Response of Fish Population
Changes to Climate Events
James Overland and Nick Bond
First Concept - Climate
• Decadal Climate Variability more “Eventlike” than regular Oscillations
• Large and long deviations from averages
• Default Model is Stochastic Red NoiseAR1 to Long Memory; Little deterministic
predictability
Arkhangel’sk
Iceland
SW Greenland
Tornedalen
2.0
Nor th Atla nti c
Ocea n
0.0
-2.0
Temp. Anomaly
(°C) (°C)
SAT anomaly
ET CW
1800
1825
1850
1875
1900
1925
1950
1975
2000
Year
Year
Extended annual mean SAT record for the northern North Atlantic region (TNA). (Wood, et al.
2010).
NO 60-80 Year Oscillations in early portion of record.
Sub-arktisk vinterklassifikation för de senaste 500 åren
0
-2
-4
-6
-8
-10
420
3
2
Maximum ice extent (x10 km )
o
Temperature ( C)
2
360
300
240
180
120
60
0
1500
1550
1600
1650
1700
1750
1800
1850
1900
1950
Perioder
1522-36
1562-76
1577-91
1597-1629
1630-62
1663-1706
1707-50
1750-1877
- 1803-20
1878-nu
- 1930-40
- 1940-42
- 1971-75
- 1985-87
2000 - 1988-93
Mild
Kall
Mild
Kall
Mild
Kall
Mild
Kall
Kall
Mild
Mild
Kall
Mild
Kall
Mild
Year
Eriksson et al, in press
No Regular Climate “Cycles”
Havsklimatgruppen
www.oceanclimate.se
3
Smoothed Air Temperature St Paul Island, Bering Sea
0
Butterworth
EEMD C6+R
-3
1900
1920
1940
1960
1980
2000
Low-pass EEMD analysis with a low-pass Butterworth filter applied to the St.
Paul air temperature time series. (From Overland et al., 2012)
2020
Can Fit a square oscillator to
Pacific Decadal Oscillation (PDO) timeseries to give “multiple stable states”
BUT: Other simple times series models without multiple stable states
(Stochastic Red Noise AR1 and Long Memeory)
also fit the PDO data equally well
CONCLUSION: Cannot determine underlying process model from
data alone for records shorter than 200 years,
CANNOT Deterministically Project, only use Statistical Scenarios
Overland et al. (2006)
George Sugihara:
“time series observations of key physical variables for the North
Pacific Ocean…are best described as linear stochastic. In
contrast, we find that time series for biological variables having
similar properties exhibit a low-dimensional nonlinear
signature*.”
*Chih-hao Hsieh, Sarah M. Glaser, Andrew J. Lucas & George
Sugihara, Distinguishing random environmental fluctuations
from ecological catastrophes for the North Pacific Ocean,
Nature 435, 336-340 (2005)
(E) Embedding dimension/Degrees of freedom: how many
independent ways a system can vary
Distinguishing random environmental fluctuations from ecological catastrophes for the North Pacific Ocean
Chih-hao Hsieh, Sarah M. Glaser, Andrew J. Lucas & George Sugihara
Nature 435, 336-340(19 May 2005)
Second Concept: Ecosystems/Fish Populations
The responses to climate shifts by biological systems are direct, but
diverse because intervening processes introduce amplifications, time
lags, hysteresis, and non-linearities, leading a variety of climate to
ecosystem transfer functions
But northern North Pacific Ecosystems have a limited
number of main interactions for each major species
Aydin
Coyle et al.
Gaichas and Francis
Southern Bering Sea Ecosystem Changes
Warm temperatures previously
favored pollock over Arctic species.
But recently, poor prey in very warm period,
more predators gave a biomass loss of 50%
Now recovery with cold temperaturesbut nearly a biological regime shift
2000
2001
Photo by W.B. Miller
ca. 1920, Alaska
Library of Congress
Cod in the
Bering Sea
1870s- 1930s
SUMMARY:
Northern North Pacific Ecosystem Response to Climate Variability
Temperature
C
Stochastic Climate Variability
on all time scales
Low Dimension
Ecosystems
Conclusions
1) In the future we can expect large excursions in the climate system that last
for multiple years, but there is as yet little predictability for when they will
occur or how long they will last. -Low Frequency Red Noise Stochastic
System with long AVERAGE return periods
2) Transitions mechanisms appear more like individual multi-year events rather
than multi-decadal oscillations. Minimal decadal (PDO) memory processes.
3) Biology shows regime shifts in response to random climate forcing. Northern
species develop to take advantage of this stochastic structure of extremes. A
few dominant species, patchiness, multi-year life spans with highly variable
year classes.