Workshop on Ecosystem Based Fisheries Management – FAME, Esbjerg 27 March 2008 – slide 1 Fisheries and species interactions Photo:Getty Images Dag Ø. Hjermann

Workshop on Ecosystem Based Fisheries Management – FAME, Esbjerg 27 March 2008 – slide 1

Fisheries and species interactions

Photo:Getty Images

Dag Ø. Hjermann Centre of Ecological and Evolutionary Synthesis (CEES)

Dept. of Biology, University of Oslo


Structure of talk• Introduction: food-webs and species interactions

• Example: Interactions among Barents Sea fish stocks

• Cod, capelin and herring

• Effects of predation on capelin

• …which affects cod recruitment

• Species interactions and fisheries

• Combining fisheries and predation: a capelin example

• Modelling ecosystems: strategies

• Ecosystem functioning and fisheries

• Bottom-up, top-down, wasp-waist

• Alternative equilibria in freshwater and marine(?) ecosystems

• Species interactions and climate















The Barents Sea, the 1980s

• ”There has been several good year-classes for cod juveniles now. Although this has no immediate effects, it will not be long until the cod fisheries will be far better than now.”

Aftenposten, 12 September 1984

• ”Never has the Lofoten sea been as black as this year. (…) The worst season since fisheries statistics started in 1859”

Aftenposten, 18 April 1988

• What happened with the cod 1984-88? …species interactions!


Species interactions

(- -) interactions: competition (both species lose)

(+ -) interactions: predation (+ parasitism, disease)

(+ +) interactions: mutualism and symbiosis

We will here focus on predation (easiest to measure) and a bit on competition

• parasitism (disease) is probably also important for marine fish

• mutualism is important in tropical reef environments


Species interactions

Food web(shows only predation)

A

B C

D E

F

A

B C

D E

+- +

-

+ - + -+

-

F+

- +-

Interactions(only signs)

A B C D E F

A

B

C

D

E

F

Interactions(quantified)


Species interactions• Example: What is the net effect of B on C?

• Seems obvious that they are competitors for food E, and therefore B has a negative effect on C

A

B C

D E

+ -

- +-

F

- +

• To find the real effect, multiply interaction effects along all paths from B to C

1. Negative: reducing the abundance of E (competing for food)

2. Negative: increasing the abundance of predator A

3. Positive: increasing the abundance of F

• To find the real net effect, the strength of all interactions must in principle be quantified

1

2

3















A case study of interactions: the Barents Sea

Spitzbergen (Svalbard) Novaya

Zemlya

Franz Josef’s land

1.4 million km2 (4 times Norway’s land area), shared Norway/RussiaAverage depth: 230 mIce coverage: used to be 50-75% i March

0-10% in September


Norwegian Sea

North Sea

GreenlandSea

ATLANTICOCEAN

Spitsbergen

Gre

enla

nd

ICELAND

Scotland

NORWAY

RUSSIA

Jan Mayen

Faroe Islands

A case study of interactions: the Barents Sea

Barents Sea

Lofoten

North Atlantic

drift current

Coastal current

Large effect of the influx of warm, Atlantic water into the Barents Sea

Large variation from year to year and on longer time scales


Main components of the Barents Sea pelagic food web

Zooplankton

Phytoplankton (Norw. Sea - Barents Sea)

Capelin

Cod

Young herring

• Relatively low rate of primary production (phytoplankton), but huge area

• "Import" of zooplankton (Calanus finmarchicus) from the Norwegian Sea

• Effective transfer of energy(the "fatty food chain")

-> High production of fish on high trophic levels (cod)


The capelin

Main zooplankton feeder

Important food source for fish, mammals and

birds

Short-lived and unstable

Zooplankton


Capelin

1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006

02

00

40

06

00

80

01

00

0 Stock 1973-2006


Capelin, Mallotus villosus

Summer

Winter

Spawning

The species best able to take advantage of the production of the Barents Sea(migrates far north in summer)


North-East Arctic cod

Cod

Most important predator of fish

Largest remaining cod stock

Zooplankton


Capelin

1913 1921 1929 1937 1945 1953 1961 1969 1977 1985 1993

01

00

00

00

20

00

00

03

00

00

00

Stock 1913-2006


SeptemberFebruaryIce edge

Spawning area

Fødeareal

Gyting

North-east arctic cod(Gadus morhua)

• Long-lived (>15 år)

• Large variation in year-classes (cohorts)


Norwegian spring-spawning herring

Cod

Young herring

Main zooplankton feeder of the

Norwegian Sea

Uses the Barents Sea as a nursery (age 0-2)

Eats capelin larvae

Zooplankton


Capelin

1907 1916 1925 1934 1943 1952 1961 1970 1979 1988 1997

0e

+0

02

e+

05

4e

+0

56

e+

05

8e

+0

5

Stock 1907-2006


overwintering

spawning

Adults, age > 3 yr (summer)

age 1-2

Spawning area age 3

larvae

Norwegian spring-spawning herring, Clupea harengus

Age 0-2 years: Eats capelin larvae


So, what did happen with cod in the late 1980s?• 1970s and 1980-81: Low recruitment of the cod stock (a cold period in the Barents Sea)

• 1982-84: Very good recruitment of cod (warm years) → optimism

• …1983: also very good herring recruitment – for the first time since 1961

• 1985: The capelin stock collapses (>95% decrease)

• massive mortality of capelin larvae during the summers 1984-85

• capelin spawns at the age of 3-4 – and dies after spawning

• two years of no recruitment = stock collapse

• …also, the capelin fishery was closed too late (spring 1986)

• delayed the recovery of capelin

• 1985-1989: food shortage for the cod (+ harp seal, guillemots)

• Slow growth → small catches, delayed maturation

• Cannibalism: the 1984 year-class "disappeared" (eaten by 1983 yr-class?)

• High mortality of premature (age 3-6) cod















1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006

02

00

40

06

00

80

01

00

0

> 1 million tons of young herring

Reproductive success in capelin:depends (for a large part) on herring


Capelin: calculating the growth rate r

Lifetime reproductive output (R): the number of matures produced per mature

Generation length (T): average age of spawners

Reproductive rate (r): r = log(R)/T

Year t t+1 t+2 t+3 t+4

Matures spawning mat. 2-yr mat. 3-yr mat. 4-yr

Hjermann, Stenseth & Ottersen, MEPS 2004


Year

r

75 80 85 90 95

-1.0

0.0

0.5

1.0

1.5

log(matures)

r

0 1 2 3 4 5

-1.0

0.0

0.5

1.0

1.5

7374

7576

7778

798081

82

8384

85

86

87

88

89

90

9192

93

9495

• 1973-1980: no herring carrying capacity around log(N) = 4.5

1973-1980

1985-1995

85-95 73-80K (carrying capacity)

1980-1984

• 1985-1995: herring has returnedcarrying capacity around log(N) = 3

Capelin: observed (1975-) growth rate r

Hjermann, Stenseth & Ottersen, MEPS 2004


A (relatively) simple capelin model

log(N1t) = a1 + (1-b1)∙ log(Nmat

t-2) – c1∙( harvaut,t-2+harvwint,t-1)/BMt-2 – d1∙log(codt-1∙BMt-1

) – e1∙herrt-1 (1a)

log(N2t) = a2 + (1-b2)∙ log(N1

t-1) – c2∙harvaut,t-1/BMt-1 – d2∙log(codt∙BMimmat

t-1) – e2∙herrt (1b)

log(N3t) = a3 + (1-b3)∙log( N2

t-1- N2,matt-1) – c3∙harvaut,t-1/BMt-1 – d3∙log(codt

∙BMimmatt-1

) – e3∙herrt (1c)

log(N4t) = a4 + (1-b4) ∙log( N3

t-1- N3,matt-1) – c4∙harvaut,t-1/BMt-1 – d4∙ log(codt

∙BMimmatt-1

) – e4∙herrt, (1d)

• Proportion of matures(age) is a function of Nage

• All matures die after spawning• Coefficients (a, b, c, d) estimated using ordinary linear regression for each age

Densitydependence

Harvest Cod Herring

Hjermann, Ottersen & Stenseth, Proc. Nat. Acad. Sci. 2004

predation on larvae


A (relatively) simple capelin model

log(N1t) = a1 + (1-b1)∙ log(Nmat

t-2) – c1∙( harvaut,t-2+harvwint,t-1)/BMt-2 – d1∙log(codt-1∙BMt-1

) – e1∙herrt-1 (1a)

log(N2t) = a2 + (1-b2)∙ log(N1

t-1) – c2∙harvaut,t-1/BMt-1 – d2∙log(codt∙BMimmat

t-1) – e2∙herrt (1b)

log(N3t) = a3 + (1-b3)∙log( N2

t-1- N2,matt-1) – c3∙harvaut,t-1/BMt-1 – d3∙log(codt

∙BMimmatt-1

) – e3∙herrt (1c)

log(N4t) = a4 + (1-b4) ∙log( N3

t-1- N3,matt-1) – c4∙harvaut,t-1/BMt-1 – d4∙ log(codt

∙BMimmatt-1

) – e4∙herrt, (1d)

Hjermann, Ottersen & Stenseth, Proc. Nat. Acad. Sci. 2004

-0.5

0.0

0.5

1.0

1.5

2.0

Eq. 1a(age 1.5)

Eq. 1b(age 2.5)

Eq. 1c(age 3.5)

Eq. 1d(age 4.5)

Sta

ndar

dize

d re

sidu

als

**

** *

***

** * (*)

*** ***

* *

(*)

(*) (*)

a (density-dependence) b (harvest) c (cod) d (herring)

herring predation on larvae

cod predation on age 1

• Key processes: herring predation on larvae and cod predation on age 1

• More detailed analysis (in prep.) shows that harvest has no effect on reproduction (exception: 1985-86)


Capelin model, simulations

Year

Lo

g(a

bu

nd

an

ce)

1982 1984 1986 1988 1990

34

56

YearL

og

(ab

un

da

nce

)1990 1992 1994 1996 1998 2000

34

56

7

1

0.8 0.9 1.0 1.1 1.2

0.8

0.9

1.0

1.1

1.2

Year

Lo

g(a

bu

nd

an

ce)

1982 1984 1986 1988 1990

12

34

5

Year

Lo

g(a

bu

nd

an

ce)

1990 1992 1994 1996 1998 2000

23

45

6

Observed Full model Model w ithout - Cod - Herring - Herring competition - Harvest

The 1980s

All reduced modelsperform less well- cod, herring and harvest are all important

Hjermann, Ottersen, Stenseth, Proc. Nat. Acad. Sci. 2004















Herring predation on capelin indirectly affects cod cannibalism

Capelin

Cod age3-6

AbundantHerring age 1-2

Cod age 1-3

Little capelin

→ Much cod cannibalism

some years after good

herring reproduction:

Capelin

Cod age3-6

Cod age 1-3

ScarceHerring age 1-2

Much capelin

→ Little cod cannibalism

some years after bad herring

reproduction:


Ncodage 3-6, t-t

BMcapelint-t

) + f ∙Tempt-3 – g ∙

(cannibalism)

(Beverton-Holt)

log(Ncodage 3, t) = log( a*BMcodspawners, t-3

1+c*BMcodspawners, t-3

(climate effect)

A model for cod recruitment

p = 0.01p < 0.001Fitted back to 1973 (start of capelin time series):

Hjermann et al., Proc. Roy Soc. 2007


Ncodage 3-6, t-t

BMcapelint-t

) + f ∙Tempt-3 – g ∙

(cannibalism)

(Beverton-Holt)



(climate effect)

Capelin can be replaced by herring age 1-2

F(Herring age 1-2)

9 10 11 12 13 14

56

78

9

log(HIt)

log(

Cap

elin

bio

mas

s)

BMcapelin

BM herring age 1-2(with time lags)

Now the cod model can be

fitted from 1921

(no longer limited by capelin data)



Ncodage 3-6, t-t

BMcapelint-t

) + f ∙Tempt-3 – g ∙

(cannibalism)

(Beverton-Holt)



(climate effect)

Capelin can be replaced by herring age 1-2

F(Herring age 1-2)

p = 0.003p = 0.003Fitted back to 1921:
















Species interactions and fisheries

• The classical Gordon-Schaefer model

• Predation not taken into account – but can be seen in the framework of this model

• But: predation is not constant

Pop. growth Fishing

Fishing + predation

Biomass added or removed

per year

Pop. growth + predation on recruiting stages

Population biomass


number of capelin (age >1 yr)

log ca

tch +

cod

(in ca

tch e

quiva

lents)

0 2000 4000 6000

750

1000

1500

2000

r < - 0 . 7 - 0 . 7 < r < - 0 . 4 - 0 . 4 < r < 0 0 < r < 0 . 4 0 . 4 0 0 . 4 0 . 4 0 . 4 0 .7 - 0 . 7 < r < - 0 . 4 0 . 4 < r < 0 . 7 r > 0 . 7

m u c h h e r r i n g m u c h h a r v e s t

1974

1996

19901985 1981

1993

Labels = year-class

Capelin: combining fishing and cod predation (i.e., predation after larval stage)

Harvest and cod predation high+ small capelin stock

= strong declineHarvest and cod predation high

+ large capelin stock= equilibrium

Consumption = productio

n

Pop. growth

(surplus)

N

Combined cod predation and

harvest *)

Hjermann, Ottersen, Stenseth, Proc. Nat. Acad. Sci. 2004

GROWTH

DECLINE

Harvest closed= increase


The time of huge capelin catches (3 mill. tons) is gone (hopefully)

Herring stock by age

1907 1916 1925 1934 1943 1952 1961 1970 1979 1988 1997

0e

+0

02

e+

05

4e

+0

56

e+

05

8e

+0

5

golden age of capelin fisheries


For the cod: harvest of cod makes the cod more sensitive to shortage of capelin

Population growth rateof cod (r)

Cod stock increasing

Cod stock decreasing

Without fishing: need 3 capelin per cod

With fishing: need 24 capelin per cod

Durant, Hjermann, Sabarros, StensethEcol. Appl. 2008















Fishing/harvesting in a multi-species food-web

• Food-webs are extremely complex (even in sub-arctic environments)

• Models must be strongly simplified versions of reality

• What do we lose by simplification?

• Approaches:

• Age-(and size?)-structured models of a few species

• Mass-balance models for a larger number of species(Ecopath with Ecosim)

• Work in the Barents Sea has concentrated on the first approach


A simplified Barents Sea ecosystem

capelin young herring

cod

zooplankton

phytoplankton

benthicinvertebrates

other pelagic prey

harp sealminke whale

man

Even this simplified version is extremely complex

http://www.skipsteknisk.no/rehua.html


A simplified Barents Sea ecosystem

capelin young herring

cod

zooplankton

phytoplankton

benthicinvertebrates

other pelagic prey

harp sealminke whale

man

Climate • Cod and herring reproduction

strongly correlated to climate

• Most complex existing age-structured model (Schweder et al. 1998)

Schweder et al.: each minke whale results in 5 tons less fish

http://www.skipsteknisk.no/rehua.html


The effects on the Barents Sea cod of reducing the capelin harvest (3 species model)

Red lines: Simulations with capelin harvest effort reduced by 75%

7 % increase

Year

0 10 20 30 40 50

-6-4

-20

24

Capelin matures biomass (log-scale)

Year

0 10 20 30 40 50

05*

10^5

1.5*

10^6

2.5*

10^6

Cod biomass age 5-10


Are models with few species the best possible approach – or are they dangerous simplifications?

• "In these models, the Barents Sea ecosystem, composed of at least 144 fish species, is reduced to four components: northern minke whales, herring, cod, and capelin" (Corkeron 2006)

• Peter Yodzis

• Need to take all possible indirect interactions into account (Yodzis TREE 2001)

• The form of the predators' functional and numerical response is important


The importance of knowing the complete food web

• Wrong decisions can be made based on mistaken beliefs of food web interactions

• Yodzis' example: Should fur seals in Southern Benguela be culled?

deep-water Cape hake

(Merluccius paradoxis)

fisheries fur seals fisheries fur seals

shallow-water Cape hake

(Merluccius capensis)

deep-water Cape hake

(Merluccius paradoxis)

Yodzis TREE 2001

YES(if maximising hake

fisheries is the goal)

? (depends on interaction strengths)















Which way is the ecosystem ”controlled”?• Bottom-up control

• populations are regulated by their food supply

• ultimately, the abundance of top predators (e.g., marine mammals and birds) are controlled by primary production (photosynthesis), which again is controlled by climate and nutrient levels

• Top-down control

• populations are regulated by their predators (including fisheries)

• changes in the abundance of key predators affects the entire food chain

• Wasp-waist control (Cury)

• Pelagic zooplankton-feeding fish are only made up by one or two extremely numerous species (e.g. anchovy, sardine)

• Changes in these key species affects the food web ”above” and ”below”


Bottom-Up Top- Down Wasp-Waist Phillippe Cury, Lynne Shannon

Which way is the ecosystem ”controlled”?


Which way is the ecosystem ”controlled”?• Or to rephrase the question: at different levels of the food web, how important is predation?

• less important than food limitation: bottom-up control

• more important than food limitation: top-down control

• for piscivorous fish, less important than food limitation; for zooplanktonn and phytoplankton, more important than food limitation: wasp-waist

• Bottom-up control traditionally thought to be dominating for marine fish

• Spatial variation between areas is clearly bottom-up controlled (i.e. fish abundance is controlled by primary productivity), e.g. Ware and Thomson (Science 2005)

• Freshwater: Top-down control found to be important

• Numerous manupulative studies showing ecosystem effects of removing top predators

• Even applied: Removal of top predators to decrease algae


Top-down effects in freshwater ecosystems

Fishing large pike improves the water quality

Pike

Zooplanktonpredators

Daphnia

Algae

–

+

–

Cannibal pike

+

–















Size-based predation can cause multiple equilibria: the lake Takvatn experiment

• Original top predator: trout. Charr introduced in the 1930s

• 1980: almost no trout + a lot of slow-growing charr

• Experimental fisheries 1984-1989: 666 000 charr removed

• Result: 30-fold increase in trout; stable 25 years after exp. fishing ended

• Explanation: exp. fishing decreased competition among charr, increased the amount of 10-15 cm charr → food for trout (and cannibalistic charr)

• The charr overcompensates for predation (survivors mature faster)

Charr

Trout

Persson et al., Science 2007

Charr length distribution

http://www.sciencemag.org/content/vol316/issue5832/images/large/316_1743_F1.jpeg


Can harvesting push marine ecosystems to alternative equilibria?

• Donald Strong: top-down control represents a form of biological instability caused by overfishing

• Ken Frank: overfishing is more likely to cause such changes in northern areas

• Lack of recovery in several Northwest Atlantic cod populations: overfishing has disrupted the predator's cultivation of its prey? (deRoos + Persson)

• In a pristine ecosystem, cod keeps its prey populations (pelagic fish) small and with high recruitment

• Overfishing: the pelagic fish gets an upper hand by competing with/predating on juvenile cod

• Closing of fisheries: the system stays in the new stable state

• Similar explanations suggested for the Baltic Sea and for the Black Sea


Multiple equilibria in the Newfoundland cod stocks

No or little recovery of several cod stocks since stock collapses in 1992 – in spite of closed fisheries

The ”cultivation-depensation” hypothesis:

Much piscivorous cod→ little forage fish

→ little competition+ predation effects on juv. cod

→ good cod recruitment

overfishingLittle piscivorous cod

→ much forage fish → juvenile cod heavily

affected by competition+ predation

→ bad cod recruitment















Species interactions & climate• External forcing, i.e. climate, extremely important in boreal ecosystems

• Predation can transfer climatic effects from one species to another

• Example: The recruitment of cod and herring in the Barents Sea are closely linked to climate of the spawing year

• Capelin recruitment is not much affected by climate of the spawning year

• …but is indirectly heavily affected by climate (with 1-4 years delay) through cod and herring


How does climate influence capelin biomass?Importance - and instability - of indirect pathways

Temperature

Cod Cod x Herr.

Capelin spawning stock

Herring

1981-1983

Dire

ct

Temperature

Cod Cod x Herr.


Herring

1984-1987

Dire

ct

Temperature

Cod Cod x Herr.


Herring

1988-1990

Dire

ct


Conclusions• Species interactions are very important and cannot be overlooked

• Predation effects are more visible in northern ecosystems, but only because there are fewer species?

• Strategies for modelling ecosystems: keep models simple?

• Predation can cause the existence of multiple equilibria – overfishing can cause transitions between equilibria


Thank you!

Photo:Getty Images

Dag Ø. Hjermann Centre of Ecological and Evolutionary Synthesis (CEES)

Institute of Biology, University of Oslo, Norway

Thanks to:

Geir Ottersen (Havforskningsinstituttet i Bergen)Nils Chr. Stenseth (CEES og Havforskningsinstituttet )

Mia Eikeset, Gjert Dingsør, Joel Durant (CEES)

Documents

Workshop on Ecosystem Based Fisheries Management – FAME, Esbjerg 27 March 2008 – slide 1 Fisheries and species interactions Photo:Getty Images Dag Ø. Hjermann