Global Warming
and the daunting challenge of
Climate-Ecosystem Feedbacks
The University of Toronto
October 20, 2005
John Harte
University of California, Berkeley
With graduate students:
Scott Saleska, Marc Fischer, Jennifer Dunne, Margaret Torn, Becky Shaw,
Michael Loik, Molly Smith, Lara Kueppers, Karin Shen, Perry deValpine, Ann
Kinzig, Fang Ru Chang, Julia Klein
and undergraduates:
Tracy Perfors, Liz Alter, Francesca Saavedra, Susan McDowell, Brian
Feifarek, Hadley Renkin, Chris Still, Laurie Tucker, Agnieska Rawa, Vanessa
Price, Julia Harte, Wendy Brown, Susan Mahler, Erika Hoffman, Jim Williams,
Eric Sparling, Jennifer Hazen, Jim Downing, Sheridan Pauker, Billy Barr,
Kathy Northrop, Ann and Dave Zweig, Kevin Taylor, Aaron Soule, Andrew
Wilcox, Mike Geluardi, Annabelle Singer, Sarah McCarthy
And with Financial Support from NSF, DOE, USDA, NASA, USEPA,
Outline of this talk
1.A quick overview of global warming
2. A global view of climate-ecosystem
interactions and feedbacks
3. A local view: results from the RMBL
meadow-warming experiment
Sunlight
Greenhouse
gases
Heat
Heat
EJ/year
World Energy 1850-2000
500
450
400
350
300
250
200
150
100
50
0
Gas
Oil
Coal
Nuclear
Hydro +
Biomass
1850 1875 1900 1925 1950 1975 2000
Year
The increase in atmospheric carbon dioxide is primarily due to
world energy consumption and secondarily due to deforestation.
Atmospheric CO2 (ppm)
400,000 Years of Atmospheric Carbon Dioxide Data
Hundreds of thousands of years ago
What is the effect on global temperature of doubling
the atmospheric concentration of carbon dioxide?
The direct effect of heat absorption by the CO2:
+ 1 oC
The indirect (feedback) effects:
+ 0.5 to 3.5 oC
•
melting ice and snow increases absorption of sunlight (ice-albedo effect)
•
warmer air holds more water vapor, another greenhouse gas
•
warmer air results in different cloud characteristics
TOTAL:
+ 1.5 to 4.5 oC
Should we worry about +4 oC change?
Temperatures During the Past Ice Age
oF
oC
Thousands of years ago
1000 Years of Northern Hemisphere Temperature Data
O
F
OC
year
Fingerprint of Global Warming
Models predict, and the data show that:
•Stratosphere cools as surface warms
•Temperature rises faster at night than day
•Temperature rises faster in winter than summer
•High latitudes warm more than low latitudes
If global warming were caused by a
brightening sun, then the stratosphere would
warm and temperature rise would be
greatest in daytime
IMPACTS OF GLOBAL WARMING
• Threats to Food Production (Diminished Water Supplies)
• Human Health Impacts (Heat Waves, Infectious Disease)
• Wildfire
• Sealevel Rise
• Ecological Effects:
Extinction Episode Comparable to K-T Boundary
Spread of Invasive Species
Coral Bleaching
Future energy policy will determine this
This warming has already occurred
2100
1000
Year
Part 2: A global view of
Climate-Ecosystem Feedbacks
Greenhouse
gas emissions
Climate
Change
Ecosystem Functions &
Biogeochemical Fluxes
Soil carbon
decay rates
Drought stress
Surface albedo
affected by plant
cover
Species Composition
& Diversity
Retreat of N. American Ice
Sheet
Rate of retreat
(km y-1)
The model with rock and silt surface
predicts slow retreat
Actual
rate
Rate
predicted w/
rock surface
Rock, Silt
a = 0.4
Surface Absorption
(1-albedo)
Retreat of N. American Ice
Sheet
Rate of retreat
(km y-1)
The model with spruce trees predicts actual rate
Actual
rate
Rate
predicted w/
rock surface
Rock, Silt
Spruce Trees
a = 0.4
a = 0.1
Surface Absorption
(1-albedo)
The Vostok core suggests
feedback
Milankovitch Cycles are the
time keeper but their
magnitude is too weak to
explain the huge climate
variability
CO2 release during slight
warming must cause more
warming!
And CO2 uptake during slight
cooling must cause more
cooling.
This effect is not incorporated in our current climate models (GCM’s)
HOW DO WE QUANTIFY FEEDBACK?
Output Signal (O)
Input Signal (I)
(e.g., direct warming
effect of GHG increase)
biosphere
(e.g., full warming effect)
Gain Factor (g)
g = p (T/pi)(pi/T)
i
O = I + gI + ggI + gggI + ...
= I / (1 - g) if g < 1
If
g < 0: O < I, negative feedback
If g > 0, g < 1: O > I, positive feedback, stable
If g > 1: unstable positive feedback
e.g., p1(T) = albedo of
land surface, which may
change if warming
induces a changes in
dominant vegetation
FEEDBACK (CONTINUED)
g < 0:
g > 0:
g > 1:
O < I, negative feedback
O > I, positive feedback
Unstable
Feedback process
Gain factor (g)
In current GCMs:
water vapor
ice and snow
clouds
0.40 (0.28 - 0.52)
0.09 (0.03 - 0.21)
0.22 (-0.12 - 0.29)
total
0.71 (0.17 - 0.77)
Climate change: ∆T=forcing effect/(1-g)
∆T = 1oC/(1 – 0.71) = 3.oC
Global Carbon Cycle
Small change in g causes large T
Asymmetries
30
Decrease in warming due
to ecosystem feedbacks
25
Extra warming due to
ecosystem feedbacks
20
T (ºC)
due to 15
feedback
10
2CO2 g = 0.7
(no eco feedbacks)
5
0
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
Change in g
due to ecosystem feedbacks
M.S. Torn, LBNL
An estimate of the carbon feedback from Vostok core data:
350
Holocene CO2-Temp coupling
300
CO2 (ppm)
250
General
Circulation
Models
200
150
y = 8.0913x + 266.55
R2 = 0.7513
100
50
0
-10
-8
-6
-4
-2
0
2
4
delta(T)
gCO2 =
CO2
32 ppm(CO2 )
T
4o C



 0.12
o
CO2
T
275 ppm(CO2 )
4 C
1oC/(1 - .71) = 3.4oC
1oC/(1 - .71 – .12) = 6oC !!!
But where is the carbon coming from?
How can we learn about climate-ecosystem feedback?
1.
Ecological correlations across different climates
•
natural climate variability in space (latitudinal, altitudinal)
•
natural inter-annual variability of climate
•
multi-decadal ecological trends synchronous with global warming trend
•
paleoclimatic variability, combined with pollen records and other
ecological reconstructions
2.
Climate manipulation experiments, with control,
allowing deduction of causal mechanisms
3.
Mathematical models
1.
Applicable to large spatial scales, but potentially misleading.
2.
Confined to plot-scale, but capable of identifying mechanisms.
3.
Only as good as the observations!
POSSIBLE LEVELS OF AGGREGATION
IN GLOBAL MODELS
Planet
Biomes
Single big
leaf
Coarse Functional Functional
Groups
Groups
N=1
N ~ 10
Ecosystems
N ~ 1000’s
Community Patch
Species
assemblage
N ~ millions
Part 3: A “longterm” warming experiment
Rocky Mt. Biological Laboratory, Gothic, Colorado
Infra-red heaters (22 W m-2). Soil is warmer, drier; Earlier snowmelt
Warming Treatment Effect:
Forb Production Decreases.
Sagebrush Production Increases.
Delphinium nelsonii in John Harte's global warming experiment
Number of flowering plants in plots
700
C-2005
600
C-1995
C-2004
C-1994
500
C-1998
H-2005
C-1996
C-1999
400
C-2003
H-1994
H-1997
300
200
C-1997
C-2001
H-1996
H-1998
H-2004
H-2003
H-2001 H-1999
100
Inouye
data
H-1995
C-2000
r2 = .470
p = .0004
H-2000
30
0
100
110
120
130
140
150
160
Julian date of snowmelt in upper zone
% Change in areal cover
Warming
 Control 
100%
Control
15
Forbs
0
-15
Sagebrush
-30
Harte and Shaw, Science, 1995
Feedback # 1: climate-induced change in species
composition can alter late-spring surface albedo
Albedo (%)
15
Darker
plants
cause
warming
10
5
0
Forbs
Sagebrush
A 20% change in regional plant cover will have
an effect on local summertime climate that is
comparable to that of 2 x CO2
Methane Uptake (mg CH4 m-2 d-1)
Feedback # 2: methane consumption
influenced by soil moisture
If warming → soil drying:
positive feedback
negative feedback
Soil Moisture (%)
Torn and Harte,
Biogeochemistry, 1995
Feedback # 3: warming can alter ecosystem carbon
storage, and thus change atmospheric CO2
Heated plot decline
converts to:
Experimental warming
~200-400 g C m-2
(out of ~2300 total, 0-10cm)
reduces soil carbon!
SOIL ORGANIC CARBON
Question: What caused
the decline in soil
carbon?
6.5
6
% soil carbon
5.5
5
Conventional answer:
warming-induced
enhancement of soil
respiration
4.5
4
3.5
Series1
3
Series2
2.5
2
0
2
4
6
year since 1990
8
10
(Post et al. 1982; Raich and Schlesinger, 1992;
Schimel et al. 1994; Trumbore et al., 1996)
What caused the decline in soil carbon?
 NOT a heating-induced increase in soil
respiration
y = 0.1925x + 0.1725
2.5
2 2
R
respiration, m g C g
-1soil
hr
-1
Soil drying and soil warming
have opposite effects
(evidence from laboratory soil incubation)
Full 5x5 factorial :
0, 10, 13,18, 30 deg. C
2, 10, 20, 30, 40 %H2O
= 0.92
1.5
Thus,
0.5
as T ↑, decomposition ↑
as M↓, decomposition ↓
0
5
temp * moisture
(o C g H2O/gsoil)
10
15
No overall effect!
Saleska et al., Global Biogeochemical Cycles, 2002
Soil Organic Carbon (gC/m2)
Approach 1: Simple space-for-time does NOT work here
3000
Lower
2500
Mid
Gradient
Study
Control
Heated
2000
Warming
Experiment
Upper
1500
temp differences
across space
temp difference
expected at future
time, in same space
Effect of
warming
1000
4
5
6
7
Mean Annual Temp (deg C)
Dunne et al., 2004, Ecology
Approach 2: A Simple Model of the local carbon cycle
Shrub
Forb
Grass
Litter inputs
Plant
productivity Pi:
Pshrub
Decomposition
losses
Pforb
Pgrass
CO2
soil organic
carbon, SOCi :
decomposition
rate constants ki
kgrass·SOCgrass
kforb·SOCforb
kshrub·SOCshrub
Can a simple mathematical model help make
sense of this space-time mismatch?
Let Ci = soil carbon derived from vegetation-type i,
where i = forb, shrub, grass
C = ΣiCi
dC/dt = ΣidCi/dt = Σi [Pi – kiCi]
P = net annual photosynthetic
production of litter to soil
.
k = soil carbon decomposition rate
constant
Under warming, Pforb ↓, Pshrub ↑, Pgrass →
Key insight: ki = ki(Temp, Moisture, litter quality)
factorial jar experiment determines T, M surface
chemical analysis of lignin;
k ~ nitrogen content/lignin content
At fixed T and M, kforb > kshrub > kgrass
Test of Simple Model for ambient Soil Organic Carbon
B The equilibrium solution to
SOC (g C m-2)
2600
the model accurately
predicts ambient soil carbon
levels across a climate
gradient.
R2=0.84
2400
It fails to predict the transient
response of carbon levels to
heating.
2200
2000
upper
middle
lower
warm contrl
2000 3000 4000
Predicted SOC
So let’s
look at the
dynamic
contrl
AVG
solution
heatto the differential
AVG
equations…
2000 3000 4000 5000
Predicted SOC
Saleska et al., Global Biogeochemical Cycles, 2002
Dunne et al., Ecology, 2004
Change in species distribution causes transient loss
and long-term gain of Soil C in heated plots
Forb litter input to
soil decreased
more than sage
litter input
increased
litter
quantity
effect
% Soil C remaining
110%
Sage litter has
higher lignin/N
content than forb
litter
litter
quality
effect
100%
Control plots
Heated plots
90%
we were here
at time of
prediction
80%
0
5
10
15
Years since warming
20
25
Something surprising has occurred in the past 5 years:
The heated and control plots carbon levels are converging as predicted,
but not because the heated plots have recovered.
Starting in ~ 2000, the control plots are losing soil carbon, at ~ 1/3 the rate
the heated plots did in 1991-1994!
SOIL ORGANIC CARBON
carbon loss
during 5
dry years
6.5
snowmelt date
“naturally”
average control plots
average heated plots
6
calendar day of
snowmelt
170
% soil carbon
5.5
5
4.5
4
carbon
loss due
to heating
3.5
DRY
140
110
80
control plots
3
heated plots
2.5
2
0
WET
5
10
year since 1990
15
20
0
5
10
year since 1990
15
Our model suggests that plant community composition/productivity controls soil carbon;
Above-Ground Biomass (AGB) of forbs and shrubs drives the model.
AGB in the control
plots in 1999-2004
resembles
AGB in the heated
plots in 1993-1997
Thus soil carbon in
the control plots in
99-04 behaved like
soil carbon in the
heated plots in 93-97
Species matter!
Response to climate vs. effect on soil carbon turnover
Effect on
Carbon
turnover
Response
to Climate
Shallow
rooted
(sensitive to
drought)
Deep rooted
(less sensitive
to drought)
Medium
lignin:N
Lower
lignin:N
Forb:
Forb:
Erigeron
speciosus
Delphinium
nuttallianum
Forb:
Forb:
Ligusticum
porteri
Helianthella
quinquinervis
What is a sensible set of plant
traits/categories?
Albedo
Contribution to NPP and thus carbon input to soil
Lignin:N of foliage
Sensitivity to climate change (e.g., rooting depth)
Transpiration rate
LAI or Shading of soil beneath plant
Canopy roughness
Categories: for each we might consider 3 levels: high, medium, low
Thus have 36 = 729 plant categories.
Just based on on the RMBL experience: the first 4 above or 34 = 81
The warming meadow contains about 75 plant species!
The feedback linkages in the meadow are very complex.
Yet we have not even begun here to look at:
•Other Habitats
(tundra, desert, savannah, temperate forests,
boreal forests, tropics, freshwater, marine …)
•Larger Spatial Scales
(emergent phenomena?)
•Animals
(grazers, pollinators, …)
•Other Climate Characteristics (extreme events, …)
•Carbon Dioxide increase
(water efficiency, growth stimulation)
•Nitrogen Deposition
(N addition can increase carbon storage)
•Land Use Changes
(albedo, water exchange …)
•Invasive Species
(carbon storage, albedo, water exchange)
•Genetic differences
between populations
(influences shifts in community composition)
Summary
•Analysis of the long-term climate record suggests that strong
positive feedback operates in Earth’s climate system.
•An ecosystem warming experiment provides further evidence
for mechanisms that induce strong feedback responses.
•Current climate models do not incorporate these feedback
effects and therefore are likely to be underestimating the
magnitude of future warming.
•Developing a global-scale understanding of the sign and
strength of these feedbacks is a huge challenge.