The Effect of Climate Change on Arid and Semi-Arid Ecosystems:

Directional Changes in Precipitation Amount and

Variability

Osvaldo Sala

Projected Precipitation Change

1970-99 vs 2071-99

US National Climate Assessment 2014

Projected Changes in Precipitation

Precipitation Variability is projected to increase

IPCC. 2013. Climate Change 2013: The physical science basis. Contribution of working group I to the fifth assessment report of the Intergovernmental Panel on Climate Change. Cambridge, United Kingdom and New York, NY, USA.

Grassland <-> Shrubland

Hypothesis 2 Directional changes in water availability that favor grasses over shrubs or shrubs over grasses are reinforced through time.

Jornada LTER VI Project

Our scientific approach

Observations Experiments

Data Mining Modelling

Observation Multiyear Precipitation trend

+

Peters et al (2011)

Consequences of Multiyear Precipitation trend

+

Peters et al (2011)

Experimentation

2014

ambient

+ 80%

- 80%

Solar panel

Battery

Pump

Intermediary tank

(55 gal.)

Float switchInterception

plot

Irrigation plot

Filter

ARMSautomated rainfall

manipulation system

(Plots trenched to 40-60 cm) Gherardi and Sala, Ecosphere: 2013

• Assess ecosystem sensitivity to precipitation

• Not to replicate climate change scenarios

Experimental Design Objectives

CCImpact = ʄ (Δ Climate, Ecosystem Sensitivity)

PPTImp= ʄ (Δ PPT, Ecosystem Sensitivity to PPT)

Climate Change Impact

Total Aboveground Net Primary Production

+ 80%

Ambient

- 80 %

Grass Aboveground Net Primary Production

+ 80%

Ambient

- 80 %

Shrub Aboveground Net Primary Production

Ambient

+ 80 %

- 80 %

Direct and Indirect Effects of Precipitation

Plant-Species Diversity H’

+ 80%

Ambient

- 80 %

There was an effect of time on ecosystem response variables to long-term changes in PPT

The effect of time varied for different response variables

Asymmetry Hypothesis – The absolute magnitude of the effect was different for increasing or decreasing PPT

Conclusions

Effects of Enhanced Precipitation Variability

10 reps * 5 treat = 50 plots

(2.5 x 2.5 m)

Trenched 60 cm deep

20 rainout shelters

20 irrigated plots

10 control plots

Methods

Methods

Effect of PPT Variability

Gherardi & Sala PNAS 2015

The Mechanism

Gherardi & Sala PNAS 2015

How do we explain these responses?

Sala, Gherardi, Peters Climatic Change 2015

Modelling

Gherardi & Sala PNAS 2015

Effect of PPT Variance Increases through Time

Demise of grasses under high PPT variability favors shrubs

Further explore the existence of thresholds ◦ Cumulative endogenous ◦ Stochastic exogenous◦ Interaction between endogenous and exogenous

Mechanisms for indirect effects

Future Studies

Thank youLaureano Gherardi, Lara Reichmann Courtney Currier Kelsey DuffyOwen McKennaJosh Haussler

Pulse and Press

Collins et al 2011

Smith MD et al (2009)

Press Conceptual Model

Hypothesis 1

a) Ecosystem response variables are proportional to water availability

increased water

ambient

decreased water

Time

Eco

syst

em

resp

onse

vari

able

Response variable = b0 + b1*PPT

0 +

+

Hypothesis 1

b) Ecosystem response variables are proportional to changes in water and to the time that the ecosystem has been exposed to the new condition

increased water

ambient

decreased water

TimeEco

syst

em

resp

onse

vari

able

Response variable = b0 + b1*PPT + b2*Time

+

+

-

0

Multiyear PPT trend

0 +

+

Peters et al (2011)

Consequences of Multiyear PPT trend

0 +

+

Peters et al (2011)

Hypothesis 1

c) Acclimation / exhausting of resources

in-creased water

am-bient

TimeEco

syst

em

resp

onse

vari

able

Response variable = b0 + b1*PPT + b2*Time + b3*Time*PPT

0 +

+

-

Hypothesis 2 The effect of time is asymmetric for reduced

water and increased water

increased water

ambient

decreased water

TimeEco

syst

em

resp

onse

vari

able

0 +

+

-

Hypothesis 3 The effect of time varies for different response variables

response variable A

response variable B

TimeEco

syst

em

resp

onse

vari

able

0 +

+

-

Solar panel

Battery

Pump

Intermediary tank

(55 gal.)

Float switchInterception

plot

Irrigation plot

Filter

ARMSautomated rainfall

manipulation system

(Plots trenched to 40-60 cm) Gherardi and Sala, Ecosphere: 2013

Total ANPP

Shrub ANPP

Grass ANPP

Spp Richness

Diversity

PPT

Conclusions Rejected H1a. There was an effect of

time on ecosystem response variables to long-term changes in PPT, due to legacies in the ecosystem response.

Asymmetry – The absolute magnitude of the effect was different for increasing or decreasing PPT, i.e. spp loss

with drought – no spp change with increased PPT

The effect of time varied for different response variables; may depend of the number of actors involved or the flow size relative to the pool size

Thank youLaureano GherardiLara ReichmannOwen McKennaJosh HausslerKelsey Duffy

Jose Anadon

NSF-Division of Environmental Biology

Jornada Basin LTER

Jornada Experimental Range - USDA

School of Life Sciences - ASU

AcknowledgmentsLara G. ReichmannR.C.A. GuchoOwen P.B.R. McKennaLaura YahdjianDeb PetersKelsey McGurrinJosh HausslerJohn Angel IIIShane & MiriamG. A. Gil

Funding sources

Contrasting productivity responses to interannual precipitation variability

Laureano A. Gherardi and Osvaldo E. SalaArizona State University, School of Life Sciences

Results of 6 years of precipitation manipulation at the Jornada Basin LTER

Projected Changes in Precipitation

Precipitation Variability is projected to increase

IPCC. 2013. Climate Change 2013: The physical science basis. Contribution of working group I to the fifth assessment report of the Intergovernmental Panel on Climate Change. Cambridge, United Kingdom and New York, NY, USA.

Objective: To study the effect of inter-annual precipitation variability per se

Interannual Precipitation Variability

ANPP Mean

ANPP CV

Productivity

Stability

NSF-Division of Environmental Biology

Jornada Basin LTER

Jornada Experimental Range - USDA

School of Life Sciences - ASU

AcknowledgmentsLara G. ReichmannR.C.A. GuchoOwen P.B.R. McKennaLaura YahdjianDeb PetersKelsey McGurrinJosh HausslerJohn Angel IIIShane & MiriamG. A. Gil

Funding sources

ANPP, PPT, Space

ANPP = - 45.13 + 0.67*MAP r2= 0.76

Bai et al (2008)

ANPP=-34+0.60*MAPr2=0.94

Sala et al (1988)

ANPP=-30+0.47*MAP

McNaughton et al (1989)

Gre

at P

lain

s

Sou

th

Am

eric

a

Mon

goli

an

Pla

teau

Message A simple model accounts for a large fraction

of ANPP variability across space and for most grasslands of the world

0 500 1000 15000

200

400

600

800

1000

Annual Precipitation (mm)

Temporal ModelSpatial Model

Ab

ove

gro

un

d N

et P

rod

uct

ion

(g

/m2 /

yr)

Lauenroth and Sala 1992 Ecological Applications 2:397-403

-34 + 0.60*MAPr2 = 0.94, p < 0.001

56 + 0.13*MAPr2 = 0.39, p < 0.001

Spatial vs. temporal models of net primary production

Sala et al 2012, Philosophical Transactions of the Royal Society B

Spatial vs. Temporal models of net primary production

r2=0.39

Sala et al 2012, Philosophical Transactions of the Royal Society B

Message Time and Space cannot be exchanged for

the ANPP-MAP relationship

Spatial model does not work through time

Temporal model only accounts for a small fraction of the variability explained by spatial models and has shallower slope

Hypothesis

Differences between spatial and temporal models are explained by time lags in ecosystem response to changes in water availability

Legacies Time lags result from legacies of wet and

dry years

ANPP observed = F (PPTt, Legacy)

Legacies = ANPP observed – ANPP expected

◦ ANPP expected = F (PPTt)

Magnitude of Legacy= F (PPTt-1 – PPTt)

Global patterns of LegaciesMagnitude of Legacy= F (PPTt-1 – PPTt)

What is the shape of F?

How does this relationship change across a PPT gradient?

Legacy Symmetry Hypotheses

Sala et al 2012 PTRSB

Legacy Symmetry Hypotheses

Sala et al 2012 PTRSB

H 3.2

Knapp and Smith (2001)

Effect of previous-year PPT on ANPP across sites

Sala et al 2012, PTRSB

Effect of current year PPT on ANPP

Sala et al 2012, PTRSB

Effect of previous-year PPT on ANPP across sites

Sala et al 2012, PTRSB

Effect of previous-year ANPP on current-year ANPP across sites

Sala et al 2012, PTRSB

Effect of current- and previous-year PPT along a PPT gradient

Sala et al 2012, PTRSB

Experimental ApproachChihuahuan Desert Grassland

Jornada LTER

MAP 240 mm Dominant

species:◦ Bouteloua

eriopoda C4◦ Prosopis

glandulosa C3

Jornada Experimental RangeChihuahuan Desert Grassland

Experimental design

Fixed rainout shelters intercept different amounts of rain, depending on the number of shingles

irrigation

Water was added to the increased PPT treatments after each PPT event, year around

Total 132 plots

Legacy Magnitude

Reichmann, Sala, Peters, Ecology 2013

Legacy = -2.71 + 0.05 * ∆PPTR2 = 0.42

Rainout shelters in the Patagonian steppe

Yahdjian and Sala (2002)

Precipitation input (mm/year)

AN

PP

(g.

m-2.y

r-1)

PPT mm/year

50

70

90

110

130

20 60 100 140 180 220

without drought legacy

after 80% rainfall interception

after 55% rainfall interception

after 30% rainfall interception

Yahdjian and Sala (2006)

Drought legacies in the Patagonian Steppe

Conclusion Changes in precipitation result in legacies

Magnitude of Legacies is a function of difference in precipitation of current and previous year

Legacies in the Chihuahuan desert ecosystem are symmetrical

◦ │ Positive legacy │ = │ Negative legacy│

Corollary

Positive legacies would compensate negative legacies

Increased precipitation variability would not affect average productivity

Hypotheses for the Legacy Mechanisms Structural mechanism

◦ Meristem density constrains production response to a wet year after a dry year

◦ Meristem density enhances production after wet years

Biogeochemical mechanism◦ N limitation constrains production response to a wet

year after a dry year

◦ Abundant reactive N enhances production after wet years

Soil moisture carry-over

Structural mechanism

Reichmann, Sala, Peters, Ecology 2013

Structural mechanism

Reichmann, Sala, Peters, Ecology 2013

(Reichmann and Sala, Functional Ecology 2014)

Structural mechanisms

Structural mechanisms

(Reichmann and Sala, Functional Ecology 2014)

Biogeochemical mechanism

Reichmann, Sala, Peters, Ecology 2013

Biogeochemical mechanism

N mineralization effect

Reichmann et al Ecosphere 2012

N uptake and leaf N concentration

Reichmann et al Ecosphere 2012

N stocks

Reichmann et al Ecosphere 2012

Test of the soil-moisture carry-over hypothesis

• Tiller density determines magnitude of

legacies

• Biogeochemical mechanisms do not

determine legacies

• Soil water carry-over does not determine

legacies

Conclusions

Pulse and Press

Collins et al 2011

Present Future

+

Pre

cipi

tati

onTime

Hypothetical pattern

Observational

Short term manipulations

Most studies are

Central question

Can we predict press effects of directional changes in precipitation amount and variability based upon our understanding of pulse responses?

Smith MD et al (2009)

Proposal Hypothesis 2 a

Hypothesis 1

a) Ecosystem response variables are proportional to precipitation

Response variable = b0 + b1*PPT

Increased precipitation

Ambient

Decreased precipitation

0 +

+

Eco

syst

em

resp

onse

var

iabl

e

Time

b) Ecosystem response variables are proportional to changes in precipitation and to the time that the ecosystem has been exposed to the new condition

Hypothesis 1

Response variable = b0 + b1*PPT + b2*Time

Increased precipitation

Ambient

Decreased precipitation

0 +

+

Eco

syst

em

resp

onse

var

iabl

e

Time

c) Acclimation / exhausting of resources

Hypothesis 1

Response variable = b0 + b1*PPT + b2*Time + b3*Time*PPT

Increased precipitation

Ambient

Decreased precipitation

0 +

+

Eco

syst

em

resp

onse

var

iabl

e

Time

The effect of time varies for different response variables

Hypothesis 2

Increased precipitation

Ambient

Decreased precipitation

0 +

+

Eco

syst

em

resp

onse

var

iabl

e

Time

The effect of time is asymmetric for reduced and increased precipitation

Hypothesis 3

Increased precipitation

Ambient

Decreased precipitation

0 +

+

Eco

syst

em

resp

onse

var

iabl

e

Time

Central question

Can we predict press effects of directional changes in precipitation amount and variability based upon our understanding of pulse responses?

Individual Species Response + 80%

- 80%

Species Richness

+ 80%

Ambient

- 80 %

Mechanism

Linear and non-linear ANPP responses to

precipitation

Increased precipitation variance implies a higher frequency of extremely dry and wet years

1.Linear and non-linear ANPP responses to annual precipitation

Precipitation

An

nu

al A

NP

P (

g m

2 yr

-1)

Therefore, NULL precipitation variance effect on ANPP.

Linear ANPP responses to precipitation result in negative effects of dry years equal to positive effects of wet years.

Precipitation

An

nu

al A

NP

P (

g m

2 yr

-1)

Non-linear ANPP responses to precipitation result in different effects of dry and wet years.

Therefore, POSITIVE precipitation variance effect on ANPP.

Precipitation

An

nu

al A

NP

P (

g m

2 yr

-1)

Non-linear ANPP responses to precipitation result in different effects of dry and wet years.

Therefore, NEGATIVE precipitation variance effect on ANPP.

Precipitation

Interannual Precipitation Variability

ANPP Mean

ANPP CV

Productivity

Stability

Plant-functional types show different stability response to PPT variability

Ecosystem stability results from the aggregated response of plant types

Functional diversity increases with PPT variability

Changes in relative abundance support such effect

How do we explain these responses?

Concluding summary: effects on ANPP mean

1. Inter-annual precipitation variability itself has a negative effect on ANPP

2. Non-linear responses and changes in soil water distribution explain such effect

3. Aggregated plant-functional type responses determine overall ecosystem response

Concluding summary: effects on stability

1. Interannual precipitation variability has a positive effect on ANPP CV

2. Contrasting plant-functional type responses result in relative abundance change and increased diversity

3. Aggregated response of plant-types determines ecosystem stability