Journal of Microscopy, Vol. 198, Pt 1, April 2000, pp. 24–33.

Received 3 June 1999; accepted 6 October 1999

The reliability of cryoSEM for the observation and

quantification of xylem embolisms and quantitative
analysis of xylem sap in situ

M. E. M C CULLY*, M. W. SHANE†, A. N. BAKER‡, C. X. HUANG§, L. E. C. LING§

& M. J. CANNY¶
*CSIRO Division of Plant Industry, Box 1600, Canberra 2601, Australia
†Department of Plant Sciences, University of Western Australia, Nedlands 6009, Australia
‡Biology Department, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario,
Canada K1S 5B6
§Carleton University Research Facility for Electron Microscopy, Carleton University,
1125 Colonel By Drive, Ottawa, Ontario, Canada K1S 5B6
¶Research School of Biological Sciences, Australian National University, Box 475, Canberra 2601,
Australia

Key words. Cryoanalytical scanning microscopy, cryoplaning, EDX calibration, gas

embolisms, liquid and gas space detection, xylem embolisms, xylem sap.

Summary time, allowed the detection and quantification of gas

bubbles and embolisms in individual xylem vessels in leaves
The reliability of cryoSEM for visualizing gas embolisms in
and roots while these organs were still attached to the
xylem vessels of intact, functioning roots is examined and
plant and fully functioning (e.g. Canny, 1997a, b; McCully
discussed. The possibility that these embolisms form as a
et al., 1998; Buchard et al., 1999; McCully, 1999; Pate &
result of freezing water columns under tension is discounted
Canny, 1999; Shane & McCully, 1999). Also, X-ray micro-
by a double-freeze experiment. Two regions of the same root,
analysis of cryoplaned faces of these frozen tissues has
one frozen under tension, the other isolated from the tension
yielded data previously unobtainable for the concentration
by the first freeze, had the same percentage of embolisms, as
of elements in the xylem sap in situ in these organs (Enns
did also long pieces of root frozen simultaneously along their
et al., 1998).
length. The reliability of energy-dispersive X-ray analysis to
The data obtained in these studies have, in several ways,
measure xylem sap concentration in situ in frozen tissue
differed from estimates of the same parameters obtained by
was established by measurement of KCl standard solution
a variety of indirect means (see discussions in the above
frozen on stubs, and within xylem vessels. Solute hetero-
papers). Because the new data have suggested the necessity
geneity within the vessels varied with freezing procedure;
of several paradigm shifts in the established interpretations
deep-freeze > LN2 > cryopliers > liquid ethane, but only the
of plant water relations (Canny, 1998) it is not surprising
deep-freeze method gave unsatisfactory estimates of con-
that the reliability of the cryoanalytical methods has been
centration for the standard solution. It is concluded that
questioned closely by plant physiologists. The following
cryoanalytical SEM is useful for direct observation of gas
are the commonest criticisms that have been raised: (A)
and liquid-filled compartments, and for solute analyses at
that the freezing of the water columns in the vessels forces
depth within intact plant organs.
air out of solution and that these bubbles nucleate cavi-
tations which immediately fill the vessels with gas and push
Introduction water out of the vessels during freezing; (B) that the sap
columns are under tension during transpiration so that
The use of the cryoscanning electron microscope (CSEM) to
they are more vulnerable to cavitation than static water
examine fully hydrated roots and leaves has, for the first
while they are freezing; (C) that embolisms form while the
Correspondence: M. E. McCully, Microscopy Centre, CSIRO Division of Plant frozen root pieces are being planed in the cryomicrotome,
Industry, Box 1600 Canberra 2601, Australia. Tel: þ61 (0)26246 5343; fax: and (D) that energy-dispersive X-ray analyses (EDX) of
þ61 (0)26246 5399; e-mail: [email protected] vessel contents are unreliable because large volumes of

24 q 2000 The Royal Microscopical Society

 C RYOS E M O F X Y L E M 25

water are pushed either into or out of the vessels during

freezing.
The original papers presenting the cryoanalytical work all
contain discussions of particular precautions taken and
controls done to minimize the possibility of artefacts. There
is, however, no comprehensive description and discussion of
these. This paper enlarges on the earlier descriptions and
results of the controls, and presents new data from further
control experiments. We include specific responses to
criticisms A, B, and D (C seems too improbable to pursue
further). Also included are suggestions on how to maximize
the reliability of the cryoanalytical techniques for use with
plant tissues, gleaned from our experiences of more than a
decade.

Background
To put this present paper in context, it is necessary first to
give a brief summary of the findings by cryoanalytical
microscopy about the formation and repair of gas embolisms
in vessels and xylem sap composition, and the divergence of
these observations from the established patterns in the
literature.
(a) Vessel embolisms. The cryotechniques have shown
that vessels in roots and leaves of a range of crop species
growing in well-watered soil are liquid-filled before sunrise Fig. 1. Cryoplaned transverse faces of pieces of buckwheat roots
(e.g. Fig. 1a) and develop gas embolisms in some vessels frozen in situ in the field: (a) frozen at dawn; all the large xylem
early in the morning (e.g. Fig. 1b) while transpiration is still vessels were sap-filled; (b) frozen at 08.30 hours. Many large vessels
very weak. The percentage of vessels embolized peaks about contained gas embolisms. Specimens viewed at 7 kV. Bar ¼ 50 mm.
noon and then declines while the plants are still transpiring
actively, to reach zero in the early evening. This cycle is
repeated daily throughout the life of the plant. Previously, it
Materials and methods
had been thought from the results of indirect methods of
embolism detection that vessels would embolize only rarely, General accounts of the techniques of freezing, preparing,
and only under conditions of much higher water stress, and observing and analysing biological tissues may be found in
that a vessel embolization was a disastrous event, possibly Echlin & Taylor (1986), Echlin (1992) and Marshall
not being repairable at all, and certainly not while (1987).
transpiration continued.
(b) Xylem sap analysis. At night, or at other times when a
Plant material
plant is not transpiring, the sap in the xylem is at positive
pressure, rather than under tension as it is during trans- The original CSEM observations referred to above were
piration. This pressure (root pressure) pushes sap into the made on buckwheat, maize, soybean and sunflower plants
shoots where, at high humidity, some may be seen on leaf growing in well-watered soil either in a glasshouse or in the
margins as guttation-drops. The root xylem vessels of intact, field (Canny, 1997a, b; McCully et al., 1998; Buchard et al.,
guttating plants were analysed in situ for the first time by 1999; McCully, 1999; Shane & McCully, 1999). One study
X-ray microanalysis of frozen material. These analyses was done on roots of the grass tree (Xanthorrhea preissii)
showed a uniform, low concentration of solutes along the growing in a sclerophyllous woodland in Western Australia
length of the vessels (Enns et al., 1998). This finding (Pate & Canny, 1999). The tests done for the present paper
contradicts the current explanation (e.g. Taiz & Zeiger, were all with transpiring maize plants grown in well-
1998) that the root pressure is generated by a high watered soil in pots (February and March in the glasshouse)
osmotic concentration of the sap within the vessels, and or in the field (July to September). The root regions used
that there may be a steep basipetal gradient in this were always proximal to where the large, primary xylem
concentration. vessels were mature (St. Aubin et al., 1986). For the

q 2000 The Royal Microscopical Society, Journal of Microscopy, 198, 24–33

26 M. E. M C CU LLY E T AL .

embolism determinations, the roots were frozen when plants EDX calibration
were transpiring vigorously.
The output of the spectral data is of two kinds: the live
time and the counts in the elemental peaks. The live time
Basic freezing methods
is the record of the time taken to accumulate the preset
Soil was gently pushed away to expose short lengths of counts (80 000 counts) in the Al peak, the conductive
individual axile roots. This was done quickly and, if in the coating which is used also as a reference. The live time is a
field, in the shade of an umbrella to minimize drying. The measure of the thickness of the coating, which varies from
exposed root regions were rapidly frozen in situ with specimen to specimen, and (slightly) from place to place in
specially constructed cryopliers cooled in liquid nitrogen each specimen. The thicker the coating, the more the
(LN2). These pliers had heavy, polished copper jaws and counts in the other elemental peaks are attenuated. To
were adjustable to match the width of the roots so that firm allow for this variation, the counts in each peak were
contact was made with the root surface but the tissues were standardized by dividing by the live time recorded for the
not crushed. Frozen pieces were detached with a scalpel and spectrum.
transferred in the pliers to LN2, placed in cryovials and held The counts in the elemental peaks are processed by the
in a cryostore until examined. For analysis of xylem sap built-in programme in two ways. First, they are turned from
along the length of an individual root, a complete root was total counts to net counts by subtracting from each the
frozen in a bath of LN2 while still attached to the plant. counts in the background X-radiation in that region of the
Pieces were cut from the root in order under LN2 and spectrum. The program calculates this background by
cryostored. averaging the recorded values of the background on either
side of the peak. Second, the net counts are expressed as a
percentage of the total counts, and recorded as ‘percentage
Microscopy of frozen root pieces
ratios’. The value we used for each elemental record was
A short section was cut from the middle of each of the the (percentage ratio) × 1000/(live time), which we call a
original frozen root pieces while they were under LN2. These ‘standard ratio’.
pieces were quickly affixed to stubs with Tissue Tek and To convert the recorded standard ratio to a concentra-
transferred under LN2 to a cryomicrotome where they were tion, calibration curves were constructed by analysis of
planed to a smooth transverse face at 193 K with glass standard solutions. For each element, a series of standard
and/or diamond knives. The pieces were then transferred solutions was prepared, usually six solutions ranging in
in LN2 to a cryotransfer unit (Oxford Instruments, concentration from 15 to 300 mM in distilled water, and
Eynsham, Oxford, U.K.) and hence to the cryostage of the aliquots of each solution were mixed with graphite slurry
SEM (JEOL 6400). Specimens were lightly etched by careful to make 5% carbon in the mixture (Treeby et al., 1987).
warming to 183 K while they were continuously observed This was to provide a similar (X-ray absorbing) background
uncoated at 1 kV. The instant that faint cell outlines began carbon concentration to that expected in the plant tissue
to appear, the specimen was recooled to 153 K, then coated specimens. Several drops of each solution sitting over holes
with 50 nm of evaporated high-purity Al. Coated specimens in Al stubs were quickly frozen with nitrogen slush, planed
were observed at 7–15 kV (for further details of these flat at 193 K, then etched and coated with Al in the same
methods see Canny & Huang, 1993; Huang et al., 1994; procedure as detailed for the in situ analysis of regular
McCully, 1994). Images were recorded as video prints and specimens described above (see also, Canny & Huang,
on TMax 100, 120 roll film. 1993; Huang et al., 1994). Thirty spectra were recorded for
each concentration from various parts of the frozen drops,
and the means of the standard ratios plotted against
X-ray microanalysis
concentration. Examples of such calibration curves are
Microanalysis was with a Link eXL system (Oxford Instru- shown in Figs 2(a) and (b) for K and Cl. The lines are the
ments) using the Be window. The working distance was linear regressions of standard ratio on concentration (mM).
35 mm, take-off angle 338, voltage 15 kV and the probe Similar calibrations made without added carbon gave
current was 1.00 nA (measured with a Faraday cup). The regressions indistinguishable from these.
scan raster was a 10-mm square at magnification 1000 ×. The threshold of sensitivity is determined by the signal/
During most analyses the magnification was varied so that noise ratio at the low concentration end of the line, i.e. on
the scan raster covered most of the surface area of a vessel the variability of the background X-radiation. Suppose the
lumen without touching the walls. Spectra were accumu- lowest detectable percentage ratio is about 1%. Putting this
lated for a total count for Al of 80 000. Analyses from value into the regression equation for K with a live time of
specimens for which the live time was outside the range of 150 s gives a value for [K] of 14 mM. This approximates the
120–150 s were not used. threshold of sensitivity.

q 2000 The Royal Microscopical Society, Journal of Microscopy, 198, 24–33

 C RYOS E M O F X Y L E M 27

styrofoam box with deeply knotched ends, which were then

sealed around the root with a styrofoam wedge and the box
filled with LN2. Two centimetres at each end of the frozen
pieces were discarded and the rest of the root cut into 2 cm
lengths under LN2. Each length was put into a cryovial as it
was cut off and its position along the root recorded. Pieces
were prepared and examined in the CSEM as described
above and the numbers and positions of embolized vessels
determined.

In situ analysis of standard solutions drawn into xylem

vessels
Ten-centimetre lengths of mature root regions were soaked
in water for at least 1 h under vacuum to refill any
embolized vessels. The cortex was gently removed from the
proximal 1 cm of each piece and the bare stele was sealed
into a small diameter plastic tube with Qubitac (Qubit
Systems Inc., Kingston, Ontario). A hand vacuum pump
was used to draw a 100-mM solution of KCl through the
xylem vessels until there had been about 10 000 vessel
volume changes. The loaded root pieces were then frozen in
one of five ways: (a) the mid-portion was frozen with
cryopliers as described above; (b) a similar piece was frozen
with the pliers, but where a short length of cortex had been
removed before perfusion; (c) the whole root piece was
plunged into LN2; (d) the whole root piece was frozen in a
freezer at 193 K; (e) a 2-cm length was perfused then frozen
Fig. 2. Calibration graphs for (a) [K] and (b) [Cl] (ordinates). The by plunging into liquid ethane close to its melting point.
abscissa gives measured values of the standard ratio (% ratio ×
1000/live time) derived from means of two replicate measurements
of 10 spectra from each of four frozen standard solutions. The lines
Effect of speed of freezing on uniformity of salt
are the linear regressions of standard ratio on concentration (for K, concentration across the lumens of individual vessels
y ¼ 0.880 þ 0.367[K], R2 ¼ 0.999; for Cl, y ¼ 0.604 þ 0.336[Cl], The above analyses of xylem lumens were all done using the
R2 ¼ 0.999). The standard solutions contained 5% carbon as
standard raster size (10 mm square at 1000×) changing the
graphite slurry.
magnification so that the raster covered most of the area of
the vessel face without including any of the wall. Con-
centrations thus determined for each vessel lumen loaded
Double freeze experiments
with the KCl solution were compared with concentrations
Lengths of roots in situ were quickly exposed as described measured with small rasters at five non-overlapping
above and frozen with cryopliers at the end closest to the locations in a standard quincunx pattern over the same
shoot. The pliers were held in place while a second freeze, lumen (Fig. 5a). These comparisons were done for material
also with cryopliers, was made about 15 cm closer to the tip frozen by each of the above methods.
of the same root within 1–2 s (two people were required for
this experiment). The upper and lower frozen pieces of each
Results
root were processed as above and examined in the CSEM
and the number of embolized vessels in each compared.
Embolisms in the frozen roots of transpiring plants
As in the earlier experiments referred to in the Introduction
Numbers of embolized vessels along lengths of individual
some of the large vessels were embolized in most of the roots
roots
examined. Gas-filled vessels are easily distinguished from
About 28 cm of mature regions of intact nodal roots of liquid-filled ones, even though the former occasionally
transpiring plants were gently exposed as for the freezing contain ice debris which drops into them during planing
experiments. Each exposed root portion was laid through a (Figs 1(a) and (b), Table 1). The number of large vessels

q 2000 The Royal Microscopical Society, Journal of Microscopy, 198, 24–33

28 M. E. M C CU LLY E T AL .

Table 1. Measured percentage embolisms at two places on the same To test this we made two freezes ,15 cm apart (pliers) on
root frozen with cryopliers. The upper segment was frozen first, dis- the same root, in rapid succession, the first freeze proximal
connecting the root below from the transpiration stream. The to the second. The first freeze stops the flow and relieves the
lower segment was frozen approximately 1 s later. The experiments tension in the sap distal to it. If freezing under tension
were done on hot, sunny days around 12.00 hours.
produces gas, the first freeze should have more gas bubbles
in it than the second. The roots were frozen between
% Embolisms 11.00 and 13.00 hours on hot sunny days.
In fact, the percentage of large vessels embolized in the
Sample Upper zone Lower zone
proximal segment of each root that was double-frozen was
the same or little different (Figs 3(a) and (b); Table 1) from
that in the distal segment of the same root.
1 66 83
2 67 67
In addition, we found that the percentage of embolized
3 25 25 vessels along a 24-cm length of a single root showed little
4 30 30 variation, and no systematic change with distance (N ¼ 11,
5 20 20
6 6 6
7 10 11
8 20 20
9 40 40

varies markedly between roots originating from different

regions of the stem, and also to some extent along individual
roots, so the degree of embolism is expressed here as percent
of vessels embolized.

Criticism A – Formation of gas spaces by freezing

The water in plants is saturated with dissolved air. Criticism
A suggests that the freezing process could force air out of
solution, forming large gas bubbles, which are mistaken for
embolisms. Freezing gas-saturated water can indeed produce
separation of gas from the ice (Lybeck, 1959; Echlin, 1992),
but the size of the bubbles depends on the rate of freezing, and
with rapid freezing they are not detected with our instrument.
Even with quite slow freezing the bubbles are small. Such
bubbles will not expand to form embolisms because they are
surrounded by ice and there is nowhere for the water to go.
The vessel will embolize only when it is thawed. That freezing
does not induce the embolisms in the xylem vessels is proved
by the sudden appearance of gas-containing vessels soon after
sunrise. During the night and early dawn, the frozen vessels
contained only frozen sap. As soon as the sun rose and the
plants began to transpire, a considerable proportion of
vessels contained gas (Figs 1(a) and (b)).
Fig. 3. Cryoplaned, transverse faces of pieces of a nodal root of a
transpiring maize plant. This root was frozen in situ in the field
Criticism B – Gas produced by freezing under tension with cryopliers, cryoplaned and observed in a CSEM. Some of
the large xylem vessels were liquid-filled, whereas others were
This is a refinement of Criticism A and seeks to explain the
gas-filled (embolized). Some of the embolized vessels contain frozen
sudden transition after dawn just mentioned. It was argued debris which accumulated during planing. The piece in (a) was
that while freezing static sap might not produce detectable frozen about 15 cm proximal to that in (b), which was frozen
gas bubbles, freezing sap under tension as it was drawn with a second pair of cryopliers about 1 s later. Viewed at 15 kV.
through vessels by evaporation would produce such bubbles. Bar ¼ 200 mm.

q 2000 The Royal Microscopical Society, Journal of Microscopy, 198, 24–33

 C RYOS E M O F X Y L E M 29

Table 2. Effect of freezing rate on heterogeneity of solution within

vessel lumens. Five EDX spectra were recorded from small rasters at
predetermined locations within each vessel (see Fig. 5a), and the
mean and SD for [K] and [Cl] concentrations calculated from
each set of five peaks. The heterogeneity of the solution was
assessed as SD/mean for each vessel, expressed as a percentage
for each element. n is the number of vessels from which composite
mean percentages have been calculated.

Heterogeneity
(SD/mean [%])

Freezing method K Cl n

Fig. 4. Percentage of embolized large vessels determined by cryo-

Deep freeze 63 63 10
microscopy of transverse faces at 2 cm intervals along 24 cm
LN2 plunge 31 37 6
lengths of two different maize roots (closed and open circles).
Pliers freeze 20 21 10
These root lengths were frozen with LN2 while still attached to
Ethane plunge 8 12 6
the plant.

see Fig. 4 for two typical roots with different percentages of

embolism). chloride concentrations within single vessel lumens as
determined from the spectra obtained from the five separate
small-raster samples of the lumen surface (Fig. 5a). Each
Criticism D – Alteration of concentration of sap solutes by
lumen had been filled with the 100 mM KCl standard
freezing
solution. Heterogeneity of a population may be measured by
To test the effect of freezing on solute concentrations in the the scatter of observations around the mean. For a Gaussian
vessels, we perfused the vessels with solutions of known KCl distribution the scatter can be expressed as the ratio of the
concentration and, after freezing by different methods, standard deviation to the mean value. The heterogeneity of
measured the [K] and [Cl] by EDX analysis. This allowed the concentrations determined within each lumen is
the testing of two properties of the frozen sap: the fine-scale indicated by the ratio of the standard deviations of the
heterogeneity produced by sequestering of solutes and individual values to the mean of these values (the lower
ice during freezing, and the accuracy of the estimate of the ratio the more homogeneous the lumen contents). The
concentration. greatest heterogeneity was found in the vessels of the roots
frozen at 193 K, and the least in those of the roots frozen in
ethane. Values within the lumens of roots plunge-frozen in
Heterogeneity of frozen vessel contents
LN2 or frozen with the cryopliers were markedly more
Appearance of frozen standard solution in xylem lumens. Light, homogeneous than from those frozen at 193 K. The variation
electron emissive lines of sequestered solutes were apparent of solute concentration in the vessels frozen with cryopliers
against less emissive, dark, frozen water in xylem lumens of was almost as low as with an ethane freeze, whereas that of
roots frozen by all of the methods used. There were however, plunge-frozen vessels was considerably higher (Table 2).
characteristic differences, with the thickest, most irregular
and widest-spaced solute lines in the roots frozen in a
Accuracy of EDX measurements of standard solution in
freezer at 193 K, and with these lines becoming increasingly
xylem lumens
thinner, more regular and more closely spaced in roots
frozen with LN2, cryopliers and liquid ethane, respectively Table 3 gives mean values for [K] and [Cl] calculated from
(Fig. 5(a)–(d)). The close-spaced, fine solute lines in the large-raster scans of vessel lumens filled with the standard
vessel contents after freezing with the pliers or with ethane solution. Values from the roots frozen at 193 K were
were uniformly distributed as illustrated in Figs 5(c) and (d). markedly below the concentration of the perfusion solution,
Figure 5(b) shows a typical vessel well-frozen in LN2, but whereas those from the ethane-frozen roots were essentially
some vessels in a plunge-frozen root could have less identical. Means for the cryopliers- and LN2-frozen roots
uniform, more widely spaced solute lines. were not significantly different from the known concentra-
Variance of small-raster measurements. Table 2 shows the tion of the control solution. Also there was no difference in
heterogeneity of the EDX determinations of potassium and the measured concentration between those roots frozen

q 2000 The Royal Microscopical Society, Journal of Microscopy, 198, 24–33

30 M. E. M C CU LLY E T AL .

Table 3. Reliability of EDX estimates of the concentration of stan-

dard solution (100 mM KCl) drawn through vessels in the roots of
maize, with different freezing methods. Each estimate was from a
spectrum collected from a raster filling as much as possible of the
vessel lumen, without including any cell wall. The measured con-
centrations are means 6 SD (n ¼ no. of vessels).

Estimated concentration (mM)

Freezing method K Cl n

Deep freeze 66 6 18 65 6 19 10
LN2 plunge 90 6 10 98 6 15 6
Pliers (cortex on) 91 6 15 95 6 19 26
(cortex off) 92 6 12 94 6 14 12
Ethane plunge 104 6 11 110 6 6 10

intact or with the cortical tissue removed to reduce the

distance from the vessels to the heat sink.

Discussion

Are the vessel embolisms freezing artefacts?

The most common question, Criticism A, is whether the gas
spaces form as a result of the freezing process. It is known
that small bubbles of dissolved gas will separate from liquids
as they freeze (Echlin, 1992). For example, with slow
freezing (to 265 K), gas dissolved in water in glass tubes of
30 mm diameter formed a string of bubbles (diameter not
given) in the centre of the tube, which could be distinguished
with an optical microscope (Lybeck, 1959). No gas bubbles of
any size were ever seen at the resolution of our microscope
(< 0.05 mm) in the numerous sap-filled vessels frozen before

Fig. 5. Preparations as in Fig. 3, but from roots frozen in different

ways after the vessels were filled with a 100 mM solution of KCl.
The pattern of sequestered solutes in the large vessels is characteris-
tic of the freezing method used: (a) In a freezer at 193 K; (b) plunged
into LN2; (c) with cryopliers; (d) plunged into liquid ethane. The
squares on (a) show the relative raster size and pattern used for
the small raster measurements. For this vessel [K] in the central
square was 103 mM, and for the other squares beginning in the
bottom one and moving clockwise was 70, 53, 9.5 and 35 mM,
respectively (mean 58 mM). A single measurement in which the
raster covered a maximum area of the lumen gave a value of
54 mM. [K] from similar small raster measurements from vessels
like those in (b)–(d) had much smaller variance, and measure-
ments from large rasters closely matched [K] of the perfusion solu-
tion (see Tables 2 and 3). All samples viewed at 15 kV. Bar ¼ (a)
30 mm; (b–d) 40 mm.

q 2000 The Royal Microscopical Society, Journal of Microscopy, 198, 24–33

 C RYOS E M O F X Y L E M 31

embolisms first appeared in the early morning (Fig. 1a), and small-raster measurements) decreases as speed of freezing
small bubbles were rare in unembolized vessels frozen at increases. But apart from the very slow freeze (freezer at
any time (earlier studies listed in the Introduction, and the 193 K), the estimated concentrations of K and Cl are not
present study). It is well known that xylem sap contains significantly different from the known concentration of these
dissolved gas (sap in grape vines is saturated with nitrogen ions in the loaded vessels. Thus, even our slowest routine
and more than half saturated with oxygen, which, if it all freezing method, immersion in LN2, gives a reliable estimate
came out of the sap upon freezing would occupy only 2% of of sap concentration. The pliers freeze produces more
the ice volume (Lybeck, 1959). Small gas bubbles, even if homogeneous frozen sap, and also allows precise selection
below the resolution of our preparations, would of course be of the regions to be frozen with minimum disturbance of the
expected to nucleate embolisms upon thawing, as is the case rest of the plant. Hinde et al. (1998) have shown that with
for bubbles forming in cells of Spirogyra (Fig. 6.21, McGrath, N2-slush frozen leaf pieces, the EDX analysis of vacuoles
1987) and in trees frozen in winter (Tyree & Sperry, 1989), of epidermal cells is reliable. The vessels in our roots are
but there is no opportunity for the sap to thaw in the both much larger than epidermal cells and further from
cryopreparations, because they are never warmer than the heat sink, but, although the vessels lie up to 0.6 mm
193 K (in the cryomicrotome). Thus, freezing itself has not deep in the tissue, their freezing rate was adequate for our
created the gas spaces and embolisms observed in the root purpose.
vessels. Earlier analyses of 40 mM (Enns et al., 1998) and 200 mM
The double-freeze experiment was done in response to the KCl (McCully et al., 1987) solution in xylem conduits also
concern that while freezing did not produce embolisms in showed a close match with EDX analysis of the bulk frozen
vessels at night (presumed to be under positive pressure in tissue. Even close to the detection limit for [K] in the frozen
herbaceous plants), freezing while the sap columns were tissues, comparison of solute concentrations measured in
under tension during the day might produce artefactual situ in frozen sap correlated closely with psychrometer
embolisms. The double-freeze results (Table 1; Fig. 3) show measurements of sap from the same roots (McCully, 1999).
that isolation of the root from the transpiration stream (by Thus, solute concentrations over a broad range can be
the first freeze) does not change the amount of embolism determined with confidence by EDX measurements of regions
detected by a subsequent freeze upstream from the blockage. of frozen plant tissue at distances of many micrometres from
This finding is similar to that of Pate & Canny (1999), who the surface of the tissue.
showed that a freeze block isolating the distal portion of a It is most probable that X-rays from surrounding vessel
grass-tree root from the transpiration stream did not walls, and possibly the surrounding parenchyma cells, were
decrease the volume of gas in the sap aspirated from xylem detected but their contribution to the total spectrum would
of the isolated root portion. Thus, embolisms observed after be negligible. This is supported by the agreement of the
our fast freezing are very unlikely to be artefactual, even measured values of KCl concentration with those of known
when roots are frozen when the sap is under tension. This standards. The volume from which X-rays are emitted in
conclusion is further strengthened by the finding of similar frozen-hydrated biological samples has been assessed by
amounts of vessel embolism along long pieces of individual modelling and experiment by Marshall (1982), Marshall &
roots frozen intact. Condron (1985) and Oates & Potts (1985). These investiga-
Earlier, Pate & Canny (1999) found close correspondence tions all indicate a depth and lateral resolution for X-ray
between the same daily course of embolism and repair in microanalysis in these samples of about 2 mm at 15 kV.
root vessels of the grass-tree as determined by cryomicro- Marshall & Xu (1998) have used an X-ray imaging
scopy, and the daily changes in the volume of air bubbles in technique to obtain quantitative analyses of frozen, planed
the sap aspirated from the vessels of parts of the same roots insect and plant tissue, avoiding the need for the specimen
which were not frozen. That study further validates the etching required for the selected area analysis used in our
freezing technique for determining the presence of embolisms studies.
in xylem vessels. Both techniques can produce accurate analyses of frozen
standard solutions (Figs 2 and 3 of Marshall & Xu, 1998;
Table 1 of Huang et al., 1994; Figs 2(a) and (b), present
Are EDX measurements of xylem sap reliable?
study). The X-ray imaging technique avoids any problems
Criticism D was that the concentration of solutes in the sap of non-uniform sublimation of ice from the sample and
was changed significantly during freezing. Our tests with also provides an unbiased sampling method (LeFurgey et al.,
roots where the vessels were loaded with solutions of known 1992). Although this technique avoids the need for experi-
concentration (Tables 2 and 3) show that this does not ence and great care in the etching process, it has the
happen. The freezing techniques in these tables are listed in disadvantage of the long times required to obtain the X-ray
order from the slowest to the fastest, and the heterogeneity images, thus making any study requiring multiple sampling
of the frozen solution (as judged by the declining variance of expensive for those who must pay for microscope time.

q 2000 The Royal Microscopical Society, Journal of Microscopy, 198, 24–33

32 M. E. M C CU LLY E T AL .

These encouraging results for lightly etched tissues apply vessel lumens also confirms the reliability of our etching
to the watery solutions of plant vessels and cell vacuoles, protocol.
where sublimation of the ice is fairly even. For tissues We have also found that it is important to cool the
containing much protein, the sublimation of ice from the specimen to 153 K before coating it. It is also important to
specimen may be uneven and may lead to greater variation standardize the thickness of the aluminium coat so as to
in the analytical measurements (Marshall, 1981). maintain the live times within the range of 120–150 s. No
analyses should be used if live times are outside this range.
The planing occasionally leaves behind small pieces of
Maximizing the reliability of the technique
debris on the surface which have been scraped from
To ensure that xylem embolisms are not induced by the elsewhere in the specimen. These must not be included
handling of the material prior to freezing, the organs should within the raster as they can markedly alter the value
remain intact and functioning on the plant. The cryopliers obtained. Similarly, any other irregularities on the surface
are essential to freeze such material efficiently and with should be avoided.
minimum disturbance to the rest of the plant; they are As shown in our results, it is important to adjust the
particularly useful for work in the field. The superior freezing magnification so that the raster covers as much area as
that can be obtained with liquid ethane is not a practical practical of the compartment being analysed. This is the
option for such studies because the organ must first be most efficient way to overcome the problem of heterogeneities
severed and cut into small pieces. Most of our studies have in the solute distribution within the space (Fig. 5a).
been with herbaceous roots no larger than about 1.2 mm As with all measurements, it is essential to analyse enough
diameter, but the comparative study of Pate & Canny (1999) samples to reduce the variance to an acceptable level. The
with grass-tree roots showed that reliable preservation of planing of the specimens makes it possible to measure
sap- or gas-filled vessels can be achieved with plier-freezing numbers of similar compartments in a single block face, and
of dense, woody roots as thick as 4 mm. the relatively fast aquisition (3–4 min) of spectra from each
The planing of the frozen specimens provides clear images selected area makes extensive sampling feasible.
of cell outlines and of liquid- and gas-filled spaces mostly A final word is appropriate to set this study briefly in the
unobtainable in fractured faces. It also provides a relatively context of current debates in plant water relations. Before
smooth surface essential for accurate EDX analysis (Hess, about 1939, it was well known that vessels of transpiring
1980; Echlin, 1992; Huang et al., 1994; Marshall & Xu, plants contained both gas and water (see, e.g. Haberlandt,
1998), as well as allowing analyses to be made from all 1914), but this old literature is no longer considered. The
regions of any specimen face, and from precise planes of the great advance revealed by our cryoSEM studies is not just
tissue impossible to achieve by fracturing (Huang et al., that these former workers were right, that embolisms are a
1994; McCully, 1994). It is important that the knives used normal daily occurrence in large diameter vessels, but also
for planing are sharp and set at the optimal angle for each that plants have evolved a mechanism for rapidly repairing
particular material, to avoid smearing and gouging of the embolisms and continuing transpiration. The mechanism of
block face. the repair process is a focus of hot debate (e.g. Canny, 1998;
Correct and standardized etching of standards and speci- Holbrook & Zwieniecki, 1999; McCully, 1999; Tyree et al.,
mens is crucial for accurate analyses. Although various 1999) and the cryoSEM technique is one of the best ways to
authors have noted that it is difficult to control the sub- investigate the process.
limation closely enough to ensure reproducible results, we
have not found this with our plant tissues. Samples and
standards are warmed to 183 K while they are continuously Acknowledgement
observed at 1 kV. The etching is stopped immediately and We thank the Natural Sciences and Engineering Research
the specimen recooled just beyond the time that the frost Council of Canada for support of this work from operating
disappears from the surface, when the faintest images of the grants to M.E.M. and M.J.C.
solid phase of the eutectic (in the standards), or cell walls
and/or the solid phase of the eutectic (in the tissues) appear.
This precise timing and the low temperature at which the References
etching is stopped are critical, but easily achieved with
Buchard, C., McCully, M. & Canny, M. (1999) Daily embolism and
practice. Table 1 of Huang et al. (1994) shows that there
refilling of root xylem vessels in three dicotyledonous plants.
was no difference between concentrations of standards Agronomie, 19, 97–106.
analysed after etching in this way, or without etching, while Canny, M.J. (1997a) Vessel contents of leaves after excision – a test
values for [K] and [Cl] doubled if the etching was at 193 K. of Scholander’s assumption. Am. J. Bot. 84, 1217–1222.
The agreement of measurements of the concentrations of Canny, M.J. (1997b) Vessel contents during transpiration –
these elements with those in the standard solution filling the embolisms and refilling. Am. J. Bot. 84, 1223–1230.

q 2000 The Royal Microscopical Society, Journal of Microscopy, 198, 24–33

 C RYOS E M O F X Y L E M 33

Canny, M.J. (1998) Applications of the compensating pressure Marshall, A.T. & Condron, R.J. (1985) X-ray microanalytical
theory of water transport. Am. J. Bot. 85, 897–909. resolution in frozen-hydrated biological bulk samples. J. Microsc.
Canny, M.J. & Huang, C.X. (1983) What is in the intercellular 104, 109–118.
spaces of roots? Evidence from the cryo-scanning electron Marshall, A.T. & Xu, W. (1998) Quantitative elemental X-ray
microscope. Physiol. Plantar. 87, 561–568. imaging of frozen-hydrated biological specimens. J. Microsc. 190,
Echlin, P. (1992) Low-Temperature Microscopy and Analysis. Plenum 305–316.
Press, New York. McCully, M.E. (1994) Accumulation of high levels of potassium in
Echlin, P. & Taylor, S.E. (1986) The preparation and X-ray the developing xylem elements in roots of soybean and some
microanalysis of bulk-frozen hydrated vacuolate plant tissue. J. other dicotyledons. Protoplasma, 183, 116–125.
Microsc. 141, 329–348. McCully, M.E. (1999) Root xylem embolisms and refilling. Relation
Enns, L.C., McCully, M.E. & Canny, M.J. (1998) Solute concentra- to water potentials of soil, roots, and leaves, and osmotic
tions in xylem sap along vessels of maize primary roots at high potentials of root xylem sap. Plant Physiol. 119, 1001–1008.
root pressure. J. Exp. Bot. 49, 1539–1544. McCully, M.E., Canny, M.J. & Van Steveninck, R.F.M. (1987)
Haberlandt, G. (1914) Physiological Plant Anatomy. Macmillan, Accumulation of potassium by differentiating metaxylem
London. elements of maize roots. Physiol. Plant. 69, 73–80.
Hess, F.D. (1980) Influence of specimen topography on micro- McCully, M.E., Huang, C.X. & Ling, L.E.C. (1998) Daily embolism
analysis. X-Ray Microanalysis in Biology (ed. by M. A. Hayat), and refilling of xylem vessels in the roots of field-grown maize.
pp. 241–262. University Park Press, Baltimore. New Phytol. 138, 327–342.
Hinde, P., Richardson, P., Koyro, H.-W. & Tomos, A.D. (1998) McGrath, J.J. (1987) Temperature-controlled cryogenic light
Quantitative X-ray microanalysis of solutes in individual plant microscopy – an introduction to cryomicroscpy. The Effects of
cells: a comparison of microdroplet and in situ frozen-hydrated Low Temperatures on Biological Systems (ed. by B. W. W. Grout and
data. J. Microsc. 191, 303–310. G. J. Morris), pp. 234–267. Edward Arnold, London.
Holbrook, N.M. & Zwieniecki, M.A. (1999) Embolism repair and Oates, K. & Potts, W.T.W. (1985) Electron beam penetration and
xylem tensions: do we need a miracle? Plant Physiol. 120, 7–10. X-ray excitation depth in ice. Micron Microsc. Acta, 16, 1–4.
Huang, C.X., Canny, M.J., Oates, K. & McCully, M.E. (1994) Planing Pate, J.S. & Canny, M.J. (1999) Quantification of vessel embolisms
frozen hydrated plant specimens for SEM observation and EDX by direct observations: a comparison of two methods. New
analysis. Microsc. Res. Techn. 28, 67–74. Phytol. 141, 33–44.
LeFurgey, A., Davilla, S.K., Kopf, D.A., Sommer, J.R. & Ingram, P. Shane, M.W. & McCully, M.E. (1999) Root xylem embolisms: impli-
(1992) Real-time quantitative elemental analysis and mapping: cations for water flow to the shoot in large, field-grown maize
microchemical imaging in cell physiology. J. Microsc. 165, 191– plants with only one root. Aust. J. Plant Physiol. 26, 107–114.
223. St. Aubin, G., Canny, M.J. & McCully, M.E. (1986) Living vessel
Lybeck, B.R. (1959) Winter freezing in relation to the rise of sap in elements in the late metaxylem of sheathed maize roots. Ann.
tall trees. Plant Physiol. 34, 482–486. Bot. 58, 577–588.
Marshall, A.T. (1981) Simultaneous use of EDS, windowless EDS, Taiz, L. & Zeiger, E. (1998) Plant Physiology. 2nd edn. Benjamin/
BE and SE detectors and digital real-time line-scanning for the Cummings Publishing Company, Redwood City, CA.
X-ray micro-analysis of frozen-hydrated biological specimens. Treeby, M.T., Van Steveninck, R.F.M. & de Vries, H.M. (1987)
Scanning Electron Microsc. 1981 (2), 327–343. Quantitative estimates of phosphorus concentrations within
Marshall, A.T. (1982) Application of f (rZ) curves and a window- Lupinus luteus leaflets by means of electron probe X-ray
less detector to the quantitative X-ray microanalysis of frozen- microanalysis. Plant Physiol. 85, 331–334.
hydrated bulk biological specimens. Scanning Electron Microsc. Tyree, M.T., Salleo, S., Nardini, A., Lo Gullo, M.A. & Mosca, R.
1982 (1), 243–260. (1999) Refilling of embolized vessels in young stems of laurel. Do
Marshall, A.T. (1987) Scanning electron microscopy and X-ray we need a new paradigm? Plant Physiol. 120, 11–21.
microanalysis of frozen-hydrated bulk samples. Cryotechniques in Tyree, M.T. & Sperry, J.S. (1989) Vulnerability of xylem to
Biological Electron Microscopy (ed. by R. A. Steinbrecht and cavitation and embolism. Ann. Rev. Plant Physiol. Plant Mol.
K. Zeirold), pp. 240–257. Springer Verlag, Berlin. Biol. 40, 19–38.

q 2000 The Royal Microscopical Society, Journal of Microscopy, 198, 24–33