Water Dynamics: The Experimental
Perspective
Alan K. Knapp
Graduate Degree Program in Ecology, Colorado State
University
Overview:
Experimental approach to:
Climate Change - Ecosystem Function (in grasslands)
• Soil moisture & ANPP (4 studies)
• Change in precipitation pattern (long-term)
• Comparison among grasslands (short-term)
Grasslands and similar ecosystem types cover
~ 40% of the Earth’s land surface
Most are water limited systems
Most of what we know is based on observation
Rainfall vs. Production across the Central US
Rainfall regimes
+-
CO2 effects
+
Temperature effects
Sala et al. 1988
Soil moisture
-
Organism
Water relations
Ecosystem
Process &
Function
Inputs
Outputs
Soil moisture
N
C
Central US grasslands provide a template for experiments at
sites that span key precipitation and temperature gradients…
“Seeing the grasslands beyond the grasses”
Mesic Grasslands of North America
The Flint Hills of Kansas – The Konza Prairie
KPBS
C4 grass dominance
A dynamic & tractable system
• Tallgrass prairie is the most mesic of US Great Plains
grasslands – sufficient precipitation to support forest
• Flint Hills are located at the dry edge of the original
range and on shallow soils
Abundant evidence that
rainfall regimes are
changing
Oklahoma
a.
14.0
r2=0.534
(We know amount is important)
12.0
Rain Intensity (mm/event)
Question:
16.0
b.
Given increased variability and frequency
of
12.0
extreme events, to what extent will altered
10.0
precipitation patterns affect ecological
processes
r =0.704
in “mesic” grasslands?
2
8.0
c.
8.0
Year
7.0
r2=0.638
6.0
Proportion of total rainfall in the US from large (>5
cm) rainfall events
1940
1950
1960
1970
Year
1980
1990
2000
The Rainfall Manipulation
Plot (RaMP) Experiment
Address the impact of changes in
size and timing of growing
season rainfall events and
elevated temperature
 12 Rainfall Manipulation Plots
(RaMPs) + non-sheltered
controls
 Collect, store, and reapply
natural rainfall on intact prairie
plots
 Treatments include ambient
and altered rainfall patterns
since 1998
 Elevated temperature treatment
added in 2003
Target altered rainfall pattern: 50% longer
inter-rainfall periods, larger individual rain events
Precipitation amount
Precipitation
amount
Current
Current
Predicted
Predicted
Increased dry periods
Greater storm size
Fewer events
Day
Dayof
ofyear
year
No change in
total
precipitation
quantity
Soil moisture dynamics
A - Ambient rainfall treatment
Soil water content 0-30 cm (%)
Variability in soil
moisture:
- increased by 27%
Depth (cm)
40-50
60-70
80-90
40
0
10
20
30
40
50
Root mass (% of total)
30
20
10
B - Altered rainfall treatment
Mean SWC = 24.7 ± 1.4%
50
40
10.0
7.5
5.0
2.5
0.0
30
20
10
Jul
Aug
Date
100
80
60
40
20
0
P = 0.01
Ambient Altered
Jun
100
80
60
40
20
0
Precipitation (mm)
Average soil water
content in top 30 cm:
- reduced by 12%
50
20-30
Precipitation (mm)
Altered ppt timing:
• repeated deficits
• more extreme wetting
and drying cycles
Mean SWC = 29.5 ± 1.2%
0-10
Change in SWC (%)
per sampling interval
Ambient ppt timing:
• “typical” seasonal
pattern
Sep
Oct
ANPP responses to altered rainfall timing
1000
900
*
*
12-22% reduction
in years with
significant
responses
800
700
*
600
400
11%
reduction
across years
(p=0.01)
pre-treatment
300
1998
2000
2002
2004
Year
-2
500
ANPP (g m )
-2
ANPP (g m )
*
ambient timing
altered timing
2006
700
600
500
400
300
200
100
0
Amb
2008
Alt
Across all years and all rainfall treatments
1000
Ambient timing
Altered timing
-2
ANPP (g m )
900
2
800
r = 0.598
700
600
500
400
20
30
40
50
CV of soil water content (0-15 cm)
If only the entire world worked like Konza….
Can we apply results from mesic grasslands to
other climates and grasslands?
A multi-system assessment of climate change responses
Short-grass
Tallgrass
Ojima and
Lackett 2002
Mixed-grass
Precipitation
gradient
Heisler-White et al. 2009. Global Change
Biology.
500
KPBS
tallgrass prairie, KS
ANPP (g/m2)
400
300
200
HAYS
mixed grass prairie, KS
SGS
shortgrass steppe, CO
100
0
358
578
833
Annual Precipitation (mm)
Semi-arid grassland
Extreme rainfall events,
long dry intervals =
reduced ANPP?
With the same experimental
design, how will these different
grasslands respond?
Mesic grassland
Extreme rainfall events,
long dry intervals =
reduces ANPP
Experimental manipulation of growing season rainfall regimes
Frequent, small events vs. infrequent large events
Equal rainfall amounts within each site
A = 12 events
B = 6 events
C = 4 events
Total = 191 mm
SGS
A = 12 events
B = 6 events
C = 4 events
Total = 450 mm
KNZ
Semi-arid
Mesic
Semi-arid steppe
191 mm
30% increase in
ANPP with fewer
large events
Mixed grass prairie
340 mm
70% increase in
ANPP with fewer
large events
Tallgrass prairie
450 mm
15% decrease
in ANPP with
fewer large
events
Heisler-White et al. 2009
Two key aspects of more
extreme rainfall regimes:
Current
Precipitation amount
Why do arid vs. mesic
grasslands respond
differently?
Predicted
Increased dry periods
Greater storm size
1. Larger rain events
2. Longer dry periods between events
Day of year
Fewer events
When soils are usually moist, longer dry periods negatively impact
ANPP…
When soil are usually dry, larger rain events improve soil water and
increase in ANPP…
The Sevilleta LTER – the edge of the Great Plains
Monsoon Rainfall Manipulation Experiment
During the summer monsoon July through September:
3 plots receive ambient rainfall
5 plots receive a 5mm rainfall event every week
5 plots receive a 20 mm rainfall event every month
Scott Collins
Michell Thomey
Rodrigo Vargas
Jennifer Johnson
Renee Brown
Don Natvig
Mike Friggens
0.10
a
-1
c
0.08
b
a
a
0.06
0.04
b
b
a
0.02
a
a
-2
Control
Small events
Large events
-3
0.00
2008
2008
c
b
30
-2 -1
(µmol CO2 m s )
2007
ab
b
25
b
c
a
1.0
b
b
a
20
0.8
a
a
0.6
15
10
0.4
5
0.2
0
0.0
2007
2008
Predawn water potential
(MPa)
a
2007
Anet leaf
0
b
0.12
2007
Soil respiration
(µmol CO2 m-2 s-1)
Seasonal soil water content
(m3 m-3)
Seasonal average soil water content, predawn water potentials, leaf-level Anet and
soil CO2 flux were highest in plots receiving a few large events
2008
Responses similar to shortgrass steppe, but opposite mesic grasslands
Thomey et al. 2011 GCB
Total ANPP and ANPP of black grama were highest in plots
receiving a few large events
• 25 – 40% greater ANPP of black grama with few large
events compared to many small events
• Similar trends for ANPP of all species combined
Thomey et al. 2011 GCB
Responses to drought - is there a
pattern/amount interaction?
Mixedgrass
Tallgrass
Shortgrass
SGS-LTER
(North)
Mean Annual Mean Annual
Soil Texture
Precipitation Temperature
Site
Elevation
Shortgrass-Steppe
LTER
(SGS)
1650 m
310 - 330 mm
7.2º C
(45º F)
Sandy Loam
Sand Creek
Massacre Nat'l
Historic Site
(SAND)
1219 m
330 - 360 mm
8.3º C
(47º F)
Clay
Fort Union Nat'l
Monument
(FOUN)
2043 m
410 - 450 mm
9.4º C
(49º F)
Sandy Clay
Loam
Sand Creek (Central)
Fort Union (South)
Karie Cherwin
These are observational data!
From Huxman et al (2004)
Testing this inference of high sensitivity
• 30 plots at each site
– 10 controls
– 20 rainout shelters
• 10 at 50% ReducedPPT
• 10 at 80% ReducedPPT
• Two-year drought
• Response Variables
– Net Primary Productivity
– Soil Moisture
• No change in timing/pattern
• Reduces event size & total amount
Growing season mean soil moisture
0.25
South
Central
North
a
Volumetric Soil Water Content
a
0.20
a
Control
-50%
-80%
b
b
b
c
0.15
a
a
c
0.10
a
b
c
b
b
c
c
c
0.05
0.00
SGS - 07
SGS - 08 SAND - 07 SAND - 08 FOUN - 07 FOUN - 08
Site and Year
2007 & 2008 ANPP
200
North
Central
South
180
160
Control
- 50%
-80%
ANPP (g/m2)
140
120
a
a
b
100
b
80
x
b
x
60
y
b
40
20
0
SGS '07
SGS '08
SAND '07 SAND '08 FOUN '07 FOUN '08
Site and Year
Why the differential sensitivity?
220
Shortgrass-steppe
200
180
High
Sensitivity to Drought
ANPP g/m2
160
140
120
Within biome
saturation?
100
80
60
220
40
0.20
200
0.25 180
Volumetric Soil Water Content (%)
160
0.05
0.10
0.15
ANPP g/m2
20
0.00
Sand Creek
140
Insensitive ?!?!
120
100
80
60
40
20
0.05
220
0.10
0.15
0.20
0.25
Volumetric Soil Water Content (%)
Low
Fort Union NM
180
160
ANPP g/m2
2007
2008
200
140
120
100
80
60
40
20
0.00
0.05
0.10
0.15
0.20
Volumetric Soil Water Content (%)
Latitude:
Temperature:
Precipitation:
Elevation:
Soil Texture:
40° N
38° N
35° N
7.2° C
8.3° C
9.4° C
330 - 360 mm
410 - 450 mm
1219 m
2043 m
310 - 330 mm
1650 m
Coarse
(Sandy Loam)
Fine
Intermediate
(Clay)
(Sandy Clay Loam)
0.25
Mean rain events across sites
20
10
Wettest years
Mean precip = 409.4 mm (574.0 mm)
Mean # days with rain = 40.2 (10.3 mm)
Mean # rain periods = 23.5 (11.9 mm)
Mean # days between rain = 7.2
2
20
0
Average years
10
4
Mean precip = 273.4 mm (373.3 mm)
Mean # days with rain =38.4 (7.2 mm)
Mean # rain periods = 23.1 (11.9 mm)
Mean # days between rain = 9.0
Rain days
Rain periods
2
20
0
Driest years
10
Mean precip = 133.3 mm (189.7 mm)
Mean # days with rain = 23.1 (5.9 mm)
Mean # rain periods = 15.9 (8.6 mm)
Mean # days between rain = 16.9
4
2
0
05- 5
10 1 0
15 1 5
20 2 0
25 2 5
30 3 0
35 3 5
40 4 0
45 4 5
50 5 0
55 5 5
60 6 0
65 6 5
70 7 0
-7
5
75
+
Number of events (May - September)
4
Event size (mm)
Rainfall regimes – there’s more
to them than amount…
In addition to more rain, wet
years have more large events
compared to average years.
Fewer, large events
(wet, high ANPP)
50
# of events
Event size
33.6 events
9.7 mm
Many, small events
(dry, low ANPP)
41.0 events
5.3 mm
20
15
30
10
20
5
10
0
0
SAND '07 FOUN '08 FOUN '07
SGS '08
SGS '07
Insensitive <---------------------> Sensitive
SAND '08
Mean Event Size
Number of Rainfall Events
40
What does all this mean?!?!
There is still much we need to learn about
precipitation regimes, soil moisture and ecosystem
responses if we are to forecast the future… Because
all three are likely to change.
• Much of what we know is from observation and
correlation… We need to challenge this knowledge
• Alterations in rainfall patterns (and of course
amounts) may change productivity in biome
specific ways…
• Alterations in rainfall patterns may even alter
sensitivity to drought…
As Ecologists – one of our greatest challenges is to provide
knowledge to help us forecast and cope with future ecosystem
structure and function…
The empirical perspective - future field experiments?
We need more long-term experiments where we “push”
ecosystems… in both magnitude and pattern…
We need comparative analyses of ecosystem responses to
climate change within and across biomes…
Clear need for long-term, multi-site and coordinated
experiments… and these should be designed with modelers…
Divergent responses are linked to differential effects of
extreme rainfall events on mean soil water content
↑ SWC
by 19%
in arid site
↓ SWC
by 20%
in mesic
site
Grasses and forbs respond to different aspects of soil
water dynamics
Response
Predictor variable
Partial r2
Model r2
Total ANPP
CV SWC 15cm
0.598 (92%)
0.651
mean SWC 15cm
0.053
mean SWC 15cm
0.470 (70%)
mean SWC 30cm
0.103
CV SWC 30cm
0.103
mean SWC 30cm
0.375 (82%)
CV SWC 30cm
0.078
Grass ANPP
Forb ANPP
0.676
0.453
• Grasses most responsive to mean surface SWC, while
forbs respond more to deeper SWC
• Potential for long-term changes in community
composition