Content uploaded by Anna L. Jacobsen
Author content
All content in this area was uploaded by Anna L. Jacobsen on Apr 30, 2019
Content may be subject to copyright.
The standard centrifuge method accurately measures
vulnerability curves of long-vesselled olive stems
Uwe G. Hacke
1
, Martin D. Venturas
2
, Evan D. MacKinnon
2
, Anna L. Jacobsen
2
, John S. Sperry
3
and
R. Brandon Pratt
2
1
Department of Renewable Resources, University of Alberta, Edmonton, AB T6G 2E3, Canada;
2
Department of Biology, California State University, 9001 Stockdale Hwy, Bakersfield, CA
93311, USA;
3
Biology Department, University of Utah, 257S 1400E, Salt Lake City, UT 84112, USA
Author for correspondence:
Uwe G. Hacke
Tel: +1 780 492 8511
Email: uwe.hacke@ualberta.ca
Received: 29 May 2014
Accepted: 1 August 2014
New Phytologist (2015) 205: 116–127
doi: 10.1111/nph.13017
Key words: cavitation, centrifuge method,
embolism, hydraulic conductivity, Olea
europea (olive), vessel length, vulnerability
curve, xylem.
Summary
The standard centrifuge method has been frequently used to measure vulnerability to xylem
cavitation. This method has recently been questioned. It was hypothesized that open vessels
lead to exponential vulnerability curves, which were thought to be indicative of measurement
artifact.
We tested this hypothesis in stems of olive (Olea europea) because its long vessels were
recently claimed to produce a centrifuge artifact. We evaluated three predictions that fol-
lowed from the open vessel artifact hypothesis: shorter stems, with more open vessels, would
be more vulnerable than longer stems; standard centrifuge-based curves would be more vul-
nerable than dehydration-based curves; and open vessels would cause an exponential shape
of centrifuge-based curves.
Experimental evidence did not support these predictions. Centrifuge curves did not vary
when the proportion of open vessels was altered. Centrifuge and dehydration curves were
similar. At highly negative xylem pressure, centrifuge-based curves slightly overestimated vul-
nerability compared to the dehydration curve. This divergence was eliminated by centrifuging
each stem only once.
The standard centrifuge method produced accurate curves of samples containing open ves-
sels, supporting the validity of this technique and confirming its utility in understanding plant
hydraulics. Seven recommendations for avoiding artefacts and standardizing vulnerability
curve methodology are provided.
Introduction
Centrifugal force can be used to create known negative pressure
in the xylem of woody vascular plants (Holbrook et al., 1995;
Pockman et al., 1995). Early experiments in spinning stems
revealed that xylem conduits remained water-filled to significant
negative pressures and that cavitation occurred at species-specific
pressures (Pockman et al., 1995). This first technique was then
modified and formalized in Alder et al. (1997) and this method
became the standard centrifuge technique broadly employed in
plant hydraulic research.
The centrifuge method, as outlined in Alder et al. (1997) has
now been used for nearly 20 yr to advance our understanding of
how water transport in plant xylem is affected by drought (Sperry
& Hacke, 2002; Jacobsen et al., 2007) and freeze/thaw stress
(Davis et al., 1999; Pittermann & Sperry, 2006). Centrifugal
force methods have allowed us to link xylem physiology with
structural xylem features (Hacke et al., 2001, 2006; Jacobsen
et al., 2005; Cai & Tyree, 2010; Lens et al., 2010), to characterize
xylem trade-offs (Hacke & Sperry, 2001; Pratt et al., 2007a), to
screen cavitation resistance in poplar and willow clones (Cochard
et al., 2007; Arango-Velez et al., 2011; Schreiber et al., 2011), to
link drought tolerance and life history traits (Pratt et al., 2007b),
and to study drought-induced mortality of woody plants (Choat
et al., 2012; Plaut et al., 2012; Anderegg et al., 2013; Paddock
et al., 2013).
According to the protocol originally described by Alder et al.
(1997), hydraulic conductivity (K
h
) of xylem segments is mea-
sured before and after spinning them in a centrifuge rotor to gen-
erate negative xylem pressure (Fig. 1). Cavitated conduits quickly
fill with gas and are no longer able to conduct water, which
results in a reduction in K
h
. The Alder et al. (1997) rotor design
and method (subsequently referred to as the ‘standard centrifuge
method’) was more recently modified with the goal of allowing
measurements of conductivity while stem segments were spin-
ning (Cochard et al., 2005; Li et al., 2008). These newer ‘flow-
centrifugation’ methods can use either the ‘cavitron’ rotor design
(Cochard et al., 2005), or a modification of the original Alder
et al. (1997) design (Li et al., 2008). The standard centrifuge
method fundamentally differs from the cavitron method.
Although water does not move through the stem segment during
centrifugation when the standard centrifuge method is used,
116 New Phytologist (2015) 205: 116–127 Ó2014 The Authors
New Phytologist Ó2014 New Phytologist Trust
www.newphytologist.com
Research
water movement during spinning is required for the cavitron
method to estimate conductivity (Table 1).
When testing the cavitron, Cochard et al. (2010) found that
the vulnerability curves of two species (Quercus robur and Prunus
persica) differed depending on the size of the rotor that was used
to generate the curves. When a smaller rotor was used (with stem
segments 17.5 cm in length), curves tended to be more vulnerable
than when a larger rotor was used (with segments 27.5 cm in
length). For these two species with long vessels, both rotors pro-
duced exponentially shaped curves: that is, curves that showed an
abrupt rise in cavitation at very modest xylem pressures. In two
other species with shorter xylem conduits, better agreement
between curves was found, regardless of rotor size and method
used to generate the curves. An earlier study using the cavitron
also showed inconsistent and nonreproducible curves for
Fraxinus excelsior, another long-vesselled species (Cochard et al.,
2005; their fig. 2).
The mechanism behind the cavitron-specific artifact seems to
be related to the flow of solution through segments during centri-
fugation (Wang et al., 2014). The higher susceptibility to cavita-
tion of short segments reported by Cochard et al. (2010) and the
extreme variability shown in Cochard et al. (2005) may be
explained by the presence of impurities. Micro-bubbles and other
particles may provide nucleating sites, which then grow when the
pressure drops as the fluid moves toward the center of the seg-
ment, initiating cavitation during spinning (Sperry et al., 2012;
Rockwell et al., 2014; Wang et al., 2014). During centrifugation,
pressures are not even throughout spinning segments with the
minimum water potential (maximum xylem tension) occurring
in the center of segments (Fig. 1). Hence, if vessels extend from
one cut end to the center of the segment, nuclei would not be
trapped by vessel ends and could trigger cavitation as they
approached the center of the segment. In this way, cavitation
would be triggered by introduced nuclei, rather than via air-
seeding through vessel wall pores (for a discussion of the air-seed-
ing mechanism, see Sperry & Hacke, 2004). The introduced
nuclei artifact would at least partly explain why the cavitron has
been affected by the open-vessel artifact whereas the standard
centrifuge method does not appear to be susceptible to this artifact.
Although it is not clear how the standard centrifuge method
could be prone to this potential artifact (because there is no flow
through segments during spinning), artifacts are still possible if
Distance from the rotor center (cm)
–7 –6 –5 –4 –3 –2 –1 0 1 2 3 4 5 6 7
Ψ (MPa)
–2.5
–2.0
–1.5
–1.0
–0.5
0.0
0.5
Rotation axis
Stem segment
Foam pad with
degassed solution Metal plate
Bolts
Reservoir Reservoir
Solution level
Solution level
4,686 rpm
6,627 rpm
9,372 rpm
Fig. 1 Diagram showing sample placement in the standard centrifuge rotor and the pressure profile developed during spinning. The top sketch shows how
the sample is set in the standard centrifuge rotor (Alder et al., 1997). The stem segment is placed with its center crossing the rotation axis, and the ends are
positioned within reservoirs containing degassed solution. Foam pads are placed in these reservoirs to avoid desiccation when the rotor is not spinning
(Tobin et al., 2013). The sample is secured in place with a metallic plate and bolts. The standard method requires the sample to be removed from the rotor
after spinning it at a certain velocity for measuring its conductivity. Then it is mounted again for spinning it at a higher velocity (greater tension). Below the
rotor sketch, the pressure (Ψ) profile developed within the conduits during spinning is represented for three different rotation velocities, calculated as
determined by Alder et al. (1997). The minimum water potential (maximum stress) is reached in the center of the stem segment. At the intersections of the
stem with reservoir solution the pressure is zero, and at the ends of the stem segment the pressure is positive. Note that with the standard centrifuge rotor,
the pressure at both ends of the segment is equal while spinning, and therefore, there is no pressure-driven flow through the stem.
Ó2014 The Authors
New Phytologist Ó2014 New Phytologist Trust
New Phytologist (2015) 205: 116–127
www.newphytologist.com
New
Phytologist Research 117
open vessels drain before the reservoir water reaches them when
the centrifuge starts spinning (Choat et al., 2010). In addition,
nucleating particles may be introduced into open vessels via per-
fusion with measuring solution before spinning in the centrifuge
(Rockwell et al., 2014). The standard centrifuge protocol usually
involves flushing of samples, although vacuum infiltration has
also been used to remove embolism before generating a curve.
Regardless of whether and how embolism is removed, conductiv-
ity is typically measured before centrifuging, and hence measur-
ing solution is introduced into stem segments. Both the draining
and the introduced nuclei ideas can be tested and protocols
adjusted accordingly. For instance, if vessel draining were
detected, it could be prevented in the future by using foam pads
soaked with degassed solution as described by Tobin et al.
(2013).
A recent report of erroneous curves produced by the standard
centrifuge method is based on work with olive (Torres-Ruiz
et al., 2014). According to this study, both the cavitron and the
standard method overestimated vulnerability to cavitation relative
to a dehydration method. The reported overestimations were
higher as the sample length was shorter, suggesting that all centri-
fuge methods may be prone to the open vessel artifact. Based on
such data, some have suggested that all exponential vulnerability
curves would be incorrect, regardless of the method that was used
to create the curves (Cochard & Delzon, 2013; Cochard et al.,
2013). It has been argued that ‘exponential curves are largely
artifactual’, and are likely based on ‘faulty methods and data’
(Cochard et al., 2013). Because many workers are coming to
far-reaching conclusions in opinion papers and commentaries
based on the accuracy or lack thereof of the standard centrifuge
method (Rockwell et al., 2014), it is important to verify the
reports that suggest that the standard method is prone to artifact.
Although scarcely mentioned in some commentaries (Cochard
& Delzon, 2013), the standard method has been rigorously tested
for short- and long-vesselled samples. Importantly, these studies
found no evidence for an open-vessel artifact when the Alder
et al. (1997) rotor design was used, either with the standard
method (segments removed from the rotor for conductivity mea-
surements) or with the newer approach of measuring conductiv-
ity while the rotor is spinning (Li et al., 2008; Christman et al.,
2012; Jacobsen & Pratt, 2012; Sperry et al., 2012; Tobin et al.,
2013). The most recent example of such work examined dehy-
dration and standard centrifuge curves measured on the same
species at the same sites and during the same season (Jacobsen
et al., 2014). Curves of 10 species were examined this way. In
contrast to expectations based on the open vessel hypothesis, the
use of these different methods did not result in a significant dif-
ference in xylem pressure causing 50% loss of conductivity (P
50
),
even though several of these species had long vessels (maximum
vessel length >1m).
There is currently no consensus about the validity of the stan-
dard centrifuge method and other methods (Rockwell et al.,
2014). Achieving such consensus is probably not a realistic expec-
tation, nor is it required. However, additional data derived from
clearly designed and carefully conducted experiments are likely
necessary to move scientific discourse forward. If discrepancies
are consistently found and are reproducible, then work may be
directed toward identifying the cause of the discrepancies and
toward improving the method. For instance, based on the work
by Li et al. (2008) it seems reasonable to expect that the cavitron
rotor design could be improved to eliminate or at least minimize
artifacts.
Here we tested the standard centrifuge method by measur-
ing vulnerability curves of current-year olive shoots, in part as
a follow-up to the suggested measurement artifacts reported in
Torres-Ruiz et al. (2014). A new variation of the standard cen-
trifuge method was also tested. In this new ‘single-spin’ proto-
col, samples are centrifuged only once. The standard
centrifuge method involves the removal of samples from the
rotor after each spin. To test whether repeated mounting and
Table 1 Characteristics of the main centrifuge techniques
Denomination Improvements Described in Rotor
Stem segment
length
Are stem segments ends
immersed in reservoirs
during spinning?
Hydraulic
measures
a
Original Pockman et al. (1995) Any Variable
(26–40 cm)
No, stem ends have to be
trimmed before
conductivity measurement
No flow
Standard Alder et al. (1997) Alder et al. (1997) design Constant (usually
14 or 27 cm)
Yes No flow
Foam pads added
to reservoir
Tobin et al. (2013) Alder et al. (1997) design Constant (usually
14 or 27 cm)
Yes, with foam pads No flow
Single-spin
protocol
Present study Alder et al. (1997) design Constant (usually
14 or 27 cm)
Yes, with foam pads No flow
Cavitron Cochard et al. (2005) Cochard et al. (2005)
design
Constant (usually
17.5, 27.5 or
37.5 cm)
Yes Flow
Flow
centrifuge
Li et al. (2008) Li et al. (2008) design Constant (usually
27.5 cm)
Yes Flow
a
No flow, no induced flow during centrifugation; samples are removed from the rotor for performing conductivity measurements with the conductivity
apparatus (Sperry et al., 1988) or the XY’LEM (Xylem Embolism Meter; Instrutec, Montigny les Cormeilles, France). Flow, stem segment conductivity is
measured while spinning, and therefore, segments are only placed once in the rotor.
New Phytologist (2015) 205: 116–127 Ó2014 The Authors
New Phytologist Ó2014 New Phytologist Trust
www.newphytologist.com
Research
New
Phytologist
118
removing of samples from the rotor causes reductions in con-
ductivity, individual branch segments were exposed to a single
xylem pressure as opposed to using the same segments at
multiple pressures.
We used a hydraulic method for assessing vessel length (similar
to that used by Torres-Ruiz et al., 2014), tested for draining of
open vessels, assessed whether there was a stem length effect, and
compared centrifuge results to native stem conductivities. We
took pains to eliminate as many sources of ambiguity as possible.
We report stem area-specific conductivity (K
s
) because recent
studies have shown that percentage loss of conductivity (PLC)
may be erroneously interpreted when K
s
is not also evaluated in
tandem (Jacobsen & Pratt, 2012; Sperry et al., 2012). Further-
more, we measured K
s
while accounting for passive uptake of
water by stems; not doing so is an important source of error when
dealing with low-conductivity material (Hacke et al., 2000b;
Torres-Ruiz et al., 2012). We also assessed the impact of flushing
on the shape of vulnerability curves. There is currently no agree-
ment among researchers whether samples should be flushed
before a vulnerability curve is constructed.
Three predictions of the open vessel artifact hypothesis were
tested. (1) Shorter stems with a larger proportion of open vessels
would be more vulnerable than longer stems with a lower propor-
tion of open vessels, as previously reported for olive (Torres-Ruiz
et al., 2014). (2) Centrifuge curves would be more vulnerable
than dehydration curves. (3) The centrifuge-based vulnerability
curve of a long-vesselled species, such as olive, would have an
exponential shape due to introduced nuclei or draining of open
vessels.
Materials and Methods
Plant material
Two sets of experiments were conducted for this study. The first
set of measurements was conducted in September and October
2013. Long branches (1.5–2 m) were collected on 20 September
2013 from six irrigated olive (Olea europea L.) trees growing on
or near the campus of California State University, Bakersfield,
CA, USA, and from these stems c. 45-cm-long current year seg-
ments were cut underwater. The ends of stems were then
wrapped in moist paper towels, and stems were placed in a large
sealable plastic bag. The bag was kept at c. 4°C, and express-
shipped to the University of Utah on the same day. Upon arrival
at the University of Utah, samples were stored at c. 4°C. Centri-
fuge curves were measured on 24–27 September 2013. On 4
October 2013 at dawn, large branches between 1.43 and 1.90 m
in length were cut from the same individuals described above and
were used in Bakersfield for determination of native hydraulic
conductivities and maximum vessel length. A second measure-
ment campaign took place in January and February 2014. For
this sampling, all stems were collected from one mature olive tree,
which was also sampled in the first set of experiments. A single
tree was used to reduce experimental variability for a set of
targeted experiments. Because all measurements in 2014 were
conducted in Bakersfield (between 22 January and 10 February
2014), samples were processed immediately after collection. All
measurements were conducted on current year’s shoots.
Maximum vessel length
In order to determine the maximum vessel length of current
year’s shoots, long branches were collected and injected with air
from the proximal end at 100 kPa. During air injection, the distal
part of stems was immersed in a water-filled tray. Stems were suc-
cessively cut from their distal end, in 1-cm intervals, until the first
bubbles appeared. To ensure that gas bubbles were travelling
through xylem tissue, 1 cm of bark was removed from the distal
end of stem segments following each cut so that only gas emerg-
ing from the cut xylem was observed. The end was then shaved
underwater and fitted with a grommet and water-filled wide piece
of plastic tubing so that the cut end could be carefully observed
under 910 magnification for emerging air bubbles. The longest
vessel length was estimated as the length of the sample at the time
that the first, single column of air bubbles was observed emerging
from the xylem tissue. Fifteen lateral stems were used for maxi-
mum vessel length measures and injection-end stem diameters
ranged from c. 4 to 9 mm.
(a)
(b)
Fig. 2 (a) Maximum vessel length as a function of stem diameter
measured in current-year growth of olive (Olea europea) trees (r=0.814;
P<0.001). Maximum vessel length was determined by injecting air into
the basal end of long branches. (b) Stem area-specific conductivity as a
function of stem diameter (r=0.635; P<0.001).
Ó2014 The Authors
New Phytologist Ó2014 New Phytologist Trust
New Phytologist (2015) 205: 116–127
www.newphytologist.com
New
Phytologist Research 119
Centrifuge vulnerability curves; flushing and vacuum
infiltration treatments
Stem segments 3.5–7.5 mm in diameter were prepared by succes-
sively cutting them from longer stems underwater and trimming
to the desired length with clippers and new razor blades. Hydrau-
lic conductivity was measured with a conductivity apparatus as
described previously (Sperry et al., 2012). Conductivity was
expressed per total stem cross-sectional area, calculated from
mid-segment average stem diameter. A filtered (0.2 lm) 20 mM
KCl measuring solution was made using purified water (distilled
water passed through deionizing and organic removal cartridges;
Barnstead E-Pure; Thermo Scientific, Waltham, MA, USA).
Before and after each measurement of pressure-driven flow, back-
ground flow was measured. Dehydrated stem tissues can absorb
water creating a negative background flow (water is driven off the
balance) when there is no pressure head, and because olive stems
had low conductivities, stable background flow measurements
were important (Hacke et al., 2000b). Hydraulic conductivity
was calculated as the pressure-driven flow corrected for back-
ground flows divided by the pressure gradient. Maximum K
s
was
measured after stems were vacuum infiltrated (September 2013)
or flushed at 100 kPa for 1 h (January 2014) with degassed and
filtered (0.1 lm) 20 mM KCl solution. Solution was degassed
using a membrane contactor (Liqui-Cel Minimodule 1.7 95.5,
Charlotte, NC, USA).
We tested the effectiveness of flushing and vacuum infiltration
treatments in removing embolism. A total of 18 stems were cut
in air at 07:45h. Emboli were introduced on cutting. The stems
were cut to a length of 14 cm and initial K
s
values were measured.
After that, nine stems were vacuum-infiltrated for 1 h with
degassed 20 mM KCl solution at a vacuum pressure of 91 kPa.
The remaining nine stem segments were flushed for 1 h with
degassed and filtered 20 mM KCl solution at a pressure 100 kPa.
Maximum K
s
values were then measured. After that, the nine
stems that were initially flushed were vacuum-infiltrated and vice
versa for the stems initially vacuum-infiltrated, again for 1 h.
Conductivity was remeasured. Thus, if one procedure was more
effective at removing emboli, it would be detected as a difference
in K
s
between paired stems exposed to the two methods.
Segments were spun in a custom-built rotor and vulnerability
curves were generated as described previously (Alder et al., 1997;
Plavcova & Hacke, 2011; Tobin et al., 2013). In September
2013, two different rotor diameters were used to test for an effect
of segment length and fraction of open vessels. We measured vul-
nerability curves on 14- and 27-cm stem segments where the dif-
ferent stem lengths correspond to the different sizes of rotors. In
January 2014, the smaller rotor design was used (14-cm seg-
ments). Vulnerability curves were constructed using both flushed
stems and stems that were not flushed before the initiation of the
curve. Curves were expressed by plotting PLC and/or K
s
vs xylem
pressure. Maximum K
s
values measured in September and Janu-
ary/February were not statistically different (P=0.60, t-test).
Samples were spun to nine xylem pressures. Moreover, we had
to spin samples at very high rotational speed to induce high
embolism levels. To test whether there was an effect associated
with repeatedly handling and measuring the same segments, we
constructed a ‘single-spin’ vulnerability curve. In this protocol,
segments were spun only to a single xylem pressure. The single-
spin centrifuge method required the use of more stems than the
standard method. Although the standard curve was constructed
with six to seven segments repeatedly spun at different xylem
pressures, a total of 50 branches were used to construct the sin-
gle-spin curve. For the single-spin approach, five to six stem seg-
ments were spun to each xylem pressure (Table 2).
Hydraulic detection of vessel length
We used an air-injection method for detecting the effect that em-
bolized open vessels (vessels open to the segment center or
beyond) have on hydraulic conductivity and PLC (see also Tor-
res-Ruiz et al., 2014). Current-year stem segments were vacuum
infiltrated in 20 mM KCl solution for 1 h. Infiltrated stems were
then injected at their proximal base with compressed nitrogen gas
at 80 kPa for 10 min. This pressure is sufficient to allow spread of
gas in vessels that are cut open, but is not high enough to force
gas through vessel end walls (Sperry et al., 2012). After injection,
3-cm-long stem segments at various distances from the injection
site were cut from the stem underwater, and their K
s
was mea-
sured before and after removing embolism by vacuum infiltration
Table 2 Centrifuge techniques used for evaluating Olea europaea cavitation vulnerability in Torres-Ruiz et al. (2014) and the present study
Study
Centrifuge
techniques
Measures
per stem
segment
Stem segment length
(cm)
No. stem
segments
Flushed or
vacuumed
previous to
spinning
Conductivity
measurement
Background
flowSpun Measured
Torres-Ruiz
et al. (2014)
Standard Single 14 & 27 2–3 excised
from the
center
a
3–7 Yes Xyl’em or conductivity
apparatus
Not reported
Cavitron Repeated 27.5 27.5 >3 Yes Meniscus –
Flow centrifuge Repeated 14 & 27.5 14 & 27.5 >3 Yes Meniscus –
Present study Standard (foam
pads improvement)
Repeated 14 & 27 14 & 27 6 Yes Conductivity apparatus Corrected for
Standard (foam
pads improvement)
Single 14 14 50 Yes Conductivity apparatus Corrected for
a
This is not a standard procedure.
New Phytologist (2015) 205: 116–127 Ó2014 The Authors
New Phytologist Ó2014 New Phytologist Trust
www.newphytologist.com
Research
New
Phytologist
120
for 1 h. The first 3-cm segment was located 0 cm from the
injected end (used as a control to verify that injection does indeed
cause substantial embolism), the second segment was located at
7 cm from the injected end (7–10 cm from the end), and the
third segment was located 13.5 cm from the injected end (13.5–
16.5 cm from the end). These distances correspond with half of
the segment length in the two rotors that differ in diameter. A K
s
that is more depressed at 7 vs 13.5 cm indicates that more open
vessels are present in the 14 vs 27 cm segments that were centri-
fuged. This approach of estimating vessel length is consistent
with the method used in Torres-Ruiz et al. (2014) and provides a
direct estimate of the hydraulic impact of open vessels in the cen-
ter of a segment.
Native embolism and water potential measurements; air
drying of branches
In order to measure native and benchtop-dehydrated conductiv-
ity, long stems (c. 2–3 m) were cut at predawn from the same
trees used for constructing centrifuge curves. Small plastic bags
were sealed onto four to five branchlets on each larger branch and
then the larger branches were tightly triple-bagged in large plastic
bags and rapidly transported to a laboratory. Branches were
allowed to equilibrate for c. 2 h for native measures and 8–12 h
for dehydrated branches. After equilibration, the bagged branch-
lets were collected and water potentials were measured using a
pressure chamber (Model 2000; PMS Instrument Company,
Albany, OR, USA). It was important to use multiple branchlets
because even after many hours of equilibration there was still var-
iability in water potential when stems were in a dehydrated state.
The cut end of each branch was then held underwater and succes-
sive 10-cm cuts were made from the cut end, until the basal 1 m
had been removed. Samples were then cut down further under-
water, alternating cuts from the proximal and distal stem ends,
until a central 14-cm-long segment was obtained. This gradual
excision was done to relax xylem pressures before excising the
final conductivity segment (Wheeler et al., 2013). Hydraulic con-
ductivity was then measured on these samples as described previ-
ously. PLC was calculated based on measurements of the
maximum K
s
as determined by vacuum infiltration or flushing.
Statistical analyses
Correlation analyses were used to test for an association between
stem diameter and maximum vessel length and K
s
. Vulnerability
curves were analyzed via ANOVA (JMP v9.0; SAS Institute Inc.,
Carey, NC, USA). For curves generated by repeatedly spinning
the same stems, the independent variables in the model included
xylem pressure as a fixed factor and stem as a random factor. For
curves generated by spinning different stems at each pressure, the
model included xylem pressure as a fixed factor. The standard
error from these different ANOVA models was used to generate
95% confidence intervals to compare conductivity between stems
at a given xylem pressure that were either repeatedly spun in a
centrifuge vs those spun only once at a given pressure. For com-
paring dehydration and single-spin VCs, PLC datasets were fitted
to Weibull curves by least square means (LSM), and their P
50
obtained from the curves. Bootstrapping was performed for prop-
agating the uncertainty in these fits. The datasets were randomly
resampled with replacement 1000 times to generate subsamples
equal in size to the original datasets. Each subsample was fitted to
a Weibull curve by LSM. P
50
and the PLC at different pressures
were obtained for each one of these curves, and the percentiles
2.5 and 97.5 were used to determine the 95% confidence inter-
vals (CIs).
Results
Testing for the presence of open vessels
Maximum vessel length of current-year stems was 75.3 4.0 cm
(SE, n=15) with a maximum of 116 cm (Fig. 2a). Maximum
(a)
*
(b)
Fig. 3 (a) Percentage loss of hydraulic conductivity (PLC) of 3-cm-long segments located at different distances from the olive (Olea europea) stem ends
that were injected with compressed nitrogen gas at mild pressure (80 kPa) (mean +1 SE). Segments located 13.5 cm away from the injected end (white
bar) had lower PLC (and fewer open vessels) than segments located 7 cm (gray bar) and 0 cm (black bar) away from the injected end (P<0.05). (b) PLC of
segments 14 and 27 cm in length after spinning to 0.5 MPa. If open vessels drained when the centrifuge rotor started spinning, then they should show
similar PLC values as shown in (a). The asterisk in (a) indicates that the PLC of segments located 13.5 cm away from the injected end were significantly
lower than the PLC of segments located 7 cm away from the injected end (P=0.030).
Ó2014 The Authors
New Phytologist Ó2014 New Phytologist Trust
New Phytologist (2015) 205: 116–127
www.newphytologist.com
New
Phytologist Research 121
vessel length was correlated with stem diameter; bigger stems
tended to have longer vessels (r=0.814; P<0.001). K
s
also
increased with stem diameter (Fig. 2b) (r=0.635; P<0.001).
Injecting compressed nitrogen into stems led to nearly 100%
loss in conductivity (PLC) directly at the cut end of the stem, as
expected (PLC =98.2 1.0, n=5) (Fig. 3a). Short segments
located 7 cm from the injected end had 81.8 7.0 PLC, whereas
segments located 13.5 cm from the injected end showed signifi-
cantly less PLC (62.3 6.0; P=0.030, paired t-test, n=9).
These results indicated that the centrifuged shorter 14-cm stems
had significantly more open vessels than the longer 27-cm stems.
To test if open vessels drained when the centrifuge rotor began
spinning, we plotted the PLC measured at 0.5 MPa for 14- and
27-cm-long segments (Fig. 3b). Draining of open vessels (open
to center) should result in high PLC even at modest xylem ten-
sion and higher PLC in spun 14-cm segments. There was no evi-
dence of vessel draining; PLC at 0.5 MPa was much lower than
the air-injected PLC values (Fig. 3a) and long segments were not
significantly different in PLC than short ones.
Testing for an effect of segment length on centrifuge
curves
Olive branches measured in September 2013 showed a gradual
increase in PLC with decreasing xylem pressure, regardless of seg-
ment length (Fig. 4). These data were collected by using the stan-
dard method; that is, the same branch segments were exposed to
increasingly negative xylem pressures. The P
50
was 1.8 0.3
and 2.1 0.2 MPa for 27- and 14-cm segments, respectively
(mean SE) (Fig. 4a). These P
50
values were not significantly
different (P=0.26, t-test). When vulnerability curves were plot-
ted as stem area-specific conductivity, the two curves were nearly
identical (Fig. 4b).
Native values of stem area-specific conductivity and xylem
pressure are shown in Fig. 4(b) (red triangles). These native val-
ues refer to branches collected at dawn during the same time that
the samples for the centrifuge curves were collected and mea-
sured, and they closely matched the centrifuge-based vulnerabil-
ity curves.
Testing for an effect of repeated spins on centrifuge curves
Standard and single-spin methods produced P
50
values of 2.92
and 3.34 MPa, respectively (Fig. 5). Curves showed reasonable
agreement and 95% confidence intervals of the means overlapped
over much of the pressure range tested; however, curves diverged
(a)
(b)
Fig. 4 (a) Effect of stem length on centrifuge vulnerability curves. Curves
were measured in September 2013. Mean 1SE for 27-cm-long (closed
squares, n=7) and 14-cm-long segments (open squares, n=6) of current-
year extension growth. Native values of xylem pressure and stem area-
specific conductivity (samples collected at predawn) are shown as red
triangles for current-year growth. Each triangle represents one olive (Olea
europea) tree; these same six trees were sampled for the vulnerability
curves. A Weibull function was used to fit the percent loss of hydraulic
conductivity (PLC) data. (b) Stem area-specific conductivity as a function
of xylem pressure.
Fig. 5 Vulnerability curves showing percentage loss of hydraulic
conductivity (PLC) as a function of xylem pressure. Means and 95%
confidence intervals (CIs) are shown. Black circles and solid line, data
obtained from using the standard centrifuge method, which involved
repeated spins of the same stem segments (n=6); open squares and
dashed line, the ‘single-spin method’ (n=5–6 per pressure). The single-
spin protocol involves spinning segments to a single xylem pressure to
avoid any fatigue that may be caused by repeated spins. CIs overlapped at
all xylem pressures, except at the two most negative xylem pressures that
were measured (indicated by asterisks). A Weibull function was used to fit
the PLC data. All stems were collected from one mature olive (Olea
europea) tree in January 2014.
New Phytologist (2015) 205: 116–127 Ó2014 The Authors
New Phytologist Ó2014 New Phytologist Trust
www.newphytologist.com
Research
New
Phytologist
122
at the most negative xylem pressures, indicating that repeated
centrifuging of the same samples may cause overestimation of
PLC. Both of the curves shown in Fig. 5 were slightly more resis-
tant than the curve constructed in the previous fall, although the
difference in P
50
was not significant (P=0.10, t-test using
repeated spin data).
Comparison of centrifuge and dehydration curves
Data points derived from native conductivity measurements and
dehydrated branches were in excellent agreement with the single-
spin centrifuge curve (Fig. 6). Both methods had nearly identical
P
50
and curve shapes. The P
50
(95% CI) of the dehydration and
single-spin centrifuge curves were 3.36 (3.91, 2.85) MPa
and 3.33 (3.79, 2.92) MPa, respectively. These are not sta-
tistically different because both confidence intervals overlap.
For the dehydration curve, in most cases, water potential mea-
surements from individual branchlets of the same large branch
produced similar values. In a few instances there was poor water
potential equilibration despite the fact that branches were allowed
to equilibrate for several hours. This occurred at c. 8 MPa, but
there were also two instances in the pressure range between 3
and 4 MPa (see horizontal error bars in Fig. 6). Instead of dis-
carding these data, we included them in our analysis and docu-
mented it. Removing the three data points exhibiting the greatest
variation in water potentials shifted the P
50
of the dehydration
curve to be slightly more negative (3.44 MPa), but this was still
not significantly different from the P
50
of the single-spin centri-
fuge curve (i.e. this value still falls within the 95% CI of the P
50
of the single-spin centrifuge curve).
Comparison of flushed vs nonflushed samples and
comparison of flushing vs vacuum infiltration treatments
Vulnerability curve shape differed depending on whether samples
had been flushed before measurements to remove native embo-
lism. Flushed segments had higher conductivities at xylem pres-
sures >2 MPa compared to samples that were not flushed
(Fig. 7, asterisks). Further, conductivity of flushed samples
decreased between 0 and 2 MPa indicating that olive stems had
a population of vessels that cavitate at modest xylem tension. By
contrast, at these same pressures the conductivity remained nearly
constant in samples that were not flushed. As a result, these latter
samples exhibited a sigmoidal curve shape. At xylem pressures of
≤2 MPa, curves converged. The approximate xylem pressure of
stems at the time of collection was 2.5 MPa (vertical arrow in
Fig. 7).
Both flushing and vacuum infiltration resulted in significant
increases in K
s
relative to initial measurements of K
s
on
PLC
0
10
20
30
40
50
60
70
80
90
100
Xylem pressure (MPa)
–8 –7 –6 –5 –4 –3 –2 –1 0
–8 –7 –6 –5 –4 –3 –2 –1 0
Stem area-specific conductivity
(kg s–1 MPa–1 m–1)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
(a)
(b)
Fig. 6 Vulnerability curves showing (a) percentage loss of hydraulic
conductivity (PLC) or (b) stem area-specific conductivity (K
s
) as a function
of xylem pressure using the single-spin centrifuge method (blue squares).
Symbols represent individual olive (Olea europea) stem segments to show
variability. Red circles show native PLC (a) and K
s
(b) as a function of the
xylem pressure branches experienced at the time of collection or after
benchtop dehydration. Each circle represents one branch. Horizontal error
bars show variation in xylem pressure for individual branches
(means 1 SE, n=4). A Weibull function was used to fit the PLC data in
(a); solid blue line for single spin method and solid red line for dehydration
and native PLC data, the dashed lines in the corresponding colors
represent their 95% CIs.
Fig. 7 Stem area-specific conductivity (K
s
) of olive (Olea europea) stem
segments as a function of xylem pressure. Dark gray circles and solid line,
segments that were flushed before centrifugation; open circles and dashed
line, segments that were not flushed (means 1SE, n=6). Vertical arrow
shows the approximate xylem pressure of nonflushed samples at the time
of collection. Asterisks indicate significant differences in K
s
at individual
xylem pressures (P<0.05, t-test). Curves converged at xylem pressures
≤2 MPa.
Ó2014 The Authors
New Phytologist Ó2014 New Phytologist Trust
New Phytologist (2015) 205: 116–127
www.newphytologist.com
New
Phytologist Research 123
embolized segments (Fig. 8) and flushing and vacuum infiltration
were not different in their effectiveness at removing emboli.
Discussion
No effect of stem length on vulnerability curves
The data indicate that there was a higher fraction of open-to-cen-
ter vessels in short (14 cm) compared to long (27 cm) centrifuged
stems and that they would be expected to vary in their vulnerabil-
ity to cavitation curves based on the hypothesized open vessel
artifact. However, we found no difference in the vulnerability
curves of stem samples of differing lengths and differing percent-
ages of open vessels. If the standard centrifuge method was
susceptible to the open vessel artifact hypothesis, we should have
found that shorter segments were more vulnerable to cavitation
than longer segments. Furthermore, if all open vessels became
gas-filled due to draining, vulnerability curves should have shown
an immediate rise to >60 PLC based on the gas-injection data.
Instead, curves did not exceed 60 PLC until pressures dropped
below 2 MPa. These results agree with recent papers that found
no stem-length effect in long-vesselled species such as Vitis
vinifera,Quercus wislizenii,Quercus robur,Prunus persica,Sorbus
scopulina and Quercus gambelii, when the standard centrifuge
method was used (Jacobsen & Pratt, 2012; Sperry et al., 2012;
Tobin et al., 2013). However, our findings are in striking contrast
with cavitron data for Prunus persica,Quercus robur,Fraxinus
excelsior and Robinia pseudoacacia (Cochard et al., 2005, 2010;
Wang et al., 2014) indicating that these two different centrifuge
methods do not produce similar results for long-vesselled species.
This raises the question of why different centrifuge methods
provide different results. The explanation that is most consistent
with the available data is that the cavitron is prone to the intro-
duced nuclei artifact as solution flows through a segment during
spinning (Wang et al., 2014). As outlined in the Introduction,
solution does not flow through segments in the standard method
(Table 1). In addition, the cavitron may have other design fea-
tures that make it prone to introduced nuclei, because the flow
method of Li et al. (2008) is apparently not prone to this same
artifact.
Without the introduction of nuclei or draining of water from
open vessels during initial rotor acceleration (before the stem
ends become immersed in reservoir water), there is no obvious
reason why open-to-center vessels should alter the process of cavi-
tation. The water at the ends of the stem remain at (or above)
atmospheric pressure during centrifugation (Fig. 1). Cavitation
should occur in the middle of the stem where the pressures are
most negative and where the vessel network (and air-seeding
sites) remain intact and undisturbed.
Comparing centrifuge curves with native conductivities and
data obtained from air drying
Dehydration curves are often considered as a benchmark for
other methods. In the present study, we found good agreement
between centrifuge-based data and dehydration-based data. Based
on this result and the many recent studies that have found the
same agreement between methods (Jacobsen et al., 2007, 2014;
Li et al., 2008; Jacobsen & Pratt, 2012; Sperry et al., 2012; Tobin
et al., 2013), we conclude that the standard centrifuge method
can accurately measure vulnerability curves for long-vesselled spe-
cies like olive. The accuracy of measurements can be maximized
if samples are spun to a single pressure, but this protocol is more
time-consuming and requires using many different stems, which
may introduce additional experimental variability.
In contrast to our findings, Torres-Ruiz et al. (2014) reported
P50 values of 0.9 and 2.8 MPa with their small (15 cm) and
large (28 cm) rotor, respectively, when using the standard centri-
fuge method. However, a potentially important source of error
relates to the fact that background flow was not consistently
accounted for in the Torres-Ruiz et al. (2014) study (Table 2).
We found that background flow had a large effect on K
s
(a) (b)
Fig. 8 Effectiveness of flushing and vacuum infiltration treatments to remove embolism in xylem. A total of 18 olive (Olea europea) stems were collected;
stems were cut in air. (a) Nine stem segments were prepared and the initial stem area-specific conductivity (K
s
) was determined (white bar). Segments
were then flushed for 1 h to remove embolism (blue bar). Subsequently, segments were vacuum infiltrated for 1 h to test if K
s
would increase further (black
bar). (b) A second set of nine stem segments was vacuum infiltrated (red bar) and subsequently flushed (black bar). Although not significant at P<0.05,
flushing did cause a slight increase in K
s
relative to vacuum infiltration. Unique letters indicate significant differences at P<0.05 based on a one-way
ANOVA. Values are means +SE.
New Phytologist (2015) 205: 116–127 Ó2014 The Authors
New Phytologist Ó2014 New Phytologist Trust
www.newphytologist.com
Research
New
Phytologist
124
measurements in olive. In addition, Torres-Ruiz et al. (2014)
reported that their K
s
values were highly variable and the actual
K
s
values for their vulnerability curve material were not reported.
The wide variability they observed suggests that they were analyz-
ing samples that were not directly comparable. Aside from an
uneven and minimal (n=3–7) sample size, their study is poten-
tially confounded by seasonal effects because their short segment
curves were all constructed in July, whereas their long segment
curves were all measured in September. Moreover, their imaging
data were collected in February and 3 yr after other measures. All
of these factors make it difficult to interpret the Torres-Ruiz et al.
(2014) findings and to deduce why their curves could not be
reproduced.
Data supporting concerns about the accuracy of the standard
centrifuge method also comes from work on grapevine (Choat
et al., 2010), a species with wide and relatively long vessels. The
authors assumed that dehydration curves can be used as a refer-
ence for the centrifuge method. However, a follow-up study with
long (>3 m) grapevine shoots revealed that dehydration can lead
to errors in determination of maximum K
s
with time, and that
this made it impossible to accurately use the relative measure of
PLC for constructing vulnerability curves in this species (Jacob-
sen & Pratt, 2012). Gel formation is well described for Vitis
vinifera cv Chardonnay, the variety Choat et al. (2010) examined,
especially in response to embolism, wounding and infection (Sun
et al., 2008, 2013). McElrone et al. (2012) argued that Choat
et al. (2010) avoided issues with tyloses/gel formation by dehy-
drating long stems and sampling at the center of these stems.
However, despite the implication by McElrone et al. (2012), Jac-
obsen & Pratt (2012) sampled in much the same way as Choat
et al. (2010) and their work revealed that gels formed at >1m
from cut ends of long branches. Importantly, the Choat et al.
(2010) study did not report area-specific conductivity in absolute
units, relying instead on changes in PLC relative to a maximum
(and presumably constant and reproducible) reference. As
explained, the maximum K
s
is prone to drift in grapevine. Abso-
lute units are arguably preferable for fully interpreting vulnerabil-
ity curves, especially if these curves are used to come to broad
conclusions about the validity of methods. When gel formation
was accounted for, dehydration and centrifuge curves in grape-
vine agreed (Jacobsen & Pratt, 2012).
Suggestions for future work
In vivo imaging has received interest as a new tool to assess vul-
nerability to cavitation. We caution that imaging (regardless of
whether it involves dye perfusions, cryo-SEM, in vivo imaging or
some other method) needs to be validated and paralleled by
hydraulic measurements. Imaging techniques are not immune to
artifacts (Canny, 1997; Cochard et al., 2000; Perez-Donoso et al.,
2007). In grapevine, it was found that gel-filled conduits could
not be differentiated from water- or saline solution-filled con-
duits when examined using MRI (Perez-Donoso et al., 2007),
suggesting that these images are of limited utility in estimating
vessel hydraulic functionality. Hence, the need for using and, as
necessary, improving hydraulic methods will remain. In the
interest of achieving consistency and accuracy in future plant
hydraulic studies, we suggest the following practices which may
have been important for avoiding artifacts in our study as well as
others that have also shown no open vessel artifact.
(1) The Alder et al. (1997) rotor design is preferable to the cavi-
tron rotor design, especially when long-vesselled species are stud-
ied. A potential source of error may occur by draining of cut
vessels when the rotor is stopped or moving very slowly and stem
ends are exposed to air (rather than being immersed in reservoir
water). This error can easily be addressed by using foam pads in
rotor reservoirs as described in Tobin et al. (2013), so that sam-
ples do not dry out between spins.
(2) Vulnerability curves should be reported with K
s
in addition
to PLC values. This is particularly evident for nonflushed
samples. In many cases it is more informative to focus on the
actual area-specific conductivities than on the PLC values
(Jacobsen & Pratt, 2012; Sperry et al., 2012), which are subject to
drift or error in the maximum conductivity reference value. Even
in the absence of reference drift, PLC is not the best basis for com-
parison because plants at 90 vs 10 PLC can actually have identical
K
s
and conducting capacities (Li et al., 2008; Taneda & Sperry,
2008). Pratt et al. (2008) showed that when curves were reported
with K
s
values, stems and roots were not different in conductivity
at the seasonal minimum predawn water potential. In other
words, the long held view that roots tend to be more vulnerable to
embolism may also be tied to the PLC vs K
s
issue in some cases.
(3) It is arguably preferable to flush (or vacuum-infiltrate) sam-
ples before the construction of vulnerability curves. Flushing
makes centrifuge curves comparable to dehydration data or native
embolism measurements, because these latter measurements typi-
cally depend on measuring the maximum (flushed) conductivity.
Curves reporting the PLC of nonflushed segments use native con-
ductivity as their maximum reference, but native conductivity
will vary with stem age, season, site and other factors. Such curves
can only capture the vulnerability of xylem that is still functional
under field conditions, and this may only be a small fraction,
especially if the xylem is already highly embolized at the time of
sampling. Because the curve only reflects more resistant xylem,
one may erroneously conclude that there are no vulnerable vessels
present and that there is a substantial margin of safety from cavi-
tation. The potential discrepancy between flushed and nonflu-
shed samples is particularly evident when curves are expressed as
PLC (Sperry et al., 2012; see their Fig. 5b).
As expected, the K
s
of flushed and unflushed curves converge
at the minimum xylem pressure that stems were exposed to in the
field (Fig. 7). Hence, a flushed curve can readily be scaled to a
nonflushed one for any desired xylem pressure (Pockman &
Sperry, 2000). An unflushed curve, however, cannot predict the
vulnerability of any cavitated xylem, and in this sense the unflu-
shed curve contains less information.
Many reported curve shape discrepancies and the discussion of
‘unrealistic’ P
50
values in the range between 0 and 0.5 MPa
(Cochard et al., 2013) may be tied to simple confusion over
flushed vs unflushed curves. Nonflushed curves are a priori more
likely to result in a sigmoid shape and more negative P
50
values
(Fig. 7), because they do not include the necessarily more
Ó2014 The Authors
New Phytologist Ó2014 New Phytologist Trust
New Phytologist (2015) 205: 116–127
www.newphytologist.com
New
Phytologist Research 125
vulnerable vessels that were already cavitated before the curve was
started. Meaningful comparisons of curve statistics require consis-
tent scaling of flushed curves, typically to a characteristic field
pressure of the species (Choat et al., 2012).
(4) In dehydrated samples and samples with low conductivity
(such as olive stems), it is crucial to determine the solution flow
in the absence of an applied pressure gradient (Hacke et al.,
2000b). Water uptake by samples may be driven by osmotic and
capillary uptake. Methods in which this ‘background flow’ is not
determined can lead to significant errors that can affect accurate
estimation of vulnerability to cavitation.
(5) Dehydration curves and native conductivities must be mea-
sured carefully. The relaxation of samples may be necessary to
avoid cutting artifacts described as occurring for some species
(Sperry, 2013; Wheeler et al., 2013); however, it is also important
to avoid hydrating for too long in some species as they can refill
emboli before excision (Trifilo et al., 2014). Estimates of native
stem xylem pressure must also be made with caution on equili-
brated tissue. When the xylem is highly embolized the pressure
chamber values can become unreliable and other methods may
be required (Tobin et al., 2013). Using stem psychrometers in
tandem with the pressure chamber can be used to diagnose such
cases.
(6) Comparisons between techniques must not be confounded
by inherent differences between individual plants, age, diameter,
aspect, vigor and health of segments, time of year, sampling pop-
ulation, cavitation fatigue and sampling year. All of these sources
of variation are potentially significant and should be controlled
(Matzner et al., 2001; Jacobsen et al., 2007; Taneda & Sperry,
2008; Anderegg et al., 2013).
(7) Curves for long-vesselled species should be checked, for
example by comparing stems of different length, or by compari-
son to native embolism or dehydration data. If the native and
centrifuge data agree, those data should be regarded as valid
regardless of the P
50
or curve shape observed.
Conclusion
We showed that the standard centrifuge method accurately mea-
sured vulnerability curves in stems with open vessels. The pres-
ence of long vessels was not associated with having vulnerable
xylem, suggesting that curve shape is not determined by the pres-
ence of open vessels. Many species exhibit an exponential loss in
hydraulic conductivity, including roots of conifers where trac-
heids render them immune to any possibility of an open vessel
artifact (Hacke et al., 2000a; Pittermann et al., 2006). A con-
certed attempt to standardize techniques and protocols across the
field to those that have been repeatedly shown to be robust to
artifact, such as those described in the present study, may help in
achieving more accurate results.
Acknowledgements
A.L.J. was supported by NSF (IOS-1252232). M.D.V. acknowl-
edges support from the Technical University of Madrid (Legado
Gonzalez Esparcia Grant). R.B.P., E.D.M. and M.D.V. were
supported by NSF (IOS-0845125). U.G.H. acknowledges fund-
ing from an NSERC Discovery grant and the Canada Research
Chair program. J.S.S. was supported by NSF grant IBN-
0743148. We appreciate Stefan Schreiber’s useful comments on
statistical analyses.
References
Alder NN, Pockman WT, Sperry JS, Nuismer S. 1997. Use of centrifugal force
in the study of xylem cavitation. Journal of Experimental Botany 48: 665–674.
Anderegg WRL, Plavcova L, Anderegg LDL, Hacke UG, Berry JA, Field CB.
2013. Drought’s legacy: multi-year hydraulic deterioration underlies
widespread aspen forest die-off and portends increased future risk. Global
Change Biology 19: 1188–1196.
Arango-Velez A, Zwiazek JJ, Thomas BR, Tyree MT. 2011. Stomatal factors
and vulnerability of stem xylem to cavitation in poplars. Physiologia Plantarum
143: 154–165.
Cai J, Tyree MT. 2010. The impact of vessel size on vulnerability curves: data
and models for within-species variability in saplings of aspen, Populus
tremuloides Michx. Plant, Cell & Environment 33: 1059–1069.
Canny MJ. 1997. Vessel contents during transpiration: embolisms and refilling.
American Journal of Botany 84: 1223–1230.
Choat B, Drayton WM, Brodersen C, Matthews MA, Shackel KA, Wada H,
Mcelrone AJ. 2010. Measurement of vulnerability to water stress-induced
cavitation in grapevine: a comparison of four techniques applied to a
long-vesseled species. Plant, Cell & Environment 33: 1502–1512.
Choat B, Jansen S, Brodribb TJ, Cochard H, Delzon S, Bhaskar R, Bucci SJ,
Feild TS, Gleason SM, Hacke UG et al. 2012. Global convergence in the
vulnerability of forests to drought. Nature 491: 752–755.
Christman MA, Sperry JS, Smith DD. 2012. Rare pits, large vessels and extreme
vulnerability to cavitation in a ring-porous tree species. New Phytologist 193:
713–720.
Cochard H, Badel E, Herbette S, Delzon S, Choat B, Jansen S. 2013. Methods
for measuring plant vulnerability to cavitation: a critical review. Journal of
Experimental Botany 64: 4779–4791.
Cochard H, Bodet C, Ameglio T, Cruiziat P. 2000. Cryo-scanning electron
microscopy observations of vessel content during transpiration in walnut
petioles. Facts or artifacts? Plant Physiology 124: 1191–1202.
Cochard H, Casella E, Mencuccini M. 2007. Xylem vulnerability to cavitation
varies among poplar and willow clones and correlates with yield. Tree Physiology
27: 1761–1767.
Cochard H, Damour G, Bodet C, Tharwat I, Poirier M, Ameglio T. 2005.
Evaluation of a new centrifuge technique for rapid generation of xylem
vulnerability curves. Physiologia Plantarum 124: 410–418.
Cochard H, Delzon S. 2013. Hydraulic failure and repair are not routine in trees.
Annals of Forest Science 70: 659–661.
Cochard H, Herbette S, Barigah T, Badel E, Ennajeh M, Vilagrosa A. 2010.
Does sample length influence the shape of xylem embolism vulnerability
curves? A test with the Cavitron spinning technique. Plant, Cell & Environment
33: 1543–1552.
Davis SD, Sperry JS, Hacke UG. 1999. The relationship between xylem conduit
diameter and cavitation caused by freeze–thaw events. American Journal of
Botany 86: 1367–1372.
Hacke UG, Sperry JS. 2001. Functional and ecological xylem anatomy.
Perspectives in Plant Ecology Evolution and Systematics 4:97–115.
Hacke UG, Sperry JS, Ewers BE, Ellsworth DS, Sch€afer KVR, Oren R. 2000a.
Influence of soil porosity on water use in Pinus taeda.Oecologia 124: 495–505.
Hacke UG, Sperry JS, Pittermann J. 2000b. Drought experience and cavitation
resistance in six shrubs from the Great Basin, Utah. Basic and Applied Ecology 1:
31–41.
Hacke UG, Sperry JS, Pockman WT, Davis SD, McCulloh KA. 2001. Trends in
wood density and structure are linked to prevention of xylem implosion by
negative pressure. Oecologia 126: 457–461.
Hacke UG, Sperry JS, Wheeler JK, Castro L. 2006. Scaling of angiosperm xylem
structure with safety and efficiency. Tree Physiology 26: 689–701.
New Phytologist (2015) 205: 116–127 Ó2014 The Authors
New Phytologist Ó2014 New Phytologist Trust
www.newphytologist.com
Research
New
Phytologist
126
Holbrook NM, Burns MJ, Field CB. 1995. Negative xylem pressures in plants: a
test of the balancing pressure technique. Science 270: 1193–1194.
Jacobsen AL, Ewers FW, Pratt RB, Paddock WA, Davis SD. 2005. Do xylem
fibers affect vessel cavitation resistance? Plant Physiology 139: 546–556.
Jacobsen AL, Pratt RB. 2012. No evidence for an open vessel effect in
centrifuge-based vulnerability curves of a long-vesselled liana (Vitis vinifera).
New Phytologist 194: 982–990.
Jacobsen AL, Pratt RB, Davis SD, Ewers FW. 2007. Cavitation resistance and
seasonal hydraulics differ among three arid Californian plant communities.
Plant, Cell & Environment 30: 1599–1609.
Jacobsen AL, Pratt RB, Davis SD, Tobin MF. 2014. Geographic and seasonal
variation in chaparral vulnerability to cavitation. Madrono, in press.
Lens F, Sperry JS, Christman MA, Choat B, Rabaey D, Jansen S. 2010. Testing
hypotheses that link wood anatomy to cavitation resistance and hydraulic
conductivity in the genus Acer.New Phytologist 190: 709–723.
Li YY, Sperry JS, Taneda H, Bush SE, Hacke UG. 2008. Evaluation of
centrifugal methods for measuring xylem cavitation in conifers, diffuse- and
ring-porous angiosperms. New Phytologist 177: 558–568.
Matzner SL, Rice KJ, Richards JH. 2001. Intra-specific variation in xylem
cavitation in interior live oak (Quercus wislizenii A. DC.). Journal of
Experimental Botany 52: 783–789.
McElrone A, Brodersen C, Alsina M, Drayton W, Matthews M, Shackel K,
Wada H, Zufferey V, Choat B. 2012. Centrifuge technique consistently
overestimates vulnerability to water stress-induced cavitation in grapevines as
confirmed with high-resolution computed tomography. New Phytologist 196:
661–665.
Paddock WAS, Davis SD, Pratt RB, Jacobsen A, Tobin MF, Lopez-Portillo J,
Ewers FW. 2013. Factors determining mortality of adult chaparral shrubs in an
extreme drought year in California. Aliso 31:49–57.
Perez-Donoso AG, Greve LC, Walton JH, Shackel KA, Labavitch JM. 2007.
Xylella fastidiosa infection and ethylene exposure result in xylem and water
movement disruption in grapevine shoots. Plant Physiology 143: 1024–1036.
Pittermann J, Sperry JS. 2006. Analysis of freeze-thaw embolism in conifers. The
interaction between cavitation pressure and tracheid size. Plant Physiology 140:
374–382.
Pittermann J, Sperry JS, Hacke UG, Wheeler JK, Sikkema EH. 2006.
Inter-tracheid pitting and the hydraulic efficiency of conifer wood: the role of
tracheid allometry and cavitation protection. American Journal of Botany 93:
1265–1273.
Plaut JA, Yepez EA, Hill J, Pangle R, Sperry JS, Pockman WT, McDowell NG.
2012. Hydraulic limits preceding mortality in a pi~non–juniper woodland
under experimental drought. Plant, Cell & Environment 35: 1601–1617.
Plavcova L, Hacke UG. 2011. Heterogeneous distribution of pectin epitopes and
calcium in different pit types of four angiosperm species. New Phytologist 192:
885–897.
Pockman WT, Sperry JS. 2000. Vulnerability to xylem cavitation and the
distribution of Sonoran desert vegetation. American Journal of Botany 87:
1287–1299.
Pockman WT, Sperry JS, O’Leary JW. 1995. Sustained and significant negative
water pressure in xylem. Nature 378: 715–716.
Pratt RB, Jacobsen AL, Ewers FW, Davis SD. 2007a. Relationships among
xylem transport, biomechanics and storage in stems and roots of nine
Rhamnaceae species of the California chaparral. New Phytologist 174: 787–798.
Pratt RB, Jacobsen AL, Golgotiu KA, Sperry JS, Ewers FW, Davis SD. 2007b.
Life history type and water stress tolerance in nine California chaparral species
(Rhamnaceae). Ecological Monographs 77: 239–253.
Pratt RB, Jacobsen AL, Mohla R, Ewers FW, Davis SD. 2008. Linkage between
water stress tolerance and life history type in seedlings of nine chaparral species
(Rhamnaceae). Journal of Ecology 96: 1252–1265.
Rockwell FE, Wheeler JK, Holbrook NM. 2014. Cavitation and its discontents:
opportunities for resolving current controversies. Plant Physiology 164: 1649–
1660.
Schreiber SG, Hacke UG, Hamann A, Thomas BR. 2011. Genetic variation of
hydraulic and wood anatomical traits in hybrid poplar and trembling aspen.
New Phytologist 190: 150–160.
Sperry J. 2013. Cutting-edge research or cutting-edge artefact? An overdue
control experiment complicates the xylem refilling story. Plant, Cell &
Environment 36: 1916–1918.
Sperry JS, Christman MA, Torres-Ruiz JM, Taneda H, Smith DD. 2012.
Vulnerability curves by centrifugation: is there an open vessel artefact, and
are ‘r’ shaped curves necessarily invalid? Plant, Cell & Environment 35:
601–610.
Sperry JS, Donnelly JR, Tyree MT. 1988. A method for measuring
hydraulic conductivity and embolism in xylem. Plant, Cell & Environment 11:
35–40.
Sperry JS, Hacke UG. 2002. Desert shrub water relations with respect to soil
characteristics and plant functional type. Functional Ecology 16: 367–378.
Sperry JS, Hacke UG. 2004. Analysis of circular bordered pit function –I.
Angiosperm vessels with homogenous pit membranes. American Journal of
Botany 91: 369–385.
Sun Q, Rost TL, Matthews MA. 2008. Wound-induced vascular occlusions in
Vitis vinifera (Vitaceae): tyloses in summer and gels in winter. American Journal
of Botany 95: 1498–1505.
Sun Q, Sun Y, Walker MA, Labavitch JM. 2013. Vascular occlusions in
grapevines with Pierce’s Disease make disease symptom development worse.
Plant Physiology 161: 1529–1541.
Taneda H, Sperry JS. 2008. A case-study of water transport in co-occurring ring-
versus diffuse-porous trees: contrasts in water-status, conducting capacity,
cavitation and vessel refilling. Tree Physiology 28: 1641–1651.
Tobin MF, Pratt RB, Jacobsen AL, De Guzman ME. 2013. Xylem vulnerability
to cavitation can be accurately characterised in species with long vessels using a
centrifuge method. Plant Biology 15: 496–504.
Torres-Ruiz JM, Cochard H, Mayr S, Beikircher B, Diaz-Espejo A,
Rodriguez-Dominguez CM, Badel E, Fernandez JE. 2014. Vulnerability to
cavitation in Olea europaea current-year shoots: further evidence of an
open-vessel artefact associated with centrifuge and air-injection techniques.
Physiologia Plantarum. doi: 10.1111/ppl.12185.
Torres-Ruiz JM, Sperry JS, Fernandez JE. 2012. Improving xylem hydraulic
conductivity measurements by correcting the error caused by passive water
uptake. Physiologia Plantarum 146: 129–135.
Trifilo P, Raimondo F, Lo Gullo MA, Barbera PM, Salleo S, Nardini A. 2014.
Relax and refill: xylem rehydration prior to hydraulic measurements favours
embolism repair in stems and generates artificially low PLC values. Plant, Cell
& Environment. doi: 10.1111/pce.12313.
Wang R, Zhang L, Zhang S, Cai J, Tyree MT. 2014. Water relations of
Robinia pseudoacacia L.: do vessels cavitate and refill diurnally or are
R-shaped curves invalid in Robinia?Plant, Cell & Environment. doi: 10.
1111/pce.12315.
Wheeler JK, Huggett BA, Tofte AN, Rockwell FE, Holbrook NM. 2013.
Cutting xylem under tension or supersaturated with gas can generate PLC and
the appearance of rapid recovery from embolism. Plant, Cell & Environment
36: 1938–1949.
Ó2014 The Authors
New Phytologist Ó2014 New Phytologist Trust
New Phytologist (2015) 205: 116–127
www.newphytologist.com
New
Phytologist Research 127
