REVIEW
https://doi.org/10.1038/s41586-018-0240-x
Triggers of tree mortality under drought
Brendan Choat
1
*, Timothy J. Brodribb
2
, Craig R. Brodersen
3
, Remko A. Duursma
1
, Rosana López
1,4
& Belinda E. Medlyn
1
Severe droughts have caused widespread tree mortality across many forest biomes with profound effects on the function
of ecosystems and carbon balance. Climate change is expected to intensify regional-scale droughts, focusing attention
on the physiological basis of drought-induced tree mortality. Recent work has shown that catastrophic failure of the
plant hydraulic system is a principal mechanism involved in extensive crown death and tree mortality during drought,
but the multi-dimensional response of trees to desiccation is complex. Here we focus on the current understanding of
tree hydraulic performance under drought, the identification of physiological thresholds that precipitate mortality and
the mechanisms of recovery after drought. Building on this, we discuss the potential application of hydraulic thresholds
to process-based models that predict mortality.
F
orests account for approximately 45% of global terrestrial carbon
stocks and have a key role in hydrological and nutrient cycles
1,2
.
They also provide a wide array of ecosystem services and are vital
for maintenance of biodiversity. While forests continue to face pressure
from expanding human populations, which drive changes in land use and
deforestation, the threat posed by climate change is less easily quantified.
Evidence from a range of sources suggests that rising atmospheric CO
2
concentrations have benefited forests, with CO
2
fertilization enabling an
increased leaf area index
3
, enhanced water-use efficiency
4
and greater
uptake of carbon globally
5
. However, extreme climate events, such as
heat waves, droughts, fires and storms, have the potential to offset these
benefits, causing widespread tree mortality and a net loss of CO
2
into the
atmosphere. Although forests are vulnerable to a wide range of extreme
climate events, drought and associated disturbances have the greatest
effect globally
6
. Recent projections
7
indicate that land surface warming
may lead to longer and more intense droughts, which has focused con-
cern on this area of research and the need for accurate predictions of
the effects of drought on forest ecosystems. In this Review, we examine
the physiological response of trees to drought, focusing on new insights
provided by rapid advances in our understanding of the hydraulic
function of plants.
Land plants require an efficient long-distance transport pathway to
lift water from the soil to the leaves at a rate that satisfies transpira-
tion
8
. In trees, the xylem tissue (wood) supplies water for all aspects of
plant function, including photosynthesis, growth and reproduction.
Damage to this hydraulic supply network as a consequence of severe
water stress has been identified as a key mechanism that is involved
in tree mortality during drought
9–11
. Recent experimental work has
quantitatively linked hydraulic failure thresholds to plant mortality
12,13
,
and field studies have demonstrated that hydraulic failure is a primary
pathway for extensive canopy death or plant mortality during natural
drought events
14–17
.
A number of other co-contributing factors may also have a role in the
death of trees during natural droughts
18
. In the absence of catastrophic
hydraulic failure, partial disruption of water transport and the regula-
tion of water loss from plants during drought may lead to an increased
likelihood of mortality through the depletion of carbohydrate reserves
used in respiration and increased vulnerability to pests and pathogens
11
.
Therefore, even in cases of co-morbidity, plant hydraulic traits occupy
a central role in determining survival during drought and the effects of
drought on carbon dynamics.
Here, we cover recent progress in our understanding of plant hydraulic
response to drought and the physiological mechanisms that govern recovery
of hydraulic function after drought. Although recent advances have
crystallized our understanding of plant hydraulic function and the
consequences of vascular impairment caused by drought stress, many
challenges remain. We evaluate recent attempts to integrate the hydraulic
traits of plants into process-based models of tree mortality with an
emphasis on major knowledge gaps.
Drought and forest mortality
The effect of future droughts will almost certainly be worsened by
increases in air temperature associated with global warming; when
natural droughts occur they will set in more quickly and be of greater
intensity
7
. Higher temperatures will usually result in greater evapo-
transpiration (the sum of evaporation and plant transpiration), thus
drying soil and plants more quickly than would be the case at lower
temperatures
19
. Droughts of this nature, termed ‘global change-type
droughts’, have had severe effects on exposed ecosystems including
mass tree mortality
20,21
.
Globally, drought is the most widespread stress factor that affects
forest carbon balance
6
with the potential to cause pronounced
depressions in gross primary productivity at regional and conti
-
nental scales
22,23
. The most notable effects of drought are manifested
in regional-scale forest mortality events, which can kill millions of
trees within short timescales. Recent high-profile examples include
extreme droughts in Texas and California, which are estimated to
have killed 300million and 102million trees, respectively
24–26
. Mass
tree mortality due to drought is not restricted to arid regions, having
been documented across many forest biomes including cool tem-
perate and tropical forests
14–16,27,28
. In tropical northern Australia,
the sudden die-off of more than 7,000ha of mangrove forest in 2015
was attributed to drought and extreme temperatures
28
. Although
such concentrated mortality events are yet to be observed in many
of the world’s most productive tropical ecosystems, drought events
in tropical rainforests (for example, the 2005 Amazon drought)
have resulted in marked increases in stem mortality and loss of
aboveground biomass
29
. Mortality is often skewed towards young
trees but recent evidence suggests that large, old trees are also
vulnerable
30,31
. Loss of large trees is particularly concerning because
they have a critical ecological role and have the largest biomass and
storage of carbon.
1
Hawkesbury Institute for the Environment, Western Sydney University, Richmond, New South Wales, Australia.
2
School of Biological Sciences, University of Tasmania, Hobart, Tasmania, Australia.
3
School of Forestry and Environmental Studies, Yale University, New Haven, CT, USA.
4
PIAF, INRA, Université Clermont Auvergne, Clermont-Ferrand, France. *e-mail: b.choat@westernsydney.edu.au
28 JUNE 2018 | VOL 558 | NATURE | 531
© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Review
ReSeARCH
Against this backdrop, it is essential to improve the accuracy with
which we can predict the response of trees to drought to understand
the resilience of forests under future climate regimes. At present, mor-
tality is not well-represented in vegetation models, owing mostly to
gaps in our understanding of physiological mechanisms and a lack of
appropriate thresholds with which to parameterize these models. We
therefore turn our attention to how these problems may be resolved.
Drought and hydraulic failure in trees
As with all vascular plants, trees prevent desiccation injury by using an
intricate plumbing system of hollow dead cells (vessels or tracheids)
to transport water from the soil to the leaves. Xylem transport relies
on an elegant mechanism whereby liquid water is held under tension,
enabling trees to lift vast volumes of water to the canopy at little ener
-
getic cost
32
. However, liquid water under tension exists in a metastable
state, similar to that of a superheated liquid
33
. In this state, water is
prone to cavitation, a sudden phase change from liquid water to gas
that creates a bubble (embolism). These gas emboli block water flow
through xylem conduits and reduce the delivery of water to the can-
opy and regenerative tissues (that is, apical and cambial meristems)
8
.
Drought leads to higher xylem tensions and an increased probability
that emboli will spread throughout the xylem network causing systemic
vascular dysfunction
12,34
.
Phases of drought stress and the response of plants
During drought, reduced precipitation leads to declines in soil mois-
ture, which are often accompanied by higher temperatures and
increased evaporative demand from the atmosphere. These factors
combine to induce water stress in plants, which is manifested as
increased tension in the xylem sap. Water stress is measurable in plants
as xylem water potential (Ψ
x
), a variable that is primarily determined by
pressure in the xylem fluid and becomes increasingly negative during
drought
32,35
. As plants desiccate, the loss of cell turgor causes stomatal
pores on the leaf surface to close, markedly slowing plant dehydration
and the rate of decrease in Ψ
x
. Most recent studies indicate that sto-
mata in trees close before reaching the threshold Ψ
x
at which significant
cavitation is initiated, despite the negative consequences of stomatal
closure
36–38
(Fig.1). On short time scales, these consequences include
a rapid cessation of photosynthetic CO
2
assimilation, loss of canopy
evaporative cooling through transpiration and greater probability of
photodamage
39,40
. Over longer time scales, low photosynthetic rates
associated with drought-induced stomatal closure can lead to depletion
of non-structural carbohydrate pools, which interferes with transloca-
tion of sugars through the phloem
11,41,42
and the production of chem-
ical defence compounds needed to prevent herbivory and disease
18,43
.
The fact that stomatal closure generally occurs before the initiation of
cavitation despite these costs suggests that avoidance of xylem cavita-
tion is of paramount importance for the long-term survival of trees.
After stomatal closure, Ψ
x
continues to slowly decrease, becoming
more negative as water is lost through cuticular conductance, stomatal
leakiness
44
and other tissues such as bark
45
. At the same time, hydrau-
lic conductance may decrease throughout the hydraulic pathway of
the plant through a number of biophysical and physiological mecha-
nisms, including reversible collapse of leaf veins
46
, regulation of aqua-
porins in cell membranes
47,48
and the formation of cortical lacuna in
fine roots
49
. Rates of water loss during this phase are typically in the
order of 100–1,000-fold less than when the stomata are fully open
44
and decreases in Ψ
x
are further buffered by the release of internally
stored water
50
. If drought persists, Ψ
x
will ultimately reach a critical
threshold at which emboli begin to propagate through the xylem
8,51
.
This process occurs throughout the hydraulic pathway including roots,
stems and leaves
34,48,49,52,53
(Fig.2). Because emboli greatly reduce water
delivery to the canopy, this hydraulic dysfunction can cause patchy
branch death and pronounced reductions in canopy leaf area
54
. During
intense droughts, emboli spread throughout the water transport net-
work, causing systemic failure of the vascular system
55
. In the face of
continuing drought and high evaporative demand, systemic vascular
dysfunction may cause rapid mortality of the whole plant through
desiccation
12,15,16
and death of the meristematic tissue in the cambium
and apical meristems.
Hydraulic traits of trees and adaptations to drought
The risk of hydraulic failure is an unavoidable consequence of trans-
porting water under tension, and thus forms a fundamental axis of
selection in the evolution of trees
56,57
. Strategies to preserve the integrity
of the plant vascular system in trees are diverse, but all revolve around
a simple framework defined by two constraints: (1) the physical limits
of the vascular system; and (2) the capacity to maintain plant water
potential within these functional limits. These two attributes dictate
how rapidly plant tissues will dehydrate during a drought and the spe-
cific thresholds at which water stress will translate into hydraulic failure
and mortality
37,58
.
Although it is possible to characterize a general sequence of events
that describe the response of vascular plants to drought, the traits
that define this response vary across species and environments
59–61
.
Recent studies have illustrated the enormous variation in vulnerability
to xylem cavitation across tree species, with changes in xylem
vulnerability correlated to mean annual precipitation and aridity of
their growth environment
60
. Species are typically compared by the
Ψ
x
value at which a 50% loss of hydraulic conductance occurs (Ψ
50
),
although other reference points may have more physiological impor-
tance, for example, Ψ
88
(Fig.1). Differences in vulnerability are driven
by the anatomical features of the xylem, including conduit dimensions,
network organization and the porosity of primary cell walls (pit mem-
branes) that limit the spread of gas between conduits
62,63
. These features
control the critical Ψ
x
at which gas will penetrate pit membranes, causing
cavitation in adjacent conduits and the spread of embolism through
the xylem
8
. However, vulnerability to cavitation does not determine
drought tolerance in itself. The probability of reaching the critical
threshold and the length of time it takes for this to occur are deter
-
mined by the interaction of a number of associated physiological and
morphological traits (Fig.3).
The multi-dimensional nature of such trait interactions has enabled
vascular plants to inhabit nearly every terrestrial habitat on Earth and
enabled a huge number of possible morphological and physiological
solutions to tolerating drought. For instance, variation in the vulnera-
bility of plants is often high within communities, particularly in drier
habitats, indicating that vulnerability and aridity are decoupled in some
cases
64
. This decoupling results from water-stress avoidance strategies
that are used by some species, such as deep root systems or drought
deciduousness, that allow them to maintain a higher Ψ
x
during drier
periods. Although this complexity makes the development of models
challenging, a suite of well-studied traits that are mechanistically linked
to drought tolerance have now emerged (Supplementary Table1) and
represents a promising direction for future research. Recent analyses
have suggested that these traits often vary in a coordinated fashion
that allows the benefits of photosynthetic carbon gain to be balanced
against the risks of a decrease in Ψ
x
and the occurrence of hydraulic
failure
59,65
. Thus, much of the complexity of trait interactions may col-
lapse onto a single axis that defines a spectrum of drought tolerance
strategies
66
.
Ultimately, we are interested in predicting when a plant will die as
a result of drought stress. Vulnerability to cavitation has emerged as a
key physiological trait that is associated with mortality, and hydraulic
failure represents a critical point in the drought response pathway.
Species-specific tree hydraulic limitations provide a powerful mecha-
nistic explanation for the observation that drought mortality is occur-
ring across forest biomes, independent of the mean rainfall at any site.
A recent data synthesis demonstrated that the majority of plant species
converge to narrow hydraulic safety margins, that is, the buffer between
minimum water potential experienced by the plant (Ψ
min
) and the
threshold Ψ
x
for rapid loss of vascular function caused by cavitation
60
.
Because Ψ
min
integrates many important aspects of plant structure
(for example, rooting depth) and physiology (for example, stomatal
532 | NATURE | VOL 558 | 28 JUNE 2018
© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Review
ReSeARCH
behaviour) in relation to the environment, the narrow safety margins
that are found across forest types offer an important insight into plant
ecology, one that suggests that the hydraulic strategies of plants are
finely tuned to their environment, allowing for maximum carbon gain
but exposing plants to the risk of hydraulic failure during drought. It
also suggests a generally ‘risky’ strategy in which plants have limited
physiological potential to respond to rapid changes in the environment.
This exacerbates the threat posed by increased occurrences of extreme
drought under climate change. Indeed, drought mortality events across
forests from a broad geographical and climatic range have been linked
quantitatively to hydraulic traits and xylem cavitation
67
. Examples
come from tropical rainforest
68
, temperate forests
14,16
, chaparral
15,17
and desert woodlands
11
. Many of these studies show differential species
mortality within each forest type, allowing us insights into the potential
winners and losers under future drought regimes.
Plasticity and genetic variation in hydraulic traits
Although much is known about the variability in hydraulic traits among
plant species, far fewer data are available to quantify within-species
variation. The capacity of trees to alter phenotype (that is, phenotypic
plasticity) and the amount of genetic diversity within a population are
key variables for the ability of species to cope with rapid climate change.
It is unlikely that trees will be capable of adapting to sudden increases
in the aridity of their environment through evolutionary mechanisms,
because of their long generation cycle and inability to migrate away
from stress. On the other hand, adaptive plasticity of hydraulic traits
may enable the acclimatization of entire populations within the neces-
sary timescales. Quantifying the extent of plasticity in hydraulic traits
is therefore an essential component for the prediction of the tolerance
ranges and resilience to drought of different species. However, com-
prehensive datasets that examine the genetic variation and phenotypic
plasticity of vulnerability to cavitation have only recently become avail-
able and are limited to a few species. A study of 513 genotypes of the
widespread pine species Pinus pinaster showed low genetic variation
of Ψ
50
between climatically contrasting populations and very limited
phenotypic plasticity
69
. These results suggest that Ψ
50
may be a canalized
trait in pines, with little capacity to enable short-term acclimatization
and adaptive plasticity. Angiosperm species have a higher potential for
phenotypic plasticity and adaptive variation between populations
70
,
although the observed shifts in vulnerability are often small in magni-
tude relative to changes in Ψ
x
that are expected to occur during severe
drought.
Long-term manipulative experiments suggest that structural
acclimatization, that is, changes in the allocation pattern between
50
88
Loss of xylem hydraulic conductance
Stomatal conductance
Cuticular conductance
Water loss from stomata and cuticles
Loss of hydraulic conductance
due to cavitation
C
uticular conductan
c
Wa
t
e
Stomatal conductance
Leaf shedding
a
l c
o
o
nductanc
e
Cuticular conductance
ng
Leaf sheddi
n
Cuticularcondu
cta
a
nce
nce
Cavitation
Cuticular
co
o
ndu
ndu
ct
cta
nce
C
avitatio
n
Plant water storage
Increasing drought stress
Decreasing
x
Ψ
Ψ
Ψ
a
b
c
Fig. 1 | Phases of drought response in plants. a, Time series of transverse
slices through the xylem tissue obtained by X-ray microtomograpy show
the spread of gas emboli through the xylem with increasing drought stress
(left to right). In each slice, water-filled vessels are seen as bright circles
whereas vessels that contain gas emboli are black. During severe drought,
almost all vessels become gas-filled, which leads to whole-plant mortality
(right). b, During the first phase, stomata close to limit water loss and
delay the decrease in xylem water potential (blue line). After stomata close,
water continues to be lost at a much lower rate via cuticular conductance.
At a critical threshold, cavitation increases rapidly and gas emboli spread
throughout the xylem (red line). Increasing levels of embolism are shown
as the proportional loss of xylem hydraulic conductance. ‘Vulnerability
curve’ analysis translates the physics of cavitation to a quantification of
species susceptibility to cavitation during exposure to water stress. These
mortality thresholds have been found to correspond to between 50% (Ψ
50
)
and 88% (Ψ
88
) loss of hydraulic function in conifers and angiosperms,
respectively. c, A general scheme for the magnitude and timing of response
processes with increasing drought stress.
28 JUNE 2018 | VOL 558 | NATURE | 533
© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
ReviewReSeARCH
water-absorbing, -conducting and -transpiring tissues, is almost cer-
tainly the dominant process by which plants adjust their hydraulic sys-
tems in response to drought
71
. Reductions in the leaf to sapwood area
ratio and shoot to root ratio result in a greater capacity to supply water
to the leaves and limit the drop in Ψ
min
, consistent with homeostasis of
water transport, however these changes come at the cost of reductions
in productivity
72
. Reductions in the leaf to sapwood area ratio result in
the maintenance of a higher Ψ
x
and a greater capacity to supply water
to the leaves
73
. These results are consistent with studies of intraspecific
variation in hydraulic architecture across aridity gradients, which show
changes in morphology and allocation patterns; however, little evidence
of adaptive variation in vulnerability to cavitation has been found in
these studies, even in species with a wide climate envelope
74,75
. Further
studies are clearly required to determine whether these patterns can be
generalized, particularly in angiosperms, and what role the plasticity of
hydraulic traits may have in the capacity of plants to survive increased
aridity.
Predicting mortality from hydraulic thresholds
Predictions of drought-induced forest mortality require a detailed
understanding of the physiological underpinnings of tree death.
Accordingly, this topic has received much attention in recent years and
substantial progress has been made in our understanding of the mech-
anisms of tree mortality
11,12
. It is clear that drought-associated forest
mortality is complex and a number of interdependent mechanisms
have important roles in this process. These mechanisms include failure
of water transport in the xylem, depletion of carbohydrate reserves
over prolonged drought
41,42
and increased vulnerability to pests and
pathogens
18,76
. All mechanisms of drought-associated mortality revolve
around the effects of stomatal closure and increasing xylem tension
during water shortage. Hydraulic failure is the most fully elaborated
mechanism and currently holds the most promise for predictive
models. It is a relatively well-understood biophysical process that is
amenable to modelling
77
, with failure thresholds that can be readily
established for a given species or population
69,70
. Accuracy and con-
fidence in the vulnerability thresholds that are chosen to represent
different species in predictive models are absolutely critical. Recent
technical and theoretical advances in the science of plant hydraulics
have provided new certainty in the quantitative nature of hydraulic
failure
34,55,78
. We thus focus on hydraulic traits as a means to under-
stand and predict patterns of tree mortality in response to drought
while emphasizing that other thresholds can be incorporated as our
understanding of them improves. Indeed, incorporation of hydrau
-
lics thresholds, such as turgor loss and stomatal closure, should assist
greatly in predicting carbon dynamics under drought.
Measuring hydraulic thresholds to mortality
As noted above, tree species exhibit a large range in xylem vulnerabil-
ity
60,61
. Clear links between xylem cavitation and tree death have been
established in pot studies
10,12,13
and natural systems
15,79,80
, suggesting
that xylem vulnerability should undergo strong selection in species
exposed to episodic water stress. Evidence of selection, or ecological
sorting, can be seen in the distribution of species with regard to strong
correlations between aridity and xylem vulnerability
10,56,64
. The excit-
ing implication of this work is that the vulnerability of xylem tissue to
cavitation provides a measurable index of the capacity of a species to
tolerate water stress during drought
77
.
Our ability to predict the level of water stress at which a plant will die
based on functional traits has advanced considerably in recent years.
The concept of a ‘lethal water potential’ for a given plant species or
population has existed for some time, but only recently has hydraulic
vulnerability been quantitatively linked to mortality
12
. In conifers, the
Ψ
50
of the stem xylem is strongly related to their ‘minimum recoverable
water potential’, essentially a physiological point of no return
12,81
. By
contrast, the lethal water potential of angiosperm species is correlated
with more complete hydraulic dysfunction, representing 80–100% loss
of xylem hydraulic conductance (Ψ
88
)
10,13
. The disparity in mortality
thresholds between conifers and angiosperms may be related to the
fundamental difference in xylem structure between these two groups
and goes some way to explaining the generally larger hydraulic safety
margins and more conservative stomatal behaviour exhibited by conifer
species
60,82
. Other hydraulic thresholds for mortality have been pro-
posed, including a sustained loss of hydraulic conductance greater than
60%
80,83
. We note that these studies were based on modelled thresholds
of whole-plant conductance in two conifer species and compare well
to the Ψ
50
threshold that has been established for conifers experimen-
tally. A recent data synthesis found that all studies reported a 60% or
greater loss in hydraulic conductance at death with a mean loss of 83%
9
.
Therefore, although there are some discrepancies between proposed
thresholds, it is clear that high levels of xylem embolism are linked with
mortality. Modern techniques for the measurement of water potential
and non-invasive visualization of embolism are providing more
a
b
c
Fig. 2 | Non-invasive imaging techniques have provided new insights
into embolism formation and spread in the xylem. a, Mapping the spread
of embolism in leaf vein networks during dehydration with transmitted
light. Left, transmitted light images highlighting the vein network. Middle,
image subtraction reveals embolism propagating from the midrib into
the secondary and tertiary venation. Right, a colour map of all cavitation
events recorded during desiccation. b, Three-dimensional rendering from
a X-ray micro-computed tomograph of a pine stem. Embolized tracheids
can be seen clearly as a black void space surrounded by water-filled (grey)
tracheids. c, Part of a root system rendered from a micro-computed
tomograph showing embolized xylem vessels (red) in the main root axis
and lateral roots during dehydration. Right, the root tissue has been made
transparent to illustrate the pathway of embolized vessels. Images in a were
reproduced with permission from Brodribb et al.
52
.
534 | NATURE | VOL 558 | 28 JUNE 2018
© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Review ReSeARCH
accurate measures of hydraulic failure thresholds
34,55
, which can then
be used to parameterize models of tree mortality.
Trait-based models of tree mortality
Although the vulnerability of xylem to cavitation defines a threshold
in water stress beyond which tree mortality will occur, the key issue
for predicting mortality is the ability to translate meteorological data
(for example, precipitation or evaporative demand) into plant water
content or xylem tension
58
. This calculation presents a number of chal-
lenges: it requires knowledge of the volume of water that is available to a
plant in the soil and internal reservoirs, as well as the rate of water loss
through transpiration. Calculating the amount of water in the soil that
is available to the plant is made difficult by a paucity of data relating
to rooting depth of trees and the architecture of the roots. Simulating
the rate of water loss is complicated by the active regulation of transpi-
ration by stomata, differences in cuticular transpiration after stomatal
closure and the degree of leaf shedding during drought. To date, most
attempts to model or predict mortality use empirical relationships
between observed mortality and climate extremes
84,85
. Such empirical
relationships, although they provide insights into the current drivers
of mortality, may not function well in the future if plant sensitivities
change over time or if novel climate conditions occur. For example,
rising CO
2
concentrations may alleviate drought stress, whereas rising
temperatures may exacerbate it
86
. Empirical relationships may also fail
in regions where long-term shifts towards a novel, drier climate are
occurring. Process-based models are thus highly desirable
87
.
Recent progress in the understanding of the hydraulic mechanisms
that lead to mortality, and quantification of the key plant traits that are
involved, has led to the incorporation of plant hydraulics in a range of
process-based vegetation models
83,88–90
. The key elements of such mod-
els are a description of the soil-to-leaf hydraulic pathway, incorporating
soil, root, xylem and stomatal conductances, whole-plant capacitance
and the vulnerability of the xylem to cavitation. Important plant traits
that are required for parameterization include the response of stomata
to decreasing water potential, the point at which leaf turgor is lost,
saturated xylem hydraulic conductance and water potential thresholds
of vulnerability to cavitation. Recent compilation efforts have made
data for these traits available for a wide range of species
60,91
and enabled
hydraulic traits to be related to other aspects of the plant economic
spectrum
88,89
. Incorporating trait variation in drought sensitivity
among species or genotypes, and relating this variation to plant water-
use strategies and other plant properties, promises to be an effective
way forwards.
Nonetheless, critical gaps remain in our ability to describe the
hydraulic pathway and its eventual failure. Here we draw attention to
several gaps that hinder model development and parameterization and
that have received comparatively little attention: (1) the dynamics of
canopy leaf area during drought; (2) the dependence of plant water
status on soil water potential; and (3) the process of plant desiccation
in very dry soil, when root water uptake is no longer possible.
First, leaf shedding occurs in many ecosystems during drought
92
,
and can mitigate water stress to the remaining foliage
93
, slowing the
rate of desiccation (the ‘hydraulic fuse’ hypothesis
53
). However, drought
deciduousness is, as yet, poorly captured in models
94
. Recently, it has
been demonstrated that representing vegetation as a set of compet-
ing plant types with varying degrees of drought deciduousness leads
to a marked improvement in modelled leaf area dynamics in Central
America; this hydraulics-based approach holds considerable promise
for model improvement
89
.
Second, a key component of the hydraulic pathway is the relationship
between plant water status (represented by pre-dawn plant water poten-
tial, Ψ
pd
) and soil water availability (represented by soil water potential,
Ψ
soil
). Simple models that treat soil water as a single bucket generally
fail to capture this relationship; it appears that models need to incor-
porate vertical gradients in soil moisture potential, the distribution of
roots and changing soil–root resistance with soil drying
88,95
. However,
it is also commonly observed that co-occurring species can have dif-
ferent Ψ
pd
when Ψ
soil
is the same
96
, and this difference cannot always
be explained by rooting distributions. The Ψ
pd
can be lower than Ψ
soil
if overnight equilibration is insufficient
89
, or considerable amounts of
Physiological traits (leaf)
• Stomatal regulation
• Turgor loss point
• Cuticular conductance
Xylem anatomical traits
• Xylem conduit size, number and
connectivity
• Pit membrane thickness/porosity
• Wood density
Morphological traits (shoot)
• Stomatal anatomy
• Leaf vein density
• Total leaf area
• Leaf shedding/drought deciduous
• Leaf to sapwood area ratio
Morphological traits (root)
• Root to shoot ratio
• Rooting depth
• Fine root loss
Physiological traits (root)
• Cortical lacunae formation
• Root shrinkage/hydraulic isolation
• Soil–root hydraulic conductance
Physiological traits (common)
• Vulnerability to cavitation
(
12
,
50
,
88
)
• Maximum hydraulic conductance
• Capacitance and water storage
• Cell membrane permeability
(aquaporin regulation)
ΨΨΨ
Fig. 3 | Tree hydraulic traits associated with drought-induced mortality.
Trees use a variety of interdependent and coordinated morphological,
anatomical and physiological traits to mitigate water loss and the
development of increasingly negative xylem sap pressures during
drought. This includes tissue-specific traits that function in the unique
microenvironment of roots, stems and leaves, as well as traits that are
common among most tissue types in trees. Many structure–function
relationships exist between traits, for example, variation in xylem
anatomical traits (pit membrane porosity, conduit size and connectivity)
determine species and population-level vulnerability to cavitation.
Note that this figure does not represent an exhaustive list of hydraulic
traits relevant to the response of trees to drought and drought-induced
mortality.
28 JUNE 2018 | VOL 558 | NATURE | 535
© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.