Modelling ecological tipping points and road-testing management strategies for increasing marine ecosystem resilience

OCEANS & ATMOSPHERE FLAGSHIP

Éva Plagányi, Tim Skewes, Alistair Hobday CSIRO, Brisbane, Australia and Laura Blamey, William Robinson University of Cape Town

3rd International Symposium on the Effects of Climate Change on the World’s Oceans, Santos, Brazil 2015

Modelling tipping points & MSE testing | Eva Plaganyi 2 |

‘EXTREME EVENT’ RESILIENCE’: If a system is perturbed, will it bounce back or fall over?

‘PERMANENT PRESS’ RESILIENCE’: If a system is subjected to multiple stresses, can we design climate-smart strategies to build resilience and reduce the risk of collapse?

OUTLINE OF TALK

1. Examples of the use of multispecies models to advance our ability to anticipate or deal with major ecosystem shifts

Impacts of perturbations on populations (e.g. Recovery time, change in state)

Methods to detect ecological tipping points 2. Examples of how the outputs can be used to inform monitoring

and management 3. Examples of the use of management strategy evaluation (MSE)

to test the performance of alternative marine monitoring and management strategies to detect and respond to ecological changes caused by climate change

Modelling tipping points & MSE testing | Eva Plaganyi 3 |

OUTLINE

Modelling tipping points & MSE testing | Eva Plaganyi

CATS (Complex Assessment Tools)

Pla

gany

i et a

l 201

1 M

ar.

Fres

hw. R

es.

Models of Intermediate Complexity for Ecosystem assessments

4 |

OPERATING MODEL

climate drivers

supply chain

MANAGEMENT MODELe.g. quotas, usage by sectors, trade-offs

adaptive feedback

stakeholder reviewadjust

Management Strategy Evaluation

0

0.5

1

1.5

2

2.5

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2007

2009

2011

COTS

per

tow

Cou

rtesy

: AIM

S

Spatial Multi-species Operating Model (SMOM), Scotia Sea

Plagányi & Butterworth 2012 Ecol. Appl.

KRILL

KRILL FISHERY

PENGUINS Adélie

Chinstrap Gentoo

Macaroni

SEALS Antarctic fur seal

FISH Myctophids and

Perciforms

WHALES Blue Fin

Minke Humpback

FISHERY

Delay difference equations with seasonal time-step

Plagányi & Butterworth 2011

Ecol. Appl.

FISH Myctophids and

Perciforms

0

1

2

3

4

5

6

2007 2017 2027 2037

Krill

bio

mas

sBi

llion

s

0

20

40

60

80

100

2007 2017 2027 2037

Peng

uin

num

bers

Thou

sand

s

(I) Smooth interaction curve

(II) Threshold - BR

(III) Threshold - S

0

2

4

6

8

10

12

14

16

2007 2017 2027 2037

Seal

num

bers Th

ousa

nds

PERTURBATION

Resilience ? Response? Recovery ?

Breeding success impacted

Survival impacted

How does variability in one part of system propagate through ecosystem?

Modelling tipping points & MSE testing | Eva Plaganyi 7 |

Source: Plaganyi et al (2014) MEPS; Robinson et al. (2015) IJMS

0

10

20

30

40

50

60

0

1000

2000

3000

4000

5000

6000

7000

8000

1989 1991 1993 1995 1997 1999 2001 2003Ur

chin

den

sity

Abal

one

biom

ass (

t)

Abalone biomassUrchin density

A

0

5000

10000

15000

20000

25000

30000

35000

0

500000

1000000

1500000

2000000

2500000

3000000

3500000

1987 1992 1997 2002 2007

Peng

uins

rela

tive

num

bers

bre

edin

g

Pela

gic f

ish co

mbi

ned

biom

ass (

t)

Pelagic prey biomassPenguin predator numbers

0

0.5

1

1.5

2

2.5

1994 1996 1998 2000 2002 2004 2007 2011

Cora

l cov

er (%

Acr

opor

a) a

nd

rela

tive

num

ber o

f sta

rfish

Coral coverCrown of thorns

(A) African penguin - sardine

(B) Crown of Thorns Starfish (COTS) - coral

(C) Abalone – urchins (shelter)

OBSERVED CHANGES BEST EXPLAINED* BY (III) ABRUPT CHANGE IN SURVIVAL

*AIC Model selection criterion – alternative multispecies models fitted to data

Modelling tipping points & MSE testing | Eva Plaganyi 8 |

Positive residuals: filled circles; negative residuals: open circles

There is an increase in the variance as the

penguin population starts

to decline substantially

Absolute residuals (penguins) versus relative depletion (sardine) using model output: early warning signal

Source: Plaganyi et al (2014) MEPS

9 |

Increasing variation in population numbers (such as in response to a decline in prey) may be a useful indicator that a system is approaching a tipping point

Abrupt changes in some populations can more readily be ascribed to a threshold-like response of adult survival to changing conditions, rather than breeding success or a recruitment collapse

In a nutshell

Non-linear changes in population parameters (such as survival rate) below critical prey thresholds may be contributing to the responses of predators to changes in their prey

Plag

anyi

et a

l. 20

14

OUTLINE OF TALK

1. Examples of the use of multispecies models to advance our ability to anticipate or deal with major ecosystem shifts

Impacts of perturbations on populations (e.g. Recovery time, change in state)

Methods to detect ecological tipping points 2. Examples of how the outputs can be used to inform monitoring

and management 3. Examples of the use of management strategy evaluation (MSE)

to test the performance of alternative marine monitoring and management strategies to detect and respond to ecological changes caused by climate change

Modelling tipping points & MSE testing | Eva Plaganyi 10 |

OUTLINE

Modelling tipping points & MSE testing | Eva Plaganyi 11 |

Source: Morello et al (2014) MEPS

(B) Crown of Thorns Starfish (COTS) - coral

EXAMPLE 1 – CROWN OF THORNS STARFISH (COTS)

1. Model resilience of alternative ecosystem structures

2. Tipping points to inform field management controls

Small predators

Triton – ve effect+ ve effect

Large fish

COTS Adults>15cm

COTS juv. < 15cm

Fast-growingcoral

Benthic invertebrates

MarineProtected

Area

Slow-growingcoral

Manual removal/Poison injection

COTS larvae

Nutrients

Average threshold COTS density: 7.1 (± 2.3) adult COTS ha-1 ≡ 0.028 (± 0.01) adult COTS min-1 ≡

Coral cover of ~14%

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0 0.05 0.1 0.15 0.2 0.25 0.3

Chan

ge in

COT

S nu

mbe

rs

Coral cover proportion

North reef, Lizard Island, Great Barrier Reef

COTS density corresp. to min.

Threshold of coral cover

Data from Lizard Is. (Pratchett, 2005, 2010)

Estimate an ecological threshold for COTS populations

Source: Plaganyi et al (2014) MEPS 12 | Modelling tipping points & MSE testing | Eva Plaganyi

+-

-

-

-

-

-++

Illegal fishing

Legal fishing

-

Fish

-

adults

juveniles

Rock Lobsters

Abalone

Urchins

Modelling tipping points & MSE testing | Eva Plaganyi 13 |

Source: Blamey et al (2014) Ecol. Mod.

EXAMPLE 2 – ABALONE-URCHIN-LOBSTER

1. Model resilience of alternative ecosystem structures:

Has overfishing altered the system resilience?

(C) Abalone – urchins (shelter)

0

0.2

0.4

0.6

0.8

1

1985 1990 1995 2000 2005 2010

0

0.2

0.4

0.6

0.8

1

1985 1990 1995 2000 2005 2010

Lobster

Urchin

Abalone

Urchin

Fish

Abalone

Lobster

Pro

porti

on o

f K

Year

A. With historic overfishing

B. No fishing scenario

Pro

porti

on o

f K

Urchin Abalone

Fish Lobster

Resilience to climate change

Overfishing scenario: Lobsters invade range of abalone, deplete urchins, change benthos and crash abalone population

Sustainable fishing scenario: Lobsters invade range of abalone, but are kept in check by fish hence system resilient to changes

From Blamey, Plaganyi, Branch 2014. Ecol. Mod.

OUTLINE OF TALK

1. Examples of the use of multispecies models to advance our ability to anticipate or deal with major ecosystem shifts

Impacts of perturbations on populations (e.g. Recovery time, change in state)

Methods to detect ecological tipping points 2. Examples of how the outputs can be used to inform monitoring

and management 3. Examples of the use of management strategy evaluation (MSE)

to test the performance of alternative marine monitoring and management strategies to detect and respond to ecological changes caused by climate change

Modelling tipping points & MSE testing | Eva Plaganyi 15 |

OUTLINE

CLIMATE-SMART STRATEGIES

Sea cucumber / bêche-de-mer Testing alternative management strategies:

resource impacted by fishing and climate

• Fishery: 8 bêche-de-mer species on 27 reef units (in 8 zones) in the Torres Strait, NE Australia, fished by indigenous fishers

• Medium term: 2011-2030 • Attribution. Climate change identifiable as

separate from other impacts (fishery exploitation)

Plaganyi, Skewes et al. 2013. Climatic Change

Dry

san

dfis

h >$

200

kg

Black fish2% Black teat fish

10% Leopardfish1% Prickly red fish

1%

Sand fish74%

Surf red fish12%

Modelling tipping points & MSE testing | Eva Plaganyi 17 |

Torres Straits, C

atch composition, 1993 - 2007

SEA CUCUMBER SPATIAL MULTISPECIES MODEL

• Data-poor – uncertainty re biological understanding and parameters

Location choice modelled as a simple function describing utility by zone

Modelling tipping points & MSE testing | Eva Plaganyi 18 |

SEA CUCUMBER SPATIAL MULTISPECIES MODEL

• Data-poor – uncertainty re biological understanding • Uncertainty re risks of climate change • Uncertainty re impact of climate change on population

2030 Impact

Life stage Component SST

Acid

ifica

tion

SL Curr

ents

, Tor

res

Stra

it

Stor

ms

and

Cycl

ones

Rain

fall

Phyt

opla

nkto

n pr

oduc

tivity

Seag

rass

Cora

l Ree

f

Juvenile Growth H L N N N N L L NMortality H L N N L N N M NCarrying cap. N N M N L N N L NGrowth H N N N N N N N NMortality H N N N L N N N NCarrying cap. N N M N N N N N NReproduction H N N N N N N N NGrowth H L N N N N M N NMortality H L N N N N M N NAdvection N N N N N N N N N

Climate change component

Adults

Larvae

Likelihood

Risk

Consequence

RISK MANAGEMENT NEEDS TO ACCOUNT FOR MULTI-DIMENSIONAL UNCERTAINTIES BIOLOGICAL CLIMATE

VARIABLES (and downscaling)

LIKELIHOOD OF CLIMATE IMPACTS (HIGH, MEDIUM, LOW RISK)

SEVERITY OF POTENTIAL CONSEQUENCES

Monitoring data SST & sea level rise fairly certain

High risk predictions most plausible

Growth first increases then decreases with increasing temperature

Population dynamics model

Ocean pH (acidification, bleaching, coral reef habitat)

Consider cumulative effects of high and medium risk predictions

Positive and negative effects on recruitment and larval survival

Fishing behaviour Storms & cyclone increases in intensity

Complex contributors to overall mortality rates

Future markets Phytoplankton productivity

Effect of changes in habitat

Implementation and control

Ocean currents Multispecies and ecosystem effects

19 |

‘PE

RM

AN

EN

T P

RE

SS

’ RE

SIL

IEN

CE

RISK MANAGEMENT NEEDS TO ACCOUNT FOR MULTI-DIMENSIONAL UNCERTAINTIES BIOLOGICAL CLIMATE

VARIABLES (and downscaling)*

LIKELIHOOD OF CLIMATE IMPACTS (HIGH, MEDIUM, LOW RISK)

SEVERITY OF POTENTIAL CONSEQUENCES

Monitoring data SST & sea level rise fairly certain

High risk predictions most plausible

Growth first increases then decreases with increasing temperature

Population dynamics model

Ocean pH (acidification, bleaching, coral reef habitat)

Consider cumulative effects of high and medium risk predictions

Positive and negative effects on recruitment and larval survival

Fishing behaviour Storms & cyclone increases in intensity

Complex contributors to overall mortality rates

Future markets Phytoplankton productivity

Effect of changes in habitat

Implementation and control

Ocean currents Multispecies and ecosystem effects

M1 – ave M M2 – low M H1 – h=0.7 H2 – h=0.5 R1 – High risk only

R2 – High+Medium risk

I1 – base I2 – double severity of impacts

20 | *Outside scope of study – could use multiple climate models; emission scenarios

‘PE

RM

AN

EN

T P

RE

SS

’ RE

SIL

IEN

CE

Results: local depletion per zone and species D

EP

LETI

ON

(Bsp

) RE

LATI

VE

TO

NO

FI

SH

ING

& N

O C

LIM

ATE

CH

AN

GE

b) With high and medium risk impacts

Wb

War

0

0.5

1

1.5i) H. scabra

BaCu

Da

D-C G-N

War0

0.5

1

1.5ii) H. whitmaei

Cu

Da

D-C

G-NS-R

Wb

War

0

0.5

1

1.5iii) A. Mauritiana

Ba

Cu

DaD-C

S-R

0

1

2iv) H. fuscogilva

Ba Cu DaD-C

0

1

2v) T. ananus

Ba

Cu

DaD-C

G-N S-RWb

War

0

1

2vi) A. echinites

BaCu

DaD-C G-N

S-R Wb War

0

1

2vii) A. miliaris

BaCu Da

D-C G-N S-R Wb

War0

1

2viii) B. argus

Increased risk under clim

ate change

21 |

Quantify risk (and associated uncertainty) to all 8 species in each of the 8 zones

Performance Summary - Harvest Strategies

Risk of sub-optimal management: the percentage of species for which the median 2030 spawning biomass level was less than Btarg (0.48K) Risk of depletion below Blim: percentage of species for which the lower 90% confidence limit of the 2030 RS projections was less than Blim

Harvest strategy

Risk of suboptimal

management

Risk of depletion

below BlimRisk of local

depletion

Average annual profit (US$

million)A. Current catch(status quo) 50 13 12 5.31B. No monitoring:Double catches 75 25 23 10.6Profit maximisation 50 13 12 5.31Location choice based on area and distance 50 13 16 5.31Spatial rotation (3 yr) 25 13 5 3.35Closed areas/sensitive species (Warrior, Sand 13 13 9 2.72Multi-species catch composition 13 13 6 3.08C. Adaptive feedback/monitoring:Hockey stick 38 13 9 3.65Hockey stick with spatial management 13 13 1 5.31Spatial closure (Single species in Zone) (30%K 38 13 8 5.11Spatial closure (Entire Zone) (30%K trigger) 13 13 5 3.19Spatial closure (Entire Zone) (20%K trigger) 13 13 7 4.09

22 | Modelling tipping points & MSE testing | Eva Plaganyi

Changes in Species Composition / Mixed harvest bag

Modelling tipping points & MSE testing | Eva Plaganyi

No. Common name Species name

1 Sandfish Holothuria scabra

2 Black teatfish Holothuria whitmaei

3 Surf redfish Actinopyga mauritiana

4 White teatfish Holothuria fuscogilva

5 Prickly redfish Thelenota ananus

6 Deepwater redfish Actinopyga echinites

7 Hairy blackfish Actinopyga miliaris

8 Leopardfish Bohadschia argus

Value

23 |

MSE as a risk management tool

• Climate risk assessment used as an input to dynamic model • Reference Set (rather than single model) used to capture key

uncertainties = ENSEMBLE • Demonstration of use of MSE to test the performance (and

adaptability) (especially in the face of uncertainty) of alternative harvest strategies in meeting fishery objectives, such as ensuring: • low risk of depletion (overall and local) • high probability of good catch / average profits • low risk of changing the multi-species community composition • high probability of managing through climate variability and change

• Climate-smart data poor strategy example: • 3-yr spatial rotation strategy for sea cucumbers

Modelling tipping points & MSE testing | Eva Plaganyi

MSE – Management Strategy Evaluation 25 |

A Belmont Coastal Vulnerability Theme Project

Global Understanding for local solutions: Reducing vulnerability of marine-dependent coastal communities (GULLS)

www.marinehotspots.org

NEXT STEPS – test climate-smart adaptation options

SYSTEM MODELe.g. MICE, Atlantis, EwE

climate driverse.g. data inputs or biophysical models

supply chain connecting products to consumers

ADAPTATION OPTION

adaptive feedback

stakeholder reviewadjust

Socio-economic data

Species sensitivity analyses

27 |

Model results can inform monitoring and management

1. Multispecies & ecosystem models:

2. Application examples: a. Overfishing reduces resilience b. Thresholds for management

+-

-

-

-

-

-++

Illegal fishing

Legal fishing

-

Fish

-

adults

juveniles

Rock Lobsters

Abalone

Urchins

Blam

ey e

t al.

2014

Small predators

Triton – ve effect+ ve effect

Large fish

COTS Adults>15cm

COTS juv. < 15cm

Fast-growingcoral

Benthic invertebrates

MarineProtected

Area

Slow-growingcoral

Manual removal/Poison injection

COTS larvae

Nutrients

Mor

ello

et a

l. 20

14

‘EXTREME EVENT’ RESILIENCE’: If a system is perturbed, will it bounce

back or fall over?

a. Understand responses to perturbations, recovery times, resilience

b. Increasing variation in population numbers (marine species) may be a useful indicator that a system is approaching a tipping point

28 |

Need to road-test climate-smartness of management

strategies

1. MSE can support effective risk management

2. Uncertainty in biological understanding + risk of climate change effects and their impacts

3. Even when integrating across broad range of uncertainties, possible to distinguish between performance of alternative management strategies

‘PERMANENT PRESS’ RESILIENCE’: Climate-smart strategies to build resilience to multiple stresses

Oceans and Atmosphere Flagship Dr Éva Pláganyi t +61 7 3833 5955 e eva.plaganyi-lloyd@csiro.au w www.csiro.au

WEALTH FROM OCEANS FLAGSHIP

Obrigado Thank you Plagányi, É., Punt, A., Hillary, R., Morello, E., Thebaud, O.,

Hutton, T., Pillans, R., Thorson, J., Fulton, E.A., Smith, A.D.T., Smith, F., Bayliss, P., Haywood, M., Lyne, V., Rothlisberg, P. 2014. Multi-species fisheries management and conservation: tactical applications using models of intermediate complexity. Fish Fisheries 15:1-22

Plagányi, É., Ellis, N., Blamey, L.K., Morello, E., Norman-Lopez, A., Robinson, W., Sporcic, M., Sweatman, H. 2014. Ecosystem modelling provides clues to understanding ecological tipping points. Mar Ecol Prog Ser 512: 99–113

Models of Intermediate Complexity for Ecosystem assessments

Plagányi, É.E., Skewes, T., Haddon, M. and N. Dowling. 2013. Risk management tools for sustainable fisheries management under changing climate: a sea cucumber example. Climatic Change 119:181–197

Acknowledgements:

Blamey, L.K., Plagányi, É.E. and G.M. Branch. 2014. Was overfishing of predatory fish responsible for a lobster-induced regime shift in the Benguela? Ecol. Modelling 273: 140-150

Pho

to: R

ob T

arr

With thanks: