The presently accepted theory explaining the ascent of sap inside a tree stem is the cohesion-tension theory, which is mainly based on four points: the transpirational pull, the xylem forming an interconnected network of conduits, the attractive interactions (hydrogen bonds) between water molecules, and the fact that the water is confined to relatively narrow conduits, remaining in a liquid state even under considerable tensions up to -10 MPa .
An important factor limiting the hydraulic conductivity is that water under such high tensions is in a metastable state and vulnerable to cavitation, i.e. formation of gas bubbles within the water column, which will rapidly expand and fill the whole conduit, forming an embolism that prevents the transpirational pull from being transmitted into the tissue below. Drought conditions as well as freeze-thaw cycles experienced by temperate and boreal species pose an increased threat of cavitation for trees .
Trees also have means of restoring the hydraulic conductivity lost by cavitation, either by producing new xylem to replace embolized conduits, or by refilling gas-filled cells with water. Under low evaporative demand, many species are able to generate root or stem pressures well above the atmospheric pressure throughout their vascular system . These pressures are thought to be responsible for embolism repair occurring in the spring before leaf flush, and during night time in the growing season [3, 4]. Refilling during transpiration is a more complex phenomenon, since it implies the existence of a significant pressure difference between the refilling conduit and the surrounding xylem. This local refilling is not very well understood, but most hypotheses attribute it to hydraulic isolation of the refilling conduit due to the geometry of the pits connecting the xylem elements, and to active secretion of solutes into the embolized conduit by adjacent still-living cells [5, 6]. In a recent review, the source of the refilling water was proposed to be the phloem, rather than the water-conducting xylem around the embolized conduits .
Until recently, a key limitation in the study of embolism development and repair has been the lack of methods to directly observe cavitation and refilling events in living plants. Traditional methods for quantifying xylem embolism include studying the cut surface of an excised stem or branch segment or leaf petiole under a microscope [8, 9], or measuring its percentage loss of hydraulic conductivity (PLC, [6, 10]). Also acoustic emissions have been used to detect cavitation events in vivo [11, 12]. The former two are destructive methods, and therefore only give the water status of the sample at one specific point in time. With these methods, embolism development and repair can only be observed indirectly, by sampling a group of similar plants undergoing the same environmental changes. Ultrasound observations are a direct method for detecting cavitation as it occurs, but we are unaware of any studies using acoustic emissions to detect refilling, which is presumably a much slower process. Moreover, ultrasound and PLC measurements, respectively, only give information on the number of embolized conduits and amount of conductivity lost due to embolism; the spatial distribution of embolized conduits with in the stem is not accessible with these methods.
The analysis of the effects of water stress on trees and verification of refilling theories would benefit from information on what are the water contents of individual cells. Optical or scanning electron microscopy (SEM) of vitrified tissues (e.g. ) could provide the needed spatial information, but both methods are relatively laborious. These methods also involve either freezing the sample rapidly to vitrify the cells’ water contents, or perfusing the excised segment with a staining agent to see which cells are conducting. Combined with cutting the samples, this poses a risk of artificially inducing embolisms during sample preparation (see e.g. [14–16]).
Magnetic resonance imaging (MRI) is one tool capable of non-destructively visualizing the stem water contents, and has already been successfully used to observe embolization in woody species such as grapevines and lianas [17–19]. The disadvantage of MRI is the relatively poor spatial resolution (in-slice resolution of 20-150 μm / pixel), which is adequate to resolve large embolized vessels, but not the surrounding tissue.
Recently, synchrotron-based x-ray imaging has been applied to observing xylem embolism and refilling. Lee and Kim  used phase-contrast micro-imaging to observe the refilling of vessels in bamboo, and Brodersen et al.  demonstrated the use of synchrotron-based high resolution x-ray microtomography (XMT, also known as μCT) in visualizing embolism refilling in grapevine stems. With a resolution of 4.4 μm per voxel, Brodersen and coworkers were able to visualize the growth of water droplets inside refilling vessels. Their results appear to confirm both the role of ray parenchyma cells in refilling and the importance of hydraulic isolation of the refilling vessel.
X-ray microtomography, however, is not limited to synchrotron sources: so called ‘desktop’ XMT systems, capable of sub-micron voxel resolution, are becoming increasingly popular. Although a synchrotron offers better temporal resolution and superior beam intensity, desktop scanners are far more easily available to most researchers, and require fairly little user experience to operate efficiently. A home lab –based system also allows conducting long-term studies that could not be carried out within a beam time allocation of a few days. Such systems have already demonstrated their capabilities in visualizing and quantifying xylem anatomy [22, 23], but so far there have been few attempts to image the xylem of live trees using these systems. In a larger scale, such a system has already been applied to imaging the development of maize seeds , and a similar XMT system to the one used in this study was recently used to observe the growth of Arabidopsis thaliana. However, the resolution used with live samples in these studies was not yet sufficient to determine the water contents of individual cells.
In this work, we have investigated the use of a desktop XMT setup in visualizing and quantifying embolism propagation and repair in live saplings and shoot tips of Silver and Curly birch (Betula pendula and B. pendula var. carelica). A significant part of the effort was in devising a suitable sample holder to avoid sample movement artefacts with a live sample, as well as finding suitable scan parameters to minimize radiation exposure to the sample while maintaining a sufficient signal-to-noise ratio to resolve individual xylem cells. The results show that x-ray tube –based XMT equipment can be used to follow the propagation of drought- or freezing-induced embolism and embolism refilling at the cellular level in birch saplings. As simple quantitative metrics of the degree of embolization within the stem, we propose to calculate two values from the cross-sectional tomography images: the number of embolized vessel cells, and the percentage of stem cross-section that consists of embolized xylem. The latter parameter is abbreviated PCS, or percentage of cavitated stem.