Technological change, environmental variability, and fish

stock collapses

Rögnvaldur Hannesson

Norges Handelshøyskole

Points to be raised

• Technical change increases catch per unit of effort (cpue)

• It may also make cpue less sensitive to stock size

• If cpue used as stock index, this will make stock assessment overoptimistic

• With environmental variability, this may precipitate stock collapses

The Norwegian winter herring fishery as a case in point

• Two major technical changes in the 1960s

• Power block– Mechanical hauling of nets– Bigger nets, bigger boats

• Sonar– Possible to see fish below the surface

• CPUE increased and may have become less dependent on stock

Gear a b g Constant R2 D-W

Land seine(1909-1959)

.8214(3.89**)

-.4959(1.17)

.0229(1.02)

3.2389(0.96)

0.4854 1.7589

Gill nets(1909-1968)

1.0905(10.05**)

.6362(4.61**)

.0334(9.02**)

-9.6377(12.34**)

0.9101 1.1964

Purse seine(1925-1971)

.6538(2.52*)

.4220(2.99**)

.0411(3.15**)

-3.6188(2.39*)

0.6000 1.0989

ln ln ln lnY A a E b S gt b insignificant for land seine, 0 < b < 1 for gill nets & purse seine.Lopping off the years after 1964 raises b and the t-value, butlopping off still more years in the 1950s & 60s lowers b andmakes it insignificant.Shaky evidence that b fell after 1964, may have been low allalong.a 1

Gear a b Constant R2

Land seine(1909-1959)

1.1624(3.12**)

.1077(0.08)

.0457(0.24)

0.1790

Gill nets(1909-1968)

.780956(4.31**)

1.187175(4.09**)

.0230722(0.36)

0.6204

Purse seine(1925-1971)

-.1531(0.48)

1.0871(3.60**)

.1214(1.10)

0.2496

ln ln lnY g a E b S b 1 for gill nets and purse seine, b 0 for land seine.a 0 for purse seine.

Purse seine

0

5000

10000

15000

20000

25000

1909 1919 1929 1939 1949 1959 1969

Sto

ck (

'000

mt)

00.20.40.60.811.21.41.6

Stock

Catch/man

Gill nets

0

5000

10000

15000

20000

25000

1909 1919 1929 1939 1949 1959 1969

Sto

ck (

'000

mt)

0

0.1

0.2

0.3

0.4

0.5

0.6

Stock

Catch/man

Land seine

0

5000

10000

15000

20000

25000

1909 1919 1929 1939 1949 1959 1969

Sto

ck (

'000

mt)

0

0.5

1

1.5

2

2.5

3

Stock

Catch/man

There does not seem to be astrong correlation betweenstock and cpue for purseseine even before 1964, butmore so for gill nets.

Catch per fisherman (logs, share of stock)

0

2

4

6

8

1019

09

1914

1919

1924

1929

1934

1939

1944

1949

1954

1959

1964

1969

Gill nets

Purse seine

Trawl

Define /ln ln 1 lnY E

x A b SS

ln lnln 1

dx d A db d SS b

dt dt dt dt

Technical progress & fall in b: firstterm > 0

b = 1, x constantTechnical progress,A risesS was falling after ’65,falling b may havecontributed to rising x

Stock

0

5000

10000

15000

20000

25000

1925

1928

1931

1934

1937

1940

1943

1946

1949

1952

1955

1958

1961

1964

1967

The breakthrough of the purse seine in thewinter herring fishery

Indices of catch per fisherman and boat (log, share of stock) and share of purse seiners with power block

0

0.2

0.4

0.6

0.8

1

1955

1957

1959

1961

1963

1965

1967

1969

1971

Seiners

Catch/man

Catch/boat

-3000

-2000

-1000

0

1000

2000

3000

4000

0 5000 10000 15000 20000

S

G

21 1 1 1t t t t tG X S aS bS

G surplus growth, S stock left after fishing, X stock beforefishing

ln ln ttD S

ˆt t tD G G G observed G estimated

tt

t

DU

e S environmental disturbance

0 1 1 ...t t t n t nV U a aU a U random environmentaldisturbance

a0 a1 R2

0.0050(0.03)

0.4578(4.14**)

0.2254

Further autocorrelation terms insignificant

0

24

6

8

1012

14

16

1.95

2.07

2.19

2.31

2.43

2.55

2.67

ln(v+10)

Fre

qu

ency

0

24

6

8

1012

14

16

Frequency

Normal

Transformed V lognormally distributed

0

5000

10000

15000

20000

25000

30000

0 10 20 30 40 50 60 70 80 90 100

Time

Sto

ck (

'000

to

ns)

Sample paths of stock with no fishing, Note possibilities oftrendless minor fluctuations, quasi-regular major cycles,and total collapse.

The Bohuslän herring collapsed in the 1500s, the Balticherring a bit earlier.

Stock development with fishing

• Stock managed by target escapement with S = Smsy

• X forecast by cpue as index (implies b = 1)• Technical change happens in year 10

– Once and for all rise in A– Once and for all fall in b and simultaneous rise in A

• These effects never discovered– Managers continue using cpue as if nothing

happened

0

2000

4000

6000

8000

10000

12000

14000

0 10 20 30 40 50 60 70 80 90 100

Time

Sto

ck (

'000

to

ns)

No change

Change in A

Change in b

0

2000

4000

6000

8000

10000

12000

14000

0 10 20 30 40 50 60 70 80 90 100

Time

Sto

ck (

'000

to

ns)

No change

Change in A

Change in b

0

2000

4000

6000

8000

10000

12000

14000

0 10 20 30 40 50 60 70 80 90 100

Time

Sto

ck (

'000

to

ns)

No change

Change in A

Change in b

Change in b clearly moredangerous than change inA only, but collapse can alsobe precipitated by the latterif learning is slow.