1. The study investigated ecosystem responses to drought conditions across sites in Australia and North America using satellite data from 2000-2009.
2. Results from the USDA and Australian sites were compared to previously published data from 14 Long-Term Ecological Research sites from 1980-1999 to examine productivity and water use efficiency over prolonged warm drought periods.
3. The study found convergence across biomes to a common rain-use efficiency, indicating a cross-biome sensitivity to prolonged warm drought conditions. The results provide insights into how carbon and water relationships may shift with increased drought frequency and intensity from climate change.
TERN Ecosystem Surveillance Plots Kakadu National Park
Alfredo Huete_The role of TERN in studies of ecosystem resilience across large-scale altered hydroclimatic conditions
1. The role of TERN in studies of ecosystem
resilience across large-scale altered
hydroclimatic conditions
Guillermo E. Ponce-Campos1
Susan M. Moran1
Alfredo Huete2,3
Derek Eamus2,3
Acknowledgements
Tim McVicar4
Randall Donohue4
Alex Held3,4
Kevin Davies, Natalia Restrepo-Coupe, Mark Broich2,3
!
Bureau of Meteorology
!
(1) USDA-ARS
!
(2) University of Technology, Sydney !
(3) TERN AusCover and OzFlux !
(4) CSIRO
!
TERN 4th Annual Symposium, Canberra 20 February 2013
2. Introduction
• Recent large-scale, warm droughts have occurred in
Australia, China, North America, Amazonia, Africa, and
Europe, resulting in dramatic changes in vegetation
productivity across ecosystems with direct impact on
societal needs, food security and basic livelihood and water
balance, and food security.
Extreme Australian temperatures alter face of country's heat map
Temperatures to soar past 50 C
http://www.cbc.ca/news/technology/story/2013/01/08/sci-aussie-heat-map.html
GEOPHYSICAL RESEARCH LETTERS, VOL. 39, L03404, doi:10.1029/2011GL050263, 2012
Tropical cyclones and the ecohydrology of Australia’s recent
continental-scale drought
Gavan S. McGrath,1 Rohan Sadler,2 Kevin Fleming,3,4 Paul Tregoning,5
Christoph Hinz,1,6 and Erik J. Veneklaas7
Received 8 November 2011; revised 20 December 2011; accepted 22 December 2011; published 9 February 2012.
[1] The Big Dry, a recent drought over southeast Australia, [Ummenhofer et al., 2009; Smith and Timbal, 201
3. • As Earth’s climate continues to change, the frequency and
intensity of warm droughts, extreme precipitation patterns,
and heat waves will alter in potentially different ways,
ecosystem functioning and productivity with major impacts on
3. Comparison of iEVI relative difference of years with similar annual precipitation but different
% (years in the Low and High groups, respectively; iEVI difference= (RUEHigh- RUELow)/
w*100)) across 11 sites. For each site, the years with similar annual precipitation were selected to
carbon
re the iEVI differences in the two groups. The inset shows the average iEVI difference combined
ome types. Different letters indicate significant differences at P < 0.05."
What is an extreme climate event?
All extremes are relative to some expectation. An ex- ing will probably become more frequent in
treme climate event is one that has appeared only rarely as warmer conditions mean some snowfall
in the historical record, which goes back about 100 storms in the Sierra Nevada converts to rai
years. For example, a 1-in-100 year flood is an extreme snow on the ground melts earlier in the yea
event, as is a three-day heat wave that is hotter than 95%
of all previous 3-day heat waves. Together, the frequency and intensity of we
ANPP
make up a distribution. The well-known be
As Earth’s climate continues to change, the climate example of a distribution. Extreme events a
extremes we experience will alter in potentially different fall on the ends of the distribution. One of
ways. The intensity could change, or the frequency (or climate science is to understand how the d
both). climate events is likely to change in the futu
Some extremes could become more intense. Intensity Understanding how the frequency and inte
refers to how different the climate extreme is from extremes changes in the future has implica
normal conditions. For instance, as the climate warms, we could adapt to those changes. For instan
MODIS Satellite EVI
heat waves will likely become hotter than any seen since ing becomes more intense (a larger volume
measurements began. ter), bigger flood control channels may be
flooding becomes more frequent, perhaps
On the other hand, some extremes could change their channels needed to drain roads that inconv
frequency, which is to say, how often they occur. Flood- flood during heavy rains would be needed
Old
Probability of occurrence
Climate
4. Relation of production across precipitation gradients for 11 sites for two groups (Low: R95p% <
Zhang et al 2013, Biogeosciences
High: R95p% ≥ 20%). See Table 2 for R95p% definitions. The relations were significantly different
two groups (F2, 106 = 18.51, P < 0.0001).
New
climate
34" More frequent
heat waves
1" Extreme precipitation patterns reduced terrestrial ecosystem production across
2" biomes Cold Average Hot
More intense
heat waves
As the climate changes, the distribution of events such as heat waves and
1 1 1 1
3" Yongguang Zhang *, M. Susan Moran , Mark A. Nearing , Guillermo E. Ponce Campos , Alfredo R. floods will change. Extreme events are those on the tails of the distribution,
and could change in their intensity (for example, how hot a heat wave is) or
4" Huete2, Anthony R. Buda3, David D. Bosch4, Stacey A. Gunter5, Stanley G. Kitchen6, W. Henry McNab7, their frequency (how often the event occurs). After IPCC (2001), Fig. 2.32.
5" Jack A. Morgan8,Mitchel P. McClaran9, Diane S. Montoya10, Debra P.C. Peters11, Patrick J. Starks12
4. • Understanding water and productivity relationships are key
issues in models that aim to predict how carbon and water
relationships will shift with projected changes in the frequency,
timing, amount and intensity of rainfall.
• The hydro-meteorological conditions that recently impacted N.
America and Australia are of the same order to those expected with
climate change, and thus offer an opportunity to investigate changes
and generalize vegetation responses to future climate change scenarios.
• “Natural experiments” have great power to study rainfall
variability and vegetation response.
LETTER doi:10.1038/nature11836
Ecosystem resilience despite large-scale altered
hydroclimatic conditions
Guillermo E. Ponce Campos1,2, M. Susan Moran1, Alfredo Huete3, Yongguang Zhang1, Cynthia Bresloff2, Travis E. Huxman4,
Derek Eamus3, David D. Bosch5, Anthony R. Buda6, Stacey A. Gunter7, Tamara Heartsill Scalley8, Stanley G. Kitchen9,
Mitchel P. McClaran10, W. Henry McNab11, Diane S. Montoya12, Jack A. Morgan13, Debra P. C. Peters14, E. John Sadler15,
Mark S. Seyfried16 & Patrick J. Starks17
5. Objectives
• Investigate cross-biome productivity and
vegetation functional responses to recent
contrasting hydro-meteorological conditions in
Australia and the Americas,
• Investigate rain use efficiencies (RUE) and water
use efficiencies (WUE) across biomes and over
prolonged warm drought periods,
• Conceptualize satellite- monitoring schemes for
ecosystem threshold and resilience
6. er nitrogen and light will influence ANPP more strongly. However,
n
7
ng ; both in locations with high MAP and in those with low MAP, water
Rainfall use efficiency (RUE) concept
mit
availability is tightly linked to biogeochemical constraints through
an
i-
mineralization processes and leaching20. Precipitation affects both
o-
nutrient availability through its effects on microbial activity and
ngn
or
al
rs
h
er
ty
18
nt.
ms a
ANPP g/m2
st,
ng
be
i-
17
he;
on
re
7
yP;
ps it
i-
r-
d
on
m
al
th
h
ty
i-
nt
of
ms Figure 1 Between-year variation in production across a precipitation gradient and a
V,
letters to nature
ng maximum rain-use efficiency. a, Plot of ANPP against PPT for 14 sites (see Methods for
or
i- ..............................................................
I, abbreviations). Multi-year data give site-specific relationships by using linear regression
s, for S. paradoxus versus S.
lizards.
ty
038/nature02597. Convergence across biomes to
he (see Supplementary Information). The overall relationship (bold line) derives from data
*ANPP difficult to measure
d a common rain-use efficiency
re from all sites: ANPP ¼ 1011.7 £ (1 2 exp(20.0006 £ precipitation)); r 2 ¼ 0.77;
spectives (eds Woods, C. A. & Sergile,
e of the Antillean insectivoran
PP P , 0.001. The inset shows theD.site-level slopes ,(ANPP plotted against precipitation) as a
of 1 2,3 4
Travis E. Huxman *, Melinda Smith *, Philip A. Fay Alan K. Knapp ,
m. Mus. Novit. 3261, 1–20 (1999).
e Caribbean region: implications for 6 7 8
5
9
M. Rebecca Shaw , Michael E. Loik , Stanley D. Smith , David T. Tissue ,
m- function of MAP:C.ANPP ¼ Weltzin , William T. Pockman , Osvaldo E. Sala ,precipitation)); r2 ¼ 0.51;
*First ‘fun’ test MODIS EVI data
ps
999).
John Zak , Jake F. 0.388 £ (1 2 exp(20.0022 £
s. Annu. Rev. Ecol. Syst. 27, 163–196
9
7
10
13 14
11 12
15
Brent M. Haddad , John Harte , George W. Koch , Susan Schwinning ,
r-
n- P , 0.001. b, AnSmall & David G.max derived from the slope of the minimum precipitation and
Eric E. overall RUE Williams
l history of Solenodon cubanus. Acta 16 17
ld
he the correspondingand Evolutionaryall sites (solidArizona, Tucson, Arizona 85721, þ 0.42 £ PTTmin. Closed
the Dominican Republic. 1–128,
Ecology ANPP for Biology, University of line): ANPP ¼ 86.1
1
re circles, minima; open circles, remaining2004 California95% confidence intervals.
Huxman et alSynthesis, Santa Barbara, lines,
USA
mhy: molecular evidence for dispersal
89, 1909–1913 (1992).
2
National Center for Ecological Analysis and data; dotted
93101, USA
Mus. Nat. Hist. 115, 113–214 (1958).
ar Arrows showConnecticut 06511, USAand for sites Biology, Yale University, New Haven, high precipitation.
average slopes Evolutionary with low, medium and
3
Department of Ecology
Species Level (Columbia Univ. Press,
th
(ed. Benton, M. J.) 117–141 (Oxford 4
Natural Resources Research Institute, Duluth, Minnesota 55811, USA
Department of Biology, Colorado State University, Fort Collins, | 10 JUNE 2004 | www.nature.com/nature
5
NATURE | VOL 429 Colorado 80523,
Ponce et al ISRSE, Sydney 2011
ure Publishing Group
i-
ntals (eds Szalay, F. S., Novacek, M. J.
USA
6
7. 1800
iEVI as proxy for ANPP 1600 y = 192.65x - 155.29
R = 0.8578
Mean Annual GPP (g C m-2)
1400
1200
1000
800
600
400
200
0
0 2 4 6 8 10
Mean Annual iEVI
FOREST DESERT-GRASSLAND
377" " SAVANNA SHRUB
OPEN FOREST SAVANNA WOODY SAVANNA
Fig. 3 Graph shows the technical scheme to derive the phenological metrics based on double logistic fitted EVI time
series (solid line) and corresponding curvature change rate (dashed line). The light grey area is the integral of annual
EVI subtract the integral of Base EVI, which is used as surrogate for grass layer productivity (Pg). The dark grey area is
the integral of annual Base EVI, which is used as surrogate for the woody layer productivity (Pw). The annual total
productivity (Pt) is the sum of Pw and Pg.
378" "
379" Figure 1.
380" Ponce-Campos et al 2013, Nature
8. The experimental sites encompass a range of precipitation regimes and capture similar biome
Location of study sites (with known in-situ field
types on both continents (Table A1). Results for USDA and Australia sites were compared with
previously-published results based on a dataset composed primarily of Long-term Ecological
Research (LTER) sites covering 14 sites with measurements made during the period from about
information)
1980-1999, hereafter referred to as the LTER dataset (Table A1; Figures A1-A3).
• In N America, LTER, LTAR
sites and on-site
meteorological and in- situ
data available co-located.
67° W 66° W
Luquillo (LU)
!
(
18° N 18° N
Source:
NOAA (http://www7.ncdc.noaa.gov)
67° W 66° W
NRCS (http://www.wwc.nrcs.usda.gov/climate)
Figure A1. Location of the USDA experimental sites with mean annual precipitation. • Use NVIS and TERN for key
Biome types
• Use of Google-Earth to assess
similar types but close to BOM
HQ stations. Check MODIS
1 time series for any LCC/
disturbances
Ponce et al (2013) Nature
9. This indicates a cross-biome aerosol-contaminated pixels and observations
remove low-quality, cloud- and sensitivity to prolonged warm drought for the Australia sites.
made at large sensor zenith angles (.30u). The retained high-quality pixels were The PDSI is a measure of drought and wet spells, in which PDSI 5 0 is normal,
a 1,200 LTER75–98 USDA00–09 Australia01–09
averaged to represent the EVI 2 = 0.95 that2site0.77 16-day period, resulting in a
value for R = and 23 is moderate drought, 24 is extreme drought, and excess precipitation is
R2 = 0.87 R
10-year EVI time series forUSDAsite. Australia01–09
a 1,200 LTER0.001 each 0.001 P < 0.05 represented by a positiveof plant production for the time period from
The response PDSI. We obtained the PDSI to precipitation
P < 75–98 P < 00–09 1980 to 2009 to identify the average drought conditions across the USDA, LTER
the 0.95 R2 = TIMESAT23 to smooth the quality-
The next step2was to useR2 = software tool0.77
R = 0.87
andduring the prolonged warm drought showed strong
1,000
assurance-filtered time series<data and P < 0.05
P < 0.001 P 0.001 standardize the MODIS EVI time series Australia sites. On the basis of this site-specific PDSI (see Supplementary
ANPPG orG or ANPPm(g) m–2)
analysis for consistent cross-site comparisons. The TIMESAT filtering option
1,000 agreement15–18 the ANPP/precipitation relations
with
Information and Supplementary Table 3) and reports of continental-scale drought
known 800the adaptive Savitzky–Golay filter24 was applied over the time series
as extent and severity (summarized in the main text), the period of altered
ANPP ANPPS (g S –2
800 ‘Mean’ years
for smoothing the data and suppressing noise by replacing each data value yi,
i 5 1,...,n by a linear combination of nearby values in a window, where
reported during the late twentieth century (mean
hydroclimatic conditions was determined to be 2000–2009 for USDA sites and
2001–2009 forfor each site over the naming convention USDA00–09 and
values Australia sites, reflected by multi-year study periods).
600 n Australia01–09, respectively.
X
600 cj yizj LTER75–98 ð1Þ
Evapotranspiration model. Estimates of evapotranspiration at different biomes
j~{n
The lowest mean RUE reported for biomes with
were obtained using a model of mean annual evapotranspiration formulated with
400 USDA
LTER75–9800–09
and the weights were cj 5 1/(2n 1 1), and the data value yi was replaced by the data from moremean precip can be explained largely by non-
highest than 250 catchment-scale measurements from around the world12.
400 Australia
USDA00–09 01–09 The two-parameter model offers an approach for estimation of mean annual eva-
average of the values in the window. The moving-average methodLTER preserved the
Australia01–09 biological components of the hydrological cycle.
200 00–09
area and mean position of a seasonal peak, but altered both the width and height. potranspiration (ET) on the basis of changes in annual precipitation (P) (mm yr21)
00–09 LTER and the percentage of forest cover ( f ), where
200
The latter properties were preserved by approximating the underlying data value
with the value obtained from a least-square fit to a polynomial, rather than the
average in00 window. For each data value2,000 1, 2,…, n a quadratic polynomial
the 1,000 2 yi, i 5 3,000 4,000 Fractional forest cover Fractional herbaceous plants cover
0 zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{ zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{
0 1
was fitted as f(t) 5 c1 1 c2t 1 c3t Precipitation points in3,000
0 1,000 to all 2n 1 1 (mm yr–1) moving window and
2,000 the 4,000 1,410 1,100
the value yi was replaced with the value of the polynomial at position ti (ref. 24).
Precipitation (mm yr–1) B 1z2 1z0:5 C
The advantage of this method was that it preserved features of the distribution ET~ @ Bf P z (1{f ) P CP ð2Þ
b LTER75–98 USDA00–09 Australia01–09 Average of all years 1,410 P 1,100 P A
such as relative maxima, minima and width, which01–09 Average‘flattened’ by
b LTER75–98 USDA00–09 Australia were usually of all years 1z2 z 1z z
1,200 WUEm = 0.65 WUEm = 0.73 WUEm = 0.73 P 1,410 P 1,100
other1,200 WUE = 0.65 WUE = 0.73 WUE = 0.73
adjacent-based averaging techniques.
R2 m0.88 of integrating EVI2values from TIMESAT and avoid
To simplifyRthe=0.88
2 = process
R2 = 0.98
m
R2 = 0.98
R2 m 0.77
=
R = 0.77
Zhang et al. (2001) ET model
P < 0.0001
parameterization, 0.0001 P < 0.0001 P < 0.0001 The model has two portions as depicted in equation (2), with the left side
P < we integrated over the entire year for every site. Therefore, the
P < 0.0001 P < 0.0001
1,000integrating EVI to obtain iEVI was based on using the default para-
process of
1,000 accounting for the fractional forest cover and the right side accounting for the
- Ecosystem water-use efficiency (WUEm) was
ANPPG or ANAPSS(g m–2) )
meters found when TIMESAT was initiated. After smoothing the series, we pro- fractional herbaceous plant cover (non-forested). The main advantage of this
–2
ceeded to extract an offset of 0.05 of each 16-day EVI value to reduce effects of soil model over more traditional models is theprecipitation gradient.
constant across the entire derivation from data readily available
G or ANAP (g m
800
800
exposure. Our process was standardized by applying the same procedures to each at theThere were no significant differences among
- catchment scale. For the USDA00–09 data set, the information about the
data set used. percentage of non-forested areas was obtained from contacts at each location. For
Meteorological data. Daily precipitation and temperature were measured at in the WUEm for the three dataof the percentage of non-forested areas
600 Australia01–09 data set, estimations sets - indicating all biomes
600
situ stations associated with the experimental sites. 1.0 1.0 annual precipitation were made using Google Earth.
Total
WUEm (g m–2 mm–1)
WUEm (g m–2 mm–1)
21
(sum of daily precipitation, mm yr ), mean annual precipitation (MAP)a(mean
0.8 0.8 a a a
retained their intrinsic sensitivity to water availability
a a
of annual precipitation over the study period, mm yr21) and mean maximum 23. during prolonged,TIMESAT–adrought conditions.
400
400 0.6 0.6 Jonsson, P. & Eklundh, L. warm program for analyzing time-series of
¨
0.4 0.4
temperature (mean of average monthly maximum temperature, uC) were com- satellite s ensor data. Comput. Geosci. 30, 833–845 (2004).
0.2 0.2
puted for the hydrological year, defined as the 12-month period from October– 24. - This suggestsM. J. E. Smoothing andgoverning how species
Savitzky, A. & Golay, that the rules differentiation of data by simplified
200
200 0 0 least s quares procedures. Anal. Chem. 36, 1627–1639 (1964).
September in the Northern Hemisphere and May–April in the Australia
LTER USDA Southern
LTER USDA Australia 25. are organized in terms of Weather Bureau Res. Paper no.45 (1965).
Palmer, W. C. Meteorological Drought their tolerance of
Hemisphere. The warm season was defined as April–September–0.81USDA–1.34 26. Thornthwaite, C. W. An approach toward a rational classification of climate.
~0 ~0 for–0.81
–1.34sites
PDSI
and November–April for Australia sites.
0
PDSI hydrological stress are robust despite extended
Geogr. Rev. 38, 55 (1948).
PDSI. The0 0 PDSI200 computed with800 Thornthwaite equation26 using1,800
25
was 400 600 the 1,000 1,200 1,400 1,600 a self- 27. Wells, N., Goddard, S. & Hayes, M. J. A self-calibrating Palmer Drought Severity
0 200 400 600 800 1,000 1,200 1,400 1,600 1,800
calibrating PDSI implementation that automatically calibrated the behaviour of
perturbation by low precipitation.
Index. J. Clim. 17, 2335–2351 (2004).
Evapotranspiration (mm yr–1) –1
Evapotranspiration (mm yr ) Ponce-Campos et al 2013, Nature
10. intrinsic sensitivity of plant communities to wat
shared capacity to tolerate low annual precipitatio
a LTER75–98
WUEx = 0.66
USDA00–09
WUEx = 0.79
Australia01–09 Driest years
WUEx = 1.01
Driest years to high annual precipitation. These findings p
model of ecosystem resilience at the decadal scal
1,200
R2 = 0.81 R2 = 0.94 R2 = 0.82 hydroclimatic conditions that are predicted for
P < 0.001 P < 0.001 P < 0.001
1,000
(Fig. 4). During the driest years, the high-produ
water limited to a greater extent resulting in high
ANPPG or ANPPS (g m–2)
LTER75–98
that encountered in less productive, more arid ec
800 USDA00–09
Australia01–09
that when all ecosystems are primarily water lim
LTER00–09
maximum WUEe will be reached (WUEx) that
600
1.2
with further reductions in water availability. Furt
WUEx (g m–2 mm–1)
b
1.0
a
that as cross-biome WUEe reaches that maximum
0.8 a
400 0.6
will approach WUEx because production will b
0.4 water supply and less so by nutrients and light (F
0.2
200 0
With continuing warm drought, the single linea
LTER USDA Australia
~0 –0.81 –1.34
spiration relation that forms the common cross-
PDSI collapse as biomes endure the significant drough
0
0 200 400 600 800 1,000 1,200 1,400 1,600 1,800
Evapotranspiration (mm yr–1) 1,200
b LTER75–98 USDA00–09 Australia01–09 Wettest years Predicted WUEx
1,200 1,000 WUEn Australia01–09
- Plotting the driest 2 n = 0.67each multi-year record, yield
WUEn = 0.57 WUE
years in WUEn = 0.65 a maximum ecosystem WUE (WUEx)
R2 = 0.77 R = 0.94 R2 = 0.70 Predicted
across all biomes for each of the three data sets.
1,000 P < 0.001 P < 0.001 P < 0.001 800 WUEx
ANPP (g m–2)
- Most biomes were primarily water limited during the driest years of the early twenty-first century
ANPPG or ANPPS (g m–2)
600
drought, overshadowing limitations imposed by other resources even at high- productivity sites.
800 1.2
WUEx (g m–2 mm–1)
1.0
Grassland
0.8
- This indicates a cross-biome sensitivity to prolonged warm drought where ecosystems
400 0.6
600
sustain productivity in the driest years by increasing their WUEe. 0.4
En (g m–2 mm–1)
1.0
0.8 0.2
a a
0.6 a 200 Primarily Limited by 0
400
0.4 water several
0.2 Ponce-Campos et al 2013, Nature
0
limited resources
12. collapse as biomes endure the significant drought-induced mortality
1,200
WUEx
USDA00–09
Predicted WUEx
1,000 WUEn Australia01–09 WUEx
LTER75–98
800
Predicted
WUEx WUEn Conceptual model of
75–09
ANPP (g m–2)
ecosystem resilience
600 1.2 a
WUEx (g m–2 mm–1)
1.0 b
Grassland
0.8
400 0.6
0.4
0.2 c
200 Primarily Limited by 0
LTER USDA Australia
water several ~0 –0.81 –1.34
limited resources PDSI
0
0 200 400 600 800 1,000 1,200 1,400 1,600 1,800
Evapotranspiration (mm yr–1)
Figure 4 | A conceptual model of ecosystem resilience during altered
hydroclimatic condition. A summary of WUEe results in this study (solid water limited to a greater extent resulting in
During the driest years, the high-productivity sites become
lines), overlain with the predicted behaviour of WUEx (brown dashed line) and
WUEn (blueWUEe line) along a continuum of sites limited primarily by water more arid ecosystems.
higher dashed similar to that encountered in less productive,
e and by other resources with an arbitrary distinction made here at
evapotranspiration 5 700 mm yr21 primarily water (black dashed line).
When all ecosystems are for illustration only limited, a cross-biome maximum WUEe is reached (WUEx)
9 Predictions are based on forecasts of continuing warm drought1. Thewater availability.
that cannot be sustained with further reductions in inset
illustrates the decrease in WUEx with PDSI for subsets of the LTER75–98 (n 5 4),
h USDA00–09 (n 5 5) and Australia01–09 (n 5 e would collapse as biomes endure the significant drought-induced
The common cross-biome WUE2) data sets limited to grassland
sites, where columns has been extensively documented over different decade.
mortality that labelled with the same letter are not significantly the past
(P . 0.05).
We hypothesize that this loss of resilience associated with dieback would probably occur first for
0 0 M O N T H 2 0 1 3 | VO L 0 0 0 | N AT U R E | 3
Limited. All rights reservedrespond most rapidly to precipitation variability (that is, grasslands).
ecosystems that
Ponce-Campos et al 2013, Nature
13. - future of Australia?
• Tropical Rainfall
Measuring Mission
(TRMM) satellite
3B43 Data
Product
2010 MODIS
iEVI
• Japan-USA joint project
• Launched 1997
Remote sensing methods, by
observing broadscale vegetation
responses to climatic variability,
offer potentially powerful insights
2010 TRMM into ecological questions on
Rainfall observable timescales.
18. Productivity- rainfall per year along
NATT transect Darwin Nhulunbuy
Regression lines become more linear with drier years
Halls Creek
Tennant Creek
Mount Isa
Driest year
Alice Springs
Birdsville
Warburton
Tuesday, 14 August 12
Site- based productivity - rainfall
7"
6" Wet tropical
5"
savanna
Annual&iEVI&
4"
y"="$8E$05
R²"="0
3" N10"
y"="$0.0005
N50" R²"="
2" N100" y"="$0.000
R²"=
1" Semi-arid Mulga N150"
y"="0.00
is there an inherent 0"
(Acacias)
N200"
R²
y"="0.00
R²
maximum RUE? 0" 1000"
Annual&Rainfall,&mm&
2000" 3000" 4000"
24. Conclusions
• Extreme precipitation patterns have substantial effects on vegetation
production,
• Cross-ecosystem water use efficiency (WUEe) and RUE will increase with
prolonged warm drought until reaching a threshold that will break down
ecosystem resilience,
• It is possible to monitor ecosystem resilience with a satellite-metric, but
vital to have long term experimental monitoring sites
• It is unclear and ambiguous what ecosystem collapse would look like from
space.
• An important goal would be to assess environmental and economic costs
associated with variations in ANPP.
• Better information for strategic resource management and adaptation
practices during altered hydro-meteorological conditions.
• Societal needs to detect, predict, and manage changes in complex managed
systems that threaten to undermine resource sustainability and security.