Transcript No Slide Title

"A Globally Coherent Fingerprint of Climate Change Impacts
across Natural Systems"
Friday, January 31, 2003
GARY W. YOHE
WESLEYAN UNIVERSITY
1
CAMILLE PARMESAN
UNIVERSITY OF TEXAS - AUSTIN
CIS oƒ HDGC
Carnegie Mellon
CIS oƒ HDGC
Carnegie Mellon
A Globally Coherent Fingerprint of Climate Change
Impacts across Natural Systems
Camille Parmesan & Gary Yohe
January 31, 2003
Center for Integrated Study of the Human
Dimensions of Global Change,
IPCC and Degrees of Confidence
Quantitative Scale:
3
» 95% or greater
Very High Confidence
» 67-95%
High Confidence
» 33-67%
Medium Confidence
» 5 – 33%
Low Confidence
» Less than 5%
Very Low Confidence
CIS oƒ HDGC
Carnegie Mellon
IPCC and Degrees of Confidence
Qualitative Scale:
Well Established – Lots of evidence; high consensus
Established but Incomplete – high consensus on
limited information
Competing Explanations – Lots of evidence;
alternative explanations
Speculative – Little evidence and many plausible
explanations
4
CIS oƒ HDGC
Carnegie Mellon
The IPCC Dynamic
“We have very high confidence that X might happen!”
“We have medium to low confidence that X will happen!”
Converged to the notion that the statements should
speak to the “will” alternative for a baseline.
5
CIS oƒ HDGC
Carnegie Mellon
Observed Changes in Physical and Ecological
Systems (from IPCC 2001)
hydrology /
glaciers
6
sea ice
animals
plants
study covers
large area
study based on
remote sensing
CIS oƒ HDGC
Carnegie Mellon
Key Conclusions from IPCC
Recent Regional Climate Changes, particularly Temperature Increases, have
Already Affected Many Physical and Biological Systems
(high confidence, or >67% sure)
Biotic change: 44 regional studies, 400 plants and animals, 20 to 50 years
Physical change: 16 regional studies, 100 processes, 20-150 yrs
»
»
»
»
»
»
»
7
non-polar glacier retreat
reduction in Arctic sea ice extent and thickness in summer
earlier plant flowering and longer growing season in Europe
poleward and upward (elevation) migration of plants, insects and animals
earlier bird arrival and egg laying
increased incidence of coral bleaching
increased economic losses due to extreme weather events
CIS oƒ HDGC
Carnegie Mellon
Figure 19-8-1: Summary of Lines of Evidence
Higher
Risks of large scale
discontinuities
Net Negative in All Metrics
Aggregate impacts
Observations
Very low
-0.7
Positive or Negative Monetary;
Most People Adversely Affected
Negative for
some regions
0
Past
Negative for Distribution
most regions of impacts
Increase
Large increase
Risk of extreme
weather events
Risks to Some
Risks to Many
Risks to unique &
threatened systems
1
2
3
4
Future
Increase in Global Mean Temperature after 1990 (°C)
8
5
CIS oƒ HDGC
Carnegie Mellon
The “Global Fingerprint” was a
“Reason for Concern”
The degree of confidence issue was contentious.
Lisbon authors’ meeting:
Chapter 2 – Tools discussion – how to judge?
Chapter 5 – Ecosystems – Of course this is “very high”
Confidence
Chapter 19 – Include in the burning ember, but with
confidence “medium” at best or “very high” ???
9
CIS oƒ HDGC
Carnegie Mellon
A Schematic Portrait of the
Problem
10
Study 1
Study 2
………………
Study n
Climate
Climate
Climate
Change
Change
Change
Impact
Impact
Impact
Non-
Non-
Non-
Climatic
Climatic
Climatic
Factors
Factors
Factors
CIS oƒ HDGC
Carnegie Mellon
A Thought Exercise
Let there be n separate studies.
Let n’ produce results that are contrarian with respect
to predicted climate impacts.
Let p be the probability that there are explanations
that compete with climate in any single study.
Let pi be the likelihood that climate was corrected
attributed as the cause in any study.
11
CIS oƒ HDGC
Carnegie Mellon
Contours for “More Likely than
Not”
Minimum Probability
1
0.75
Medium Confidence;
p=0
0.5
Medium Confidence;
p = 0.33
Medium Confidence;
p = 0.67
0.25
0
0
0.2
0.4
0.6
0.8
(n'/n) Proportion
12
CIS oƒ HDGC
Carnegie Mellon
Contours for Low and High
Confidence
Minimum Probability
1
Low Confidence;
p=0
High Confidence;
p=0
Low Confidence;
p = 0.33
High Confidence;
p = 0.33
Low Confidence;
p = 0.67
High Confidence;
p = 0.67
0.75
0.5
0.25
0
0
0.2
0.4
0.6
0.8
(n'/n) Proportion
13
CIS oƒ HDGC
Carnegie Mellon
Complications
• Common Drivers of Non-climatic Drivers – Raises the
contours.
• Studies that Show Sign Changes – Lowers the
contours.
• Publication Bias against Contrary or Insignificant
results – Increases the (n’/n) ratio.
14
CIS oƒ HDGC
Carnegie Mellon
New Analyses of changes from literature
• Studies with > 20 years data
• Primarily multi-species
- Counteracts publishing bias highlighting
only species which show significant change
• Mostly multi-site (moderate to large geographic
coverage)
• Conducted in nature reserve or rural natural area
- minimizes chances of confounding factors
(brings p closer to 1)
15
CIS oƒ HDGC
Carnegie Mellon
Biotic Changes are Systematically in accord with Climate Change
Predictions
Type of Analys is
Phenological
N = 484 / ( 678)
Distributional change s:
At pole w ard/upper range boundarie s
At equatorial/low e rrange boundarie s
Com munity (abundance) change s:
Cold-adapted spe cies
Warm-adapte d spe cies
N = 460 / ( 920)
Me ta-analys is
Range-boundarie s (n=99)
Phenologies (n=172)
Changed as
predicted
(n)
Changed oppos ite
to prediction
(n)
P
87 %
13 %
< .1 x10-12
81 %
75 %
19 %
25 %
74 %
91 %
26 %
9%
81 %
19 %
6.1 km-m/decade
northward/upward shift
2.3 d/decade advancement
< .1 x10-12
.013
< 0.05
Diverse species of: trees, herbs, shrubs, reptiles, amphibians, fish, marine
zooplankton & invertebrates, mammals, birds butterflies
16
(Parmesan & Yohe, Nature 2003)
CIS oƒ HDGC
Carnegie Mellon
Edith’s Checkerspot (Euphydryas editha)
17
CIS oƒ HDGC
Carnegie Mellon
52° N
Edith’s Checkerspot
butterfly:
Patterns of Population
Extinctions in natural
areas (good habitat)
48° N
44° N
40° N
36° N
% extinctions
> 70 %
32° N
35 - 55 %
< 20 %
18
CIS oƒ HDGC
Carnegie Mellon
Climatic Connections: USA
•
0.7º C warming over Western USA climatic shift of
105 km North & 105 m up
(Karl et al. 1996)
E. editha: mean location shifted
92 km North & 124 m up
•
Both snowpack & E. editha extinction trends
(Parmesan 1996)
shift at 2400 m:
% snow/50 yrs
»
»
Below 2400 m
Above 2400 m
14 % less snowpack ; melt 7 d earlier
8 % more snowpack ; no  melt
% extinctions
46 %
14 %
(T. Johnson, 1998)
19
CIS oƒ HDGC
Carnegie Mellon
Climate and E. editha : literature
Ehrlich, Evolution ‘65
Singer, Ph.D. dissertation Stanford ‘71
Singer, Science ‘72,
Singer & Ehrlich, Fortschritte der Zoologie ‘79
Parmesan, Ph.D dissertation ‘95
Boughton, Ecology ‘99
Ehrlich et al., Oecologia ‘80
Boughton, Ph.D dissertation ‘99
White & Levin, Amer. Midland Natur. ‘80
Weiss et al, Oecologia ‘93
MacKay Ecology ‘85
Foley et al.,
MacKay, Res. Pop. Ecol. ‘85
Singer, Evolution ‘83
Murphy & White, Pan. Pacific Entomol. ‘84
Singer & Thomas, American Naturalist ‘96
Thomas et al., American Naturalist ‘96
Dobkin et al., Oecologia ‘87
Parmesan, Nature ‘96
Weiss et al, Oikos ‘87
Hellman, book ch.
Weiss et al, Ecology ‘88
Fleishman et al. J. Res. Lep.
Moore, Ecology ‘89
Singer, in press
McLaughlin et al. 2000
20
CIS oƒ HDGC
Carnegie Mellon
ATTRIBUTION by INFERENCE
Example: Euphydras editha butterfly
•
Correlational Patterns
»
»
Long-term patterns (100 years) --- range shift matches temperature
isotherm shift and matches patterns of snowpack dynamics
(Parmesan 1996, Karl et al. 1996, Johnson 1998)
“natural experiments” (40 years) --- below 2400 m, population extinctions
occur in drought years and following false springs (light snowpack). Above
2400m, booms occur with heavy snowpack
(Singer & Ehrlich 1979, Singer & Thomas 1996, McLaughlin et al. 2002)
•
•
21
Field Manipulations
»
manipulating thermal environment (slope aspect, habitat type) affects
larval growth rates, pupal times, synchrony with host plant, and
colonization success
(Singer 1972, Weiss et al. 1988, 1993,, Boughton 1999)
Laboratory Experiments
»
temperature increases larval growth rates
(Weiss et al 1988, Hellmann 2000)
CIS oƒ HDGC
Carnegie Mellon
Thought Exercise revisited
p = Probability of competing explanations (confounding factors)
π = Probability that observed change is really due to climate (mechanistic link)
n’/n = Proportion of species going in opposite direction to climate change predictions
Binomial probability model with each factor varying from 0 to 1
Here, p=0
1
Confide nc e Re gions
0. 75
Ver y high
High

0. 5
Me dium
Low
Minimum
Probabilit y
( š)
es tima ted confidence
0. 25
fromliterature review
0
0
0. 2
0. 4
(n'/n) Pr opor tion
22
0. 6
0. 8
CIS oƒ HDGC
Carnegie Mellon
Diagnostic Biological Fingerprint
• Temporal
- Advancement of timing or northward expansion in warm
decades (1930s/40s & 1980s/'90s)
- Delay of timing or southward contraction in cool decades
(1950s/'60s)
• Spatial
Different behaviors at extremes of range boundary during particular climate phase,
e.g. expansion at northern range boundary simultaneous with contraction at
southern range boundary during warming period
• Community
Abundance changes have gone in opposite directions for cold-adapted vs. warmadapted species.
e.g. lowland birds increasing and montane birds decreasing at
mid-elevation site.
23
CIS oƒ HDGC
Carnegie Mellon
Diagnostic Biological Fingerprint
• “Sign-Switching” found for 294 species
»
»
»
80% of abundance shifts in communities
100% follow decadal trends in temperatures
(up & down)
100% show geographic contraction at equatorial boundary
coupled with expansion at poleward boundary of species range
• Increases confidence from either perspective
(Parmesan & Yohe 2003)
24
CIS oƒ HDGC
Carnegie Mellon
Conclusions
• We have high to very high confidence that regional climate
changes (resulting from global warming) have had impacts on
wild species
• Observed changes are typically small in magnitude, but are
likely to be an important factor in long-term persistence of
species and stability of ecosystems
25
CIS oƒ HDGC
Carnegie Mellon
More Conclusions from a Skeptical
Perspective
• The thought exercise allows an approach that
accommodates maximum skepticism.
• Even then, the Medium Confidence can be claimed.
• Indeed, Schneider’s “more likely than not” benchmark
is satisfied.
• Adding sign-switching adds to the power and moves
even a skeptical interpretation in the High Confidence
range.
26
CIS oƒ HDGC
Carnegie Mellon