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

Heat

Greenhouse gases

Heat

The increase in atmospheric carbon dioxide is primarily due to world energy consumption and secondarily due to deforestation.

World Energy 1850-2000

050

100150200250300350400450500

1850 1875 1900 1925 1950 1975 2000

YearE

J/ye

ar

GasOilCoalNuclearHydro +Biomass

Hundreds of thousands of years ago

400,000 Years of Atmospheric Carbon Dioxide DataA

tmos

pher

ic C

O2

(ppm

)

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

Temperatures During the Past Ice Age

oF oC

Thousands of years ago

Should we worry about +4 oC change?

O CO F

1000 Years of Northern Hemisphere Temperature Data

year

Fingerprint of Global Warming

•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

Models predict, and the data show that:

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

1000 2100

Year

Part 2: A global view of Climate-Ecosystem Feedbacks

Climate Change

Species Composition & Diversity

Ecosystem Functions & Biogeochemical Fluxes

Surface albedoaffected by plant cover

Greenhouse gas emissions

Soil carbon decay rates

Drought stress

Retreat of N. American Ice Sheet

The model with rock and silt surface predicts slow retreat

Rate predicted w/ rock surface

Rat

e of

retr

eat

(km

y-1

)

Surface Absorption(1-albedo)

Actualrate

Rock, Silta = 0.4

Retreat of N. American Ice Sheet

The model with spruce trees predicts actual rate

Rate predicted w/ rock surface

Rat

e of

retr

eat

(km

y-1

)

Surface Absorption(1-albedo)

Spruce Treesa = 0.1

Actualrate

Rock, Silta = 0.4

Vegetation effect

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)

Slide 16

WB3 Wendy Brown, 2003-10-13

biosphere

HOW DO WE QUANTIFY FEEDBACK?

O = I + gI + ggI + gggI + ...

= I / (1 - g) if g < 1

If g < 0: O < I, negative feedbackIf g > 0, g < 1: O > I, positive feedback, stableIf g > 1: unstable positive feedback

Gain Factor (g)

Output Signal (O)

(e.g., full warming effect)

Input Signal (I)

(e.g., direct warming effect of GHG increase)

g = Σ (∂T/∂pi)(∂pi/∂T)

e.g., p1(T) = albedo of land surface, which may change if warming induces a changes in dominant vegetation

pi

FEEDBACK (CONTINUED)g < 0: O < I, negative feedback g > 0: O > I, positive feedback g > 1: Unstable

Feedback process Gain factor (g)

In current GCMs:water vapor 0.40 (0.28 - 0.52)ice and snow 0.09 (0.03 - 0.21)clouds 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

0

5

10

15

20

25

30

-0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4

2×CO2 g = 0.7(no eco feedbacks)

Extra warming due to ecosystem feedbacks

Decrease in warming due to ecosystem feedbacks

Global Carbon Cycle

ΔT (ºC)due to

feedback

Change in gdue to ecosystem feedbacks

Small change in g causes large ΔT Asymmetries

M.S. Torn, LBNL

gCO2 = 12.04

)(32)(275

4 2

2

2

2

=•=∂

∂•

∂∂

CCOppm

COppmC

TCO

COT

o

o

An estimate of the carbon feedback from Vostok core data:

1oC/(1 - .71) = 3.4oC

1oC/(1 - .71 – .12) = 6oC !!!

y = 8.0913x + 266.55R2 = 0.7513

0

50

100

150

200

250

300

350

-10 -8 -6 -4 -2 0 2 4

delta(T)

CO2

(ppm

)

General Circulation Models

Holocene CO2-Temp coupling

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

Single big leaf

N=1

Biomes

Coarse Functional Groups

N ~ 10

Ecosystems

Functional Groups

N ~ 1000’s

Community Patch

Species assemblage

N ~ millions

Part 3: A “longterm” warming experiment

Rocky Mt. Biological Laboratory, Gothic, ColoradoInfra-red heaters (22 W m-2). Soil is warmer, drier; Earlier snowmelt

Warming Treatment Effect: Forb Production Decreases. Sagebrush Production Increases.

% Change in areal cover

Warming− Control( )Control

•100%

Forbs

Harte and Shaw, Science, 1995

Sagebrush

-30

-15

0

15

30

Delphinium nelsonii in John Harte's global warming experiment

Julian date of snowmelt in upper zone100 110 120 130 140 150 160

Num

ber o

f flo

wer

ing

plan

ts in

plo

ts

0

100

200

300

400

500

600

700

H-1994H-1995

H-1996

H-1997

H-1998

H-1999

H-2000

H-2001H-2003

H-2004

H-2005

C-1994

C-1995

C-1996

C-1997

C-1998

C-1999

C-2000

C-2001

C-2003

C-2004

C-2005

r2 = .470p = .0004

Inouye data

Feedback # 1: climate-induced change in species composition can alter late-spring surface albedo

Forbs Sagebrush

Alb

edo

(%)

A 20% change in regional plant cover will have an effect on local summertime climate that is

comparable to that of 2 x CO2

0

5

10

15

Darker plants cause warming

Met

hane

Upt

ake

(mg

CH

4m

-2d-1

)

Soil Moisture (%)

Feedback # 2: methane consumption influenced by soil moisture

negative feedbackpositive feedback

If warming → soil drying:

Torn and Harte, Biogeochemistry, 1995

SOIL ORGANIC CARBON

2

2.5

3

3.5

4

4.5

5

5.5

6

6.5

0 2 4 6 8 10

year since 1990

% s

oil c

arbo

n

Series1

Series2

Feedback # 3: warming can alter ecosystem carbon storage, and thus change atmospheric CO2

Question: What caused the decline in soil carbon?

Conventional answer: warming-induced enhancement of soil respiration (Post et al. 1982; Raich and Schlesinger, 1992; Schimel et al. 1994; Trumbore et al., 1996)

Heated plot decline converts to:

~200-400 g C m-2

(out of ~2300 total, 0-10cm)

Experimental warming

reduces soil carbon!

What caused the decline in soil carbon? NOT a heating-induced increase in soil

respiration

y = 0.1925x + 0.17252R2 = 0.92

0.5

1.5

2.5

0 5 10 15

temp * moisture (o C g H2O/gsoil)

resp

iratio

n, μ

g C

g-1

soil

hr-1

Soil drying and soil warming have opposite effects

(evidence from laboratory soil incubation)

Full 5x5 factorial :0, 10, 13,18, 30 deg. C2, 10, 20, 30, 40 %H2O

Thus,

as T ↑, decomposition ↑

as M↓, decomposition ↓

No overall effect!

Saleska et al., Global Biogeochemical Cycles, 2002

Upper

LowerMid

Heated

1000

1500

2000

2500

3000

4 5 6 7Mean Annual Temp (deg C)

Soil

Org

anic

Car

bon

(gC

/m2)

Control

Approach 1: Simple space-for-time does NOT work here

GradientStudy

WarmingExperiment

Effect of warming

temp differences across space

temp difference expected at future time, in same space

Dunne et al., 2004, Ecology

soil organic carbon, SOCi :

decomposition rate constants ki

Shrub Forb Grass

Pshrub Pforb Pgrass

kshrub·SOCshrub

kforb·SOCforb

kgrass·SOCgrass

Plant productivity Pi:

Litter inputs

Decomposition losses

Approach 2: A Simple Model of the local carbon cycle

CO2

Key insight: ki = ki(Temp, Moisture, litter quality)

chemical analysis of lignin; k ~ nitrogen content/lignin content

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]

. Under warming, Pforb ↓, Pshrub↑, Pgrass→

factorial jar experiment determines T, M surface

At fixed T and M, kforb > kshrub > kgrass

P = net annual photosynthetic production of litter to soil

k = soil carbon decomposition rate constant

2200

2000 3000 4000

2400

2600

SOC

(g C

m-2

)

2000 3000 4000 5000Predicted SOC

2000

B

uppermiddlelowerwarm contrl

contrlheat

AVGAVG

R2=0.84

Test of Simple Model for ambient Soil Organic Carbon

Predicted SOC

The equilibrium solution to the model accurately predicts ambient soil carbon levels across a climate gradient.

It fails to predict the transient response of carbon levels to heating.

So let’s look at the dynamic solution to the differential equations…

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

Years since warming

% S

oil C

rem

aini

ng

80%

90%

100%

110%

0 5 10 15 20 25

litter quality effect

litter quantity

effect

we were here at time of prediction

Forb litter input to soil decreased more than sage litter input increased

Sage litter has higher lignin/N content than forblitter

Heated plots

Control plots

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

2

2.5

3

3.5

4

4.5

5

5.5

6

6.5

0 5 10 15 20

year since 1990

% s

oil c

arbo

n

control plots

heated plots

carbon loss during 5 “naturally” dry years

carbon loss due to heating

snowmelt date

80

110

140

170

0 5 10 15

year since 1990

cale

ndar

day

of

snow

mel

t

average contro l plo ts

average heated plots

DRYWET

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

Our model suggests that plant community composition/productivity controls soil carbon;

Above-Ground Biomass (AGB) of forbs and shrubs drives the model.

Effect on Carbon

turnover

Response to Climate

Medium lignin:N

Lower lignin:N

Shallow rooted(sensitive to drought)

Forb:Erigeron

speciosus

Forb:Delphinium nuttallianum

Deep rooted(less sensitive to drought)

Forb:Ligusticum

porteri

Forb:Helianthella

quinquinervis

Species matter! Response to climate vs. effect on soil carbon turnover

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 (influences shifts in community composition) between populations

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.