Using ecosystem modeling for fisheries management

Using ecosystem modeling for fisheries management

Cape Town, September 2006Cape Town, September 2006IncoFish WP4 WorkshopIncoFish WP4 Workshop

Villy Christensen

2

Are ecosystem models useful

for fisheries management?

Are ecosystem models useful

for fisheries management?

3

“One of those really smart quotes”“One of those really smart quotes”

“We believe the food web modelling approach is hopeless as an aid to formulating management advice; the number of parameters and assumptions required are enormous.”

Hilborn and Walters (1992, p. 448)

“We believe the food web modelling approach is hopeless as an aid to formulating management advice; the number of parameters and assumptions required are enormous.”

Hilborn and Walters (1992, p. 448)

4

Willie asked the right question...

• Why don’t the fish eat them all, dad?

5

6

• Prey behavior limits predation (foraging arena assumptions)

• Prey behavior limits predation (foraging arena assumptions)

A key aspect of EwE modeling:A key aspect of EwE modeling:

7

Organisms are not chemicals!Organisms are not chemicals!Ecological interactions are highly organizedEcological interactions are highly organized

Big effects from small changes in space/time scale

Reaction vat model Foraging arena model

Preyeaten

Prey density

Preyeaten

Prey density

Prey behaviorlimits ratePredator handling

limits rate

8

Unavailable prey B-V

Unavailable prey B-V

Available prey, VAvailable prey, V

vVvV

Predator, PPredator, P

Foraging arenaForaging arena

v = behavioral exchange rate (‘vulnerability’); predator-prey specific;based on foraging arena theory (Walters and Juanes, 1993)

v = behavioral exchange rate (‘vulnerability’); predator-prey specific;based on foraging arena theory (Walters and Juanes, 1993)

aVPaVP

v(B-V)v(B-V)

9

Time predictions from an ecosystem model of the Georgia Strait, 1950-2000

With mass-action (Lotka-Volterra) interactions only:

With foraging arena interactions:

10

A critical parameter: vulnerabilityA critical parameter: vulnerability

Top-down/bottom-up “control” & carrying capacityTop-down/bottom-up “control” & carrying capacity

11

Predator abundancePredator abundance

Predicted predation mortality ‘T

radi

tiona

l’

‘Tra

ditio

nal’

EcosimEcosim

Predation mortality: effect of vulnerabilityPredation mortality: effect of vulnerability

Bottom-upBottom-upTop-DownTop-Down

High vHigh v Low vLow v

Carrying capacity

Carrying capacity

00 Ecopath baselineEcopath baseline

V = = 2

12

So how do we get estimates of carrying capacity?

So how do we get estimates of carrying capacity?

• Surveys• Assessments

– Stock reduction analysis

• Surveys• Assessments

– Stock reduction analysis

Num

bers

(x

1000

)N

umbe

rs (

x 10

00)

Fin whalesBlue whales

Christensen, LB, 2006Christensen, LB, 2006

YearYear YearYear

13

Evaluation of simulationsEvaluation of simulations

• Can the model – replicate historic trends?– make plausible extrapolations to novel

situations?

• Can the model – replicate historic trends?– make plausible extrapolations to novel

situations?

14

Fitting to time series: learning from ecosystem history

Fitting to time series: learning from ecosystem history

• A proliferation of ecosystem modeling activities has in recent years produced many apparently credible models that fit historical data well and make reasonable policy predictions

• A proliferation of ecosystem modeling activities has in recent years produced many apparently credible models that fit historical data well and make reasonable policy predictions

15

Ecosystems where EwE models have been tested using historical trend data Ecosystems where EwE models have been tested using historical trend data

• E Bering Sea• Aleutian Islands• W&C GoAlaska• E GoAlaska• W Vancouver Island• Hecate Strait• British Columbia Shelf• Strait of Georgia • NE Pacific• CN & ET Pacific • NWHI, Hawaii• Gulf of California • Central Chile

• E Bering Sea• Aleutian Islands• W&C GoAlaska• E GoAlaska• W Vancouver Island• Hecate Strait• British Columbia Shelf• Strait of Georgia • NE Pacific• CN & ET Pacific • NWHI, Hawaii• Gulf of California • Central Chile

• Bay of Quinte • Oneida Lake• Scotian Shelf• Chesapeake Bay• Tampa Bay• S Brazil Bight• Norwegian Sea• North Sea • Baltic • S Benguela• Gulf of Thailand• South China Sea

• Bay of Quinte • Oneida Lake• Scotian Shelf• Chesapeake Bay• Tampa Bay• S Brazil Bight• Norwegian Sea• North Sea • Baltic • S Benguela• Gulf of Thailand• South China Sea

16

Formal estimation

Ecosystem model (predation,

competition, mediation,

age structured)

Ecosystem model (predation,

competition, mediation,

age structured)

ClimateClimate NutrientloadingNutrientloading

FishingFishing

Predicted C, B, Z, W, dietsPredicted C,

B, Z, W, diets

ObservedC,B,Z,W, diets

ObservedC,B,Z,W, diets

Log Likelihood

Log Likelihood

( BCC/B0)( BCC/B0)

(Diet0)(Diet0)

(Z0)(Z0)

Habitat area

Habitat area

Errorpattern

recognition

Errorpattern

recognition

Choice of parametersto include in final

estimation (e.g., climate anomalies)

Choice of parametersto include in final

estimation (e.g., climate anomalies)

Judgmental evaluationJudgmental evaluation

Modeling process: fitting & drivers

Search Search

17

Confounding of fishery, environment, and trophic effects: monk seals in NWHI

Confounding of fishery, environment, and trophic effects: monk seals in NWHI

Initial Ecosim runs: fishing &trophic interactions togethercould not explain monk sealdecline. Predicted lobster recovery

Initial Ecosim runs: fishing &trophic interactions togethercould not explain monk sealdecline. Predicted lobster recovery

Satellite chlorophyll dataindicate persistent ~40%decline in primary production around 1990. ‘Explains’ both continued monk seal decline and persistent low lobster abundance

Satellite chlorophyll dataindicate persistent ~40%decline in primary production around 1990. ‘Explains’ both continued monk seal decline and persistent low lobster abundance

Low Chl

Fishing effort:

1970 2000

18

Are seals causing fish declines in the Georgia Strait? Is it fishing?

Is it environ-mental change?

Or, is it all three?

Are seals causing fish declines in the Georgia Strait? Is it fishing?

Is it environ-mental change?

Or, is it all three?

1950 19502000 2000

19

Strait of Georgia

• EwE PP & Index of Fraser River runoff (March-April salinity at two measuring stations)

Dave Preikshot, UBC FC

0.75

0.85

0.95

1.05

1.15

1.25

1950 1960 1970 1980 1990 2000

prod

uctio

n an

omal

y

30.4

30.6

30.8

31.0

31.2

31.4

31.6salinity (‰

)Model PhytoplanktonRace Rocks Salinity

20

BC Shelf biomass changes

0.0

1.0

2.0

3.0

4.0

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

herring EwEherring SA

0.0

0.1

0.2

0.3

0.4

0.5

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

P. cod EwEP. cod SA

0.1

0.2

0.3

0.4

0.5

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

halibut EwEhalibut SA

0.0

0.3

0.5

0.8

1.0

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

coho EwE

coho B from catch

0.0

0.5

1.0

1.5

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

chinook EwE

chinook B from catch

0.00

0.02

0.04

0.06

0.08

0.10

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

seals EwEseal SA

0.0

1.0

2.0

3.0

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

hake EwEhake SA

0.0

0.1

0.2

0.3

0.4

0.5

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

POP EwEPOP SA

0.1

0.2

0.3

0.4

0.5

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

sablefish EwE

sablefish SA

sablefish B=C/F

Dav

e P

reik

shot

, UB

C F

C

21

BC shelf: Upwelling index in May, June, and July. ≥10 year period

0.7

0.9

1.1

1.3

1950 1960 1970 1980 1990 2000

prod

uctio

n an

omal

y

-20

-10

0

10

upwelling (m

3/100m/s)

Model Phytoplankton54ºN Upwelling

Dave Preikshot, UBC FC

22

Northeast Pacific biomass changes

0.0

0.2

0.4

0.6

0.8

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

arrowtooth EwE

arrowtooth SA

1

2

3

4

5

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

pollock EwEpollock SA

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

P. cod EwE

P. cod SA

0

0.2

0.4

0.6

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

sockeye EwE

sockeye B from catch

0

0.1

0.2

0.3

0.4

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

halibut EwE

halibut SA

0

0.1

0.2

0.3

0.4

0.5

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

pink EwE

pink B from catch

0.2

0.4

0.6

0.8

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

plaice EwE

plaice SA

0.0

0.4

0.8

1.2

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

POP EwEPOP SA

0.2

0.4

0.6

0.8

1950 1960 1970 1980 1990 2000

biom

ass

(t/k

m2 )

chum EwE

chum B from catch

Dave Preikshot, UBC FC

23

Northeast Pacific: PDO index (Pacific Decadal Oscillation), April to July. 50 year period

0.7

0.9

1.1

1.3

1.5

1950 1960 1970 1980 1990 2000

pro

duct

ion a

nom

aly

-1.0

-0.5

0.0

0.5

1.0

1.5

PD

O in

dex

Model PhytoplanktonPDO index

Dave Preikshot, UBC FC

24

Why have Steller sea lions declined?

Guenette, Heymans, Christensen & Trites (CJFAS Nov 2006)

1000 0 1000 2000 3000 4000 Kilometers

N

Competitive Interactions

Fishing

Ocean Climate Change

Predation

Alaska

Aleutian Islands

Guénette, Heymans, Christensen & Trites (MS)

1960 1980 20000

10,000

20,000

30,000

40,000

Ab

un

da

nc

e

Competitive Interactions

Predation

Fishing

General finding: multiple factors impact ecosystem resources

(in all but the easiest cases)

General finding: multiple factors impact ecosystem resources

(in all but the easiest cases)

27

Evaluating trendsEvaluating trends

1. Fishing pressure

2. Trophic impact, including competition

3. Environmental impact

4. Nutrient loading

As a rule: All of the above contribute

1. Fishing pressure

2. Trophic impact, including competition

3. Environmental impact

4. Nutrient loading

As a rule: All of the above contribute

28

• It’s beginning to look like it;

• We can with some credibility describe agents of mortality and trophic interdependencies;

• Evaluation of relative impact of fisheries and environmental factors is progressing;

• As a rule we need to invoke fisheries and environmental drivers to fit models.

• It’s beginning to look like it;

• We can with some credibility describe agents of mortality and trophic interdependencies;

• Evaluation of relative impact of fisheries and environmental factors is progressing;

• As a rule we need to invoke fisheries and environmental drivers to fit models.

Are we finally able to develop useful predictive models for ecosystem

management?

Are we finally able to develop useful predictive models for ecosystem

management?

• When we have a modelthat can replicate development over time we can (with some confidence) use it for ecosystem-based policy exploration.

• When we have a modelthat can replicate development over time we can (with some confidence) use it for ecosystem-based policy exploration.

29

Report card: Using models to address ecosystem management questions

CONCERN GRADE COMMENT

Bycatch impacts A- We are not bad at predicting direct effect of fishing in general

Top-down effects(of predator culling or protection)

C Trophic effects of fishing can be classified as ‘top down’ or ‘bottom up’ with respect to where management controls are exerted

- on valued prey B Changes in M for prey species already subject to assessment

- on ‘rare’ prey F Outbreaks of previously rare species

30

Modeling report card (cont.)

CONCERN GRADE COMMENT

Bottom-up effects(effects of prey harvesting on predator stocks)

C Uncertainty here is about flexibility of predators to find alternative food sources when prey are fished

Multiple stable states

B ‘Cultivation-depensation’ mechanism appears to be main mechanism that could cause ‘flips’

Habitat damage D Lack of understanding about real habitat dependencies, bottlenecks

Selective fishing practices/policies

F We have not yet looked closely at options in this area!

Production regime changes

B Models look good when fitted to data, but have not stood test of time

Regime shifts C Policy adjustments in response to ecosystem-scale productivity change

31

So are ecosystem models actually used for fisheries management?

So are ecosystem models actually used for fisheries management?

32

Use of EM for fisheries management

• Multispecies models– Estimating predation mortality for stock assessment;– Limit harvest of prey species to meet consumer

demands;– Impact of changing mesh size, North Sea roundfish;– Minke whale and harp seal culling?– Environmental Impact Assessment (EIA), Alaska

groundfish;– Target species response to TACs, Bering Sea.

• Multispecies models– Estimating predation mortality for stock assessment;– Limit harvest of prey species to meet consumer

demands;– Impact of changing mesh size, North Sea roundfish;– Minke whale and harp seal culling?– Environmental Impact Assessment (EIA), Alaska

groundfish;– Target species response to TACs, Bering Sea.

33

Use of EM for fisheries management

• EwE– Evaluate impact of shrimp trawling, GoCalifornia;– Evaluate impact of bycatch, GoCalifornia;– Evaluate impact of predators on shrimp, GoMexico;– Demonstrate ecological role of species, GoMexico;– Impact of proposed fisheries interventions, Namibia?– EIA of proposed fisheries interventions, Bering Sea;– EIA of alternative TAC’s, Bering Sea and GoAlaska;– Target species response to TACs, Bering Sea– Closed area sizing, Great Barrier Reef, Australia– Valuation of cormorant impact, Ortobello, Italy– South Africa pelagic fisheries: in progress.

• EwE– Evaluate impact of shrimp trawling, GoCalifornia;– Evaluate impact of bycatch, GoCalifornia;– Evaluate impact of predators on shrimp, GoMexico;– Demonstrate ecological role of species, GoMexico;– Impact of proposed fisheries interventions, Namibia?– EIA of proposed fisheries interventions, Bering Sea;– EIA of alternative TAC’s, Bering Sea and GoAlaska;– Target species response to TACs, Bering Sea– Closed area sizing, Great Barrier Reef, Australia– Valuation of cormorant impact, Ortobello, Italy– South Africa pelagic fisheries: in progress.

34

So why aren’t ecosystem models used more for management?

So why aren’t ecosystem models used more for management?

• Lack of experience using ecosystem models for predictive purposes;

• Ecosystem modeling is for strategic management, and supplements the tactical single species assessment;

• Fisheries management process is trapped in tactical management;

• Strategic decisions are virtually non-existing.

• Lack of experience using ecosystem models for predictive purposes;

• Ecosystem modeling is for strategic management, and supplements the tactical single species assessment;

• Fisheries management process is trapped in tactical management;

• Strategic decisions are virtually non-existing.

35

• We need longer-term data than typical in assessments to avoid shifting baselines, e.g., 1950-present;– Data mining is required;– There is much more information out there:

Catches, CPUE, w, …• Assessments should be expanded back in time:

– Stock Reduction Analysis;• Biggest information gaps for:

– Mid-TL forage fishes;– Novel conditions (vampires in the basement)– Estimates of mortality rates.

Data gap for modeling

36

Our empirical knowledge is limitedOur empirical knowledge is limited

• Habitat and environmental changes (including those caused by fishing) and intensive fishery removals are creating novel situations, which we can only handle with difficulty:– We do not to understand the ‘mechanics’ of

ecological response well enough to be able to predict all important responses to these novel situations;

– Make models one can play with;

• Habitat and environmental changes (including those caused by fishing) and intensive fishery removals are creating novel situations, which we can only handle with difficulty:– We do not to understand the ‘mechanics’ of

ecological response well enough to be able to predict all important responses to these novel situations;

– Make models one can play with;

37

Our capability to provide advice about large-scale dynamics is limited

Our capability to provide advice about large-scale dynamics is limited

• We cannot resolve uncertainty about how ecosystems change based on models and time-series data only;

• We cannot resolve uncertainty about how ecosystems change based on models and time-series data only;

38

Predictive approaches are uncertain, for some obvious reasons

Predictive approaches are uncertain, for some obvious reasons

• Lack of long-term monitoring data on non-target species and life stages;

• Concentration of interaction effects (trophic, habitat) on early life stages (recruitment) that are difficult to monitor;

• Confounding of fishery, environmental, and trophic effects in historical data;

• Failure to anticipate new problems (‘vampires in the basement’) due to unpredictable changes in system structure, (exotic invasions, fisheries inventions);

• Unpredictable pre-adaptations to habitat alterations.

• Lack of long-term monitoring data on non-target species and life stages;

• Concentration of interaction effects (trophic, habitat) on early life stages (recruitment) that are difficult to monitor;

• Confounding of fishery, environmental, and trophic effects in historical data;

• Failure to anticipate new problems (‘vampires in the basement’) due to unpredictable changes in system structure, (exotic invasions, fisheries inventions);

• Unpredictable pre-adaptations to habitat alterations.

39

Ecosystem modeling for adaptive management requires a very different

approach to assessment

Ecosystem modeling for adaptive management requires a very different

approach to assessment

• Modelers must attempt to uncover alternative models that equally well explain historical data but imply different policy choices:– Environmental vs. fisheries vs. trophic effects;

• Policy options would include diagnostic management experiments to distinguish between the alternative models:– Spatial closures to test recovery predictions;

– Ecosystem modification to test trophic interaction effects.

• Modelers must attempt to uncover alternative models that equally well explain historical data but imply different policy choices:– Environmental vs. fisheries vs. trophic effects;

• Policy options would include diagnostic management experiments to distinguish between the alternative models:– Spatial closures to test recovery predictions;

– Ecosystem modification to test trophic interaction effects.

40

Models are not like religion

– you can have more than one

41

The new Ecopath with Ecosim

• Four year project funded through Lenfest Ocean Program

• Lenfest Ocean Futures Project:– New generation of EwE to be

released Sep 07– Single-player game version 2008– Multi-player game version 2009

• Customized versions facilitated• User Ownership

• Customized versions facilitated• User Ownership