The Effects of Drought on Amazonian Rain Forests
P. Meir,1 P. M. Brando,2,3,4 D. Nepstad,5 S. Vasconcelos,6 A. C. L. Costa,7 E. Davidson,4
S. Almeida,8 R. A. Fisher,9 E. D. Sotta,10 D. Zarin,2 and G. Cardinot11
The functioning of Amazonian rain forest ecosystems during drought has become
a scientific focal point because of associated risks to forest integrity and climate.
We review current understanding of drought impacts on Amazon rain forests by
summarising the results from two throughfall exclusion (TFE) experiments in
old-growth rain forests at Caxiuanã and Tapajós National Forest Reserves, and
an irrigation experiment in secondary forest, near Castanhal, Brazil. Soil physical
properties strongly influenced drought impacts at each site. Over years 1 to 3 of
soil moisture reduction, leaf area index declined by 20–30% at the TFE sites. Leaf
physiology and tree mortality results suggested some species-based differences
in drought resistance. Mortality was initially resistant to drought but increased
after 3 years at Tapajós to 9%, followed by a decline. Transpiration and gross
primary production were reduced under TFE at Caxiuanã by 30–40% and 12–
13%, respectively, and the maximum fire risk at Tapajós increased from 0.27 to
0.47. Drought reduced soil CO2 emissions by more than 20% at Caxiuanã and
Castanhal but not at Tapajós, where N2O and CH4 emissions declined. Overall, the
results indicate short-term resistance to drought with reduced productivity, but that
increased mortality is likely under substantial, multiyear, reductions in rainfall.
These data sets from field-scale experimental manipulations uniquely complement
existing observations from Amazonia and will become increasingly powerful if
the experiments are extended. Estimating the long-term (decadal-scale) impacts
of continued drought on Amazonian forests will also require integrated models to
couple changes in vegetation, climate, land management, and fire risk.
1
School of Geosciences, University of Edinburgh, Edinburgh,
UK.
2
Department of Botany, University of Florida, Gainesville, Florida, USA.
3
Instituto de Pesquisa Ambiental da Amazônia, Belém, Brazil.
Amazonia and Global Change
Geophysical Monograph Series 186
Copyright 2009 by the American Geophysical Union.
10.1029/2009GM000882
4
Woods Hole Research Center, Falmouth, Massachusetts, USA.
Gordon and Betty Moore Foundation, Palo Alto, California,
USA.
6
EMBRAPA-Amazônia Oriental, Belém, Brazil.
7
Centro de Geociências, Universidade Federal do Pará, Belém,
Brazil.
8
Museu Paraense Emilio Goeldi, Belém, Brazil.
9
Los Alamos National Laboratory, Los Alamos, New Mexico,
USA.
10
EMBRAPA-Amapa, Macapa, Brazil.
11
Instituto de Pesquisa Ambiental da Amazônia, Brazil.
5
429
430 EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
1. Introduction
Over the last decade, the possible impacts of drought have
become touchstone issues for environmental science and
governance in Amazonia. The geographical size, biophysical
properties, and species diversity of Amazonian forests have
led to analysis of their role as providers of key environmental services at all scales. Increased moisture limitation in the
region is likely this century, and some of the services provided by forests, including the storage and sequestration of
carbon, evapotranspiration, and the maintenance of species
diversity are potentially at risk. Human natural resource use
intersects with these roles because of rapid land use change
and tends to magnify the likelihoods of drought and forest
degradation through fire.
There is uncertainty with respect to all of these outcomes,
governed by two core questions: (1) How likely, widespread,
and severe is future drought? (2) What is the likely impact
of drought on forest ecosystem properties? In this chapter,
we address the latter question considering alterations to the
carbon balance, transpiration, tree mortality, species-based
differences in drought responses, and vulnerability to fire.
Evidence of the response by forests to drought is often based
on observation of forest processes during natural droughts.
However, unusually for any region, three large-scale soil
moisture manipulation experiments have been implemented
in Amazonia during the Large-Scale Biosphere-Atmosphere
(LBA) Experiment in Amazonia. These experiments are pro­
viding new tests of the modeled response by rain forest ecosystems to levels of drought that are beyond the bounds of
recent climatic variation, but in line with some future climate scenarios. A comparison and analysis of the first results emerging from these experiments forms the core of this
chapter and is used to summarize a view of the risks to, and
responses by, Amazonian rain forests experiencing drought
during the twenty-first century.
2. Background
2.1. Future Drought?
Two main lines of evidence suggest that Amazonian
drought may become more frequent and more severe during
this century. First, episodic drought has been associated with
the occurrence of the El Niño–Southern Oscillation (ENSO),
caused by the warming of the tropical eastern Pacific Ocean,
and more recently in 2005, with abnormal warming of the
northern tropical Atlantic Ocean, relative to the south [Cox
et al., 2008; Marengo et al., 2008]. Future increases in greenhouse gas concentrations coupled with reductions in global
aerosol emissions may increase the likelihood of 2005-like
drought events [Cox et al., 2008] and possibly also ENSO
events [Timmermann, 1999]. More generally with respect
to climatic change over the twenty-first century, the multimodel data set (MMD) used in the IPCC Fourth Assessment
Report projected an annual mean warming in Amazonia of
1.8°–5.1°C over this century, with rainfall reductions in parts
of central and eastern Amazonia likely as a result, especially
in the dry season [IPCC Working Group I, 2007]. In a recent
analysis involving 23 MMD models, Malhi et al. [2008] reported 20–70% agreement among models in the prediction
of substantial dry season reductions in precipitation across
Amazonia, with the greatest likelihood of drought in the east
of the region.
Second, land use change is likely to exacerbate the effects
of climatic warming. Widespread forest conversion to pasture and agriculture is expected to reduce rainfall in the region through differential effects on latent and sensible heat
transfer [Werth and Avissar, 2002; Chagnon and Bras, 2005;
Costa et al., 2007; Nobre et al., 1991] and increased regional
atmospheric aerosol loading could cause widespread reductions in precipitation [IPCC Working Group I, 2007]. Overall, these results imply an increased frequency of extreme
events at the seasonal and interannual timescale, and a secular shift to drought at decadal timescales, the strength of any
shift strongly influenced by land use change.
2.2. Modeling Drought Impacts
Estimates of the impact of drought on Amazonian forests
have historically been made with limited field data, and this
has contributed to uncertainty in estimates of future global
atmospheric CO2 concentrations [Meir et al., 2006; IPCC
Working Group I, 2007; Huntingford et al., 2009]. Recent
analysis of the bioclimatic distribution of current natural
Amazonian vegetation and the predictive output from 19
global circulation models (GCMs) suggests that twenty-first
century climate change is most likely to lead to drier conditions more appropriate for seasonal forest in eastern Amazonia [Malhi et al., 2009], although edaphic conditions may
in reality favor a transition to degraded forest or savanna. In
contrast, smaller impacts are likely on vegetation in western
Amazonia [Malhi et al., 2009]. Driving a vegetation model
with a variety of climate scenarios, Sampaio et al. [2007]
predicted a similar but more extreme range of forest-to-savanna switches following climatic warming and drying. In
both cases, these decadal-century timescale scenarios were
based on some form of “equilibrium” vegetation response
to drought. In reality, the actual vegetation response will
be determined (rapidly or gradually) by multiple ecological and physical processes, as acknowledged by the same
authors.
meir et al. 431
By contrast, process-based dynamic vegetation models
have the structure to capture the relevant ecology, enabling
land surface-atmosphere interactions to be modeled on a
continuous basis. However, they are computationally expensive and ultimately require observation-based, ecological parameterization [Prentice and Lloyd, 1998; Meir et al.,
2008]. The first predictions of substantial Amazonian dieback in response to warming and drought emerged from a
global dynamic vegetation model (DGVM) framework and
were made at the decadal-century timescale, with relatively
simplistic representations of canopy structure, soil processes,
and functional diversity among species [White et al., 1999;
Cox et al., 2000]. Large differences in parameterization and
mathematical description of core ecosystem processes have
since been identified among DGVMs [Dufresne et al., 2002;
Meir et al., 2006; Friedlingstein et al., 2006; B. Poulter et
al., Managing uncertainty of tropical Amazon dieback, submitted to Global Change Ecology, 2009; D. Galbraith et al.,
Quantifying the contributions of different environmental
factors to predictions of Amazonian rain forest dieback in
three dynamic global vegetation models (DGVMs), submitted to Global Change Biology, 2009], emphasizing the need
for improvements using field-based measurements and experimentation. Nondynamic, but process-based, vegetation
and biogeochemical models have also been used to advance
the representation of Amazonian forest functioning [Lloyd
et al., 1995; Williams et al., 1998; Potter et al., 2004], although differences in process description have been apparent [Tian et al., 1998; Prentice and Lloyd, 1998; Foley
et al., 2002; Zeng et al., 2005; Meir et al., 2008]. Only recently have detailed validations been made of the modeled
response to drought and with some success over seasonal
to interannual timescales [Fisher et al., 2007; Baker et al.,
2008].
The validation of longer-term vegetation dynamics is
more difficult because it requires data sets at the scale of
years to decades, representing multiple ecological processes.
The multiyear soil moisture manipulation experiments described here provide unique insight into some of the relevant
physiological and ecological processes. Also, they indicate
the potential first steps in quantifying natural lags in vegetation change as a function of tree species’ drought resistance and regeneration characteristics. However, vegetation
change under climatic drying is influenced strongly by fire
and deforestation. The area of burnt understory forest during
the ENSO of 1998 (2.9 × 106 ha) was more than 10 times
greater than during an average rainfall year and twice the
area of annual deforestation [Nepstad et al., 1999; Alencar
et al., 2006], while a burnt area up to 2800 km2 was attributed to fire leakage alone in the 2005 drought [Aragão et
al., 2007]. Fire is more probable in forest that has burned
previously [Cochrane, 2003], and the fragmentation effects
of deforestation tend to increase fire risk strongly [Uhl and
Kauffman, 1990; Cochrane and Laurance, 2002]. Deforestation has repeatedly been shown to affect regional climate,
though the effects vary with scale [Werth and Avissar, 2002;
Chagnon and Bras, 2005]. At large scale, deforestation
scenarios result in modeled reductions in precipitation and
relative humidity, and increases in temperature [e.g., Hoffmann et al., 2003; Costa et al., 2007]. Strong increases in
fire risk (20–120%) have thus been associated with scenarios
of partial and complete deforestation [Cardoso et al., 2003].
While we need to understand the responses to drought by
Amazonian rain forest, the interactions between drought and
increased fire risk must also be considered, as fire may well
be the agent that, in the context of secular or episodic climatic drying, triggers the switch from a forest to a savanna
[Hutyra et al., 2005; Aragão et al., 2007].
2.3. Basin-Wide Observations of the Forest Carbon
Balance During Drought
Inversion studies have provided the largest scale of observation-based information on drought effects on Amazonia.
Despite the limitations of this method [see Houghton et al.,
this volume], a net emission of CO2 from the region of up
to 1.5 Pg C a−1 has been reported for dry and warm periods
during strong ENSOs [Bousquet et al., 2000, Rödenbeck et
al., 2003, Zeng et al., 2005, see also Houghton et al., this
volume], although it is unclear if these higher emissions are
principally the result of changes in ecosystem carbon cycling or increased fire occurrence [Langenfelds et al., 2002;
Meir et al., 2008]. Results from biogeochemical modeling
of the forest carbon cycle have been consistent with the observation of net regional CO2 emissions during ENSO [e.g.,
Tian et al., 1998; Foley et al., 2002; Zeng et al., 2005], but
the mechanisms underpinning these modeled results have
tended to overemphasize the role of the temperature sensitivity of soil respiration during drought [Meir et al., 2008].
Phillips et al. [2009] recently reported basin-wide long-term
observations of Amazonian forest productivity, estimating
an overall negative impact of 1.2–1.6 Pg C in old growth
forests during the 2005 drought, which was driven by lower
growth and (spatially patchy) increases in tree mortality.
This outcome is consistent with the notion that gross primary production (GPP) declines under drought, but without
measurements of concurrent soil processes (e.g., CO2 efflux
from soil), it does not resolve the question of the short-term
impact of the 2005 drought on net ecosystem productivity
(NEP). [NEP is the difference between carbon that is photo­
synthesised and respired by an ecosystem. NEP = GPP −
Reco, (Reco = total ecosystem respiration).]
432 EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
In summary, analysis of basin-wide atmospheric and tree
growth data has highlighted the importance of understanding NEP at different timescales, and identified large changes
in possible ecosystem functioning during interannual-scale
drought, including reductions in aboveground productivity,
increases in mortality in response to drought, and differential
tree species’ survival [cf. Engelbrecht et al., 2007]. These
data will be invaluable to validate estimates of tree growth
from DGVMs or finer-scale vegetation models, but alone,
they do not pinpoint the mechanisms determining large-scale
CO2 exchange observed during intermittent severe drought
[e.g., Rödenbeck et al., 2003]. In particular, CO2 emission
processes have been hard to pin down, first because respiration terms are often poorly quantified [Meir et al., 2008],
and second, because the increases in fire occurrence identified during ENSO or the 2005 drought [Nepstad et al., 1999;
Aragão et al., 2007] are hard to quantify in terms of CO2
emissions [van der Werf et al., 2004].
2.4. Observations at the Stand Scale
Some of the first eddy covariance measurements in Amazonia, quantifying the forest water and carbon cycles at the
scale of a few square kilometers, suggested little seasonality
in rain forest carbon exchange capacity [Grace et al., 1995].
However, although not universal, seasonal drought effects on
NEP have since been observed at some sites, with observed
changes in gross photosynthesis, respiration, and transpiration across sites differing in seasonality yielding new insights into the functional basis of the response [Malhi et al.,
1998; Carswell et al., 2002; Saleska et al., 2003; Vourlitis
et al., 2005; see da Rocha et al., this volume; Saleska et al.,
this volume; da Rocha et al., 2009].
Uniquely, experimental field manipulation of soil moisture
enables the separation of the effects of otherwise naturally
covarying environmental drivers (i.e., edaphic and atmo­
spheric variables) of the processes determining NEP and has
the potential to provide powerful insight into process-level
responses to short-term and extended drought. Three such
field experiments have been performed within LBA, all in
drought-threatened eastern Amazonia. Strong soil moisture
deficit has been imposed over 1 ha of old-growth forest
at two experimental sites, in the National Forest Reserves
of Caxiuanã (near Portel, State of Pará) and Tapajós (near
Santarém, State of Pará) [Nepstad et al., 2002; Fisher et
al., 2007; Meir et al., 2008], and at a third site, dry-season
moisture stress has been reduced though the irrigation of
replicated 0.04-ha plots in secondary regrowth forest near
Castanhal, State of Pará [Vasconcelos et al., 2004].
The remainder of this chapter is concerned primarily with
how the results of these experiments inform our understand-
ing of drought impacts on Amazonian rain forest, mostly in
relation to the cycling of carbon and water, but also including the emission of other trace gases such as CH4, NO, N2O,
and isoprene. Where possible, we make a first comparison
of the still-emerging experimental results. We ask three
questions: (1) What have we discovered about the impact
of drought at seasonal to interannual timescales? (2) What
have we discovered about the impact of drought at multiyear
to decadal timescales? (3) How does the combined impact
of fire and drought influence the risk of widespread loss of
forest in favor of smaller-stature vegetation types, such as
savanna or degraded forest?
3. Soil Moisture Manipulation Experiments
at Caxiuanã, Tapajós, and Castanhal
Species diversity is substantial (>150–200 species ha−1)
in the old-growth rain forests at both Caxiuanã and Tapajós, while only four tree species (=70% of stems) dominate
the regrowth forest at Castanhal. Standing biomass is much
larger at the rain forest sites (approximately 300 t ha−1 at
Caxiuanã, 240 t ha−1 at Tapajós, and 50 t ha−1 at Castanhal)
[Baker et al., 2004; Vasconcelos et al., 2004; Brando et al.,
2008]. Rain forest canopy heights are similar at 30–40 m,
while the height of the secondary forest at Castanhal is approximately 5 m [Coelho et al., 2004]. The soils at Caxiuanã
and Tapajós are highly weathered Oxisols, and at Castanhal, the soil is a dystrophic yellow Latosol. At Caxiuanã,
the soil composition is a sandy loam (70–83% sand) with
the water table at 15–20 m depth; at Tapajós, the composition is more clay-rich and the soil profile is much deeper
(60–80% clay; >100 m in depth); and at Castanhal, the soil
is shallow and concretionary, with a high sand content (20%
clay, 74% sand). Annual precipitation at Caxiuanã, Tapajós,
and Castanhal is approximately 2300, 2000, and 2500 mm
a−1, respectively. Further site descriptions are provided elsewhere [Davidson, 1992; Nepstad et al., 2002; Ruivo et al.,
2003; Fisher et al., 2007; Vasconcelos et al., 2004].
The method of physically excluding rainfall that penetrates
the canopy (“throughfall exclusion” (TFE)) was replicated
at Caxiuanã and Tapajós using approximately six thousand
4.5 m2 plastic panels and guttering placed at 2 m above the
ground. The infrastructure removed approximately 50% of
incoming precipitation [Nepstad et al., 2002; Fisher et al.,
2007] (Table 1) and was installed at the beginning of 2000 at
Tapajós and 2002 at Caxiuanã. Each experiment comprised
1 ha of TFE forest and 1 ha of undisturbed (nonmanipulated)
“control” forest. The large scale of the manipulation was necessary because of substantial lateral extension of the surface
roots of large trees. Treatment replication at both sites was
limited by financial resources, but pretreatment calibration
meir et al. 433
Table 1. Total Incoming Precipitation for Each Year of the Drought Treatment by Dry (Panels Off)
and Wet Season (Panels On)a
Precipitation, mm
Year
Period of Exclusion
Total Incoming
Panels Off
2000
2001
2002
2003
1 Feb to 8 July
1 Jan to 31 July
1 Jan to 31 July
21 Jan to 14 Aug
Totals
2517
1882
1958
1690
8047
830
171
292
394
1687
Panels On
(Throughfall Excluded)
1687 (844)
1711 (856)
1665 (833)
1295 (648)
6358 (3179)
a
Approximately 50% of incoming precipitation was excluded when the panels were on; those
volumes are presented in parentheses.
measurements were made in all plots to enable replication
over time [Davidson et al., 2004], and the method follows
the design of other unreplicated large-scale ecosystem manipulation experiments [e.g., Likens et al., 1970], whose
strength is acknowledged, especially where large treatment
effects are expected [Hurlbert, 2004]. The perimeter of the
TFE plots was trenched to 1–2 m depth to prevent the horizontal ingress of water from adjacent normally watered soil,
and the control plot perimeters were also trenched to avoid
confounding treatment effects. Soil and plant measurements
were made more than 20 m inside the perimeter of each plot
to further eliminate confounding treatment effects. Management of the experiments was similar, except that at Tapajós,
the panels were removed during the peak dry season, while
at Caxiuanã, this procedure was not followed because of the
prevailing late dry season storm risk [Carswell et al., 2002].
Litterfall was manually returned to the soil, where it fell on
the TFE paneling. Full canopy access was provided using 40
m towers in all plots, also enabling the installation of on-site
automatic weather stations. Principal access to the soil was
provided by four soil shafts per plot, excavated to a maximum depth of 10 m at Caxiuanã and 14 m at Tapajós.
The irrigation experiment at Castanhal was designed to
remove dry season moisture stress and began in 1999 [Vasconcelos et al., 2004]. Plot size was 20 m × 20 m (0.04 ha),
and four replicate plots were used to contrast irrigated and
undisturbed vegetation; adjacent plots were placed 10 m or
more apart and a nested 10 m × 10 m plot in each 0.04 ha
main plot was used for measurements. An additional treatment removing litter was also implemented at Castanhal, but
is not discussed here. Ground-level irrigation was provided
using perforated microtape installed at 2-m spacings; water
was applied at a rate of approximately 5 mm d−1 for 30 min
during the dry season. The amount of irrigation was selected
to approximately replace daily evapotranspiration estimated
regionally (660–790 mm) [Jipp et al., 1998, Vasconcelos et
al., 2004]. Surface soil moisture availability was measured
in all plots, and the relatively small stature of the vegetation
allowed lower-canopy access for leaf water potential mea­
surements [Fortini et al., 2003].
4. Seasonal to Interannual
Drought Impacts
4.1. Soil Moisture and Its Supply to Plants
The change in soil moisture, in relation to adjacent undisturbed (control) forest, was the main experimentally
manipulated parameter in each experiment. The TFE infrastructure at Caxiuanã and Tapajós resulted in reductions in
plant available water (PAW) of 80–200 mm in the top 3 m
of soil (Table 1 and Figure 1a) [Fisher et al., 2007, 2008;
Brando et al., 2008], while the irrigation at Castanhal nearly
completely removed the dry season moisture constraints
[Vasconcelos et al., 2004]. A strong seasonality was evident
in PAW at the TFE experiments, and the rate of soil moisture
drawdown was larger over the first 1–3 years at Caxiuanã
than at Tapajós.
The impact on PAW of manipulating precipitation input
to the soil was strongly modified by the soil properties at
each site. At Caxiuanã, the sandy-loam composition created
a relatively high moisture holding capacity per unit volume
[Carswell et al., 2002; Fisher et al., 2008], while the deep
clay-rich soil at Tapajós potentially held substantial water
reserves mainly because of its exceptional soil volume [Nepstad et al., 2002; Belk et al., 2007]. Detailed mea­surements
of soil hydraulic properties [e.g., Tomasella and Hodnett,
1997] remain rare in Amazonia, but variation in the key soil
parameters determining PAW, soil water potential, hydraulic conductivity, and moisture volume content, can cause
large and basin-wide differences in the supply of water to
plants under moisture stress [Fisher et al., 2008]. Analogously, the shallow concretionary soil structure at Castanhal
likely contributes to moisture limitation of root and micro-
434 EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
(a)
(b)
Figure 1. (a) Plant available water (PAW) over wet and dry seasons in the control (filled symbols) and throughfall exclusion (TFE) plots (open symbols) for both experiments Tapajós (circles) and Caxiuanã (triangles). The experiment in
Caxiuanã began in 2001, and in Tapajós, it started in 1999 (pretreatment). PAW is the water available to plants between
field capacity and a soil moisture potential of −1500 kPa. The arrow shows when TFE infrastructure was installed at each
site. The data points are derived from 48 fortnightly measurements made per soil access shaft, four access shafts per plot
[Brando et al., 2008; Fisher et al., 2007]. (b) Monthly natural rainfall, soil water balance, and annual soil water deficit
at Caxiuanã, 1999–2004. Maximum PAW at Caxuiana is ~400 mm water (a) [Fisher et al., 2008]; under normal rainfall,
soil moisture supply is sufficient to support transpiration, but under strong drought, moisture limitation is likely.
meir et al. 435
bial activity during the natural dry season [Vasconcelos et
al., 2004]. More soil hydraulics data from across the region
are required to make satisfactory PAW calculations for the
modeling of vegetation activity.
The supply of moisture from soil to leaf is also affected
by site differences in rooting properties. Rooting depth can
be substantial, enabling increased access to soil water: roots
have been detected at depths >14 m at Tapajós and at Caxiuanã, to the approximate maximum depth of excavation,
9 m [Nepstad et al., 2004; Fisher et al., 2007]. As well as
enabling fuller exploration of the soil profile, deep rooting
can also facilitate hydraulic redistribution, thus further helping to maintain rhizosphere moisture availability and plant
function under reduced rainfall, as observed at Tapajós in
some instances [Oliveira et al., 2005]. Initial data have not
identified major changes in root dynamics at Tapajós in the
TFE treatment [Brando et al., 2008]. However, responses
in the modes of root growth at Caxiuanã are partially consistent with theory [Joslin et al., 2000; Schymanski et al.,
2008]. As the drought progressed at Caxiuanã, surface roots
(0–30 cm) tended to increase in length per unit mass, thus
increasing the explored soil volume [Metcalfe et al., 2008],
but it is less clear how plastic root growth responses were to
changes in the vertical distribution of soil moisture availability within the soil profile, and answering this question will
have implications for modeling resilience to drought. Species variation in these root properties differentially affects
water acquisition among species and between sites, and variation in resistance to drought was also observed in the foliage. Leaves generally experience maximum daily moisture
limitation at high atmospheric moisture deficit, soon after
midday. Early afternoon measurements of the minimum leaf
water potential tolerated by tree species at Caxiuanã, Tapajós, and Castanhal demonstrated differences among species
(Table 2), although the minimum value measured in both
TFE plots was similar (−3.2 and −2.7 MPa, respectively),
indicating a possible maximum tolerance to moisture limitation in eastern Amazonian rain forest trees. Stem hydraulic
conductance did not appear to be the major constraint to leaf
water supply at low soil PAW, but species-based differences
in this parameter were also observed at Caxiuanã [Fisher et
al., 2006] potentially further influencing species-based differences in resistance to soil moisture limitation [cf. Franks
et al., 2007].
In summary, substantial differences in soil properties at
each experimental site strongly influenced the storage of water in soil and its supply to plant roots under drought. The
TFE results from Tapajós and Caxiuanã also highlighted
mechanisms conferring tolerance to drought in terms of
root, stem, and leaf hydraulic properties. Variation among
tree species in these responses to drought, including addi-
Table 2. Minimum Leaf Water Potential (Min ψl) Measurements
for Tree Species at the Caxiuanã and Tapajós TFE Experiments
and the Castanhal Irrigation Experimenta
Site
Species
Min ψlb, MPa
T
T
T
T
C
C
C
C
Cs
Aparisthmium cordatum
Astrocaryum gynacanthum
Coussarea racemosa
Poecilanthe effusa
Licaria ameniaca
Hirtela bicornis
Lecythis confertiflora
Swartzia racemosa
Miconia ciliate
−1.5
−1.5
−2.7
−2.4
−2.2
−2.1
−2.9
−3.2
−3.0
a
Abbreviations are C, Caxiuanã; T, Tapajós; Cs, Castanhal.
Values for T and C are taken from trees in the TFE plots, and
values for Cs are from an understorey species (data from Fisher et
al. [2006], Fortini et al. [2003], and G. Cardinot et al., manuscript
in review, 2009).
b
tional evidence of direct uptake of dew in two species at the
Tapajós experiment (G. Cardinot et al., manuscript in review, 2009) suggest likely differences in survivorship under
drought at multiyear or decadal time scales.
4.2. Canopy Structure and Productivity
Although some savanna tree species have a deciduous or
brevi-deciduous phenology [Furley et al., 1992], this strategy is relatively unusual in rain forests, where leaf area index
(LAI, m2 leaf area per m2 unit ground area) is maintained under normal climatic variation. There is some ground-based
evidence for dry-season increases in LAI [Carswell et al.,
2002] and albedo [Culf et al., 1995], but irrespective of how
this may affect forest functioning, it seems that PAW can
often be maintained by forest trees through extensive and
sometimes exceptionally deep rooting systems [Nepstad
et al., 1994; Bruno et al., 2006]. However, under moisture
constraints in excess of normal climatic variation, we have
limited understanding of the limits to moisture access by forest trees. The TFE experiments thus provide a direct way of
determining the thresholds in PAW that may lead to changes
in canopy productivity, LAI, and fire vulnerability.
The LAI at both sites was resistant to strong artificial
soil moisture deficit for about a year following installation
of the TFE infrastructure, remaining at 5–6 m2 m−2. After
12 months, and a reduction in PAW of about 150–200 mm,
LAI declined to 70–80% of the control (and original) values
at each site (Figure 2), and this reduction was maintained
subsequently following three more years of TFE (Brando
et al., 2008, D. B. Metcalfe et al., Impacts of experimentally imposed drought on leaf respiration and morphology
436 EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
Figure 2. Variation in leaf area index (LAI) during the TFE experiments at Caxiuanã and Tapajós. LAI is expressed as
the quotient of (TFE plot LAI) / (Control plot LAI). LAI in undisturbed forest at both sites is 5–6 m2 m−2. The arrow
denotes when the TFE infrastructure was installed at each site. (LAI was measured using a Li2000 (Licor, USA) at fixed
100 points [Brando et al., 2008; Fisher et al., 2007; Metcalfe et al., submitted manuscript, 2009]).
in an Amazon rain forest, submitted to Functional Ecology,
2009). However, litter production patterns differed among
sites, perhaps partially reflecting differences in tree community responses to moisture limitation. At Caxiuanã, litterfall
in the TFE plot declined within the first 12 months following
installation of the TFE and remained lower than undisturbed
forest over the following dry seasons. At Tapajós in the
TFE plot, and at Castanhal under irrigation, litterfall did not
change significantly and tracked the control forest litter flux
rates closely in the first 2–3 years of rainfall manipulation
(Figure 3). The reduction in LAI and subsequent litterfall at
Caxiuanã is consistent with reduced leaf regrowth, while the
maintenance of litterfall at Tapajós and Castanhal suggests a
stronger limitation on leaf replacement. However, observed
increases in leaf mass per unit area during experimental
drought also contributed substantially to changes in LAI at
Caxiuanã and Tapajós [cf. Wright et al., 2006; Metcalfe et
al., submitted manuscript, 2009; Tohver et al., unpublished
data, 2007]. Finally, changes in leaf turnover rates, and the
temporary impacts of mortality events after three years or
more (data not shown) may help explain litterfall patterns
over the longer term at Tapajós and Castanhal [Brando et al.,
2008; Vasconcelos et al., 2008].
In summary, initial resistance to change in LAI during the
first 12 months of soil drought at both TFE experiments was
followed by substantial reductions in LAI of 20–30% over
the following 2 years, and this was maintained subsequently.
Dry season litterfall declined at Caxiuanã relative to undisturbed forest, but artificial droughting and irrigation had only
small effects on litterfall at Tapajós and Castanhal. Drought
impacts on forest canopy physiology and structure have also
been examined using remote sensing data products [Asner
et al., 2004; Saleska et al., 2007; Huete et al., 2008, see
also Saleska et al., this volume]. It remains unclear if these
data can be used to reliably quantify changes in productivity
during drought, but the TFE experiments provide a way of
specifying models to test such estimates. The consequences
of reduced LAI for alterations in gas exchange capacity and
fire vulnerability are considered below.
4.3. Trace Gas Emissions From Soil
The flux of CO2 from soil (“soil respiration,” Rs), comprises the largest single respiratory flux in the terrestrial
carbon cycle and derives from the combined respiration
of heterotrophic (microbial, faunal) and autotrophic (root)
meir et al. 437
Figure 3. Seasonal litterfall (±SE) at Caxiuanã, Tapajós, and Castanhal from 1999 to 2004. The arrows denote when
the TFE or irrigation treatments were installed. The data are derived from 20 litter traps (1 m2) per plot; litterfall was
collected fortnightly or monthly at each site [Brando et al., 2008; Vasconcelos et al., 2008; Meir et al., manuscript in
preparation, 2009].
components of the soil biological community [Trumbore,
2006; see also Trumbore and de Camargo, this volume]. Rs
has been reported for all three experiments, and additional
measurements of NO, N2O, and CH4 have also been made at
Tapajós and Castanhal.
Consistent with biophysical expectation [Howard, 1979;
Meir et al., 2008], seasonal declines in Rs under reduced
soil moisture have been observed widely in Latin American
rain forests [Davidson et al., 2000; Schwendenmann et al.,
2003]. While the experimental manipulation of soil moisture
lowered Rs strongly during the first 2 years at Caxiuanã (by
>20%, equivalent to >2 t C ha−1 a−1; Figure 4) [Sotta et al.,
2007], the response of Rs to temperature was small and non-
significant [Sotta et al., 2007]. After 3–4 years, dry season
Rs in the Caxiuanã TFE plot remained lower than for soil
in undroughted forest, although overall between-plot differences in Rs were smaller, perhaps because of increased
wet-season root respiration rates in the TFE plot [Metcalfe
et al., 2007]. At Castanhal, the difference in Rs between irrigated and undisturbed soil reached maxima during the dry
seasons, and annual Rs in irrigated plots was 13–27% larger
than on undisturbed (drier) plots [Vasconcelos et al., 2004].
In contrast, at Tapajós, although the gross fluxes were not
unusual in magnitude, Rs was similar between the TFE and
control plots throughout the experiment, even after 5 years
of TFE treatment [Davidson et al., 2008]. This outcome was
438 EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
Figure 4. Soil respiration (CO2 efflux from soil, Rs) at Caxiuanã, Tapajós, and Castanhal. TFE or irrigated plots (gray
triangles) are plotted against control plots (black filled circles) at each site. Measurements were made by infrared gas
analysis, using permanently installed collars. Data are presented as monthly means ±SE (n = 20 at Caxiuanã and Tapajós,
n = 12 at Castanhal), with the experimental soil moisture manipulations beginning at month 0. Inset: the relationship
between soil moisture potential and Rs from Caxiuanã. Data are from Vasconcelos et al. [2004], Sotta et al. [2007], and
Davidson et al. [2008]. Inset from Sotta et al. [2007], reprinted with permission from Wiley-Blackwell.
surprising given the seasonal variation in PAW at Tapajós,
and the close tracking of Rs with seasonal PAW in the TFE
plot at Caxiuanã (Figures 1 and 4). The similarity in Rs between TFE and control plots at Tapajós is consistent with
the similarity in total litterfall and the radiocarbon-derived
ages of respired carbon from both plots. This has led to the
suggestion that the unusually large soil and rooting depths at
Tapajós explains the unexpected maintenance of normal Rs
fluxes under drought, although increased root mortality and
decomposition in the TFE plot may also have been responsible [Brando et al., 2008; Davidson et al., 2008]. An additional
possible explanation is that the role of hydraulic redistribution has been underestimated. For example, if nighttime recovery of rhizosphere moisture content occurred at Tapajós,
but was not detected using the time domain reflectometry
instruments installed there [Nepstad et al., 2002; Oliveira
et al., 2005], then autotrophic and heterotrophic respiration
may have continued in the roots and rhizosphere, respectively, maintaining relatively high Rs even at the low PAW
measured in the TFE plot, although how (or if) Rs and GPP
were differentially affected by such levels of rhizosphere
moisture availability remains unexplored at Tapajós.
Emissions of NO, N2O, and CH4 were also measured at
Tapajós and Castanhal. Moisture limitation effects on each
followed biophysical expectation [Forster et al., 2007], although NO emissions were more resistant to change than
expected. At the Tapajós TFE experiment, N2O emissions
declined, and CH4 consumption increased at low soil mois-
meir et al. 439
ture availability. Consistent with this, irrigation at Castanhal
had the reverse effects on fluxes of both trace gases; the thin
concretionary soils and secondary regrowth history of this
site were associated with lower overall fluxes [Vasconcelos
et al., 2004]. Emissions of NO were not altered by irrigation
at Castanhal, and they did not increase substantially at the
Tapajós TFE until soil moisture availability was very low,
a result also partially attributed to the soil texture at this site
[Vasconcelos et al., 2004; Davidson et al., 2008]. Following permanent removal of the TFE treatment after 5 years
at Tapajós, Davidson et al. [2008] reported a return of all
trace gas soil fluxes to pre-TFE treatment levels and argued
that the observed effects of the drought treatment at Tapajós most likely reflected changes in soil aeration rather than
substrate supply.
In summary, experimental manipulation quantified the
strong influence of soil moisture on Rs in two of the three
experiments (at Caxiuanã and Castanhal). These results
demonstrated the primary importance of moisture limitation
over temperature on respiration processes during drought,
consistent with observations elsewhere in the region and,
contradicting earlier, widely employed, modeling assumptions [Tian et al., 1998; Zeng et al., 2005; Peylin et al., 2005].
The maintenance of relatively high Rs at low soil moisture
content observed in the TFE plot at Tapajós, possibly explained by an exceptionally deep soil profile, serves to highlight the existence of spatial variation in drought responses
across the basin, although other drought-related trace gas
emissions of NO, N2O, and CH4 responded to reduced soil
moisture availability as expected, with temporary reductions
in CH4 and N20 emission and increases in NO production
under severe moisture limitation.
developed from experimental fires conducted in the vicinity
of the Tapajós TFE experiment [Ray et al., 2005], we calculated daily probabilities of a fire spreading in both control
and TFE plots for the Tapajós experiment, from July 2000 to
December 2004 [equation (1), Figure 5]. We ran two simulations: one in which both precipitation and LAI varied and
another in which precipitation was set to zero while LAI was
not constrained.
P = 1-
1
,
1 + e5.35-0.3*cwp-0.131*CH-0.36*LAI
(1)
where cwp is the sum of precipitation in time (t) in days
and the cumulative precipitation in time (t − 1) × 0.5; CH is
canopy height (kept constant); and LAI is monthly leaf area
index.
During the dry season of 2002, the maximum probability
of fire spread (P) in the control forest was 0.27, while in
the TFE, plot P was 0.47. This increase in fire susceptibility
occurred especially after the pulse in tree mortality, 3 years
following installation of the TFE infrastructure. By simulating forest flammability based only on understory microclimate, we assumed that fine fuel loads would not limit fire
spread, although we note that the increase in coarse woody
debris resulting from higher mortality would have increased
fire intensity.
4.4. Fire Risk
One of the most obvious ways in which drought affects
the flammability of tropical forests is through temporary
changes in the understory microclimate: drier and warmer
conditions increase the risk of fire [Alencar et al., 2004,
2006]. But less obvious are the indirect and lasting effects of
drought on forest flammability through reduced PAW [Nepstad et al., 1999, 2001, 2004]. As PAW reaches deficits large
enough to induce leaf shedding, solar radiation penetrating
the forest canopy increases and leads to higher air temperatures near the forest floor [Uhl and Kauffman, 1990]. This
process speeds the rate at which the drying of fine fuel (e.g.,
leaves and small twigs) occurs, one of the best proxies of
forest flammability [Hoffmann et al., 2003].
The observed LAI reductions in both TFE experiments
were sufficient to increase forest flammability in the TFE
plots. Based on an LAI and precipitation-driven model
Figure 5. Probability of fire spread for the Tapajós TFE (gray) and
control (black) plot forests. Dashed black and gray lines represent
where weighted precipitation was set to zero, so LAI is driving forest flammability. Fire spread probability is calculated using equation (1) and data from the TFE experiment [Ray et al., 2005].
440 EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
In summary, the experimental drought treatment substantially increased the susceptibility of the forest to fire, even
over the short term (Figure 5), converting it from a state
where fire was unlikely to one where fire was probable in
the presence of appropriate ignition sources. The risk of fire
thus increases markedly under extended drought, and this
has implications for long-term carbon storage and emissions
[van der Werf et al., 2004]. The decline of LAI in both TFE
experiments, and the particularly high mortality at Tapajós,
signify that the positive feedbacks between drought and fire
may be stronger than previously hypothesized.
5. Medium- and Long-Term Drought Impacts:
Responses in Physiology and Mortality
Physiological responses to drought observed over the short
term have impacts over the long term, most notably through
their negative impacts on growth and mortality. Tree death
delivers carbon to the decomposer pool, committing future
emissions of CO2 to the atmosphere, and in the absence of
replacement through recruitment and regrowth, overall net
primary production and transpiration decline, while the risk
to fire rises.
McDowell et al. [2008] review proposed plant physiological responses to drought [Tardieu and Simonneau, 1998]
and distinguish between plants under moisture limitation
that exert strong stomatal control to maintain leaf water
potential above a minimum value (“isohydry”) and those
that exert less stomatal control of transpiration, relying on
continued water supply from the soil (“anisohydry”). Under
strong moisture limitation (and high tension), the transpiration stream may cleave (embolize), and if this happens to
a large extent, hydraulic failure can occur, leading to plant
death [Tyree and Sperry, 1988; West et al., 2008]. Anisohydric plants exert minimal stomatal control under moisture
limitation and thus risk mortality by hydraulic failure under severe drought unless this risk is reduced, for example,
through the construction of resistant xylem vessels. By contrast, isohydric plants reduce the risk of hydraulic failure
through stomatal closure, although this risk is not entirely
avoided, and other resistance terms in the soil-to-atmosphere
hydraulic path may also change [Franks et al., 2007]. Stomatal closure can lead to high leaf temperatures, to reduced
photosynthetic carbon gain, and under extended drought to
increased risk of mortality through carbon starvation of metabolism (mainly respiration) and/or increased susceptibility
to pathogen attack.
Leaf water potential measurements from the Caxiuanã
TFE experiment were consistent with trees responding isohydrically to drought [Fisher et al., 2006]. This proposed
mode of response is further consistent with: (1) declines in
GPP and stem growth rates following TFE treatment [Fisher
et al., 2007; Brando et al., 2008]; (2) the initial resistance
to mortality observed over the first 2 years of experimental drought at Caxiuanã and Tapajós [Nepstad et al., 2007;
A. C. L. Costa et al., manuscript in preparation, 2009]; and
(3) pantropical observations of small average increases in
mortality following ENSO events over the last three decades (the median increase in mortality following an ENSO
event across 45 tropical forest plots was 1.2%) [Meir et al.,
2008].
Further consistent with the notion that isohydry may be
common in rain forest trees that are not adapted to long-term
and severe droughts, resistance in mortality rates to the TFE
treatment gave way after 2–3 years to substantial mortality
increases [Nepstad et al., 2007; Brando et al., 2008], possibly as a result of carbon starvation. The Tapajós TFE experiment revealed highly variable background rates of mortality
in both the TFE and control plots (1–3%), and this was then
followed by exceptionally high mortality (9% in trees with
dbh > 10 cm) after 3 years of experimental drought [Nepstad
et al., 2007; Brando et al., 2008]. In years 4 and 5 of the
experimental drought at Tapajós, mortality declined to just
above (the relatively high) background levels, but 1 year after
removal of the TFE infrastructure at Tapajós, mortality rose
again to 7%, suggesting longer term and possibly speciesspecific impacts. Indications of a correlation between species differences in stomatal control and mortality at Tapajós
[Ehleringer et al., 2004] further support the notion of species-specific variation in mortality risk during drought [cf.
Fisher et al., 2006] (see Table 2), and increased regrowth
during years 4 and 5 by understory species released via mortality impacts on the canopy will also have influenced the
range of species-based responses at Tapajós [Brando et al.,
2008]. The immediate influence of increased mortality on
CO2 emissions is likely to be small during drought because
of the desiccation constraint on organic matter breakdown
[Meir et al., 2008], although the effect of such strong pulses
of tree death on ecosystem-level GPP is less clear because
mortality reduces LAI while simultaneously increasing radiation availability within the canopy. However, over the long
term, mortality clearly commits substantial carbon to the atmosphere through the breakdown of additional necromass. If
the high mortality pulse observed at the Tapajós TFE experiment after 3 years of drought (5.4 Mg C ha−1) [Brando et al.,
2008] occurred widely over the region, it would represent a
large net committed emission to the atmosphere. Although
mortality at Tapajós declined to 2–4% under the following
2 years of TFE treatment [Brando et al., 2008], the live biomass removed during this single mortality event represented
up to 8.5 years of aboveground growth under normal rainfall
[Nepstad et al., 2007].
meir et al. 441
The findings from the TFE experiments are consistent with
recent observational reports of natural drought effects on
Amazonian forest trees. For example, following the severe
drought event of 2005, tree growth observations in 55 1-ha
plots distributed across Amazonia demonstrated droughtinduced reductions in stem growth (especially in larger trees)
and significant, but spatially patchy, increases in mortality
[Phillips et al., 2009]. The 2005 drought in Amazonia was
less severe or prolonged than the soil moisture limitation
imposed in the TFE experiments and comprised additional
climatic impacts on temperature, atmospheric humidity, and
radiation, yet Phillips et al. [2009] estimated an overall reduction in aboveground growth of 5.3 Mg ha−1, in addition to
a substantial increase in mortality-committed carbon emissions. The regional-scale spatial variation in mortality observed during 2005 was dependent on differences in climate,
soil-type, and species, with a tendency for higher mortality
in species with lower density wood [Phillips et al., 2009].
Such spatial variation in the response to drought was also
evident in a recent pantropical survey of recent ENSO impacts on mortality [Meir and Grace, 2005; Meir et al., 2008].
Taken together with evidence of species differences in photosynthesis, growth, and reproduction from the TFE experiments [Ehleringer et al., 2004; Fisher et al., 2006; Brando et
al., 2006], these results imply initial, but spatially variable,
resistance to short-term soil moisture limitation, followed by
increased mortality and likely alterations in tree community
composition as the severity of drought deepens and extends.
The longer term (multiyear to decadal) effects of drought
on NEP are not well constrained by any current data sets,
but may contain surprises. As observed in the TFE experiments and elsewhere, short-term reductions in GPP and Rs
are likely, and increased mortality and fire incidence will increase losses of carbon to the atmosphere. However, recent
observations of significant medium-term (5 year) increases
in leaf respiration at the Caxiuanã TFE experiment [Meir et
al., 2008; Metcalfe et al., submitted manuscript, 2009] suggest unexpected additional foliar emissions of CO2 during
drought, even after correcting for temperature. Although
previously not considered in NEP calculations, this effect
has been reported for other trees experiencing low rainfall
[Turnbull et al., 2001; Wright et al., 2006] and, as well as
reducing NEP, may also increase mortality risk through excessive metabolism of carbon reserves. Other unexpected
drought response processes may need to be considered in
the future, including the potential for changes in isoprene
emissions, currently a small component of the carbon cycle
[Pegoraro et al., 2004], in pathogen attacks, known to be
substantial during drought stress in other forest ecosystems
[Ayers and Lombardero, 2000] and in soil fungal activity
[Meir et al., 2006].
6. Modeling Twenty-First Century Drought
Impacts on Amazonian Rain Forests
6.1. Short-Term Effects: Seasonal to Interannual
Notwithstanding the possible discovery of new long-term
drought responses, correct attribution of physiological pro­
cesses at seasonal or interannual timescales is needed to interpret the effects of climate anomalies, such as ENSO or
the 2005 drought in Amazonia, and to provide the basis for
robust model predictions.
The TFE experiments simulated rainfall reductions similar
to that of a severe ENSO, such as the 1997/1998 event [Meir
et al., 2008], but the results are only beginning to be incorporated into vegetation modeling frameworks. Fisher et al.
[2007] successfully simulated the effects of the Caxiuanã
TFE manipulation on GPP, specifying a detailed multilayer
soil and canopy physiological model [Williams et al., 1996]
using measurements from the TFE experiment of soil hydraulic properties, leaf biochemical photosynthetic capacity
and LAI, and meteorological data (Figure 6). Gas exchange
was validated at leaf and canopy scales using independent
leaf-scale stomatal conductance [Fisher et al., 2006] and
tree-scale sap flux [Fisher et al., 2007] measurements. The
analysis demonstrated that GPP and transpiration at Caxiuanã are not constrained by moisture supply under normal
climatic variation [see Saleska et al., this volume], but that
more severe moisture limitation imposed strong constraints
upon transpiration and GPP. Transpiration declined by 30–
40% (300–418 mm a−1) and GPP by 13–14% (4.0–4.3 t C
ha−1 a−1) during the first 2 years of experimental drought at
Caxiuanã [Fisher et al., 2007]. Changes in LAI and hydraulic (rather than biochemical) properties were the principal
determining parameters: maximum foliar stomatal conductance declined by more than 50% and belowground hydraulic resistance increased more than 10-fold. Combined with
changes in heterotrophic and autotrophic respiration [Meir et
al., 2008], the impact of drought on NEP is probably finely
balanced and closer to zero than the large carbon emissions
predicted by earlier model analyses of the effects on NEP of
the 1997/1998 ENSO drought [e.g., Tian et al., 1998; Zeng
et al., 2005; Peylin et al., 2005].
Coarser-scale models have also been used to simulate
Amazonian drought impacts. Potter et al. [2004] used the
Carnegic Ames Stanford Approach (CASA) model to quantify rain forest ecosystem functioning during drought, partially driving simulations with remotely sensed data. In this
work, Rs was more moisture sensitive than in earlier model
analyses, and in a subsequent development, the spectral signal from the Tapajós canopy to drought was also successfully incorporated [Asner et al., 2004], offering the future
442 EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
Figure 6. The modeled response to normal rainfall and TFE treatment by GPP and canopy (stomatal) conductance to
water vapor, at Caxiuanã: (top) incident rainfall, (middle) stomatal conductance at canopy scale, and (bottom) GPP. Open
circles denote TFE treatment forest; solid circles denote control forest. From Fisher et al. [2007], reprinted with permission from Wiley-Blackwell.
prospect of detecting drought stress from space. Recently, in
a third-generation development of the SiB ecosystem model
[Sellers et al., 1986], Baker et al. [2008] incorporated new
soil depth and rooting properties observed at Tapajós [Oliveira et al., 2005; Lee et al., 2005], together with a seasonal
soil moisture response function for Rs observed at a separate
site near the Tapajós TFE experiment [Saleska et al., 2003].
The new model formulation successfully replicated mea­
surements of the seasonal variation in forest-atmosphere carbon exchange made using eddy covariance at Tapajós. Only
with the combination of several newly observed mechanisms
was it possible to simulate seasonality in carbon exchange
correctly with this model [Baker et al., 2008]. However,
getting water supply mechanistically correct should ideally
start with correctly parameterized soil hydraulic properties,
as highlighted by Harris et al. [2004] for a central Amazonian site near Manaus [cf. Fisher et al., 2008]. Whether
the successful reformulation of SiB by Baker et al. [2008]
included a representation of site-corrected soil hydraulic parameters is unclear, but this could strongly influence modeled PAW and gas exchange at low soil moisture contents,
and underlines the importance of ongoing data-model validation efforts.
6.2. Longer-Term Effects
At longer (decadal-to-century) timescales, it has been
necessary to derive the physiological impacts on forest carbon metabolism of twenty-first century increases in drought
stress, temperature, and atmospheric CO2 concentration
from physiological principles, as well as from measurements
made over shorter time periods [Betts, 2004]. In a recent
review, Lloyd and Farquhar [2008] argued that the positive
effect of increased atmospheric CO2 concentration on photosynthesis is likely to outweigh any negative impacts of concurrent warming, and this will probably balance in favor of
a positive impact on NEP. In addition, the tendency in most
plants to reduce stomatal conductance at high atmospheric
CO2 concentration makes possible reductions in water loss
through transpiration without diminishing carbon acquisition, thus further increasing the resistance of vegetation
to climatic drying. However, there remains uncertainty in
transpiration estimates, as under drought and/or warming, a
drier atmosphere will impose a bigger atmospheric demand
on evaporation potentially leading instead to higher rates
of evapotranspiration [Salazar et al., 2007] irrespective of
reductions in stomatal conductance. Plant respiration [e.g.,
meir et al. 443
Meir et al., 2001] now seems likely to acclimate at higher
temperatures [Atkin and Tjoelker, 2003; Atkin et al., 2008],
and this would confer drought resistance through reduced
use of plant carbon reserves. But whether stomatal conductance responses to increased atmospheric CO2 concentration
over the long term remain similar to short-term measurements is uncertain, and in any case, the carbon economy of
trees and ecosystems may be further influenced by changes
in LAI and the drought response in leaf respiration [Atkin
and Macherel, 2009; Meir et al., 2008]. Of course, over such
longer timescales, any resistance to drought based on plant
physiology may also be strongly and negatively impacted by
pest or pathogen attack [Ayers and Lombardero, 2000; Meir
et al., 2006], or by an increase in the frequency of extreme
weather events [IPCC Working Group I, 2007].
Incorporating many of these responses into vegetation
models operating over the time periods required to simulate
vegetation change is still at an early stage [Meir et al., 2006;
Ostle et al., 2009]. However, analysis of the drought response
mechanisms specified in different dynamic global vegetation models (DGVMs) has identified the need for corrections
to some process representations. The mortality risks recently
quantified for natural and more severe drought [Nepstad et
al., 2007; Meir et al., 2008; Phillips et al., 2009] have not
yet been incorporated into current modeling frameworks and
require specific model structures or parameterization to do
so [Moorcroft, 2006]. Furthermore, differences exist among
different DGVMs in the allocation of fixed carbon to aboveand belowground ecosystem components, and this has a significant impact on the response in Rs to climatic warming
and hence to changes in NEP [Dufresne et al., 2002]. Recent new insight into carbon allocation pro­cesses [see Malhi
et al., this volume) should inform this issue further. More
surprisingly, Galbraith et al. (submitted manuscript, 2009),
analyzing three widely used DGVMs [Cox et al., 2000; Levy
et al., 2004; Sitch et al., 2003], have shown that the modeled “dieback response” in Amazonian vegetation [Cox et
al., 2000; Sitch et al., 2008] is more strongly dependent on
the specified temperature responses in respiration and photosynthesis than on the direct effects of moisture limitation,
despite observational evidence elsewhere of acclimation to
temperature in plant respiration [Atkin and Tjoelker, 2003],
and the effects of drought summarized here. Thus, the challenge now is to incorporate the range of observed moisture
limitation effects correctly into DGVMs and other vegetation
modeling frameworks. Getting the balance right between
the drought-buffering effects of above- and belowground
ecosystem components and representing them at the correct
scale will require a two-way interaction between data providers and modelers. This work will improve the modeling
of vegetation-atmosphere interactions during drought, but to
understand the overall effects of drought on Amazonian rain
forests, further outward links to models of fire risk, and land
use change are also needed.
6.3. Drought and Fire
Switches in vegetation cover under climatic drying and
warming have been predicted using both dynamic and equilibrium vegetation model analyses [Oyama and Nobre, 2003;
Salazar et al., 2007; Sitch et al., 2008], but none of these
simulations is likely to be realistic without the inclusion of
fire risk. Lags in the development of natural vegetation under climatic change outside the fundamental niche of many
tree species may occur because trees are large, resistant,
and long-lived organisms. However, fire has the potential to
switch forest to savanna or grassland, short-circuiting these
lags and rapidly accelerating natural rates of climate-driven
vegetation change.
The networks of positive feedbacks among climatic
warming and drying, deforestation, forest fragmentation,
and fire have been described in detail previously [Nepstad
et al., 1999, 2001; Soares-Filho, 2006], and the southeastern sector of Amazonia seems most vulnerable to forest loss,
as high drought risk and high rates of deforestation overlap
[Malhi et al., 2008]. However, dynamic integrated models of
climate, fire risk, vegetation, and deforestation have not yet
been developed very far [Nepstad et al., 2008]. In one such
early study, Golding and Betts [2008] demonstrated substantially increased fire risk across the region by 2020, rising
to a “high” risk of fire across 50% of the region by 2080.
This analysis superimposed deforestation scenarios [SoaresFilho et al., 2006; van Vuuren et al., 2007] and a simple
fire model [a forest fire danger index (FFDI), parameterized
in Australia] [Noble et al., 1980; Hoffmann et al., 2003] on
an ensemble climate model analysis using HADCM3 that
incorporated the vegetation response in a simplified way
through altered GCM parameter sets [Golding and Betts,
2008]. The next steps in this process will be to incorporate
the flammability estimates and vegetation responses from
the TFE and other observational data into a fully functioning
GCM-DGVM-fire vulnerability framework.
The Australia-derived FFDI model used by Betts [2008]
does not consider fire vulnerability in the forest understory,
a frequent precursor to subsequent full-canopy fire events
in Amazonian forests [Nepstad et al., 2001]. Hence, the results from the TFE experiments (Figure 5) probably indicate a higher vulnerability to fire than specified by the FFDI
model: under scenarios of stronger drying or greater deforestation, the biophysical component of the risk to forest loss
estimated by Betts et al. [2008] may prove conservative.
However, future fire risk is also strongly dependent on the
444 EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
nature of environmental governance in forested regions because of its impact on deforestation and other modes of land
use change [Malhi et al., 2009]. In this regard, there may
be some room for optimism. In particular, the potential to
mitigate the risk to fire and forest degradation through local
and regional governance mechanisms is growing rapidly in
parts of Amazonia [Nepstad et al., 2006; Soares-Filho et al.,
2006]. The added possibility of national and international
agreements that may permit and help finance sustainable
land use based on payments for forest ecosystem services
[Mitchell et al., 2008; Daily et al., 2009] may further reduce the risk of drought-related forest loss, thus delaying or
avoiding some of the more extreme “dieback” scenarios that
have been modeled for the region.
7. Conclusions
Drought in Amazonia cannot be represented as a single
parameter, and current modeling trends are moving toward
the representation of the suite of changes in climate, vegetation, soil, fire, and land use that map to this term. The LBA
observational network has enriched our capacity to specify
such modeling frameworks. The manipulative experiments
described here form part of this network and comprise an
important means by which we can test the mechanistic basis of the relevant ecological processes, as well as validate
measured gross ecosystem fluxes, and thus provide confidence in model predictions of future scenarios.
Validating process representation in models at multiyear
or longer timescales is becoming increasingly important,
and the value of long-term experimental data is increasingly
recognized [Sitch et al., 2008]. Combining the experimental
results summarized here with observational data from across
the basin, a picture of the impacts of drought on Amazonian
vegetation is beginning to emerge. This understanding needs
to be incorporated into the next generation of DGVMs and
can currently be summarized as follows. (1) Resistance to
seasonal drought in rain forest functioning is likely at most
locations, especially climatically marginal sites. (2) Even under severe drought, such as that imposed in the TFE experiments, a surprising degree of resistance is probable at first,
although spatially patchy increases in mortality were observed
during the 2005 drought. In the short term, while strong responses in GPP, transpiration, respiration, and LAI can occur during periods of strong moisture limitation of up to 24
months, overall NEP may change only slightly. (3) As severe
drought extends to 3–5 years and beyond, mortality increases
markedly, and species differences may emerge in terms of
loss, survival, reproduction, and regrowth, with substantial
negative impacts on NEP and transpiration, and substantial
increases in vulnerability to fire. (4) At longer timescales,
our ability to tightly constrain the tempo and mode of any
drought-driven tipping point in forest function is limited by
data availability, but the increased risk of fire associated with
drought means that switches in vegetation type at decadal or
greater timescales will rely as much on human activities as on
climate.
New results are still emerging from the three soil moisture manipulation experiments at Caxiuanã, Tapajós, and
Castanhal. The rainfall exclusion at Tapajós has now been
halted, and the processes governing recovery from drought
are under analysis [e.g., Davidson et al., 2008]; at Caxiuanã and Castanhal, the experimental treatments continue.
The possibility of long-term data sets offered by these experiments represents a uniquely powerful way by which we
can begin to understand how multiyear and decadal-scale
drought will impact the species composition, vulnerability, and gas exchange properties of Amazonian vegetation.
The standard science funding cycle of 3 or 5 years is insufficient to fully address ecological questions of this sort,
and although this issue has been recognized within LBA
and elsewhere [Hobbie et al., 2003], the case for supporting
long-term ecological studies is urgent and needs to be made
more widely. As the experiments, data, and model analyses are extended, some of the key emerging science challenges are likely to include at least some of the following
questions:
1. What is the minimum set of rooting and soil depth
properties required to model vegetation and soil function
adequately during drought [Woodward and Osborne, 2000;
Bruno et al., 2006; Metcalfe et al., 2008]?
2. How can plant hydraulic and biochemical sensitivities be represented accurately at large scale, including their
role in affecting tree mortality [Fisher et al., 2006, 2007;
McDowell et al., 2008]?
3. How can respiration in plants and microbes be represented over short and long timescales under climatic warming and drying? [Trumbore, 2006; Meir et al., 2008]
4. Can mortality risks and alterations to reproductive output be modeled to predict change in vegetation properties
and species composition under drought [Brando et al., 2008;
Meir et al., 2008; Phillips et al., 2009]?
5. What are the sensitivities in the components of NEP
and the allocation of net primary production to PAW and
temperature [Brando et al., 2008; Galbraith et al., submitted
manuscript, 2009]?
6. Can we model recovery from drought, and how long
does recovery take [Brando et al., 2008]?
7. How will twenty-first century land use change, fire incidence, and drought interact to affect rain forest functioning
or a transition from rain forest to different vegetation types
[Soares-Filho et al., 2006; Betts et al., 2008]?
meir et al. 445
Acknowledgments. We would like to thank several funding bodies for the initiation and continuation of research funding support
at all three sites from 1999 to 2009: LBA, Brazil; NERC, UK; EU
5th Framework Programme; NASA, USA; NSF, USA; CNPq, Brazil. We also thank the institutes responsible for maintenance of the
reserves at each site and for providing permission and support for
this work: MPEG, Belem, Pará; Santarem, Pará; UFRA, Castanhal,
Pará. A large number of Brazilian students have been trained at B.S.,
M.S., and Ph.D. level during the running of the three experiments,
and we thank LBA for making this training support available.
References
Alencar, A. A. C., L. A. Solorzano, and D. C. Nepstad (2004),
Modeling forest understory fires in an eastern Amazonian landscape, Ecol. Appl., 14, S139–S149.
Alencar, A. A. C., D. Nepstad, and M. D. V Diaz (2006), Forest
understory fire in the Brazilian Amazon in ENSO and non-ENSO
years: Area burned and committed carbon emissions, Earth
Interact., 10(6), EI150, doi:10.1175/EI150.1.
Aragão, L. E. O. C., Y. Malhi, R. M. Roman-Cuesta, S. Saatchi,
L. O. Anderson, and Y. E. Shimabukuro (2007), Spatial patterns
and fire response of recent Amazonian droughts, Geophys. Res.
Lett., 34, L07701, doi:10.1029/2006GL028946.
Asner, G. P., D. Nepstad, G. Cardinot, and D. Ray (2004), Drought
stress and carbon uptake in an Amazon forest measured with
spaceborne imaging spectroscopy, Proc. Natl. Acad. Sci. U. S.
A., 101, 6039–6044.
Atkin, O. K., and D. Macherel (2009), The crucial role of plant mitochondria in orchestrating drought tolerance, Ann. Bot., 103(4),
581–597, doi:10.1093/aob/mcn094.
Atkin, O. K., and M. G. Tjoelker (2003), Thermal acclimation and
the dynamic response of plant respiration to temperature, Trends
Plant Sci., 8, 343–351.
Atkin, O. K., L. J. Atkinson, R. A. Fisher, C. D. Campbell,
J. Zaragoza-Castells, J. W. Pitchford, F. I. Woodward, and V.
Hurry (2008), Using temperature-dependent changes in leaf
scaling relationships to quantitatively account for thermal acclimation of respiration in a coupled global climate-vegetation
model, Global Change Biol., 14, 2709–2726.
Ayers, M. P., and M. J. Lombardero (2000), Assessing the consequences of global change for forest disturbances for herbivores
and pathogens, Sci. Total Env., 262, 263–286.
Baker, I. T., L. Prihodko, A. S. Denning, M. Goulden, S. Miller,
and H. R. da Rocha (2008), Seasonal drought stress in the Amazon: Reconciling models and observations, J. Geophys. Res.,
113, G00B01, doi:10.1029/2007JG000644.
Baker, T. R., et al. (2004), Variation in wood density determines
spatial patterns in Amazonian forest biomass, Global Change
Biol., 10, 1–18, doi:10.1111/j.1529-8817.2003.00751.x.
Belk, E. L., D. Markewitz, T. C. Rasmussen, E. J. M. Carvalho,
D. C. Nepstad, and E. A. Davidson (2007), Modeling the effects
of throughfall reduction on soil water content in a Brazilian
Oxisol under a moist tropical forest, Water Resour. Res., 43,
W08432, doi:10.1029/2006WR005493.
Betts, R. A. (2004), Global vegetation and climate: Self-beneficial
effects, climate forcings and climate feedbacks, J Phys. IV, 121,
37–60.
Bousquet, P., P. Peylin, P. Ciais, C. Le Quéré, P. Friedlingstein,
and P. P. Tans (2000), Regional changes in carbon dioxide fluxes
of land and oceans since 1980, Science, 290, 1342–1346.
Brando, P., et al. (2006), Effects of partial throughfall exclusion on
the phenology of Coussarea racemosa (Rubiaceae) in an eastcentral Amazon rainforest, Oecologia, 150, 181–189.
Brando, P. M., D. C. Nepstad, E. A. Davidson, S. E. Trumbore,
D. Ray, and P. Camargo (2008), Drought effects on litterfall,
wood production and belowground carbon cycling in an Amazon
forest: Results of a throughfall reduction experiment, Philos.
Trans. R. Soc., Ser. B, 363, 1839–1848.
Bruno, R. D., H. R. da Rocha, H. C. De Freitas, M. L. Goulden, and
S. D. Miller (2006), Soil moisture dynamics in an eastern Amazonian tropical forest, Hydrol. Processes, 20, 2477–2489.
Cardoso, M. F., G. C. Hurtt, B. Moore, C. A. Nobre, and E. Prins
(2003), Projecting future fire activity in Amazonia, Global Change
Biol., 9, 656–669, doi:l 0.1046/j.1365-2486.2003.00607.x.
Carswell, F. E., et al. (2002), Seasonality in CO2 and H2O flux at
an eastern Amazonian rain forest, J. Geophys. Res., 107(D20),
8076, doi:10.1029/2000JD000284.
Chagnon, F. J. F., and R. L. Bras (2005), Contemporary climate change in the Amazon, Geophys. Res. Lett., 32, L13703,
doi:10.1029/2005GL022722.
Cochrane, M. A. (2003), Fire science for rainforests, Nature, 421,
913–919.
Cochrane, M. A., and W. F. Laurance (2002), Fire as a largescale edge effect in Amazonian forests, J. Trop. Ecol., 18,
311–325.
Coelho, R. F. R., D. G. Zarin, I. S. Miranda, and J. M. Tucker
(2004), Análise florística e estrutural de uma floresta em diferentes estágios sucessionais no município de Castanhal, Pará,
Acta Amazonica, 33, 563–582.
Costa, M. H., S. N. M. Yanagi, P. J. O. P. Souza, A. Ribeiro, and
E. J. P. Rocha (2007), Climate change in Amazonia caused by
soybean cropland expansion, as compared to caused by pastureland expansion, Geophys. Res. Lett., 34, L07706, doi:10.1029/
2007GL029271.
Cox, P. M., R. A. Betts, C. D. Jones, S. A. Spall, and I. J.
Totterdell (2000), Acceleration of global warming due to carbon-cycle feedbacks in a coupled climate model, Nature, 408,
184–187.
Cox, P. M., P. P. Harris, C. Huntingford, R. A. Betts, M. Collins,
C. D. Jones, T. E. Jupp, J. A. Marengo, and C. A. Nobre (2008),
Increasing risk of Amazonian drought due to decreasing aerosol
pollution, Nature, 453, 212–215.
Culf, A. D., G. Fisch, and M. G. Hodnett (1995), The albedo of
Amazonian forest and ranchland, J. Clim., 8, 1544–1554.
Daily, G. C., et al. (2009), Ecosystem services in decision
making: Time to deliver, Front. Ecol. Environ., 7, 21–28,
doi:10.1890/080025.
da Rocha, H. R., et al. (2009), Patterns of water and heat flux across
a biome gradient from tropical forest to savanna in Brazil, J.
Geophys. Res., 114, G00B12, doi:10.1029/2007JG000640.
446 EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
da Rocha, H. R., A. O. Manzi, and J. Shuttleworth (2009), Evapotranspiration, Geophys. Monogr. Ser., doi:10.1029/2008GM000744,
this volume.
Davidson, E. A. (1992), Sources of nitric oxide and nitrous oxide
following wetting of dry soil, Soil Sci. Soc. Am. J., 56, 95–102.
Davidson, E. A., L. V. Verchot, J. H. Cattanio, I. L. Ackerman, and
J. E. M. Carvalho (2000), Effects of soil water content on soil
respiration in forests and cattle pastures of eastern Amazonia,
Biogeochemistry, 48, 53–69.
Davidson, E. A., F. Y. Ishida, and D. C. Nepstad (2004), Effects
of an experimental drought on soil emissions of carbon dioxide,
methane, nitrous oxide, and nitric oxide in a moist tropical forest, Global Change Biol., 10, 718–730.
Davidson, E. A., et al. (2008), Trace gas emissions Tapajos 5 years,
Global Change Biol., 14, 2582–2590.
Dufresne, J. L., P. Friedlingstein, M. Berthelot, L. Bopp, P. Ciais,
L. Fairhead, H. Le Treut, and P. Monfray (2002). On the magnitude of positive feedback between future climate change and
the carbon cycle, Geophys. Res. Lett., 29(10), 1405, doi:10.1029/
2001GL013777.
Ehleringer, J. R., et al. (2004), The Third International Conference of the Large-Scale Biosphere Atmosphere Program, Available at ftp://lba.cptec.inpe.br/presentations/LBA-III-ConferenceJuly2004-Brasilia/July28,2004/S15/12_7%20Ehleringer%20Sta
ble%20isotope%20analyses.pdf
Engelbrecht, B. M. J., L. S. Comita, R. Condit, T. A. Kursar,
M. T. Tyree, B. L. Turner, and S. P. Hubbell (2007), Drought
sensitivity shapes species distribution patterns in tropical forests,
Nature, 447, 80–82.
Fisher, R. A., M. Williams, R. Lobo do Vale, A. C. Lola da Costa,
and P. Meir (2006) Evidence from Amazonian forests is consistent with isohydric control of leaf water potential, Plant Cell
Environ., 29, 151–165.
Fisher, R. A., M. Williams, A. L. Costa, Y. Malhi, R. F. da Costa,
S. Almeida, and P. Meir (2007), The response of an Eastern
Amazonian rain forest to drought stress: Results and modelling analyses from a throughfall exclusion experiment, Global
Change Biol., 13, 2361–2378.
Fisher, R. A., M. Williams, M. de Lourdes Ruivo, A. L. Costa,
and P. Meir (2008), Evaluating climatic and soil water controls
on evapotranspiration at two Amazonian rainforest sites, Agric.
For. Meteorol., 148, 850–861.
Foley, J. A., A. Botta, M. T. Coe, and M. H. Costa (2002), El
Niño–Southern oscillation and the climate, ecosystems and
rivers of Amazonia, Global Biogeochem. Cycles, 16(4), 1132,
doi:10.1029/2002GB001872.
Fortini, L. B., S. S. Mulkey, D. J. Zarin, S. S. Vasconcelos and C. R.
J. D Carvalho (2003), Drought constraints on leaf gas exchange by
Miconia ciliata (Melastomataceae) in the understory of an eastern
Amazonian regrowth forest stand, Am. J. Bot., 90, 1064–1070.
Forster, P., et al. (2007), Changes in atmospheric constituents and
in radiative forcing, Climate Change 2007: The Physical Science
Basis, edited by S. Solomon et al, pp. 129–234, Cambridge Univ.
Press, Cambridge, U. K.
Franks, P. J., P. L. Drake, and R. H. Froend (2007), Anisohydric
but isohydrodynamic: Seasonally constant plant water gradient
explained by a stomatal control mechanism incorporating variable plant hydraulic conductance, Plant Cell Environ., 30, 19–30.
Friedlingstein, P., et al. (2006), Climate-carbon cycle feedback
analysis: Results from the C4MIP model intercomparison, J.
Clim., 19, 3337–3353.
Furley, P. A., J. Proctor, and J. A. Ratter (1992), Nature and Dynamics of Forest-Savanna Boundaries, CRC Press, London.
Golding, N., and R. Betts (2008), Fire risk in Amazonia due to
climate change in the HadCM3 climate model: Potential interactions with deforestation, Global Biogeochem. Cycles, 22,
GB4007, doi:10.1029/2007GB003166.
Grace, J., J. Lloyd, J. McIntyre, A. C. Miranda, P. Meir, H. S.
Miranda, C. Nobre, J. Moncrieff, J. Massheder, and Y. Malhi
(1995), Carbon dioxide uptake by an undisturbed tropical rain
forest in southwest Amazonia, 1992 to 1993, Science, 270, 778.
Harris, P., C. Huntingford, P. M. Cox, J. H. C. Gash, and Y. Malhi
(2004), Effect of soil moisture on canopy conductance of Amazonian rainforest, Agric. For. Meteorol., 122, 215–227.
Hobbie, J. E., S. R. Carpenter, N. B. Grimm, J. R. Gosz, and T. R.
Seastedt (2003), The US Long Term Ecological Research program, BioScience, 53, 21–32.
Hoffmann, W. A., W. Schroeder, and R. B. Jackson (2003), Regional feedbacks among fire, climate, and tropical deforestation,
J. Geophys. Res., 108(D23), 4721, doi:10.1029/2003JD003494.
Houghton, R. A., M. Gloor, J. Lloyd, and C. Potter (2009), The
regional carbon budget, Geophys. Monogr. Ser., doi:10.1029/
2008GM000718, this volume.
Howard, P. J. A. (1979), Respiration of decomposing litter in relation to temperature and moisture, Oikos, 33, 457–465.
Huete, A., K. Didan, W. van Leeuwen, T. Miura, and E. Glenn
(2008), MODIS vegetation indices, Land Remote Sensing and
Global Environmental Change: NASA’s Earth Observing System and the Science of ASTER and MODIS.
Huntingford, C., J. A. Lowe, B. B. B. Booth, C. D. Jones, G. R.
Harris, and P. Meir (2009), How large is the effect of model
uncertainty in the global carbon cycle compared to model uncertainty in thermal properties of the Earth system when predicting temperature increases by year 2100?, Tellus, Ser. B, 61,
355–360.
Hurlbert, S. H. (2004), On misinterpretations of pseudoreplication
and related matters: A reply to Oksanen, Oikos, 104, 591.
Hutyra, L. R., J. W. Munger, C. A. Nobre, S. R. Saleska, S. A.
Vieira, and S. C. Wofsy (2005), Climatic variability and vegetation vulnerability in Amazonia, Geophys. Res. Lett., 32, L24712,
doi:10.1029/2005GL024981.
IPCC Working Group I (2007), Regional Climate Projections. in
Climate Change 2007: The Physical Science Basis. Contribution
of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, edited by S. Solomon et
al., pp. 847–940, Cambridge Univ. Press, Cambridge, U. K.
Jipp, P. H., D. C. Nepstad, D. K. Cassel, and C. Reis de Carvalho (1998),
Deep soil moisture storage and transpiration in forests and pastures of seasonally-dry Amazonia, Clim. Change, 39, 395–412.
Joslin, J. D., M. H. Wolfe, and P. J. Hanson (2000), Effects of altered water regimes on forest root systems, New Phytol., 147,
117–129.
meir et al. 447
Langenfelds, R. L., R. J. Francey, B. C. Pak, L. P. Steele, J.
Lloyd, C. M. Trudinger, and C. E. Allison (2002), Interannual
growth rate variations of atmospheric CO2 and its δ13C, H2,
CH4, and CO between 1992 and 1999 linked to biomass burning, Glo­bal Biogeochem. Cycles, 16(3), 1048, doi:10.1029/
2001GB­001466.
Lee, J. E., R. S. Oliveira, T. E. Dawson, and I. Fung (2005), Root
functioning modifies seasonal climate, Proc. Nat. Acad Sci. U. S.
A., 102, 17,576–17,581.
Levy, P. E., M. G. R Cannell, and A. D. Friend (2004), Modelling
the impact of future changes in climate, CO2 concentration and
land use on natural ecosystems and the terrestrial carbon sink,
Global Environ. Change, 14, 21–30.
Likens, G. E., F. H. Bormann, N. M. Johnson, D. W. Fisher, and
R. S. Pierce (1970), Effects of forest cutting and herbicide treatment on nutrient budgets in the Hubbard Brook watershedecosystem, Ecol. Monogr., 40(1), 23–47.
Lloyd, J., and G. D. Farquhar (2008), Effects of rising temperatures and [CO2] on the physiology of tropical forest trees, Philos.
Trans. R. Soc. Ser. B, 363, 1811–1817.
Lloyd, J., J. Grace, A. C. Miranda, P. Meir, S. C. Wong, B. S. Miranda, I. R. Wright, J. H. C. Gash, and J. McIntyre (1995), A simple calibrated model of Amazon rain forest productivity based on
leaf biochemical properties, Plant Cell Environ., 18, 1129–1145.
Malhi, Y., A. D. Nobre, J. Grace, B. Kruijt, M. G. P. Pereira, A. Culf,
and S. Scott (1998), Carbon dioxide transfer over a Central Amazonian rain forest, J. Geophys. Res., 103(D24), 31,593–31,612.
Malhi, Y., J. T. Roberts, R. A. Betts, T. J. Killeen, W. Li, and C. A.
Nobre (2008), Climate change, deforestation, and the fate of the
Amazon, Science, 319, 169.
Malhi, Y., et al. (2009), Exploring the likelihood and mechanism
of a climate-change-induced dieback of the Amazon rainforest.
Proc. Natl. Acad. Sci. U. S. A., doi:10.1073/pnas.0804619106.
Malhi, Y., S. Saatchi, C. Girardin, and L. E. O. C. Aragão (2009),
The production, storage, and flow of carbon in Amazonian forests, Geophys. Monogr. Ser., doi: 10.1029/2008GM000779, this
volume.
Marengo, J. A., C. A. Nobre, J. Tomasella, M. D. Oyama, G. S.
De Oliveira, R. De Oliveira, H. Camargo, L. M. Alves, and I. F.
Brown (2008), The drought of Amazonia in 2005, J. Clim., 21,
495–516.
McDowell, N., et al. (2008), Mechanisms of plant survival and
mortality during drought: Why do some plants survive while
others succumb to drought?, New Phytol., 178, 719–739.
Meir, P., and J. Grace (2005). The response to drought by tropical
rain forest ecosystems, in Tropical Forests and Global Climate
Change, edited by Y. Malhi and O. Phillips, pp. 71–80, Oxford
Univ. Press, Oxford, U. K.
Meir, P., J. Grace, and A. C. Miranda (2001), Leaf respiration in
two tropical rain forests: constraints on physiology by phosphorus, nitrogen and temperature, Funct. Ecol., 15, 378–387.
Meir, P., P. Cox, and J. Grace (2006), The influence of terrestrial
ecosystems on climate, Trends Ecol. Evol., 21, 254–260.
Meir, P., D. B. Metcalfe, A. C. L. Costa, and R. A. Fisher (2008), The
fate of assimilated carbon during drought: Impacts on respiration in
Amazon rainforests, Philos. Trans. R. Soc. Ser. B, 363, 1849–1855.
Metcalfe, D. B., P. Meir, L. E. O. C. Aragão Y. Malhi, A. C. L.
da Costa, A. Braga, P. H. L. Gonçalves, J. de Athaydes, S. S.
de Almeida, and M. Williams (2007), Factors controlling spatio-temporal variation in carbon dioxide efflux from surface litter, roots, and soil organic matter at four rain forest sites in the
eastern Amazon, J. Geophys. Res., 112, G04001, doi:10.1029/
2007JG000443.
Metcalfe, D. B., P. Meir, L. Aragão, A. C. L. da Costa, A. P. Braga,
P. H. L. Gonçalves, J. de Athaydes Silva Jr., S. S. de Almeida,
L. A. Dawson, and Y. Malhi (2008), The effects of water availability on root growth and morphology in an Amazon rainforest,
Plant Soil, 311, 189–199.
Mitchell, A. W., K. Secoy, N. Mardas, M. Trivedi, and R. Howard
(2008), Forests now in the fight against climate change, Forest
Foresight Report 1.v3, Global Canopy Programme, Oxford.
Moorcroft, P. R. (2006), How close are we to a predictive science
of the biosphere?, Trends Ecol. Evol., 21, 400–407.
Nepstad, D. C., C. R. de Carvalho, E. A. Davidson, P. H. Jipp, P.
A. Lefebvre, G. H. Negreiros, E. D. da Silva, T. A. Stone, S. E.
Trumbore, and S. Vieira (1994), The role of deep roots in the hydrological and carbon cycles of Amazonian forests and pastures,
Nature, 372, 666–669.
Nepstad, D. C., et al. (1999), Large-scale impoverishment of Amazonian forests by logging and fire, Nature, 398, 505–508.
Nepstad, D. C., G. Carvalho, A. C. Barros, A. Alencar, J. P. Capobianco, J. Bishop, P. Moutinho, P. Lefebvre, U. L. Silva, and E.
Prins (2001), Road paving, fire regime feedbacks, and the future
of Amazon forests, For. Ecol. Manage., 154, 395–407.
Nepstad, D. C., et al. (2002), The effects of partial throughfall exclusion on canopy processes, aboveground production, and biogeochemistry of an Amazon forest, J. Geophys. Res., 107(D20),
8085, doi:10.1029/2001JD000360.
Nepstad, D. C., P. Lefebvre, U. L. Da Silva, J. Tomasella, P. Schlesinger, L. Solorzano, P. Moutinho, D. Ray, and J. G. Benito (2004),
Amazon drought and its implications for forest flammability and tree
growth: A basin-wide analysis, Global Change Biol., 10, 704–717.
Nepstad, D., et al. (2006), Inhibition of Amazon deforestation and
fire by parks and indigenous lands, Conserv. Biol., 20, 65–73.
Nepstad, D. C., I. Tohver, D. Ray, P. Moutinho, and G. Cardinot
(2007), Mortality of large trees and lianas following experimental drought in an Amazon forest, Ecology, 88, 2259–2269.
Noble, I. R., A. M. Gill, and G. A. V. Bary (1980), McArthur’s fire
danger meters expressed as equations, J. Ecol., 5, 201–203.
Nobre, C. A., P. J. Sellers, and J. Shukla (1991) Amazonian deforestation and regional climate change, J. Clim., 4, 957–988.
Oliveira, R. S., T. E. Dawson, S. S. O. Burgess, and D. C. Nepstad (2005), Hydraulic redistribution in three Amazonian trees,
Oecologia, 145, 354–363.
Ostle, N. J., et al. (2009), Integrating plant-soil interactions into
global carbon cycle models, J Ecol., 97, 851–863.
Oyama, M. D., and C. A. Nobre (2003), A new climate-vegetation equilibrium state for tropical South America, Geophys. Res.
Lett., 30(23), 2199, doi:10.1029/2003GL018600.
Pegoraro, E., et al. (2004), Effect of drought on isoprene emission
rates from leaves of Quercus virginiana Mill, Atmos. Environ.,
38(36), 6149–6156.
448 EFFECTS OF DROUGHT ON AMAZONIAN RAIN FORESTS
Peylin, P., P. Bousquet, C. Le Quéré, S. Sitch, P. Friedlingstein,
G. McKinley, N. Gruber, P. Rayner, and P. Ciais (2005), Multiple constraints on regional CO2 flux variations over land and
oceans, Global Biogeochem. Cycles, 19, GB1011, doi:10.1029/
2003GB002214.
Phillips, O. L., et al. (2009), Drought sensitivity of the Amazon
rainforest, Science, 323, 1344–1347.
Potter, C., et al. (2004), Understanding global teleconnections of
climate to regional model estimates of Amazon ecosystem carbon fluxes, Global Change Biol., 10, 693–703.
Prentice, C., and J. Lloyd (1998), C-quest in the Amazon, Nature,
396, 619.
Ray, D., D. Nepstad, and P. Moutinho (2005), Micrometeorological and canopy controls of fire susceptibility in a forested Amazon landscape, Ecol. Appl., 15, 1664–1678.
Rodenbeck, C., S. Houweling, M. Gloor, and M. Heimann (2003),
CO2 flux history 1982–2001 inferred from atmospheric data using a global inversion of atmospheric transport, Atmos. Chem.
Phys., 3, 1919–1964.
Ruivo, M., S. B. Pereira, B. Quanz, and P. Meir (2003), Characterization and classification of soils of the LBA experimental sites,
Caxiuana, PA, Revista de Ciencias Agrarias, 75–85.
Salazar, L. F., C. A. Nobre, and M. D. Oyama (2007), Climate
change consequences on the biome distribution in tropical
South America, Geophys. Res. Lett., 34, L09708, doi:10.1029/
2007GL029695.
Saleska, S. R., S. D. Miller, D. M. Matross, M. L. Goulden, S. C.
Wofsy, H. R. da Rocha, P. B de Camargo, P. Crill, B. C. Daube,
and H. C. de Freitas (2003), Carbon in Amazon forests: Unexpected seasonal fluxes and disturbance-induced losses, Science,
302, 1554–1557.
Saleska, S. R., K. Didan, A. R. Huete, and H. R. da Rocha (2007),
Amazon forests green-up during 2005 drought, Science, 318,
612.
Saleska, S., H. da Rocha, B. Kruijt, and A. Nobre (2009), Ecosystem carbon fluxes and Amazon forest metabolism, Geophys.
Monogr. Ser., doi:10.1029/2008GM000728, this volume.
Sampaio, G., C. Nobre, M. H. Costa, P. Satyamurty, B. S. SoaresFilho, and M. Cardoso (2007), Regional climate change over
eastern Amazonia caused by pasture and soybean cropland
expansion, Geophys. Res. Lett., 34, L17709, doi:10.1029/
2007GL030612.
Schwendenmann, L., E. Veldkamp, T. Brenes, J. J. O’Brien, and
J. Mackensen (2003), Spatial and temporal variation in soil CO2
efflux in an old-growth neotropical rain forest, La Selva, Costa
Rica, Biogeochemistry, 64, 111–128.
Schymanski, S. J., M. Sivapalan, M. L. Roderick, J. Beringer, and
L. B. Hutley (2008), An optimality-based model of the coupled
soil moisture and root dynamics, Hydrol. Earth Syst. Sci., 12,
913–932.
Sellers, P. J., Y. Mintz, Y. C. Sud, and A. Dalcher (1986), A simple
biosphere model (SiB) for use within general circulation models,
J. Atmos. Sci., 43, 505–531.
Sitch, S., et al. (2003), Evaluation of ecosystem dynamics, plant geography and terrestrial carbon cycling in the LPJ dynamic global
vegetation model, Global Change Biol., 9, 161–185.
Sitch, S., et al. (2008), Evaluation of the terrestrial carbon cycle,
future plant geography and climate-carbon cycle feedbacks using five Dynamic Global Vegetation Models (DGVMs), Global
Change Biol., 14, 25.
Soares-Filho, B. S., D. C. Nepstad, L. M. Curran, G. C. Cerqueira,
R. A. Garcia., C. A. Ramos, E. Voll, A. McDonald, P. Lefebvre,
and P. Schlesinger (2006), Modelling conservation in the Amazon basin, Nature, 440, 520–523.
Sotta, E. D., E. Veldkamp, L. Schwendenmann, B. R. Guimarães, R. K. Paixão, M. L. P. Ruivo, A. C. L Costa, and P. Meir
(2007), Effects of an induced drought on soil CO2 efflux and
soil CO2 production in an Eastern Amazonian rainforest, Brazil,
Global Change Biol., 13(10), 2218–2229, doi:10.1111/j.13652486.2007.01416.x.
Tardieu, F., and T. Simonneau (1998), Variability among species
of stomatal control under fluctuating soil water status and evaporative demand: Modelling isohydric and anisohydric behaviours,
J. Exp. Bot., 49, 419–432.
Tian, H., J. M. Melillo, D. W. Kicklighter, A. D. McGuire, J. V. K.
Helfrich, B. Moore, and C. J. Vorosmarty (1998), Effect of interannual climate variability on carbon storage in Amazonian
ecosystems, Nature, 396, 664–667.
Timmermann, A. (1999), Detecting the nonstationary response of
ENSO to greenhouse warming, J. Atmos. Sci., 56, 2313–2325.
Tomasella, J., and Hodnett, M. G. (1997), Estimating unsaturated
hydraulic conductivity of Brazilian soils using soil-water retention data, Soil Sci., 162, 703.
Trumbore, S. U. S. A, (2006), Carbon respired by terrestrial ecosystems—Recent progress and challenges, Global Change Biol.
12, 141–153.
Trumbore, S., and P. B. de Camargo (2009), Soil carbon dynamics, Geophys. Monogr. Ser., doi:10.1029/2008GM000741, this
volume.
Turnbull, M. H., et al. (2001), Responses of leaf respiration to temperature and leaf characteristics in three deciduous tree species
vary with site water availability, Tree Phys., 21, 571–578.
Tyree, M. T., and J. S. Sperry (1988), Do woody plants operate
near the point of catastrophic xylem dysfunction caused by dynamic water stress—Answers from a model, Plant Phys., 88,
574–580.
Uhl, C., and J. B. Kauffman (1990), Deforestation, fire susceptibility, and potential tree responses to fire in the eastern Amazon,
Ecology, 71, 437–449.
van der Werf, G. R., J. T. Randerson, G. J. Collatz, L. Giglio, P.
S. Kasibhatla, A. F. Arellano, S. C. Olsen, and E. S. Kasischke
(2004), Continental-scale partitioning of fire emissions during
the 1997 to 2001 El Niño/La Niña period, Science, 303, 73–76.
van Vuuren, D. P., P. L. Lucas, and H. Hilderink (2007), Downscaling drivers of global environmental change: Enabling use of
global SRES scenarios at the national and grid levels, Global Environ. Change, 17(1), 114–130, doi:10.1016/j.gloenvcha.2006.
04.004.
Vasconcelos, S. S., et al. (2004), Moisture and substrate availability
constrain soil trace gas fluxes in an eastern Amazonian regrowth
forest, Global Biogeochem. Cycles, 18, GB2009, doi:10.1029/
2003GB002210.
meir et al. 449
Vasconcelos, S. S., D. J. Zarin , M. M. Araújo , L. G. T RangelVasconcelos, C. J. R. De Carvalho, C. L. Staudhammer, and
F. A. Oliveira (2008), Effects of seasonality, litter removal and
dry-season irrigation on litterfall quantity and quality in eastern
Amazonian forest regrowth, Brazil, J. Trop. Ecol., 24, 27–38.
Vourlitis, G. L., J. de Souza Nogueira, N. P. Filho, W. Hoeger, F.
Raiter, M. S. Biudes, J. C. Arruda, V. B. Capistrano, J. L. Brito
de Faria, and F. de Almeida Lobo (2005), The sensitivity of diel
CO2 and H2O vapor exchange of a tropical transitional forest to
seasonal variation in meteorology and water availability, Earth
Interact., 9, 1–23, doi:10.1175/EI124.1.
Werth, D., and R. Avissar (2002), The local and global effects
of Amazon deforestation. J. Geophys. Res., 107(D20), 8087,
doi:10.1029/2001JD000717.
West, A. G., K. R. Hultine, J. S. Sperry, S. E. Bush, and J. R. Ehleringer (2008), Transpiration and hydraulic strategies in a pinonjuniper woodland, Ecol. Appl., 18, 911–927.
White, A., M. G. R. Cannell, and A. D. Friend (1999), Climate
change impacts on ecosystems and the terrestrial carbon sink: A
new assessment, Global Environ. Change, 9, S21–S30.
Williams, M., et al. (1996), Modelling the soil-plant-atmosphere
continuum in a Quercus-Acer stand at Harvard forest: The regulation of stomatal conductance by light, nitrogen and soil/plant
hydraulic properties, Plant Cell Environ., 19, 911–927.
Williams, M., Y. Malhi, A. D. Nobre, E. B. Rastetter, J. Grace, M.
G. P. Pereira (1998), Seasonal variation in net carbon exchange
and evapotranspiration in a Brazilian rainforest: A modelling
analysis, Plant Cell Environ., 21, 953–968.
Woodward, F. I., and C. P. Osborne (2000), The representation of
root processes in models addressing the responses of vegetation
to global change, New Phytol., 147, 223–232.
Wright, I. J., P. B. Reich, O. K. Atkin, C. H. Lusk, M. G. Tjoelker,
and M. Westoby (2006), Irradiance, temperature and rainfall
influence leaf dark respiration in woody plants: Evidence from
comparisons across 20 sites, New Phytol., 169, 309–319.
Zeng, N., A. Mariotti, and P. Wetzel (2005), Terrestrial mechanisms of interannual CO2 variability, Global Biogeochem. Cycles, 19, GB1016, doi:10.1029/2004GB002273.
S. Almeida, Museau Paraense Emilio Goeldi, Av. Perimetral
1901, Terra Firme, CEP 66077-830, Belém, PA, Brazil.
P. M. Brando and D. Zarin, Department of Botany, University of
Florida, Gainesville, FL 32611, USA.
G. Cardinot, Instituto de Pesquisa Ambiental da Amazônia, Av.
Nazaré 669, 66035-170, Belém, PA, Brazil.
A. C. L. Costa, Centro de Geociências, Universidade Federal do
Pará, Belém, CP 1611 66017-970, Brazil.
E. Davidson, Woods Hole Research Center, Falmouth, MA
02540, USA.
R. A. Fisher, Los Alamos National Laboratory, Los Alamos, NM
87545, USA.
P. Meir, School of Geosciences, University of Edinburgh, Edinburgh EH8 9XP, UK. ([email protected])
D. Nepstad, Gordon and Betty Moore Foundation, Palo Alto, CA
94304, USA.
E. D. Sotta, EMBRAPA Amapa, Rod. Juscelino Kubitscheck
km 05, 2600, CEP 68903-419, Macapa, AP, Caixa Postal 10 CEP
68906-970, Brazil.
S. Vasconcelos, EMBRAPA-Amazônia Oriental, Belém, PA
CEP 66095-100, Brazil.