13
Fisheries Research 94 (2008) 238–250 Contents lists available at ScienceDirect Fisheries Research journal homepage: www.elsevier.com/locate/fishres Beyond biological performance measures in management strategy evaluation: Bringing in economics and the effects of trawling on the benthos C.M. Dichmont a,, A. Deng a , A.E. Punt b,c , N. Ellis a , W.N. Venables d , T. Kompas e,f , Y. Ye a , S. Zhou a , J. Bishop a a CSIRO Marine and Atmospheric Research, 233 Middle Street, Cleveland, Qld, Australia b CSIRO Marine and Atmospheric Research, GPO Box 1538, Hobart, Tas, Australia c School of Aquatic and Fishery Sciences, Box 355020, University of Washington, Seattle, WA 98195-5020, USA d CSIRO Mathematical and Information Statistics, 233 Middle Street, Cleveland, Qld, Australia e ABARE, GPO Box 1563, Canberra, Australia f Australian National University, Crawford School of Economics and Government, ANU, Canberra, Australia article info Article history: Received 20 November 2007 Received in revised form 30 April 2008 Accepted 10 May 2008 Keywords: Australia Benthic impacts Management strategy evaluation Maximum Economic Yield Technical interaction Bio-economic assessment Prawns abstract The performance of management strategies for a prawn fishery in northern Australia is evaluated using the management strategy evaluation (MSE) approach. The operating model on which the analyses are based includes population dynamics models for four prawn species and five stocks of each species, an effort allocation model and a benthic impacts model. Management is implemented through controls on the fishing effort that targets the two main target species (Penaeus semisulcatus and Penaeus esculentus) and the technical interactions between the two species are also taken into account. The total effort set by management is distributed to regions and grid cells in each region through effort allocation models. The performance measures used in this study cover conservation of the target species, economic returns and the impact of fishing on benthic communities. Two classes of management strategy are evaluated. One class seeks to move stocks towards the target spawning stock size which is a pre-specified fraction of the spawning stock size at which Maximum Sustainable Yield (MSY) is achieved using a threshold control rule, while the other class selects time-trajectories of future effort to maximize discounted profit. Management strategies that control effort levels to maximize the total profit over the long-term outperform those which aim to move the spawning stock size toward S MSY in terms of most performance measures. For example, even when the target stock size for the MSY-based management strategy is selected to be the same as that which maximizes profits, selecting effort to maximize profits leads to lower variability in catches and profits. This study illustrates how broader ecosystem considerations can be included in MSE analyses without the need for the development and implementation of full ecosystem models and hence provides a “middle road” between single-species MSEs and full ecosystem MSEs. Crown Copyright © 2008 Published by Elsevier B.V. All rights reserved. 1. Introduction A management strategy is a fully specified set of rules for deter- mining tactical management regulations, and generally includes specifications for a monitoring system, an assessment proce- dure, and a decision rule. The management strategy evaluation (MSE) approach allows the trade-offs among the (pre-agreed and pre-specified) management objectives achieved by different man- agement strategies to be evaluated, taking account of various sources of uncertainty (e.g. uncertainty in the assessment, imple- mentation error, etc.), and with the aim of identifying management Corresponding author. Tel.: +61 7 3827 7219; fax: +61 7 3826 7222. E-mail address: [email protected] (C.M. Dichmont). strategies that are robust to uncertainty and achieve desired trade- offs among the management objectives. MSE has been applied to several single and multispecies fish- eries (Punt, 1992; De la Mare, 1996; Butterworth et al., 1997; Punt and Smith, 1999; Smith et al., 1999; Punt et al., 2002; Dichmont et al., 2006a) and to ecosystems (Sainsbury et al., 2000; Fulton et al., 2007). However, most MSE applications have focused on yield or stock status objectives for one or several species. Few MSE studies have explicitly considered economics and/or incor- porated performance measures related to environmental effects not directly linked to the fished populations. Furthermore, MSEs taking an explicit ecosystem approach (e.g. Fulton et al., 2007) are uncommon and in many cases impractical owing to lack of data and resources. This study shows that it is possible to undertake an MSE which considers several high-level objectives and multiple 0165-7836/$ – see front matter. Crown Copyright © 2008 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.fishres.2008.05.007

Beyond biological performance measures in management strategy evaluation: Bringing in economics and the effects of trawling on the benthos

  • Upload
    csiro

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

Fisheries Research 94 (2008) 238–250

Contents lists available at ScienceDirect

Fisheries Research

journa l homepage: www.e lsev ier .com/ locate / f i shres

Beyond biological performance measures in management strategy evaluation:Bringing in economics and the effects of trawling on the benthos

C.M. Dichmonta,∗, A. Denga, A.E. Puntb,c, N. Ellis a, W.N. Venablesd, T. Kompase,f, Y. Yea, S. Zhoua, J. Bishopa

a CSIRO Marine and Atmospheric Research, 233 Middle Street, Cleveland, Qld, Australiab CSIRO Marine and Atmospheric Research, GPO Box 1538, Hobart, Tas, Australiac School of Aquatic and Fishery Sciences, Box 355020, University of Washington, Seattle, WA 98195-5020, USAd CSIRO Mathematical and Information Statistics, 233 Middle Street, Cleveland, Qld, Australiae ABARE, GPO Box 1563, Canberra, Australiaf Australian National University, Crawford School of Economics and Government, ANU, Canberra, Australia

a r t i c l e i n f o

Article history:Received 20 November 2007Received in revised form 30 April 2008Accepted 10 May 2008

Keywords:AustraliaBenthic impactsManagement strategy evaluationMaximum Economic YieldTechnical interactionBio-economic assessmentPrawns

a b s t r a c t

The performance of management strategies for a prawn fishery in northern Australia is evaluated usingthe management strategy evaluation (MSE) approach. The operating model on which the analyses arebased includes population dynamics models for four prawn species and five stocks of each species, aneffort allocation model and a benthic impacts model. Management is implemented through controls onthe fishing effort that targets the two main target species (Penaeus semisulcatus and Penaeus esculentus)and the technical interactions between the two species are also taken into account. The total effort set bymanagement is distributed to regions and grid cells in each region through effort allocation models. Theperformance measures used in this study cover conservation of the target species, economic returns andthe impact of fishing on benthic communities. Two classes of management strategy are evaluated. Oneclass seeks to move stocks towards the target spawning stock size which is a pre-specified fraction of thespawning stock size at which Maximum Sustainable Yield (MSY) is achieved using a threshold control rule,while the other class selects time-trajectories of future effort to maximize discounted profit. Managementstrategies that control effort levels to maximize the total profit over the long-term outperform those whichaim to move the spawning stock size toward SMSY in terms of most performance measures. For example,even when the target stock size for the MSY-based management strategy is selected to be the same asthat which maximizes profits, selecting effort to maximize profits leads to lower variability in catches

and profits. This study illustrates how broader ecosystem considerations can be included in MSE analyses

evelosingle

1

msd(pasm

so

eaee

0d

without the need for the da “middle road” between

. Introduction

A management strategy is a fully specified set of rules for deter-ining tactical management regulations, and generally includes

pecifications for a monitoring system, an assessment proce-ure, and a decision rule. The management strategy evaluationMSE) approach allows the trade-offs among the (pre-agreed and

re-specified) management objectives achieved by different man-gement strategies to be evaluated, taking account of variousources of uncertainty (e.g. uncertainty in the assessment, imple-entation error, etc.), and with the aim of identifying management

∗ Corresponding author. Tel.: +61 7 3827 7219; fax: +61 7 3826 7222.E-mail address: [email protected] (C.M. Dichmont).

yMpntuaa

165-7836/$ – see front matter. Crown Copyright © 2008 Published by Elsevier B.V. All rigoi:10.1016/j.fishres.2008.05.007

pment and implementation of full ecosystem models and hence provides-species MSEs and full ecosystem MSEs.

Crown Copyright © 2008 Published by Elsevier B.V. All rights reserved.

trategies that are robust to uncertainty and achieve desired trade-ffs among the management objectives.

MSE has been applied to several single and multispecies fish-ries (Punt, 1992; De la Mare, 1996; Butterworth et al., 1997; Puntnd Smith, 1999; Smith et al., 1999; Punt et al., 2002; Dichmontt al., 2006a) and to ecosystems (Sainsbury et al., 2000; Fultont al., 2007). However, most MSE applications have focused onield or stock status objectives for one or several species. FewSE studies have explicitly considered economics and/or incor-

orated performance measures related to environmental effects

ot directly linked to the fished populations. Furthermore, MSEsaking an explicit ecosystem approach (e.g. Fulton et al., 2007) arencommon and in many cases impractical owing to lack of datand resources. This study shows that it is possible to undertaken MSE which considers several high-level objectives and multiple

hts reserved.

ries Re

smmcntfitct

taaifJaoTC2atsfi

posefiitaafsatpatpt

fbsittaim

aatomiaffi

idfirotlTt

2

3

4

2

2

wtitmFM

2

m

2

auessestbette

o

C.M. Dichmont et al. / Fishe

pecies and stocks, incorporates economics into both performanceeasures and management strategies, and includes performanceeasures related to environmental effects where the primary con-

ern is not on how this influences the fished population—withouteeding to base the analyses on a full ecosystem model. Althoughhis study is tailored to the specifics of Australia’s northern prawnshery (NPF), including its management system, it illustrates howhe conventional MSE approach, with its traditional fisheries focus,an be extended to deal with a broader set of management objec-ives, but still remain relatively tractable.

The NPF is a multi-species, multi-stock prawn fishery in theropical region of northern Australia (Fig. 1). It is managed usingtradeable input control system, presently based on gear size. Thennual amount of fishing effort can therefore be adjusted by chang-ng the amount of gear available to the fleet. The fishery occursrom circa April–November with a mid-season closure from roughlyune–August (the exact dates for the length of the whole seasonnd the dates separating the first and second sub-seasons dependn the assessed status of spawning stocks or in-season catch rates).he fishery was worth over AU$160M and was one of the Australianommonwealth’s most valuable fisheries in 2001 (Galeano et al.,004). However, cheap imports of aquaculture prawns, the appreci-tion of the Australian dollar, and increasing fuel prices have meanthat the fishery has been much less profitable in recent years. Aftereveral industry and government funded buy-back schemes, theshery now consists of 52 vessels and 19 operators.

The overall fishery consists of two parts: a banana and a tigerrawn fishery. The tiger prawn fishery captures mainly two speciesf tiger prawns (Penaeus semisulcatus, Penaeus esculentus) and twopecies of endeavour prawn (Metapenaeus endeavouri, Metapenaeusnsis) (Venables and Dichmont, 2004), while the banana prawnshery targets the banana prawns Penaeus merguiensis and Penaeus

ndicus (Venables et al., 2006). This study focuses on managing theiger prawn fishery only, although banana prawns are included toccount for the impact of fishing for banana prawns on the effortvailable for the tiger prawn fishery. Tiger prawns have been theocus for quantitative stock assessments and management mea-ures for many years (Somers, 1990; Somers and Wang, 1997; Wangnd Die, 1996; Dichmont et al., 2001, 2003), owing to the percep-ion that these species can be recruitment overfished. Endeavourrawns are predominantly a bycatch of targeting tiger prawns,nd management measures for tiger prawns therefore also tendo impact endeavour prawns. In contrast to the situation for tigerrawns, assessments of endeavour prawns were only conducted forhe first time in 2007.

Dichmont et al. (2006a,b,c) evaluated management strategiesor the two tiger prawns based on reference points derived fromiological and yield considerations (SMSY: the spawning stockize corresponding to MSY; and EMSY: the effort at which MSYs achieved) for the two tiger prawn species. Their results showhat management strategies based on SMSY and EMSY are unableo maintain both species at SMSY simultaneously, primarily due tossessment bias and lack of consideration of the spatial character-stics of the resource in the NPF-wide stock assessments on which

anagement advice has been based.Although the work of Dichmont et al. (2006a,b,c) provided guid-

nce on management strategies for the target species, it did notdequately address the full range of management objectives forhe NPF. Specifically, endeavour prawns were not included, andbjectives related to the economic performance of the fishery (a

anagement objective for the NPF is now an achievement of Max-

mum Economic Yield (MEY) by 2014, where MEY can be defineds “a sustainable catch or effort level that creates the largest dif-erence between total revenue and the total costs of fishing for theshery as a whole” Grafton et al., 2006; Kompas, 2005), and fishing

weiRT

search 94 (2008) 238–250 239

mpact on the benthos were not considered. The consequences ofirect physical disturbance of the seabed caused by towed bottom-shing gear have been studied extensively (see, for example, theeview by Kaiser et al., 2006). The results from these studies dependn the gear used and the habitat in which it was deployed. However,he impacts have a reputation of being substantial and have beenikened to clear-cutting of virgin forest (Watling and Norse, 1998).his study therefore extends the earlier MSE analyses to addresshese issues as follows:

1. Account is taken of the population status, yield and economiccontribution of the two endeavour prawn species.

. The impact of fishing on the benthos is quantified by using amodel (Pantus et al., 2007) of how prawn trawling influencesbenthic species.

. The performance measures used to summarize the ability to sat-isfy the management objectives include profit as well as yield(the previous MSE analyses approximated profit using tempo-rally discounted yield).

. Management strategies that are based on a bio-economic assess-ment model that attempts to find the time-trajectory of effortwhich maximizes profit, are considered in addition to man-agement strategies of the type considered by Dichmont et al.(2006a,b,c).

. Methods

.1. General MSE structure

The basic MSE framework has been described extensively else-here (e.g. Kell et al., 2006). Conceptually, however, it includes

hree key components: (a) the operating models that describe “real-ty”; (b) the management strategies that are to be evaluated; and (c)he performance measures that will be used to evaluate the perfor-

ance of each management strategy in relation to the objectives.ig. 2 indicates how these three components are linked for the NPFSE.

.2. The operating model

The operating model consists of three parts: a biological prawnodel, a benthic impacts model, and an effort allocation model.

.2.1. Prawn biological modelThe population dynamics of prawns was modelled in the oper-

ting model in the same way as in Dichmont et al. (2006a), althoughnlike Dichmont et al. (2006a) account was taken of two species ofndeavour prawns (M. endeavouri and M. ensis) as well as the twopecies of tiger prawns (P. esculentus and P. semisulcatus). All fourpecies are represented by multiple stocks. Although hypothesesxist which suggest that there may be up to seven stocks of eachpecies in the NPF, each associated with one of the regions in Fig. 1,he three westernmost regions (potential stocks) have been com-ined in the operating model owing to low densities of tiger andndeavour prawns in these regions. All four species occur in three ofhe five regions, whereas two of the easternmost regions only con-ain three species, with P. semisulcatus absent from one area and P.sculentus from the other.

The population dynamics of each species in each region that theyccur are governed by a delay-difference model that operates on a

eekly time step. The conditioning of the operating model involved

stimating annual recruitments from data on catches and standard-sed catch-rates and using these to estimate the parameters of aicker stock-recruitment relationship (cf. Dichmont et al., 2003).he values for natural mortality, growth, and weekly availability,

240 C.M. Dichmont et al. / Fisheries Research 94 (2008) 238–250

F used io

ro

oandepmu

2

ra

Feo

icpbrrt

ba

ig. 1. Map of northern Australia showing the seven regions and the 6-min gridsperating model.

ecruitment and spawning proportion were based on tagging data,n analyses of experimental data, and on analyses of survey data.

Targeted effort for one species in a given region leads to catchesf all species that are found in that region (i.e. technical interactionsre included in the operating model). Unlike tiger prawns, there iso directed (target) fishery for endeavour prawns (although, anec-otally, operators sometimes target endeavour prawns towards thend of the fishing season) so the fishing mortality for endeavourrawns is the fishing effort targeted at the two tiger prawn speciesultiplied by species-specific catchability coefficients (computed

sing the methods of Wang, 1999).

.2.2. Benthic impacts modelThe benthic impacts model estimates the primary effects of

epeated trawling on the biomass of benthic organisms (ignoringny long-term consequences of that removal on the ecosystem,

ig. 2. Conceptual overview of the MSE for the NPF. The dark boxes denote the bio-conomic component and the hashed boxes denote the benthic impacts componentf the MSE.

wtar

lopwasotvabttsrri(

n the benthic impact model. The three westernmost regions are combined in the

ncluding any effect on prawn productivity). It consists of threeomponents given input on effort by 6-min grid cell: (i) a com-onent that calculates the depletion of the biomass of a range ofenthic species (or taxonomic groups) that occur in the NPF givenepeated trawling, (ii) a component that determines the rate ofecovery of each of the species, and (iii) a component that dis-ributes the biomass (initially) uniformly over space.

The model consists of the following differential equation for theiomass Bs,g

t of benthic species group s in grid cell g at time t (Ellisnd Pantus, 2001):

dBs,gt

dt= rsBs,g

t

(1 − Bs,g

t

Bs,g0

)− Bs,g

t Egt �s (1)

here rs is the large-scale recovery rate for species group s; Egt is

he effort in grid cell g at time t in units of proportion of grid cellrea swept per unit time; and; �s is the large-scale trawl depletionate for species group s.

This biomass-dynamic logistic equation, which operates on thearge scale (6 nautical mile squares), arises from the scaling upf impacts operating at the small scale (20 m squares within theath of the trawl net). It is assumed that the small-scale effortithin the cell is a stochastic point process that results in a neg-

tive binomial distribution for the counts of trawl passes over 20 mquares. The large-scale effort Eg

t is directly related to the meanf this negative binomial distribution and the degree of aggrega-ion is related to its variance. For example, in random trawling theariance equals the mean, whereas in aggregated trawling the vari-nce exceeds the mean. The small-scale biomass dynamics operatey removing a proportion d of the biomass for every pass of therawl net and then allowing the biomass to recover along a logisticrajectory parameterized by a small-scale recovery rate rsmall. The

caling-up procedure converts the discrete process of depletion andecovery on 20 m squares, characterized by small-scale parameterssmall, d, and the parameters of the negative binomial distribution,nto a continuous process described by a differential equation (Eq.1)) with large-scale parameters r, � and Eg

t . The differential equa-

C.M. Dichmont et al. / Fisheries Re

Table 1Recovery and depletion rates by benthic group (Haywood et al., 2005)

Benthic group Small-scale recoveryrate, rsmall (year−1)

Large-scale recoveryrate, r (year−1)

Depletion rate, �

Ascideans 0.40 0.38 0.107Asteroids 0.97 0.92 0.109Bivalves 0.52 0.49 0.109Bryozoans 0.40 0.36 0.201Crinoids 0.56 0.53 0.105Crustaceans 0.52 0.48 0.136Echinoids 0.40 0.39 0.031Gastropods 0.41 0.37 0.200Gorgonians 0.71 0.66 0.150Holothuroids 0.56 0.51 0.164Hydrozoans 0.56 0.50 0.198OPS

tc

aT2sTteP2ofpbsnoi2od

2

fimotteatPam

b(cacsntb

iitCaf

tVroatctTeofi

eaiws

3iwtitotabsitsf

wbtdrtDe

2

iafmmo

phiuroids 0.63 0.62 0.033haeophyta 0.97 0.78 0.373oft corals 0.40 0.38 0.090

ion removes any need to compute impacts on the 20 m scale byapturing the net effect of the small-scale process at the large scale.

In common with the rest of the operating model, the benthos isssumed to be unfished at the start of the prawn fishery in 1970.he benthic model is restricted to the area that was fished between000 and 2004 (years in which the fleet was smaller than 100 ves-els, and the length of the fishing season was similar to the present).he values for the parameters of Eq. (1) for each group of ben-hic species (Table 1) were based on a series of depletion–recoveryxperiments on the Great Barrier Reef (GBR) (Burridge et al., 2003;oiner et al., 1998; Pitcher et al., 2007a,b), and the NPF (Hill et al.,002; Haywood et al., 2005), with the NPF values taking precedencever the GBR values. In the application of the benthic impact modelor this paper, we assumed a completely spatially random pointrocess for fishing within a grid cell, resulting in a Poisson distri-ution for the counts of trawl passes over 20 m squares. This is apecial case of the point process model in which the variance of theegative binomial distribution equals the mean. The assumptionf random fishing is conservative because, for this fishery, trawl-ng is somewhat aggregated at the sub-grid cell level (Deng et al.,005), which means that grid-cell scale impacts will be slightlyver-estimated by the benthic impacts model. When fishing is ran-om the small- and large-scale depletion rates are the same.

.2.3. Effort allocation modelThe effort allocation model consists of two components. The

rst component takes the total annual tiger prawn effort set byanagement, allows for implementation error (i.e. scientific rec-

mmendations for changes in effort limits may be ignored so theotal effort for year y may be same as that for year y − 1 even thoughhe management strategy suggested a change in effort; Dichmontt al., 2006a), generates a banana prawn season, calculates the totalvailable effort on both banana and tiger prawns, and allocates thisotal effort to week, and then target species (P. semisulcatus and. esculentus, banana prawns), and region. The second componentllocates this effort to the grid cells on which the benthic impactsodel is based.In recent years, the fishery has consisted of two sub-seasons split

y a mid-year spawning closure. The first sub-season is targeted atcommon) banana prawns and the effort during this sub-seasononsequently depends very little on tiger or endeavour prawnbundance. The fleet switches to the tiger prawn fishery, for which

atch-rates are lower but less variable, during the second sub-eason although, if banana prawns are still available in large enoughumbers, they will be fished by some of the fleet. As the length ofhe second sub-season is generally fixed, any effort expended onanana prawns is effort effectively taken away from tiger prawns. It

rtaob

search 94 (2008) 238–250 241

s not possible to model the dynamics of banana prawns and hencenclude it in the operating model explicitly owing to an inabilityo characterize the stock-recruitment relationship for this species.onsequently, the amount of effort expended on banana prawns inny future year is selected by choosing a fishing pattern at randomrom those for past years.

The movement of individual vessels from 1 day of the seasono the next is governed by a time-inhomogeneous Markov chain.essels are independently assigned to regions on a daily basis andegional effort is aggregated to week to match the time-step of theperating model. The transition probabilities of the Markov chainre functions of a number of key drivers, principally the absoluteime of year, the elapsed fishing time within the season, aggregateatch rate measures for each region for the preceding week, andhe average cost of travel that the transition in question implies.he parameters used to determine the transition probabilities arestimated from logbook data, using transitions that have actuallyccurred. Separate models are used for, and calibrated from, therst and second sub-seasons.

Once the transition for a particular virtual vessel has been gen-rated by the Markov chain, a decision is made whether the effortpplies to the banana or tiger fishery based on empirical probabil-ties. This process imitates the action of the real fleet in a realisticay, and can respond to changes in local abundance in time and

pace, as well as more gradual changes in the cost of travel.The total effort in a region is subsequently allocated to about

000, 6-min grid cells on an annual time step for use by the benthicmpacts model. Instead of a probabilistic transition model, which

ould be unworkable at this scale, an empirical approach is usedhat bases the effort distribution on historical patterns. Benthicmpacts are sensitive not only to the total trawling effort, but also tohe degree of aggregation (Die and Ellis, 1999): higher effort obvi-usly results in higher impacts, but increasing aggregation tendso reduce the impact because effort is applied to areas that havelready been depleted. Historically the degree of aggregation haseen negatively correlated with total effort, implying that the ves-els take fewer risks and stay close to known hot-spots when efforts low, and vice versa. Using the area index of Prince (pers. comm.)o quantify the degree of aggregation, we fit a power–law relation-hip between area index and total regional effort from effort dataor 1970–2004.

To allocate effort by region to grid cells in each projection year,e selected a spatial distribution of effort from a randomly chosenut recent historical year, applied a power–law to the effort dis-ribution to reflect the relationship between total effort and theegree of spatial aggregation of the effort, and then scaled theesulting “effort field” so that the total effort by grid cell equaled theotal effort from the first component of the effort allocation model.orn (2001) and Hutton et al. (2004) outline related approaches toffort allocation using probabilistic methods.

.3. Management strategies

Management strategies are either empirical (i.e. based on trendsn directly observable quantities) or model-based. Dichmont etl. (2006b,c) considered both types of management strategy andound that management strategies based on the stock assessment

ethod actually used to provide management advice performeduch better than an alternative stock assessment method based

n a biomass dynamic model. Although basing catch limits on the

esults of a regression of catch-rate on year performed as well ashe best model-based management strategy, it did not have thedded advantage of estimating spawning stock size—an indicatorf value to management. This paper is therefore restricted to model-ased management strategies using the current stock assessment

2 ries Re

mltumsttadtta

apEprttePweb

2

pot2

E

w0f(rotsiatovst(

2

sieo

c

swtfiptn(esismftu

2

3

tc

t

wkeptoaoea

ˇ

ws

fiRtasfiw

42 C.M. Dichmont et al. / Fishe

ethod because these types of management strategies are mostikely to be used as the basis for actual management advice forhe NPF tiger and endeavour prawns. The stock assessment methodnderlying all of the management strategies therefore involves esti-ating annual recruitments and hence the parameters of a Ricker

tock-recruitment relationship by fitting a delay-difference modelo weekly catch and standardised catch rate data from the start ofhe fishery in 1970 (see Dichmont et al., 2003 for full details). EMSYnd MSY for each species are calculated under the assumption ofeterministic dynamics and allowing for the technological interac-ion between the two tiger prawn fleets by selecting effort by fleeto maximize the sum of the long-term catches of all species beingssessed.

None of the management strategies examined by Dichmont etl. (2006b) were able to leave the spawning stock size of both tigerrawn species near SMSY. Reducing the target effort level to belowMSY increased the spawning stock size at the end of the projectioneriod, but the reduced risk came at a cost of reduced catches. Inecent years, the price of prawns has decreased substantially andhe price of fuel has increased dramatically. Consequently, evenhough the tiger prawn species are recovering to SMSY, the fish-ry remains in economic difficulty. Furthermore, the Environmentrotection and Biodiversity Act 1999 and the Australian Common-ealth Harvest Strategy Policy (Anon, 2007) have increased the

mphasis on precaution including the impact of the fishery on theenthos.

.3.1. MSY strategiesTwo categories of management strategy are considered in this

aper. The first set of strategies (the “MSY” strategies) is basedn setting the annual effort by target (i.e. tiger) species using ahreshold management strategy (Punt et al., 2008; Dichmont et al.,006b)1, i.e. the target effort for each tiger species is calculated as:

t =

⎧⎪⎨⎪⎩

0 if St ≤ Slimit

EtargetSt − Slimit

Starget − Slimitif Slimit < St ≤ Starget

Etarget if St > Starget

(2)

here Et is the effort for year t, Etarget is the target effort (either EMSY,.8 EMSY or 0.4 EMSY), St is the estimate of the spawning stock sizeor year t, Starget is the estimate of the “target” spawning stock sizeeither SMSY, 1.2SMSY or 1.6 SMSY), Slimit is the estimate of the limiteference point (in all scenarios, 0.5 SMSY), and EMSY is the estimatef the effort at which MSY is achieved. This means that effort is seto zero if the stock is depleted to below Slimit, increases linearly withpawning stock size until the spawning stock size reaches Starget ands constant for spawning stock sizes of Starget and larger. Note thatlthough Eq. (2) is applied to calculate effort for each tiger species,he impact of technical interactions is such that effort targeted atne species will lead to catches of all four species. Moreover, thealue of EMSY is calculated taking technical interactions into accounto that the predicted fishing mortality corresponds to FMSY for eachiger species if St = Starget for both tiger species (see Dichmont et al.2001) for additional details).

.3.2. MEY strategiesThe second set of strategies (the “MEY” strategies) is based on

etting effort to maximize profit thereby moving the fishery to Max-mum Economic Yield (MEY). Given a new assessment (undertakenvery alternate year), this involves first determining whether anyf the species is “overfished” (defined as the average spawning

1 The impact of implementation error is such that the effort may not actuallyhange each year.

fira2fuac

search 94 (2008) 238–250

tock size over the most recent 5 years being below 0.5 SMSY), inhich case the fishery is closed until that species has recovered

o above 0.5 SMSY. If none of the species is assessed to be over-shed, the sequence of future effort levels is calculated so thatrofit over a 50-year projection period is maximized (althoughhe estimation has a lower bound so that effort in a given year isot less than half of that directed towards P. esculentus in 2006a value recommended by industry and management). The futureffort for the first 7 years (with the effort for the eighth and allubsequent years set to that for the 7th year) are selected to max-mize total profit. Future harvest levels and the spawning stockizes for each of the four prawn species are projected from theost recent stock assessment and determining future recruitment

rom the deterministic component of the stock-recruitment rela-ionship. As a result, each year that a bio-economic assessment isndertaken:

1. the MEY is recalculated, i.e. the target is able to change over timebased on the biological and economic inputs,

. the most profitable pathway (as opposed to a linear pathway)to the target is calculated, i.e., the optimal pathway is able tochange over time, and

. the effort for the next 2 years is set based on selecting the firstand second year’s effort from the 7-year effort series, since anassessment is calculated biennially.

The profit function, which forms the basis for estimating theime-trajectory of future effort, accounts for costs due to labour,apital, fuel and other causes, i.e.∑

>tcur

�t = ˇt

∑w

∑s

[vs

t,wHst,w −

(cLvs

t,wHst,w + cMHs

t,w

+cKEst,w + cFEs

t,w

)](3)

here �t is the profit in future year t, vst,w is the average price per

ilogram for species s during week w of (future year) t (assumedxogenous as the product is exported); Hs

t,w is the harvest (kg) ofrawns of species s during week w of year t; Es

t,w is the fishing effortargeted at species s during week w of year t; cL, cM is the share costf labour and other variable costs per weight of output; cK, cF is theverage repairs and maintenance, and fuel and grease costs per unitf effort; ˇt is a discount factor (the rate at which future income orxpenditures is discounted relative to the present value; Grafton etl., 2006):

= 1(1 + i)t−tcur (4)

here i is the rate of interest (assumed to be 5% per annum in thistudy); and; tcur is the current year.

Cost parameters were derived from economic surveys of theshery undertaken by the Australian Bureau of Agricultural andesource Economics (ABARE). The economic survey does not dividehe NPF into tiger and banana prawn fisheries. Therefore, the aver-ge revenue and costs per vessel are computed from the NPFample as a whole and then recalculated for the tiger prawnshery considering that the banana fishery fishes for 24 h a dayhereas the tiger prawn fishery is restricted to 12 h of nightshing. The values for the economic parameters from the mostecent two surveys and those used in the analyses of this paperre summarized in Table 2. Most of the costs were similar in

004–2005 and 2005–2006, although the increase in fuel costsrom 2004–2005 to 2005–2006 is particularly noteworthy. Valuessed in the analyses were based on the survey data for 2004–2005nd 2005–2006, but also accounted for recent industry advice onosts (D. Carter, Austral Fisheries Pty Ltd., pers. comm.). Fuel and

C.M. Dichmont et al. / Fisheries Research 94 (2008) 238–250 243

Table 2Economic parameters of the tiger prawn fishery in the NPF, Australia (average per boat)

2004–2005 (number of observations, 25) 2005–2006 (number of observations, 24) Used in model

Share of labour per $1 income, cL (cents/$1 income) 0.288 0.257 0.24Share of other variable costs per 1 kg prawn, cM ($/kg) $0.68 $0.67 $0.67Repair and maintenance cost per unit effort, cK ($/day) $663 $780 $922Fuel cost, c ($/day) $1,401 $2,190 $2,190AA

S

gfe

iabop

2

aocttsaoaoaifiAos

23

4

tst(itgaptty

eao

TT

“ntts

F

verage price for tiger prawns ($/kg) $16.33verage price for endeavour prawns ($/kg) –

ource: ABARE (2007); all values are in 2006–2007 AU$.

ear costs (cF) per unit of effort are estimated by dividing totaluel and grease costs by total fishing effort (total tiger fishing dayquivalent).

Endeavour prawns are essentially caught as a bycatch of target-ng tiger prawns. This means that management is essentially aimedt tiger prawns and endeavour prawns are only considered in theio-economic model as added revenue while the catches of endeav-ur prawns only increase costs through catch-associated costs (i.e.ackaging, labour, etc.).

.4. Scenarios considered

Dichmont et al. (2006c) explored which factors in the oper-ting model or management strategy most affected the resultsf the MSE for tiger prawns. These were the catchability coeffi-ient used to convert from fishing effort to fishing mortality inhe operating model and the management strategy, implementa-ion error, and whether recruitment is spatially correlated amongtocks. Given recent changes in the management of this fishery,nd the previous sensitivity analysis, much of this paper focusesn the choice of target (MSY or MEY) in the management strategy,nd only two operating model variants differing in the true valuef catchability (denoted as “q” and “2q” here and by Dichmont etl. (2006a,b,c)) are considered. The focus on the choice of targets important because of the current poor economic state of theshery and the desire for more precaution by the Managementuthority. The scenarios considered in the analyses (combinationsf operating model and management strategy specifications) areummarised in Table 3.

Several variants of the MEY strategy are also considered:

1. Should effort be set based on a tiger prawn assessment only (i.e.assessments of endeavour prawns are not undertaken—an optionsome may argue is appropriate given the uncertainty associ-

teo

w

able 3he scenarios (in each row) considered in the management strategy evaluation of this pap

OpQ” and “ManQ” are respectively, the assumptions about catchability for endeavour prumber of species in the assessment model (3: the two tiger prawn species and endeavouhe extent of additional precaution in the MEY strategies, “MEY target” indicates whetherarget reference point of the management strategy, and “LRP Spp” indicates the number otock is “overfished” (3 and 2 are defined as for “Spp”). The shading indicates differences

$20.38 $20.00– $10.00

ated with the values for the biological parameters for endeavourprawns)?

. Should the limit reference point be applied to tiger prawns only?

. Should effort be reduced (by multiplying by a factor p) toaccount for uncertainty (and hence prevent spawning stocksfrom becoming “overfished”)?

. Should different effort levels be estimated for each of the next 7years, i.e. a sliding 7-year window (with the effort levels for allfuture years set to that for the 7th year), or should it be assumedthat fishing effort will equal EMEY from 2014 (as was intended byindustry)?

The Base Case scenario therefore sets the catchability values inhe operating model and management strategies to the presenttandard assessment values from Wang (1999)—“q”. In this case,he management strategy considers endeavour prawns as a groupi.e. the two species are combined) even though they are separatedn the operating model. The biological parameters for M. endeavouri,he most numerous and studied species, is used for this combinedroup. An assessment of M. ensis would involve too many unknownsnd uncertainties for a rigorous assessment for direct managementurposes. The biological limit reference point is only applied to thewo tiger prawn species. The target is SMEY but there is no constrainthat effort must equal EMEY by 2014 as is attempted in the “fixedear” scenario (Table 3).

Other management strategies that aim for SMEY include consid-ring only tiger prawns in the management strategy (“tigers only”),dding levels of precaution so as to achieve SMEY for all speciesr only tiger prawn species by 2014 (“p = 0.1” and “p = 0.2” respec-

ively), investigating the effects of reducing effort substantially ifndeavour prawns as well as tiger prawns are considered to beverfished (“End. Limit”).

Different target levels in the MSY strategies are examined: SMSY,hich is still a target in many countries, 1.2 SMSY, a common proxy

er

awns in the operating model and the management strategies, “Spp” indicates ther prawns as a group; 2: only the two tiger prawns), “p” is the value that determinesthe application of Eq. (2) should assume that effort equals EMEY in 2014, “TRP” is thef species to use when deciding to set effort to minimum level because a spawning

from the Base Case scenario.

2 ries Re

fsm

2

r1s(c

1

s

iiotftibrNc

3

3

brt

Fslt

44 C.M. Dichmont et al. / Fishe

or SMEY in Australia (Anonymous, 2007), and 1.6 SMSY, the spawningtock size which matches the actual values of SMEY in the operatingodel most closely.

.5. Performance measures

The performance measures are chosen to capture objectiveselated to conservation of target species (performance measures–3), achievement of high economic returns (performance mea-ures 4–6), and minimization of impacts on benthic communitiesperformance measures 7–10). All of the performance measures areomputed using the output from the operating model:

1. the ratio of S2014 to SMSY for each species;2. the ratio of S2014 to SMEY, where SMEY is the spawning stock size

at which Maximum Economic Yield is achieved (each speciesseparately);

3. the probability of the spawning stock size being above the limitreference point (quantified as the average spawning stock sizeover the most recent 5 years being above 0.5 SMSY);

4. the profit over the first 2 years of the projection period (2007and 2008; AU$M);

5. the total cumulative discounted profit over the projectionperiod (2007–2014 AU$M),

6. the inter-annual variability in profit (PAAV) over 2007–2014:

PAAV = 100∑2014

t=2007|�t − �t−1|(5)

8 median2007≤t≤2014

(�t)

7. the total effort (boat days) in 2014;8. the proportion of 6-min grid cells fished between 2000 and

2004 that are fished for more than 1 boat day in 2014;

tecpb

ig. 3. “True” (i.e. operating model) time-trajectories (medians, and 5th and 95th percentitock size relative to SMEY (centre panels), and 5-year moving average spawning stock sizeines in the right panels are the limit reference point of 50% of SMSY. The vertical dotted linhe Base Case management strategy.

search 94 (2008) 238–250

9. the total benthic biomass in the 6-min grid cells fished between2000 and 2004 in 2014 relative to unfished levels; and

0. the total gastropod biomass in the 6-min grid cells fishedbetween 2000 and 2004 in 2014 relative to unfished levels.

All of the performance measures are summarized by among-imulation medians, and 5th and 95th percentiles.

Performance measures 4–6 are based on Eq. (3), except that Hst,w

s set to the actual catch removed from the populations representedn the operating model rather than being projected from the resultsf a stock assessment. Performance measures 7 and 8 summarizehe absolute amount of effort and its patchiness, respectively. Per-ormance measures 9 and 10 summarize the relative impact onhe benthic biomass with performance measure 9 examining thempact on all species and performance measure 10 focusing on theenthic group that is the most vulnerable to trawling in the Tor-es Strait (Pantus et al., 2007) and second most vulnerable in thePF (Haywood et al., 2005). The benthic impact model is used toalculate the values of these two performance measures.

. Results and discussion

.1. Biological and economic consequences—Base Case strategy

All the four prawn species from the operating model are belowoth SMSY and SMEY at the start of the projection period (2007) if theesults for all stocks of each species are amalgamated. Fig. 3 showshe time-trajectories of spawning stock size when effort is set using

he Base Case MEY strategy (p = 1; both tiger prawn species and thendeavour prawn group used when calculating MEY and the asso-iated effort levels; effort trajectories based on a 7-year projectioneriod; fishery closed only if one of the tiger prawns is assessed toe “overfished”). Fig. 4 shows the time-trajectories of effort by tar-

les) by prawn species of spawning stock size relative to SMSY (left panels), spawningrelative to SMSY (right panels) for the Base Case simulations. The horizontal dottedes indicate the start of the projection period. The results in this figure are based on

C.M. Dichmont et al. / Fisheries Research 94 (2008) 238–250 245

F ercent panel)p

gaAssrteepssetpltsfiMMr

tFnietahs(2

3

fadttRt0t

iCsiSiltsnaitieh

ig. 4. “True” (i.e. operating model) time-trajectories (medians, and 5th and 95th potal profit (3rd right panel), and depletion of total benthic biomass (lower rightrojection period.

et (tiger) species, catch by species, total fishery discounted profit,nd relative benthic biomass for this strategy and operating model.s expected given the aim of the MEY strategy to move spawningtock size towards SMEY by 2014 (the year the Management Advi-ory Committee set as the target year by which tiger prawns shouldecover to SMEY) by maximizing profits over the period, the trajec-ories of spawning stock size all trend upward. However, only P.sculentus actually recovers to SMEY by 2014, although all species,xcept M. ensis, recover to above SMSY by the end of the projectioneriod. There is a probability of greater than 90% that the averagepawning stock size over 2010–2014 is greater than 0.5 SMSY for allpecies even though M. ensis does not recover to SMSY before thend of the projection period (Fig. 3). M. ensis is actually depletedo below the “overfished” threshold at the start of the simulationeriod (i.e. the 5-year moving average spawning stock size in 2007 is

ess than 0.5 SMSY). Despite endeavour prawns not being included inhe “overfished” rule in the Base Case strategy, the spawning stockize of M. ensis recovers almost immediately to above the “over-shed” threshold. It should be noted, however, that the catches of. ensis are small relative to those of the two tiger species and. endeavouri (Fig. 4, left panels) so recovering this species has a

elatively minor impact on the economic value of the fishery.The recovery of the four prawn species is enhanced by an ini-

ial reduction in effort (particularly that targeted at P. semisulcatus;ig. 4, upper right panel) to the minimum level allowed if a species isot declared “overfished” (half of that directed towards P. esculentus

n 2006). Had the minimum effort level not been imposed, the bio-conomic model optimisation would have resulted in a closure ofhe fishery for a year. Discounted profit increases steadily until 2012

nd then declines afterwards. This is due primarily to the low initialarvest rate recovering the spawning stock size sooner which thenupports more catch later, and the impact of economic discountingEq. (3)) as profits in absolute terms remain fairly constant after012.

t

fdw

tiles) of catch by species (left panels), effort by tiger fleet (upper two right panels)for the Base Case simulations. The vertical dotted lines indicate the start of the

.2. Biological and economic consequences—alternative strategies

Fig. 5 provides a summary of five biological and economic per-ormance measures for the Base Case strategy, three MSY strategies,nd five variants of the Base Case MEY strategy (see Table 3 foretails). As expected, the size of the spawning stock increases withhe value for Starget in the MSY strategies and (particularly) whenhe effort levels from the MEY strategy are multiplied by 0.1 and 0.2.ecovery to SMEY by 2014 occurs for all species when p = 0.1 and forhe two tiger prawn species when p = 0.2. However, setting p = 0.1 or.2 leads to much lower catches and hence profits (particular overhe longer term).

Virtually the same short-term and cumulative discounted prof-ts (and other performance measures) are achieved by the Basease MEY strategy and the MSY strategy with a target spawningtock size of 1.6 SMSY (SMEY is roughly 1.6 SMSY). Although not eas-ly discerned from the performance measures in Fig. 5, the 1.6MSY strategy leads to more variation in catches (and hence prof-ts) than the Base Case MEY strategy (Fig. 6). This is because theevel of effort set by the management strategy can vary substan-ially from 1 year to the next, particularly if the spawning stockize is assessed to be close to 1.6 SMEY, owing to the thresholdature of the control rule for the MSY strategies (Eq. (1)). It shouldlso be noted that the choice of 1.6 SMSY as the target spawn-ng stock size was based on where SMEY is roughly located fromhe bio-economic model, so this strategy is actually using morenformation than would be available in reality for an “MSY” strat-gy. The strategy is only realistic if past bio-economic analysesave been undertaken which gave an indication of SMEY relative

o SMSY.

The strategies that (a) assume that effort will be set to EMEYrom 2014 and not adjusted annually using a 7-year sliding win-ow (“Fixed” in Table 3 and Fig. 5), (b) which include M. endeavourihen considering whether spawning stocks are “overfished” (“End.

246 C.M. Dichmont et al. / Fisheries Research 94 (2008) 238–250

Fig. 5. Biological, economic, and ecosystem performance measures (medians, and 5th and 95th percentiles) for the Base Case operating model and a variety of managements w S201

M lative2 anels2 opods

LjTs

trategies (see Table 3 for scenario descriptions). The left and left-centre panels sho. endeavouri (“MEd”) and M. ensis (“MEs”). The centre-right panels show the cumu

007 to 2014 (AU$ millions), and the profit variability over that period. The right p014, the total benthic biomass relative to unfished levels, and the biomass of gastr

imit” in Table 3 and Fig. 5), and (c) which only calculate effort tra-ectories based on assessments for tiger prawns (“Tiger only” inable 3 and Fig. 5)), perform very similarly to the Base Case MEYtrategy. In contrast, the performances of the MSY strategies with

tlts

Fig. 6. As for Fig. 4, except that the res

4/SMSY and S2014/SMEY (expressed as %) for P. semisulcatus (“PS”), P. esculentus (“PE”),profit over the first 2 years of the projection period (AU$ millions), the profit from

show the total effort in 2014, the proportion of grids fished for more than 1 day inin 2014 relative to unfished levels.

argets of SMSY and 1.2 SMSY are particularly poor. These strategieseave the spawning stock size of all of the species below SMEY (andhose of the two endeavour prawns below SMSY), and also achievehort-term and cumulative discounted profits well below those for

ults are for the 1.6 SMSY strategy.

ries Re

tv

(aramdot(aphst

3

uibsose

euiToh

barBo1r

ftFtb

(lr(teet

oaethFw

C.M. Dichmont et al. / Fishe

he Base Case MEY strategy. These two strategies also lead to higherariability in profits than the other strategies.

The Australian Commonwealth Harvest Strategy PolicyAnonymous, 2007) states that the limit reference point forll Australian fisheries should, in principle, ensure that theesource is kept above 20% of unfished stock size (0.2 SVIR) withprobability of 90%. Due to the variable nature of annual recruit-ent of prawns, the NPF Management Advisory Committee set a

ifferent limit reference point to that of the policy (right panelsf Fig. 3). Nevertheless, all of the management strategies leavehe two tiger prawns species and M. endeavouri above 0.2 SVIRresults not shown). Although most scenarios leave M. ensis with

greater than 80% probability of being above 0.2 SVIR, the leastrecautionary of the management strategies (SMSY and 1.2 SMSY)ave a probability of 28% and 42% of leaving the spawning stockize of M. ensis below 0.2 SVIR respectively, but, as noted above,hese strategies are inferior in several ways to the other strategies.

.3. Benthic consequences—alternative strategies

Time trajectories for the total benthic biomass relative to itsnfished level are shown in Fig. 4 for the Base Case strategy and

n Fig. 6 for the 1.6 SMSY strategy. In both cases, the total benthiciomass starts at 94% of the unfished level and then slowly andmoothly recovers to about 96% of this level. These results perhapsverstate the impact on the benthos because the performance mea-ures are based on 6-min grid cells fished recently rather than thentire suite of grid cells that have been fished at any point in time.

As expected, there is a clear relationship between the level offfort and the number of 6-min grid cells that are fished (Fig. 5,

pper right panels). Increased effort leads to both more fishing

n “hotspots” and an increase in the spatial extent of the fishery.he effect of trawling on the benthos increases with the numberf times it is fished (Poiner et al., 1998; Haywood et al., 2005), soigher effort results in more grid cells being fished and grid cells

fbewt

Fig. 7. Mean relative biomass in 2014 (relative to unfished levels) for several benthic g

search 94 (2008) 238–250 247

eing fished more often. This, in turn, leads to lower total benthicnd gastropod biomass. However, this is not a simple proportionalelationship—effort increases by a factor of about three between thease Case and SMSY strategies, but the number of grid cells fishednly increases by a factor of about two. Except for the SMSY and.2 SMSY strategies, the gastropods (a group vulnerable to trawling),emain above 85% of unfished levels.

Fig. 7 explores the impact of prawn trawling on benthic biomassurther by showing the median relative biomass of a range of ben-hic groups in 2014 for the management strategies considered inig. 5. Most of the groups are above 90% of unfished levels, even forhe 1.2 SMSY strategy. However, gastropods and echinoderms areelow 90% of unfished levels for the SMSY strategy.

The impacts on benthic groups are relatively small in Figs. 4–6and spatially confined owing to the restriction of the fishery to aimited number of grid cells). This is perhaps unexpected given theesults of previous reviews on the impact of bottom-trawl gearse.g. Kaiser et al., 2006). However, an analysis of impacts of prawnrawling on GBR using Ecospace (Gribble, 2003), and repeat trawlxperiments in the GBR (Poiner et al., 1998) and the NPF (Haywoodt al., 2005), also concluded that impacts in this region from prawnrawling were relatively minor.

The benthic biomass performance measures 9 and 10 improvenly modestly even when effort is reduced substantially (the p = 0.1nd 0.2 scenarios). This can be attributed to the high degree offfort aggregation in this fishery, i.e. some grid cells have rela-ively high effort, some have intermediate levels of effort, but mostave relatively low effort irrespective of the total effort expended.urthermore, the rate of recovery of benthic biomass depends onhere on the logistic curve the biomass lies. The recovery can be

ast in grid cells with intermediate levels of effort (where relativeiomass is close to 50%), but slow for cells with either high or lowffort. However, effort reduction scenarios are indiscriminate as tohich grid cells are affected, and so the benefits to the benthos

end to be less marked than the reductions in effort and conse-

roups for a range of management strategies and the Base Case operating model.

2 ries Re

qabimo

3

taDmafgt(Mmrfti

3

op2bbm

eiG(bso“

t((a2erimelisnom

rosu

Fp

48 C.M. Dichmont et al. / Fishe

uently catch. The dilution of the benefits of effort reduction canlso be seen in the results of Hiddink et al. (2006) where changes iniomass tended to be roughly ten times smaller than the changes

n effort. To obtain bigger improvements in benthic performanceeasures, management actions targeted at the spatial distribution

f benthos would need to be implemented.

.4. Alternative operating models

The results in Figs. 3–7 are based on the assumption thathe catchability coefficients in the operating model are the sames those that underlie the management strategies. However,ichmont et al. (2006c) note that the performances of manage-ent strategies in the NPF are very sensitive to violation of this

ssumption. Fig. 8 therefore examines the impact of such violationor endeavour and tiger prawns (see Table 3 for details). As expectediven the results in Dichmont et al. (2006b) basing assessments onwice the default catchability value is more precautionary. Howeverand somewhat surprisingly), this also leads to the highest profits.

oreover, contrary to the results of Dichmont et al. (2006c), a mis-atch between the assumed and true value for catchability has

elatively little impact on the results. Part of the reason for the dif-erence in results can be attributed to lower effort levels leadingo lower costs and hence (in general) higher profits, and to lessermpacts on the benthos.

.5. General discussion

Globally, there has been a broadening of fisheries managementbjectives from single-species considerations to objectives that

ertain to various aspects of the ecosystem (Sainsbury and Sumaila,001; Garcia et al., 2003), and hence a move to “Ecosystem-ased Fisheries Management” (EBFM). In Australia, EBFM haseen adopted both as part of the Australian Fisheries Manage-ent Authorities (AFMA) objectives, and included in the Australia’s

slsst

ig. 8. As for Fig. 5, except that the operating models and management strategies are serawns on the performance measures.

search 94 (2008) 238–250

nvironmental protection legislation. Moreover, it has also beenncluded in Australia’s Commonwealth Harvest Strategy Policy anduidelines (Anonymous, 2007). Managing fisheries with multiple

and often conflicting) objectives leads to increased complexityoth at the technical level and when making management deci-ions. Simply arriving at an appropriate suite of managementbjectives can be extremely difficult; selecting a definition forecosystem overfishing” may be even more difficult.

A variety of ways have been suggested as the technical basiso support EBFM, including the use of multi-species modelse.g. Magnusson, 1995), full ecosystem models such as EcosimChristensen and Walters, 2004) and Atlantis (Fulton et al., 2005),nd ecological risk assessment (e.g. Fletcher, 2005; Smith et al.,007). We, along with many others (e.g. Marasco et al., 2007),xpect that the transition to EBFM will be evolutionary rather thanevolutionary in that we do not anticipate moving to models whichnclude ecosystem considerations for tactical fisheries manage-

ent advice in the short- to medium-term. Rather, we believe thatxisting single-species approaches to management should be tai-ored appropriately to address a broader set of objectives. We havellustrated this principle for the NPF through our MSE analyses. Theelection of management strategies in this paper accounts for tech-ical interactions among species, defines “fishery success” in termsf profits rather than simply yields, and quantifies some (but by noeans all) of the broader impacts of fishing.Although the length of a standard scientific paper precludes

eporting all of the results for many management strategies andperating models (or even the risk-related results at the level oftock rather species), the analyses provide results that are of generalse to other fisheries. Specifically, management strategies which

elect effort levels to maximize profits (without added precaution)ead to the same or better outcomes in all the performance mea-ures than those which aim at MSY. In particular, the Base Case MEYtrategy achieves higher (and less variable) cumulative profits thanhe SMSY strategy and has a lower impact on benthic communities.

lected to assess the impact of uncertainty regarding the catchability of endeavour

ries Re

AmiedeSesr

mcedmcb

taktaitm

A

MBm

R

AA

B

B

C

D

D

D

D

D

D

D

D

D

E

F

F

F

G

G

G

G

H

H

H

H

K

K

K

M

M

P

P

P

C.M. Dichmont et al. / Fishe

s such, simply including maximization of profit as an objective in aulti-species fishery can lead to the reduction of first-order benthic

mpacts. An additional benefit of lower fishing effort, which was notxplicitly included in the analyses of this paper, is the decrease ofiscarded bycatch. It should be noted that the Base Case MEY strat-gy does not leave the spawning stock sizes of all of the species atMEY by 2014 as desired. This goal can only be achieved by reducingffort substantially (e.g. p = 0.1 strategy in Fig. 5), although such atrategy will also reduce fishery profits to virtually zero over theecovery period.

This study does not evaluate the impact of fishing on the com-unity, either in the form of changes in mean size or on other

omponents of the ecosystem, as this would require a full-scalecosystem model such as Ecosim or Atlantis. It does, however,emonstrate that precautionary management strategies aimed ataximizing profit for a fishery that has high daily fishing costs

an reduce effort to levels that are arguably precautionary for theenthos as well.

It could be argued that to move entirely into EBFM will requirehe use of full ecosystem models. However, such models are notlways feasible for all fisheries due to limited data availability, poornowledge, or complexity of modelling methods. In this context,he middle ground as outlined in this paper is both more practicalnd achievable and allows smooth progress towards EBFM. Finally,t is perhaps pertinent that unlike many applications of full ecosys-em models, all of the values for the parameters of the operating

odel are grounded in data for the NPF (or related areas).

cknowledgements

This work was supported by FRDC project 2004/022 and CSIROarine and Atmospheric Research. Sean Pascoe, Richard Little, Jim

ence and two anonymous reviewers are thanked for their com-ents on earlier versions of this paper.

eferences

BARE, 2007. Australian Commodity Statistics 2006, Canberra, 349 pp.nonymous, 2007. Commonwealth Fisheries Harvest Strategy: Policy and Guide-

lines. Department of Agriculture, Fisheries and Forestry, Canberra, 55 pp.urridge, C.Y., Pitcher, C.R., Wassenberg, T.J., Poiner, I.J., Hill, B.J., 2003. Mea-

surement of the rate of depletion of benthic fauna by prawn (shrimp)otter trawls: an experiment in the Great Barrier Reef. Aust. Fish. Res. 60,237–253.

utterworth, D.S., Cochrane, K.L., De Oliveira, J.A.A., 1997. Management procedures:a better way to manage fisheries? The South African experience. In: Pikitch,E.K., Huppert, D.D., Sissenwine, M.P., Duke, M. (Eds.), Global trends: FisheriesManagement. Am. Fish. Soc. Symp. 20, 83–90.

hristensen, V., Walters, C.J., 2004. Ecopath with Ecosim: methods, capabilities andlimitations. Ecol. Model. 172, 109–139.

e la Mare, W.K., 1996. Some recent developments in the management of marineliving resources. In: Floyd, R.B., Shepherd, A.W., De Barro, P.J. (Eds.), Frontiers ofPopulation Ecology. CSIRO Publishing, Melbourne, Australia, pp. 599–616.

eng, R., Dichmont, C., Milton, D., Haywood, M., Vance, D., Hall, N., Die, D., 2005.Can vessel monitoring system data also be used to study trawling intensity andpopulation depletion? The example of Australia’s northern prawn fishery. Can.J. Fish. Aquat. Sci. 62, 611–622.

ichmont, C.M., Die, D., Punt, A.E., Venables, W., Bishop, J., Deng, A., Dell, Q., 2001.Risk Analysis and Sustainability Indicators for the Prawn Stocks in the North-ern Prawn Fishery. Report of FRDC Project No. 98/109. CSIRO Marine Research,Cleveland, 187 pp.

ichmont, C.M., Deng, A., Punt, A., Venables, W., Haddon, M., 2006a. ManagementStrategies for short lived species: the case of Australia’s Northern Prawn Fish-ery. 1. Accounting for multiple species, spatial structure and implementationuncertainty when evaluating risk. Fish. Res. 82, 204–220.

ichmont, C.M., Deng, A., Punt, A., Venables, W., Haddon, M., 2006b. ManagementStrategies for short lived species: the case of Australia’s Northern Prawn Fishery.

2. Choosing appropriate management strategies using input controls. Fish. Res.82, 221–234.

ichmont, C.M., Deng, A., Punt, A., Venables, W., Haddon, M., 2006c. ManagementStrategies for short lived species: the case of Australia’s Northern Prawn Fish-ery. 3. Factors affecting management and estimation performance. Fish. Res. 82,235–245.

search 94 (2008) 238–250 249

ichmont, C.M., Punt, A.E., Deng, A., Venables, W., 2003. Application of a weeklydelay-difference model to commercial catch and effort data for tiger prawns inAustralia’s Northern Prawn Fishery. Fish. Res. 65, 335–350.

ie, D., Ellis, N., 1999. Aggregation dynamics in penaeid fisheries: banana prawns(Penaeus merguiensis) in the Australian Northern Prawn Fishery. Mar. FreshwaterRes. 50, 667–675.

orn, M.W., 2001. Fishing behavior of factory trawlers: a hierarchical model of infor-mation processing and decision-making. ICES J. Mar. Sci. 58, 238–252.

llis, N., Pantus, F., 2001. Management Strategy Modelling: Tools to Evaluate trawlManagement Strategies with Respect to Impacts on Benthic Biota within theGreat Barrier Reef Marine Park Area. CSIRO Marine Research, Cleveland, Aus-tralia, 111 pp.

letcher, W.J., 2005. The application of qualitative risk assessment methodology toprioritize issues for fisheries management. ICES J. Mar. Sci. 62, 1576–1587.

ulton, E.A., Smith, A.D.M., Punt, A.E., 2005. Which ecological indicators can robustlydetect effects of fishing? ICES J. Mar. Sci. 62, 540–551.

ulton, E.A., Smith, A.D.M., Smith, D.C., 2007. Alternative Management Strategies forSoutheastern Australian Commonwealth Fisheries. Stage 2. Quantitative Man-agement Strategy Evaluation. Report to the Australian Fisheries ManagementAuthority. CSIRO Marine and Atmospheric Research, Hobart, 400 pp.

aleano, D., Langenkamp, D., Shafron, W., Levantis, C., 2004. Australian FisheriesSurvey Report 2003. Australian Bureau of Agricultural and Resource Economics,Canberra, 68 pp.

arcia, S.M., Zerbi, A., Aliaume, C., Do Chi, T., Lasserre, G., 2003. The ecosystemapproach to fisheries. Issues, terminology, principles, institutional foundations,implementation and outlook. FAO Fisheries Technical Paper No. 443. Rome, FAO,71 pp.

rafton, Q.R., Kirkley, K., Kompas, T., Squires, D., 2006. Economics for Fisheries Man-agement. Ashgate, London, 161 pp.

ribble, N.A., 2003. GBR-prawn: modelling ecosystem impacts of changes in fisheriesmanagement of the commercial prawn (shrimp) trawl fishery in the far northernGreat Barrier Reef. Fish. Res. 65, 493–506.

aywood, M., Hill, B., Donovan, A., Rochester, W., Ellis, N., Weina, A., Gordon, S.,Cheers, S., Forcey, K., McLeod, I., Moeseneder, C., Smith, G., Manson, F., Wassen-berg, T., Thomas, S., Kuhnert, P., Laslett, G., Burridge, C., Thomas, S., 2005.Quantifying the effects of trawling on seabed fauna in the Northern PrawnFishery. FRDC Project No. 2002/102, CSIRO Marine Research, Cleveland, 462 pp.

iddink, J.G., Hutton, T., Jennings, S., Kaiser, M.J., 2006. Predicting the effects of areaclosures and fishing effort restrictions on the production, biomass, and speciesrichness of benthic invertebrate communities. ICES J. Mar. Sci. 63, 822–830.

ill, B.J., Haywood, M., Venables, B., Gordon, S., Condie, S., Ellis, N.R, Tyre, A., Vance,D., Dunn, J., Mansbridge, J., Moeseneder, C., Bustamante, R., Pantus, F., 2002.Surrogates. 1. Predictors, impacts, management and conservation of the benthicbiodiversity of the Northern Prawn Fishery. FRDC Project No. 2000/160, CSIROMarine Research, Cleveland, 437 pp.

utton, T., Mardle, S., Pascoe, S., Clark, R.A., 2004. Modelling fishing location choicewithin mixed fisheries: English North Sea beam trawlers in 2000 and 2001. ICESJ. Mar. Sci. 61, 1443–1452.

aiser, M.J., Clarke, K.R., Hinz, H., Austen, M.C.V., Somerfield, P.J., Karakassis, I., 2006.Global analysis of response and recovery of benthic biota to fishing. Mar. Ecol.Prog. Ser. 311, 1–14.

ell, L.T., De Oliveira, J.A.A., Punt, A.E., McAllister, M.K., Kuikka, S., 2006. Opera-tional management procedures: an introduction to an evaluation framework.In: Motos, L., Wilson, D. (Eds.), The Knowledge Base for Fisheries Management.Elsevier Limited, pp. 379–407.

ompas, T., 2005. Fisheries management: economic efficiency and the concept of‘Maximum Economic Yield’. Aust. Commod. 12, 152–160.

agnusson, K.G., 1995. An overview of the multispecies VPA—theory and applica-tions. Rev. Fish. Biol. Fish. 5, 195–212.

arasco, R.J., Goodman, D., Grimes, C.B., Lawson, P.W., Punt, A.E., Quinn II, T.J., 2007.Ecosystem-based fisheries management: some practical suggestions. Can. J. Fish.Aquat. Sci. 64, 928–939.

antus, F., Ellis, N., Browne, M., Okey, T., Robinson, M., Rochester, W., Welna, A., 2007.Torres Strait Management Scenario Evaluation: Effects of Trawling. Report onCRC Torres Strait task T3.3. CSIRO, Cleveland, 131 pp.

itcher, C.R., Dogerty, P., Arnold, P., Hooper, J., Gribble, N., Bartlett, C., Browne,M., Campbell, N., Cannard, T., Cappo, M., Carini, G., Chalmers, S., Cheers, S.,Chetwynd, D., Colefax, A., Coles, R., Cook, S., Davie, P., De’ath, G., Devereux,D., Done, B., Donovan, T., Ehrke, B., Ellis, N., Ericson, G., Fellegara, I., Forcey, K.,Furey, M., Gledhill, D., Good, N., Gordon, S., Haywood, M., Jacobsen, I., Johnson,J., Jones, M., Kinninmoth, S., Kistle, S., Last, P., Leite, A., Marks, S., McLeod, I.,Oczkowicz, S., Rose, C., Seabright, D., Sheils, J., Sherlock, M., Skelton, P., Smith,D., Smith, G., Speare, P., Stowar, M., Strickland, C., Sutcliffe, P., Van der Geest, C.,Venables, W., Walsh, C., Wassenberg, T., Welna, A., Yearsley, G., 2007a. SeabedBiodiversity on the Continental Shelf of the Great Barrier Reef World HeritageArea. AIMS/CSIRO/QM/QDPI CRC Reef Research Task Final Report. CSIRO MarineResearch, Cleveland, 315 pp.

itcher, C.R., Haywood, M., Hooper, J., Coles, R., Bartlett, C., Browne, M., Cannard,T., Carini, G., Carter, A., Cheers, S., Chetwynd, D., Colefax, A., Cook, S., Davie, P.,

Ellis, N., Fellegara, I., Forcey, K., Furey, M., Gledhill, D., Jacobsen, I., Johnson, J.,Jones, M., Last, P., Marks, S., McLeod, I., Sheils, J., Sheppard, J., Smith, G., Strick-land, C., Sutcliffe, P., Van der Geest, C., Venables, W., Wassenberg, T., Yearsley,G., 2007b. Mapping and Characterisation of Key Biotic & Physical Attributes ofthe Torres Strait Ecosystem. CSIRO/QM/QDPI CRC Torres Strait Task number T2.1Final Report. CSIRO Marine Research, Cleveland, 145 pp.

2 ries Re

P

P

P

P

P

S

S

S

S

S

S

V

V

W

50 C.M. Dichmont et al. / Fishe

oiner, I.R., Glaister, J., Pitcher C.R., Burridge, C., Wassenberg, T., Gribble N., HillB., Blaber, S.J.M., Milton, D.A., Brewer D., Ellis, N., 1998. The environmentaleffects of prawn trawling in the far northern section of the Great Barrier ReefMarine Park: 1991–1996. Final Report to GBRMPA and FRDC. CSIRO Divisionof Marine Research—Queensland Department of Primary Industries Report,554 pp.

unt, A.E., 1992. Selecting management methodologies for marine resources, withan illustration for southern African hake. S. Afr. J. Mar. Sci. 12, 943–958.

unt, A.E., Dorn, M.W., Haltuch, M.A., 2008. Simulation evaluation of threshold man-agement strategies for groundfish off the U.S. west coast. Fish. Res. 94, 251–266.

unt, A.E., Smith, A.D.M., 1999. Harvest strategy evaluation for the eastern stock ofgemfish (Rexea solandri). ICES J. Mar. Sci. 56, 860–875.

unt, A.E., Smith, A.D.M., Cui, G., 2002. Evaluation of management tools for Aus-tralia’s South East Fishery. 1. Modelling the South East Fishery taking account oftechnical interactions. Mar. Fresh. Res. 53, 615–629.

ainsbury, K.J., Punt, A.E., Smith, A.D.M., 2000. Design of operational managementstrategies for achieving fishery ecosystem objectives. ICES J. Mar. Sci. 57, 731–741.

ainsbury, K., Sumaila, U.R., 2001. Incorporating ecosystem objectives into manage-

ment of sustainable marine fisheries, including ‘best practice’ reference pointsand use of marine protected areas. Paper to the Reykjavic Conference on Respon-sible Fisheries in the Marine Ecosystem, Reykjavic, Iceland, October 1–4, 2001,19 pp.

omers, I.F., 1990. Manipulation of fishing effort in Australia’s penaeid prawn fish-eries. Aust. J. Mar. Freshwat. Res. 41, 1–12.

W

W

search 94 (2008) 238–250

omers, I., Wang, Y., 1997. A simulation model for evaluating seasonal closures inAustralia’s multispecies Northern Prawn Fishery. N. Am. J. Fish. Manage. 17,114–130.

mith, A.D.M., Hobday, A.J., Webb, H., Daley, R., Wayte, S., Bulman, C., Dowdney, J.,Williams, A., Sporcic, M., Dambacher, J., Fuller, M., Furlani, D., Griffiths, S., Kenyon,R., Walker, T., 2007. Ecological Risk Assessment for the Effects of Fishing: FinalReport R04/1072 for the Australian Fisheries Management Authority, Canberra,300 pp.

mith, A.D.M., Sainsbury, K.J., Stevens, R.A., 1999. Implementing effective fisheriesmanagement systems—management strategy evaluation and the Australianpartnership approach. ICES J. Mar. Sci. 56, 967–979.

enables, W.N., Kenyon, R.A., Bishop, J.F.B., Dichmont, C.M., Deng, R.A., Burridge, C.,Taylor, B.R., Donovan, A.G., Thomas, S.E., Cheers, S.J., 2006. Species Distributionand Catch Allocation: data and methods for the NPF, 2002–2004. Report to AFMANo. R01/1149, CSIRO Publishers, Canberra, 171 pp.

enables, W., Dichmont, C.M., 2004. A generalized linear model for catch allocation:an example from Australia’s northern prawn fishery. Fish. Res. 70, 405–422.

ang, Y.-G., 1999. A Maximum-likelihood method for estimating natural mortalityand catchability coefficient from catch-and-effort data. Aust. J. Mar. Freshwat.

Res. 50, 307–311.

ang, Y.-G., Die, D., 1996. Stock-recruitment relationships of the tiger prawns(Penaeus esculentus and Penaeus semisulcatus) in the Australian northern prawnfishery. Mar. Freshwat. Res. 47, 87–95.

atling, L., Norse, E.A., 1998. Disturbance of the seabed by mobile fishing gear: acomparison to forest clear-cutting. Conserv. Biol. 12, 1180–1197.