0.178(20) 0.489(20), 0.122(20), 0.245(20) 0.043(20), 0.011(20), 0.022(20)30 0.435(23), 0.109(20), 0.218(17) 0.356(23), 0.089(20), 0.178(17) 0.489(23), 0.122(20), 0.245(17) 0.043(23), 0.011(20), 0.022(17)


yield/biomass’’ (denoted as ‘‘1-Y/B’’) that can also be considered asa measure of fishing pressure with a highest value of 1 when thereis no fishing pressure.

Changes in average annual biomass from one regime to anotherare the consequences of both climate changes and fishery impacts.Biomass reduction from the 1st regime of higher productivity tothe 2nd regime of lower productivity (Bred) was calculated as:

�B1��B2�B1

, and biomass increase from the 2nd regime of lower to the

3rd regime of higher productivity (Binc) was calculated as:�B3��B2

�B2,

where�B1,�B2, and�B3 are the second-10-year-average-biomass forthe 1st, 2nd, and 3rd regime, respectively. The index Bred captureschanges in biomass during the 2nd regime as a function of the bio-mass during the 1st regime. A lower value of this index indicates areduced decline in biomass, whereas a value close to 1 indicates asubstantial drop in biomass during the 2nd regime. The index Bincmeasures changes in biomass during the 3rd regime as a functionof the biomass during the 2nd regime. A value close to zero indi-cates no change in biomass between the last two regimes and a va-lue close to or greater than one indicates a doubling or greaterincrease in biomass during the 3rd regime. The ratio of Binc to Bred,denoted as Binc-red ratio, reflects the relative recovery of biomassduring the 3rd regime of higher productivity. The Binc-red ratio val-ues from the different fishing mortality levels within each fisheryare used to compare the relative advantage or disadvantage of dif-ferent fishing mortality levels.

In addition to changes in biomass and yield, the variability ofthese quantities can also function as performance indicators. Thedynamics of B can be captured by the indicator: 1/coefficient of var-iation (CV) of B, denoted as 1/CVB. This indicator reflects overall eco-system stability (Shin et al., 2010b) and is assumed to be affected byfishing (Blanchard and Boucher, 2001; Fulton et al., 2004; Hsiehet al., 2006) with a low 1/CVB indicating a low ecosystem stabilityand low resistance to perturbations (Shin et al., 2010b). However,

in some cases high 1/CVB does not imply high ecosystem resistancewhen its high level is caused by populations staying at low levels; incontrast, low 1/CVB can be caused by increasing abundance of anumber of species in the ecosystem which may not reflect low resis-tance to perturbations. Therefore, we propose a companion perfor-mance indicator: 1/CVBs, the inverse of the CV of each speciesbiomass averaged over all species. Furthermore, it is desirable to de-velop an indicator that encompasses not only the variability of bio-mass but also that of yield, since one of the principal goals offisheries management is to achieve low interannual variability infishery production (Punt and Ralston, 2007; Punt, 2011), i.e. low CVof yield (CVY). Therefore, we developed a more comprehensive indi-cator 1=CVBsY ¼ 2=ðCVBsþ CVYÞ, to reflect the stability of both eco-system and fishery productions.

3. Results

3.1. Productivity regimes and fishing scenarios

When there was no fishing for 60 years, each species respondedto simulated climate regime changes in plankton productivity (i.e.,imposed after year 20) differently due to different life history traitsand different trophic interactions (Fig. 2a). The biomass of theshort-lived euphausiids declined quickly in response to the declineof plankton productivity in the 2nd regime. However, with the de-cline of its predators, euphausiid biomass rebounded slightly dur-ing this low plankton regime. As plankton increased to half itsoriginal biomass during the 3rd regime, so did euphausiid biomass.The biomass changes for Pacific herring followed similar patternsexcept that both the decline and recovery following changes inplankton productivity lagged behind euphausiids by seven years.The decline in Pacific hake biomass during the 2nd regime ap-peared to be most dramatic. In contrast, the decline of walleye pol-

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6euphausiidsPandalid shrimpPacific herringPacific hakewalleye pollockPacific codlingcodspotted ratfishspiny dogfishharbour seal

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

0 5 10 15 20 25 30 35 40 45 50 55 60

euphausiidsPandalid shrimpPacific herringPacific hakewalleye pollockPacific codlingcodspotted ratfishspiny dogfishharbour seal

0

0.2

0.4

0.6

0.8

1

1.2

1.4

Time (year)

(b)

Bio

mas

s re

lativ

e to

yea

r 0

(a)

0 5 10 15 20 25 30 35 40 45 50 55 60

Fig. 2. Biomass relative to year 0 under (a) no fishing on any species, (b) fishing Pacific herring, Pacific hake, walleye pollock, and spiny dogfish at natural mortality levels. Aclimate regime shift affecting the productivity of the plankton was imposed at years 20 (decrease) and 40 (increase) – see Section 2 for details.


lock biomass in the 2nd regime was less dramatic and during the3rd regime pollock biomass increased to a level 40% higher thanthe original biomass. Similarly, Pacific cod and lingcod declinedduring the 2nd regime but started to rebound toward the end ofthe 2nd regime, with Pacific cod experiencing an almost linear in-crease during the 3rd regime and lingcod recovering to about itsoriginal level during the last 12 years. Harbour seal biomass de-clined throughout the last two regimes due to reduced food avail-ability. In contrast, the biomass decline of Pandalid shrimp was lessdrastic during the 2nd regime due to less drastic reduction of foodsources (only subject to diatom reduction) and reduction of its pre-dators (walleye pollock, Pacific cod, and lingcod). Spotted ratfishwas least sensitive to the changes of plankton productivity andthe biomass of other species due to its relatively small interactionswith other species included in the model. Similarly, spiny dogfishbiomass slowly decreased during the 2nd regime and slowly recov-ered in the 3rd regime.

Under the 4-species fishery (herring + hake + pollock + dogfish)at F = M, Pacific hake and walleye pollock biomass remained atlow levels after the first 6 years (Fig. 2b). In contrast, Pacific herringrecovered to 65% of its initial biomass during the 1st regime evenwith high fishing mortality (F = M). The decline of Pacific herringpredators (Pacific hake and walleye pollock) under F = M allowedherring to be more robust to fishing pressure. As plankton produc-tivity was reduced during the 2nd regime, Pacific herring biomassfluctuated around 20% of its initial biomass. As plankton productiv-ity increased during the 3rd regime, Pacific herring biomass alsoincreased and was sustained around 25% of its initial biomass.Spiny dogfish biomass decreased steadily during the first two re-gimes but its biomass stabilized during the last 12 years. The de-cline of spiny dogfish resulted in an increase in spotted ratfishbiomass which consequently resulted in a reduction in the biomassof its prey (Pandalid shrimp) as well as other shrimp predators(e.g., lingcod and Pacific cod).

3.2. Yield and fishery closures

When Pacific herring alone was fished (scenarios 1–5, Table 3),yield over the 60 years increased as F increased from 0.25M, 0.5M,to M (Fig. 3a); however, the number of years with fishery closuresalso increased (scenarios 2 and 3, Fig. 3b). Applying regime-specificF (e.g. scenario 4) reduced the total yield compared to fishing atF = M, but the number of fishery closures was also reduced. Re-gime-specific F with 3-year lags (e.g., scenario 5) slightly increasedyield as well as the number of years of fishery closures comparedwith the regime-specific F without a time lag. Fishing Pacific her-ring along with Pacific hake (scenarios 6–10, Table 3) resulted insimilar patterns in yield and number of years with fishery closures.When Pacific hake was also fished, the yield of Pacific herring in-creased yet the number of years with fishery closures decreasedfor all fishing scenarios (6–10) in comparison with scenarios 1–5.Similar results occurred when Pacific herring was fished along withwalleye pollock (scenarios 11–15) or with spiny dogfish (scenarios16–20). Compared with the 2-species fisheries (herring + hake,herring + pollock), the 3-species fishery (herring + hake + pollock)resulted in higher yield but fewer fishery-closure years for eachof the three species. Similarly, the 4-species fishery resulted inhigher yield but fewer fishery-closure years than the 3-speciesfishery for each of the four species, which indicated a positiveinfluence on herring from fishing all higher trophic level species.Compared with constant F = 0.5M, the regime-specific F resultedin higher yields but fewer fishery-closure years when fishing Paci-fic herring alone, fishing herring and spiny dogfish, and with the 4-species fishery.


Under each fishery, the total biomass B during the 1st produc-tivity regime decreased when F increased from 0.25M, 0.5M, to M

0

500

1000

1500

2000

2500

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Tota

l yie

ld (1

000

t)spiny dogfishwalleye pollockPacific hakePacific herring

0

10

20

30

40

50

60

70

80

Tota

l num

ber o

f yea

rsw

ith fi

sher

y cl

osur

es

(a)

(b)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Fig. 3. Total yield (a) and number of years with fishery closures (b) for the identified species under 30 different fishing scenarios (Table 3).


(Fig. 4a). During the 2nd regime, when spiny dogfish was not in-cluded in the fishery, B changed only slightly among the differentF levels; B did not show an inverse relationship with F due toextensive fishery closures at higher F levels. For the 2-species fish-ery of herring + dogfish and the 4-species fishery, B decreased withincreasing F due to limited fishery closures for spiny dogfish. Dur-ing the 3rd regime, B decreased with increasing F from 0.25M to M,but increased when the regime-specific F without and with a timelag was applied, which indicated an advantage of a regime-specificF for population recovery after productivity regimes turned tomore favourable conditions. The indicator B/Y decreased when Fincreased from 0.25M to M during the 1st climate regime(Fig. 4b). When the regime-specific F was applied during the 3rdregime with or without a time lag, B/Y increased relative to theconstant F = M scenarios indicating less fishery impact from the re-gime-specific F. For each fishery, B/Y was much higher during the2nd regime due to more frequent fishery closures during this re-gime, and in all cases, the regime-specific F scenarios resulted inthe highest B/Y. The indicator ‘‘1-Y/B’’ had similar patterns to B/Yfor all fisheries and productivity regimes, but were less sensitiveto small Y values during the 2nd regime, which made the real dif-ferences in this indicator among the different fishing scenarios dur-ing the 1st and 3rd regimes more obvious (Fig. 4c).

From the 1st to the 2nd productivity regime, Bred in all fishingscenarios was higher at lower F levels (Fig. 5), indicating that morebiomass was lost to adverse climate conditions when fisheriesoperated at lower F levels, however, the regime-specific F tendedto result in lower Bred. On the other hand, Binc from the 2nd tothe 3rd productivity regime were also higher at lower F levels,and the regime-specific F tended to result in higher Binc than whenF = M. Operating at higher F undermined biomass increases whenproductivity regimes turned to favourable condition. Applying re-gime-specific F enhanced Binc when the climate turned favourable,even when the regime shift was not detected until 3 years later.Under all six fisheries, fishing at F = M always resulted in the low-est Binc-red ratio, indicating biomass recovery was the least under

F = M; regime-specific F scenarios were always advantageous com-pared with fishing at 0.5M in terms of biomass recovery.

During the 1st productivity regime, for each fishery, 1/CVB washighest when F = 0.25M, however, there was no consistent patternamong the other F levels (Fig. 6a). For the 3- or 4-species fisheries,fishing at 0.5M resulted in the lowest 1/CVB. During the 2nd re-gime, 1/CVB tended to be higher at regime-specific F than atF = M except for the 2-species fishery with Pacific herring and hakewhere hake experienced extensive fishery closures. During the 3rdregime, the single- or 2-species fisheries provided consistent pat-terns with 1/CVB being highest at F = 0.25M and lowest at F = Mindicating higher B variability at higher F. In contrast, the 3- or4-species fisheries with F = M produced rather high 1/CVB whichwas the result of all fished species being stabilized at low bio-masses through fishery closures. The 1/CVBs values for the 1st re-gime were as expected, i.e., 1/CVBs was highest at F = 0.25M andremained equal or was similar when F = M and for regime-specificF with and without lags (Fig. 6b). During the 2nd regime, the 1/CVBs value remained stable across all fishing scenarios. Duringthe 3rd regime, 1/CVBs remained at a consistently low level forthe single- or 2-species fishery; for the 3- or 4-species fishery, 1/CVBs tended to be high when F = M. During the 1st regime, F = Mor the two regime-specific F scenarios resulted in similar 1/CVBsY;but during the 2nd or 3rd regimes, regime-specific F without a timelag always resulted in higher 1/CVBsY than when F = M (Fig. 6c).

4. Discussion

4.1. Ecosystem model and management application

As we move toward developing ecosystem-based approaches tomanaging marine ecosystems, it is important to consider climateregime shifts and how ecosystems might react to these shifts incombination with fishery exploitation or other anthropogenic im-pacts (Lees et al., 2006; deYoung et al., 2008; Barange et al.,

0

50

100

150

200

Biom

ass

(100

0 t)

0

5

10

15

20

25

30

35

B/Y

0

0.2

0.4

0.6

0.8

1

1-Y/

B

0

50

100

150

200

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

0

5

10

15

20

25

30

35

0

0.2

0.4

0.6

0.8

1

(a)

(b)

(c)

Fishing scenario1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Fig. 4. Ecological indicators (a) total biomass (B), (b) total biomass and yield ratio (B/Y), and (c) 1-yield/biomass (1-Y/B) under 30 fishing scenarios (Table 3) averaged over thelast 10 years of each 20-year productivity regime (1st regime: black bar, 2nd regime: grey bar, and 3rd regime: bar with stripes).


2010; Stock et al., 2011). Comprehensive mechanistic end-to-endecosystem models such as OSMOSE and Atlantis have been appliedto integrate the combined effects of climate dynamics, speciesinteractions, and human activities on living marine resources(e.g., Fulton et al., 2004; Travers et al., 2009). These end-to-endmodels provide useful indications of potential unanticipated con-sequences of management actions, system responses to variousnatural and anthropogenic drivers, and trade-offs between the var-ious ecological, economic and social objectives within ecosystem-based management. However, it remains inappropriate to usethese models to provide tactical management decisions, such assetting catch quotas, because these models are still in development(Fulton, 2010; Fulton et al., 2011; Stock et al., 2011). Nevertheless,this end-to-end modelling approach is needed to avoid manage-ment actions that may appear sound but have undesirable out-comes within a complex ecosystem (Fulton, 2010).

Along the same line, we do not recommend our simulation re-sults for direct management application, but rather for examiningsystem-level ‘what-if’ fisheries management and climate impactscenarios. By imposing three different productivity regimes, ourOSMOSE model revealed how species at different trophic levelsand with different life histories may respond to these productivityregimes changes under different fishing scenarios. In addition, themodel demonstrated the different performances of different fish-

ing scenarios during different productivity regimes. These simula-tion results form the bases for further evaluations of alternativehypotheses on climate changes, ecosystem structures, and particu-larly management strategies in support of ecosystem-based adap-tive management.

4.2. Yield and fishery closures

The potential rise of water temperature during the next fewdecades predicted by the Intergovernmental Panel on ClimateChange are likely to have an impact on total phytoplankton andzooplankton production and plankton community composition(e.g. Richardson and Schoeman, 2004) and subsequent fish produc-tivity (Ware and Thomson, 2005). Incorporating productivity re-gimes into our ecosystem simulation model, we derivedconclusions similar to King and McFarlane (2006a) that regime-specific fishing mortalities that reflected the relative levels of pro-ductivity by regime (i.e., high, low, moderate) produced higherbenefits (higher yield) and lower trade-offs (fewer fishery closures)compared to the traditional approach of a constant F of 0.5M forthe one-species fishery (herring), a 2-species fishery (her-ring + dogfish), and the 4-species fishery. For the two 2-speciesfisheries (herring + hake, herring + pollock), the regime-specific Fresulted in higher Pacific herring yield and fewer number of years

0

0.2

0.4

0.6

0.8

1

1.2

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

BredBincBinc-red ratio

Fig. 5. Proportion of 10-year-average-biomass reduction from 1st to 2nd productivity regime (Bred), proportion of 10-year-average-biomass increase from 2nd to 3rdproductivity regime (Binc), and the ratio of Binc to Bred under 30 fishing scenarios (Table 3). 1st regime: black bar, 2nd regime: grey bar, and 3rd regime: bar with stripes.

020

4060

80100

120140

160180

1/C

VBsY

050

100150200250300350400450

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

1/C

VB

0

20

40

60

80

100

120

140

160

1/C

VBs

1/C

VB

Fishing scenario

(a)

(b)

(c)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 300

20

4060

80100

120140

160180

1/C

VBsY

050

100150200250300350400450

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

1/C

VB

0

20

40

60

80

100

120

140

160

1/C

VBs

1/C

VB

Fishing scenario

(a)

(b)

(c)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Fig. 6. Ecological indicators (a) 1/CV of total biomass (1/CVB), (b) the inverse of CV of species biomass averaged over all species (1/CVBs), and (c) 1 over the average of CV ofspecies biomass and CV of yield (1/CVBsY) under 30 fishing scenarios (Table 3) calculated over the last 10 years of each 20-year productivity regime (1st regime: black bar, 2ndregime: grey bar, and 3rd regime: bar with stripes).



with herring fishery closures than fishing at 0.5M. This conclusionfavours efforts to detect changes in productivity related to regime-shifts and to implement regime-specific F levels. Although the timelag to detect regime shifts and implement regime-specific F maycompromise the performance of regime-specific fishing strategies,the benefits of implementing regime-specific fishing effort areobvious.

What was not addressed by King and McFarlane (2006a) werethe species interactions and the nature of a multi-species fishery.Within an ecosystem with various species interacting throughpredator–prey relationships, the response of one population tothe changes in another population can be nonlinear and non-intu-itive (Gamble and Link, 2009). By fishing the predators of Pacificherring, i.e., Pacific hake, walleye pollock, and spiny dogfish under2-species fisheries, the fishery yield of Pacific herring was in-creased and the number of years with herring fishery closureswas reduced under all fishing levels. In comparison with the 2-spe-cies fisheries, the 3-species fishery (herring + hake + pollock)achieved higher yield but fewer number of years with fishery clo-sures, for each of the three species. In the same way, the 4-speciesfishery resulted in higher yield but fewer fishery-closure yearsthan the 3-species fishery for each of the four species. These find-ings support the conclusion that more conservative fisheries man-agement might involve balanced exploitation across all trophiclevels (e.g., Bundy et al., 2005; Zhou et al., 2010; Garcia et al.,2011). Therefore, it is necessary to consider which species are tobe fished or not to be fished in order to achieve the most economicbenefits yet maintain ecosystem health and function in the contextof ecosystem-based fishery management.


To support the implementation of ecosystem-based fisherymanagement, it is important to develop and monitor ecologicalindicators to assess ecosystem status and the effectiveness of man-agement strategies (Shin et al., 2010a,b). Travers et al. (2006)showed that indicators did not always vary as theoretically pre-dicted because of indirect effects of fishing on the different compo-nents of the ecosystem and thus the fishing configuration (speciestargeted, fishing intensity) and the type of community studiedshould be carefully specified. Our simulations of 30 fishing scenar-ios under three productivity regimes indicated that the way indica-tors vary depends not only on the fishing configuration but also onthe different environmental conditions (high, low and mediumplankton productivity in the system). Overall, the indicators re-sponded to the 30 different fishing scenarios in a consistent andpredictable manner during the 1st productivity regime with lowerF resulting in higher system biomass B and fish community stabil-ity captured by the inverse of biomass coefficient of variation (1/CVB and 1/CVBs) as well as higher 1/CVBsY, the indicator that con-siders variability of fishery yield along with that of B. In contrast,the indicators 1/CVB and 1/CVBs had strikingly different patternsduring the 3rd regime under the 3- or 4-species fishery (Fig. 6),i.e., fishing at F = M tended to result in higher indicator valuesdue to the fact that all fished species stabilized at low B throughfishery closures. Nevertheless, for 1/CVBsY, regime-specific F with-out lag produced the highest indicator values (Fig. 6c), which im-plied that 1/CVBsY was a more comprehensive indicator than 1/CVB and 1/CVBs for evaluating fishery management strategies.

4.4. Caveats and conclusions

In this study, we modelled climate regimes as changes in theproduction of phytoplankton and zooplankton, while ignoringother potential changes such as physiology, geographic range andphenology at different scales (e.g., population, community, ecosys-

tem). We acknowledge that while this sort of simplistic expressionof possible climate change effects is common in modelling studies,other climate change effects might produce conclusions that differfrom the ones portrayed here and that future studies should con-sider analyses of sensitivity to the exclusion of other mechanisms.Furthermore, although the fixed regime duration of 20 years iswithin the range that Gedalof and Smith (2001) identified, otherlengths of duration may generate different impacts especially forthe longer-lived species such as spiny dogfish, lingcod, and harbourseals. Sensitivity analysis with different lengths of duration will beworth considering in future studies.

Match and/or mismatch between such processes as the initia-tion of the spring phytoplankton bloom, the timing of the peakzooplankton biomass in the upper layers during spring, and thespawning of fish species is critical for determining the dynamicsof fish populations in the SoG (Johannessen and Macdonald,2009; Perry et al., 2012; Schweigert et al., 2013). In this study,the biomass of phytoplankton and zooplankton was assumed tobe distributed homogenously throughout all the grid cells and allthe time steps, an assumption which is much simpler than the real-ity. Although this assumption affected the ability to model the ef-fects of spatial and temporal dynamics in plankton productivity,we believe it does not affect the conclusions about how each spe-cies would respond to the overall changes of plankton productivity.More simulation modelling that incorporates spatial and seasonalpatterns of plankton productivity derived from the lower trophiclevel NPZD model is currently being undertaken, which would pro-vide a more realistic spatial and temporal resolution to the model.However, the lack of time series of observed (Johannessen andMacdonald, 2009) and simulated plankton biomass from the cur-rent NPZD model for the SoG will continue to confine us to theexercises of exploring hypothetical ‘what-if’ scenarios concerningthe effects of lower trophic level productivity.

Pacific herring is a dominant species in the SoG ecosystemduring their residency in the strait and heavily impacted by pro-ductivity regimes. While assuming average growth and naturalmortality during its migration out of the system in summerand fall, the model can potentially undermine the impacts of re-gime variability. A more finely tuned model in the future wouldinclude productivity regimes for modelling these two biologicalprocesses when Pacific herring is outside the studied ecosystem.In addition, Pacific herring can be subject to predation by thevarious salmon species, although the degree of this predationimpact is dependent on the duration of salmon residency inthe strait, which may have varied through time (Preikshotet al., 2012). Future modelling efforts would be directed to in-clude more taxa such as salmon.

Through the simulations, we concluded: (1) under no fishing,each species responded to the regime changes in plankton bio-masses differently due to different life history traits and differenttrophic interactions; (2) for each of the six fisheries, when F in-creased from 0.25M to M, yield increased as did the number ofyears with fishery closures. In general, regime-specific F was morefavourable than fishing at F = 0.5M in terms of higher yield and alower number of fishery-closure years; (3) under all six fisheries,fishing at F = M always resulted in the highest Bred from 1st to2nd regime and lowest Binc from 2nd to 3rd regime, and regime-specific F were always advantageous compared with F = 0.5M,achieving the highest Binc-red ratio; (4) compared with 1/CVBand 1/CVBs, the indicator 1/CVBsY was more comprehensive andtherefore better suited to evaluating fishery management strate-gies. Overall, while this model was parameterized for the SoG eco-system, the conclusions drawn here are relevant to similarecosystems and can help to better understand how an ecosystemmight work if there were fishing for the various species under dif-ferent climate regimes.


Acknowledgements

This publication is the result of research sponsored by Fisheries& Oceans Canada under the Ecosystem Research Initiatives. Wethank two anonymous reviews and guest editor Dr. Dave Mackasfor their thoughtful and constructive comments. We also wouldlike to thank the many individuals who contributed their knowl-edge and data to this research, especially Sandy McFarlane, PeterOlesiuk, Norm Olsen, Kate Rutherford, Lingbo Li, and Morgane Tra-vers. Special acknowledgement goes to Hai Nguyen who producedspecies distribution maps.

References

A’mar, T.Z., Punt, A.E., Dorn, M.W., 2009. The impact of regime shifts on theperformance of management strategies for the Gulf of Alaska walleye pollock(Theragra chalcogramma) fishery. Canadian Journal of Fisheries and AquaticSciences 66, 2222–2242.

Barange, M., Cheung, W.W.L., Merino, G., Perry, R.I., 2010. Modelling the potentialimpacts of climate change and human activities on the sustainability of marineresources. Current Opinion in Environmental Sustainability 2, 1–8.

Beamish, R.J., Sweeting, R.M., 2009. Spiny dogfish in the pelagic waters of the Straitof Georgia and Puget Sound. In: Gallucci, V., McFarlane, G.A., Bargmann, G.C.(Eds.), Biology and Management of Dogfish Sharks. American Fisheries Society,Bethesda, Maryland, pp. 101–119.

Beamish, R.J., Benson, A.J., Sweeting, R.M., Neville, C.M., 2004. Regimes and thehistory of the major fisheries off Canada’s west coast. Progress in Oceanography60, 355–385.

Blanchard, F., Boucher, J., 2001. Temporal variability of total biomass in harvestedcommunities of demersal fishes. Fisheries Research 49, 283–293.

Bundy, A., Fanning, P., Zwanenburg, K.C.T., 2005. Balancing exploitation andconservation of the eastern Scotian Shelf ecosystem: application of a 4Decosystem exploitation index. ICES Journal of Marine Science 62, 503–510.

Cass, A.J., Beamish, R.J., McFarlane, G.A., 1990. Lingcod (Ophiodon elongatus).Canadian Special Publication of Fisheries and Aquatic Sciences 109, Fisheriesand Oceans, Ottawa.

Cheung, W.W.L., Dunne, J., Sarmiento, J., Pauly, D., 2011. Integrating eco-physiologyand plankton dynamics into projected changes in maximum fisheries catchpotential under climate change. ICES Journal of Marine Science 68, 1008–1018.

Christensen, V., Walters, C., 2004. Ecopath with Ecosim: methods, capabilities andlimitations. Ecological Modelling 172, 109–139.

deYoung, B., Barange, M., Beaugrand, G., Harris, R., Perry, R.I., Scheffer, M., Werner,F., 2008. Regime shifts in marine ecosystems: detection, prediction andmanagement. Trends in Ecology and Evolution 23, 402–409.

Fu, C., Shin, Y.-J., Perry, R.I., King, J., Liu, H., 2012. Exploring climate and fishingimpacts in an ecosystem framework. In: Kruse, G.H., Browman, H.I., Cochrane,K.L., Evans, D., Jamieson, G.S., Livingston, P.A., Woodby, D., Zhang, C.I. (Eds.),Global Progress in Ecosystem-based Fisheries Management. Alaska Sea Grant,University of Alaska Fairbanks, Fairbanks, pp. 65–85.

Fulton, E.A., 2010. Approaches to end-to-end ecosystem models. Journal of MarineSystems 81, 171–183.

Fulton, E.A., Fuller, M., Smith, A.D.M., Punt, A.E., 2004. Ecological indicators of theecosystem effects of fishing: final report. AFMA Final Research Report R99(1546), 239.

Fulton, E.A., Link, J., Kaplan, I.C., Johnson, P., Savina-Rolland, M., Ainsworth, C.,Horne, P., Gorton, R., Gamble, R.J., Smith, T., Smith, D., 2011. Lessons inmodelling and management of marine ecosystems: the Atlantis experience. Fishand Fisheries 2, 171–188.

Gamble, R.J., Link, J.S., 2009. Analyzing the tradeoffs among ecological and fishingeffects on an example fish community: a multispecies (fisheries) productionmodel. Ecololgical Modelling 220, 2570–2582.

Garcia, S.M., Kolding, J., Rice, J., Rochet, M.-J., Zhou, S., Arimoto, T., Beyer, J., Borges,L., Bundy, A., Dunn, D., Graham, N., Hall, M., Heino, M., Law, R., Makino, M.,Rijnsdorp, A.D., Simard, F., Smith, A.D.M., Symons, D., 2011. Selective Fishingand Balanced Harvest in Relation to Fisheries and Ecosystem Sustainability.Report of a Scientific Workshop Organized by the IUCN-CEM Fisheries ExpertGroup (FEG) and the European Bureau for Conservation and Development(EBCD) in Nagoya (Japan), 14–16 October 2010. Gland, Switzerland andBrussels, Belgium: IUCN and EBCD, pp. iv + 33.

Gedalof, Z., Smith, D.J., 2001. Interdecadal climate variability and regime-scale shiftsin Pacific North America. Geophysical Research Letters 28, 1515–1518.

Haltuch, M.A., Punt, A.E., 2011. The promises and pitfalls of including decadal-scaleclimate forcing of recruitment in groundfish stock assessment. Canadian Journalof Fisheries and Aquatic Sciences 68 (5), 912–926.

Hare, S.R., Mantua, N.J., 2000. Empirical evidence for North Pacific regime shifts in1977 and 1989. Progress in Oceanography 47, 103–145.

Hobson, L.A., McQuoid, M.R., 2001. Pelagic diatom assemblages are good indicatorsof mixed water intrusions into Saanich Inlet, a stratified fjord in VancouverIsland. Marine Geology 174, 125–138.

Hsieh, C.-H., Reiss, C.S., Hunter, J.R., Beddington, J.R., May, R.M., Sugihara, G., 2006.Fishing elevates variability in the abundance of exploited species. Nature 443,859–862.

Hughes, L., 2000. Biological consequences of global warming: is the signal alreadyapparent? Trends in Ecology and. Evolution 15, 56–61.

Johannessen, S., Macdonald, R., 2009. Effects of local and global change on an inlandsea: the Strait of Georgia, British Columbia, Canada. Climate Research 40, 1–21.

King, J.R., McFarlane, G.A., 2006a. A framework for incorporating climate regimeshifts into the management of marine resources. Fisheries Management andEcology 13, 93–102.

King, J.R., McFarlane, G.A., 2006b. Shift in the size-at-age of the Strait of Georgiapopulation of Pacific hake (Merluccius productus). CCOFI Report 7, 111–118.

Lees, K., Pitois, S., Scott, C., Frid, C., Mackinson, S., 2006. Characterizing regime shiftsin the marine environment. Fish and Fisheries 7, 104–127.

Li, L., Ainsworth, C., Pitcher, T., 2010. Presence of harbour seals (Phoca vitulina) mayincrease exploitable fish biomass in the Strait of Georgia. Progress inOceanography 87, 235–241.

Li, L., Mackas, D., Hunt, B., Schweigert, J., Pakhomov, E., Perry, R.I., Galbraith, M.,2013. Zooplankton communities in the Strait of Georgia, British Columbia, tracklarge-scale climate forcing over the Pacific ocean. Progress in Oceanography115, 90–102.

Link, J.S., 2002. Ecological considerations in fisheries management: when does itmatter? Fisheries 27 (4), 10–17.

Logan, G., de la Mare, W., King, J., Haggarty, D., 2005. Management Framework forStrait of Georgia Lingcod. Can. Sci. Ad. Sec. Res. Doc. 2005/048.

Mackas, D.L., Galbraith, M., Faust, D., Masson, D., Shaw, W., Romaine, S., Trudel, M.,Dower, J., Campbell, R., Sastri, A., El-Sabaawi, R., 2013. Zooplankton time seriesfrom the Strait of Georgia: results from year-round sampling at deep waterlocations, 1990–2010. Progress in Oceanography 115, 129–159.

Mantua, N.J., Hare, S.R., Zhang, Y., Wallace, J.M., Francis, R.C., 1997. A Pacificinterdecadal climate oscillation with impacts on salmon production. Bulletin ofthe American Meteorological Society 78, 1069–1079.

Masson, D., Cummins, P., 2007. Temperature trends and interannual variability inthe Strait of Georgia, British Columbia. Continental Shelf Research 27, 634–649.

McCarty, J.P., 2001. Ecological consequences of recent climate change. ConservationBiology 15, 320–331.

McFarlane, G.A., Beamish, R.J., 1985. Biology and Fishery of Pacific hake (Merlucciusproductus) in the Strait of Georgia. Marine Fisheries Review 47 (2), 23–34.

McFarlane, G.A., King, J.R., Beamish, R.J., 2000. Have there been recent changes inclimate? Ask the fish. Progress in Oceanography 47, 147–169.

Olesiuk, P.F., 1993. Annual prey consumption by harbor seals (Phoca vitulina) in theStrait of Georgia, British Columbia. Fishery Bulletin 91, 491–515.

Parma, A.M., Deriso, R.B., 1990. Experimental harvesting of cyclic stocks in the faceof alternative recruitment hypotheses. Canadian Journal of Fisheries andAquatic Sciences 47, 595–610.

Perry, R.I., 2013. An integrated analysis of the marine social-ecological system of theStrait of Georgia, Canada, over the past four decades, and development of aregime shift index. Progress in Oceanography 115, 14–27.

Perry, R.I., Masson, D., Preikshot, D., Araujo, H.A., 2012. A Synthesis of the Outcomesfrom the Strait of Georgia Ecosystem Research Initiative, and Their Ecosystem-based Management and Policy Implications. DFO Can. Advis. Sec. Res. Doc.2012/P59, vi + xx p.

Pikitch, E.K., Santora, C., Babcock, E.A., Bakun, A., Bonfil, R., Conover, D.O., Dayton, P.,Doukakis, P., Fluharty, D., Heneman, B., Houde, E.D., Link, J., Livingston, P.,Mangel, M., McAllister, M., Pope, J., Sainsbury, K.J., 2004. Ecosystem-basedfishery management. Science 305, 346–347.

Plagányi, E.E., 2007. Models for an Ecosystem Approach to Fisheries. FAO FisheriesTechnical Paper 477, Rome, p. 108.

Polovina, J.J., 2005. Climate variation, regime shifts, and implications for sustainablefisheries. Bulletin of Marine Science 76, 233–244.

Preikshot, D., Neville, C.M., Beamish, R.J., 2012. Data and Parameters Used in a Straitof Georgia Ecosystem Model. Canadian Technical Report of Fisheries andAquatic Sciences 3005, viii + 58p.

Punt, A.E., 2011. The impact of climate change on the performance of rebuildingstrategies for overfished groundfish species on the U.S. west coast. Fish andFisheries 109 (2–3), 320–329.

Punt, A.E., Ralston, S.V., 2007. Management strategy evaluation of rebuildingrevision rules for overfished rockfish species. In: Heifetz, J., DiCosimo, J.,Gharrett, A.J., Love, M.S., O’Connell, V.M., Stankey, R.D. (Eds.), Biology,Assessment and Management of North Pacific Rockfishes. Alaska Sea GrantCollege Program. University of Alaska Fairbanks, Fairbanks, USA, pp. 329–351.

Richardson, A.J., Schoeman, D.S., 2004. Climate impact on plankton ecosystems inthe Northeast Atlantic. Science 305, 1609–1612.

Schweigert, J.F., Thompson, M., Fort, C., Hay, D.E., Therriault, T.W., Brown. L.N., 2013.Factors linking Pacific herring (Clupea pallasi) productivity and the springplankton bloom in the Strait of Georgia, British Columbia, Canada. Progress inOceanography 115, 103–110.

Schweigert J., Christensen, L.B., Haist, V., 2009. Stock Assessment for BritishColumbia Herring in 2008 and Forecasts of the Potential Catch in 2009. DFOCan. Sci. Advis. Sec. Res. Doc. 2009/019. pp. iv + 61.

Shaw, W., McFarlane, G.A., 1986. Biology, distribution and abundance of walleyepollock (Theragra chalcogramma) off the west coast of Canada. InternationalNorth Pacific Fisheries Commission Bulletin 45, 262–283.

Shin, Y.-J., Cury, P., 2001. Exploring fish community dynamics through size-dependent trophic interactions using a spatialized individual-based model.Aquatic Living Resoures 14, 65–80.

Shin, Y.-J., Cury, P., 2004. Using an individual-based model of fish assemblages tostudy the response of size spectra to changes in fishing. Canadian Journal ofFisheries and Aquatic Sciences 61, 414–431.


Shin, Y.-J., Bundy, A., Shannon, L.J., Simier, M., Coll, M., Fulton, E.A., Link, J.S., Jouffre,D., Ojaveer, H., Mackinson, S., Heymans, J.J., Raid, T., 2010a. Can simple be usefuland reliable? Using ecological indicators to represent and compare the states ofmarine ecosystems. ICES Journal of Marine Science 67, 717–731.

Shin, Y.-J., Shannon, L.J., Bundy, A., Coll, M., Aydin, K., Bez, N., Blanchard, J.L., Borges,M.F., Diallo, I., Diaz, E., Heymans, J.J., Hill, L., Johannesen, E., Jouffre, D., Kifani, S.,Labrosse, P., Link, J.S., Mackinson, S., Masski, H., Möllmann, C., Neira, S., Ojaveer,H., Mohammed Abdallahi, K., Perry, I., Thiao, D., Yemane, D., Cury, P.M., 2010b.Using indicators for evaluating, comparing, and communicating the ecologicalstatus of exploited marine ecosystems. 2. Setting the scene. ICES Journal ofMarine Science 67, 692–716.

Sissenwine, M., Murawski, S., 2004. Moving beyond ‘intelligent thinking’: advancingan ecosystem approach to fisheries. In: Perspectives on Ecosystem-basedApproaches to the Management of Marine Resources. Marine Ecology ProgressSeries, vol. 274, pp. 269–303.

Stock, C.A., Alexander, M.A., Bond, N.A., Brander, K.M., Cheung, W.W.L., Curchitser,E.N., Delworth, T.L., Dunne, J.P., Griffies, S.M., Haltuch, M.A., Hare, J.A., Hollowed,A.B., Lehodey, P., Levin, S.A., Link, J.S., Rose, K.A., Rykaczewski, R.R., Sarmiento,J.L., Stouffer, R.J., Schwing, F.B., Vecchi, G.A., Werner, F.E., 2011. On the use ofIPCC-class models to assess the impact of climate on living marine resources.Progress in Oceanography 88 (1–4), 1–27.

Stout, H.A., Gustafson, R.G., Lenarz, W.H., McCain, B.B., VanDoornik, D.M., Builder,T.L., Methot, R.D., 2001. Status Review of Pacific Herring in Puget Sound,Washington. NOAA Tech. Memo. NMFS-NWFSC-45, p. 175.

Thomson, R.E., 1981. Oceanography of the British Columbia Coast. Canadian SpecialPublication of Fisheries and Aquatic Sciences 56, Fisheries and Oceans, Ottawa.

Travers, M., Shin, Y.-J., Shannon, L., Cury, P., 2006. Simulating and testing thesensitivity of ecosystem-based indicators to fishing in the southern Benguelaecosystem. Canadian Journal of Fisheries and Aquatic Sciences 63, 943–956.

Travers, M., Shin, Y.-J., Jennings, S., Machu, E., Huggett, J.A., Field, J.G., Cury, P.M.,2009. Two-way coupling versus one-way forcing of plankton and fish models topredict ecosystem changes in the Benguela. Ecological Modelling 220, 3089–3099.

Vasseur, D.A., McCann, K.S., 2005. A mechanistic approach for modelingtemperature-dependent consumer-resource dynamics. American Naturalist166, 184–198.

Ware, D.M., McFarlane, G.A., 1995. Climate-induced changes in Pacific hake(Merluccius productus) abundance and pelagic community interactions in theVancouver Island upwelling system. In: Beamish, R.J. (Ed.), Climate Change andNorthern Fish Populations. Canadian Special Publication of Fishereis andAquatic Sciences 121. Fisheries and Oceans, Ottawa, pp. 509–521.

Ware, D.M., Thomson, R.E., 2005. Bottom-up ecosystem trophic dynamicsdetermine fish production in the Northeast Pacific. Science 308, 1280–1284.

Westrheim, S.J., Foucher, R.P., 1987. Stock Assessment of Pacific Cod, (Gadusmacrocephalus) in Georgia and Juan de Fuca Straits. Can. MS Rep. Fish. Aquat.Sci. 1905, p. 84.

Yang, M.-S., 1993. Food Habits of the Commercially Important Groundfishes in theGulf of Alaska in 1990. NOAA Tech. Memo. NMFS-AFSC-22, p. 150.

Zhou, S., Smith, A.D.M., Punt, A.E., Richardson, A.J., Gibbs, M., Fulton, E.A., Pasco, S.,Bulman, C., Bayliss, P., Sainsbury, K., 2010. Ecosystem-based fisheriesmanagement requires a change to the selective fishing philosophy.Proceedings of the National Academy of Sciences of the United States ofAmerica 107, 9485–9489.

Documents

An ecosystem modelling framework for incorporating climate regime shifts into fisheries management