Variability in xylem anatomy is of interest to plant scientists because of the role water transport plays in plant performance and survival. Insights into plant adjustments to changing environmental conditions have mainly been obtained through structural and functional comparative studies between taxa or within taxa on contrasting sites or along environmental gradients. Yet, a gap exists regarding the study of hydraulic adjustments in response to environmental changes over the lifetimes of plants. In trees, dated tree-ring series are often exploited to reconstruct dynamics in ecological conditions, and recent work in which wood-anatomical variables have been used in dendrochronology has produced promising results. Environmental signals identified in water-conducting cells carry novel information reflecting changes in regional conditions and are mostly related to short, sub-annual intervals. Although the idea of investigating environmental signals through wood anatomical time series goes back to the 1960s, it is only recently that low-cost computerized image-analysis systems have enabled increased scientific output in this field. We believe that the study of tree-ring anatomy is emerging as a promising approach in tree biology and climate change research, particularly if complemented by physiological and ecological studies. This contribution presents the rationale, the potential, and the methodological challenges of this innovative approach.
Long records of environmental conditions are essential for evaluating scenarios of climate change and the consequences for species and plant performance. As instrumental records are not long enough to provide a complete picture of dynamics in past climates, they need to be supplemented by proxy records. Trees, as long-living organisms, record ecologically relevant information in their annual rings and hence represent important natural archives for the study of global changes throughout the last millennium (Esper et al., 2002; Cook et al., 2004; Treydte et al., 2006; Trouet et al., 2009). Tree-ring variables such as ring width or maximum latewood density have been shown to be strongly influenced by environmental conditions, especially where temperature or precipitation limits tree growth. They therefore play a prominent role in the study and reconstruction of climate variation (IPCC, 2007, Jones et al., 2009).
There are, however, other less widely studied characteristics of wood, for example its anatomical structure, which can encode additional and novel ecological information. Variation in wood-anatomical characteristics represents adaptive structural solutions adopted by the tree in order to achieve an optimal balance among the competing needs of support, storage and transport under changing environmental conditions and phylogenetic constraints (Chave et al., 2009). Consequently, studies of variations in xylem anatomy have already been an important source of information in plant sciences (Larson, 1994; Gartner, 1995). Until recently, wood anatomists have advanced the understanding of phylogenetic adaptations in plants by analysing and interpreting variation of wood structures across taxa and climatic zones (e.g. Carlquist, 1988; Wheeler & Baas, 1993; Wiemann et al., 1998). Intraspecific variation across climatic zones, along environmental gradients, or between contrasting sites supplied additional information about the linkage between ecology (habitat) and functioning (derived from xylem anatomy) (e.g. Carlquist, 1975; Baas, 1986; Villar-Salvador et al., 1997; Wheeler et al., 2007). There is, however, another source of variation, that is, the wood-anatomical variability along tree-ring sequences – which is the focus of this review – which has been less widely studied by wood anatomists, and which we believe can be used to elucidate how individual trees and species respond to changing environmental conditions (Schweingruber, 1996, 2006). The ability of a genotype to adjust the phenotype over the life of a tree is a result of short-term to long-term physiological responses to environmental variability and can be used to link environment with xylem structure.
Tree-ring anatomy is a methodological approach based on dendrochronology and quantitative wood anatomy to assess cell anatomical characteristics (such as conduit size and density, cell wall thickness and tissue percentage) along series of dated tree-rings and to analyse them through time (at the intra- and/or inter-annual level) in order to characterize the relationships between tree growth and various environmental factors. This approach supplements tree-ring based reconstructions of past environmental conditions with novel understanding about the range and strategies of species’ responses and their chances of success, and thus contributes to the evaluation of the impact of predicted climate change on future vegetation dynamics.
Because of the importance of water in all physiological processes, its availability and the efficiency and safety of its transport are often the factors most limiting plant growth (Tyree & Zimmerman, 2002; Lambers et al., 2008). Considering that > 90% of the water taken up by plants is lost by transpiration through the leaf, while CO2 is absorbed at the same time, the importance of water becomes apparent (Kramer & Boyer, 1995). Consequently, to evolve into tall and self-supporting land plants, trees had to develop the ability to easily access and economically transport water and to regulate water loss through their leaves (Koch et al., 2004). Long-distance water transport in trees occurs passively through the lumina of nonliving conductive cells in the xylem (Carlquist, 1975) and is transferred between conduits through bordered pits, that is, through openings in the cell walls regulated by a pit membrane. In conifers, water flows from tracheid to tracheid through bordered pits. In angiosperm trees, water is transported through longitudinally connected vessel elements that form pipes up to several metres in length. Vessel elements are longitudinally connected by dissolved end walls (perforation plates) and adjacent vessels are laterally connected by pits in the longitudinal cell walls to form a vessel network.
The major force for water transport in the conducting xylem is generated by transpiration of water from the leaves, which creates a negative vapour pressure in the cells surrounding the stomata. This causes a negative hydrostatic pressure in the conducting cells that literally pulls the water through the continuous network of conduits. As a result of the cohesive forces among water molecules, this suction force is transmitted downwards into the root system, where water is taken up via the root hairs along the fine roots (see the cohesion-tension theory of the ascent of sap in vascular plants; Dixon & Joly, 1895; Tyree & Zimmerman, 2002).
However, the need to supply water to the canopy at a high rate has to be balanced against mechanical stability while minimizing the risk of xylem dysfunction by cavitation (Hacke & Sperry, 2001; Sperry, 2003). This places an important constraint on the architecture of stems and represents an important trade-off in plant function (Baas et al., 2004).
At the conduit level, according to the Hagen–Poiseuille law, water conductivity approximately corresponds to the fourth power of the conduit diameter. However, maximum gain in transport efficiency can only be realized if the end wall conductivity increases in concert with diameter. On its way through the tracheid network, water travels not only through the lumina, but also through the bordered pits connecting adjacent cells. Physiological studies have demonstrated that pit membranes are responsible for at least 50% of the hydraulic resistance in the xylem (Hacke et al., 2006). Changes in the thickness and porosity of the pit membranes therefore have the potential to exert significant influences on the total hydraulic resistance in the plant. The longer and wider the conduit and the thinner and more porous the pit membrane, the lower is its resistance to water flow. Consequently, hydraulic conductivity can be considerably increased by slightly increasing the cross-sectional lumen area of the conduits and bordered pits, but increased conduit diameter greatly decreases the safety of water transport against cavitations (Tyree & Zimmerman, 2002). Cavitations are caused by nucleation forming air emboli in conduits that interrupt upwards water movement when the conduits come under high tension. Vulnerability to cavitation is increased by greater conduit size (see reviews by Hacke & Sperry, 2001; Cochard, 2006) and by weak pit structures (Jansen et al., 2003). Drought-induced cavitations propagate by air seeding at interconduit pit membranes (Hacke & Sperry, 2001). Pit morphology may differ widely between tree species; the correlation between pit membrane size and conduit diameter in different taxa has been found to be weak, but differences in pit structure and total area of pits per conduit seem to strongly influence embolism resistance (Wheeler et al., 2005; Hacke & Jansen, 2009). Within a single stem, however, conduit diameter correlates with vulnerability to drought-induced cavitation, as wider conduits have a greater surface area of pit membranes and therefore a higher probability of having a large pit membrane pore (Gartner, 1995). By contrast, frost-induced cavitations occur when xylem sap freezes and dissolved gases create air bubbles in the wider conduits. Wider conduits trap larger bubbles in the ice, which are more likely to trigger cavitation during thawing (e.g. Lemoine et al., 1999; Field & Brodribb, 2001). This risk appears to be dependent also on the sugar content of the sap, the minimum temperature experienced, and the rate and number of freeze–thaw cycles (Mayr et al., 2007). In some cases it was observed that cavitations could be actively removed. This process of water refilling under negative pressure is not fully understood, but appears to involve living cells and to require energy (Cochard et al., 2001; Holbrook et al., 2001; Salleo et al., 2004).
Ecological relevance of xylem hydraulic architecture
The characteristics of xylem hydraulic architecture, such as the arrangement, frequency, length, diameter, wall thickness and pit characteristics of conduits, not only regulate the efficiency of water transport but also affect the margins of safety against hydraulic system failures (Comstock & Sperry, 2000; Hacke et al., 2001, 2006; Pittermann et al., 2006; Sperry et al., 2006; Choat et al., 2008). Inter- and intraspecific differences in xylem hydraulic architecture reflect not only size- or age-related trends but also differences in the way trees adapt or adjust to environmental variability, and can provide information about the plasticity of a species under changing environmental conditions. A more direct approach is to assess temporal plasticity in xylem hydraulic architecture in a tree-ring sequence of a single tree. As within the same tree and species resistance to cavitation is related to conduit diameter, the risk of system failure is higher in tree rings where a large amount of the total hydraulic conductivity is contributed by a few wide conduits. This holds especially true for ring-porous species where water transport is assumed to take place in the outermost tree ring only. Figure 1 shows an example of how the risk of a system failure can vary along the annual rings of one individual. In this case, because of higher cavitation risk under similar stress conditions, at least 50% loss of conductivity are more likely to occur in years such as 1988 than as 2002 (see Fig. 1).
The developmental success and the competitiveness of trees depend on their ability to adjust and optimize their hydraulic architecture to their specific environment. Major hydraulically relevant properties, such as ring-porous or diffuse-porous xylem structure, leaf stomatal behaviour or the kind of root system, generally define the range of a species’ tolerance and competitiveness and thus the ecological setting to which a species is adapted. However, the ecological amplitude and thus the species distribution within given ecological settings may be partly limited by the species’ plasticity in relevant traits in response to the environmental variability, not only in spatial terms, but also over the lifetime of a tree (Sultan, 2000; Valladares et al., 2007). Moving outside these ranges can have detrimental consequences for the plant.
However, a gap exists regarding the study of dynamic hydraulic adjustments through the lifetimes of individuals or groups of trees. Coping with temporal environmental variability is the most critical challenge for the survival of an individual tree. Because trees undergo a continuous process of ontogenetic adjustments to respond to stress situations caused by a changing environment, and changing size and age, valuable ecological information can be extracted from the temporal reconstruction of these responses.
Environmental imprinting in wood cell anatomy
Meristems generate new functional structures during the entire life-span of an organism. Secondary growth of the woody stem in particular is a dynamic process and is influenced in a complex way by whole-tree physiology, which in turn is controlled by environmental conditions. The effect of factors that strongly influence secondary growth are permanently registered within the anatomical characteristics and reflected in the tree-ring structure. During wood formation, xylem cells differentiate through a complex process encompassing cell-type determination, cell division, cell differentiation and programmed cell death (see reviews in Fukuda, 1996; Plomion et al., 2001; Scarpella & Meijer, 2004). These processes are genetically controlled and depend on the ontogenetic status of the tree, but are also influenced, directly and indirectly, by environmental conditions (Denne & Dodd, 1981).
On the one hand, an environmental event such as a frost can directly influence cells undergoing differentiation and thus leave an imprint of weakly lignified and crumpled conduits inside a band of dead cell tissue in the tree ring (Glerum & Farrar, 1966). Analogously, spring conditions occurring at the time of early wood vessel formation determine cell size by influencing the rate of cell division and differentiation, as observed for some ring-porous species (García-González & Eckstein, 2003;Fonti & García-González, 2004, 2008). In these cases, the susceptible period of xylem formation to directly perceive and encode environmental signals is the time window during which cells are developing. As the periods of division, expansion and maturation of xylem cells range from several days to a few weeks (e.g. Rossi et al., 2006), concurrent weather conditions are likely to directly leave imprints of their occurrence in the ring structure.
On the other hand, prevailing environmental conditions such as persistent drought periods can also indirectly induce adjustments in the wood structure through tree physiological modifications to adapt to the new environmental demands. Cambial activity and wood cell development are strongly dependent on the availability of photoassimilates. In this case, the photosynthesis rate is reduced and assimilate translocation is adjusted, which ultimately influences cambial activity and xylogenesis, even in subsequent seasons, as observed for Quercus pubescens and Pinus sylvestris growing under contrasting water supplies (Eilmann et al., 2009). The resulting wood-anatomical modifications can greatly differ depending on tree metabolism and species-specific wood structure, but also depending on the timing of the season when the environmental event occurs. A drought event early in the growth season can induce different wood-anatomical modifications from a drought event at the end of the summer, when trees might merely respond by ceasing wood formation early (Arend & Fromm, 2007).
Through the means of wood formation, trees are thus able to perceive directly and indirectly environmental changes which leave permanent environmental imprints on xylem cells and wood structures, representing a valuable archive for environmental scientists.
Principles and challenges for decoding cell-based information
Reconstruction of past environmental conditions using the variability of datable tree-ring structures is an important area in dendrochronology. The study of the variation of cell-anatomical characteristics across series of annual rings started in the 1960s and 1970s (Knigge & Schulz, 1961; Eckstein et al., 1974) but has intensified in the last two decades as a result of improvements in digital image analysis. Formerly, measurements were made visually on microscope slides, with attendant constraints in terms of the objectivity of quantification and the sample size that could be used. At present, if the cells are large enough, for example in the early wood vessels of ring-porous species, digital images can be directly captured from the wood surface, allowing a more efficient survey to be performed (e.g. Munro et al., 1996; Fonti et al., 2009a; Fig. 2). In these cases, specific surface preparation techniques are required (e.g. Spiecker et al., 2000). In general, cutting is preferred to sanding as it keeps cell walls clean and cell lumina open. Another necessity for these procedures is to obtain a high contrast between target objects and background. This contrast can be enhanced by darkening the wood surface with ink or a stain and subsequently filling the cell lumina with a bright substance such as white chalk, plasticine or wax. Continuous progress in the development of image analysis systems involving powerful digital cameras, scanners and sophisticated software, as well as new techniques for wood surface preparation using specific microtomes (Gärtner & Nievergelt, in press), suggests that in the future it will probably be possible to examine smaller cells, such as the vessels of diffuse-porous wood, tracheids, fibres and parenchyma cells, and even subcellular features such as bordered pits.
The extraction of information from series of wood- anatomical characteristics of xylem cells has been based on well-established dendrochronological principles, such as the existence of similar environmentally driven responses in individuals growing under similar environmental conditions. This assumes the existence of common variability in the time series of different individuals (common signal), caused by the influence of a given environmental factor (the signal). Moreover, the processes linking current environmental conditions with responses must have been the same as those operating in the past (James Hutton’s principle of uniformitarianism; Britannica Concise Encyclopædia, 2009). In order to extract this information, a widely accepted set of specific sampling principles (selection of sites, species and trees) and methodological procedures (definition of tree-ring variables, cross-dating, replication, standardization for noise reduction and detrending of ageing trend) has been established for which only variables such as ring width and maximum latewood density were initially considered (Cook & Kairiukstis, 1990;Fritts, 2001).
The major differences between these traditional and wood-anatomical variables are the scale (moving from mm to μm), the larger number of observations for each ring, and the higher temporal (intra-annual) resolution of the measurements of the wood-anatomical variables. While ring-width-based dendrochronology usually extracts one value per ring, integrating radial growth throughout the growing season, measurements of wood-anatomical variables yield much more data from different parts of the tree ring which are highly variable along both the radial (time) and tangential (spatial) positions within a tree ring (see images in Fig. 1). From these data, meaningful wood-anatomical variables (mean values, density values, and tissue proportions) have to be calculated for each tree ring to build annual time series. As a consequence of the changing environmental conditions throughout the year and especially during the growing period, radial files of consecutive cells produced at different times during the year encode seasonal information. But even cells formed at the same time must be measured in sufficient numbers to account for tangential variability in the xylem. If too few cells are considered, or cells encoding different environmental information at different times are mixed, the ecological information can be obscured or reduced. A higher time resolution of the climate signal can often be achieved by using features of subgroups of cells that are formed at the same time (García-González & Fonti, 2006). In these cases the signal encoded can reflect climatic conditions that prevail for short periods of from 1 to 2 months.
Studies on ring-porous early wood vessels have shown that all vessels along a 12-mm-thick tangential band have to be measured to stabilize the extractable environmental signal (García-González & Fonti, 2008). Moreover, the environmental signal can be maximized, reduced, or even absent depending on the criteria applied to select different vessel-area categories (Fig. 1) or vessel positions (e.g. early wood vessels of the first row) within the rings.
In conifers, specific standardization procedures (normalized tracheidogram; Vaganov, 1990; or Gompertz function; Rossi et al., 2003) have been developed to transform the absolute radial position of a radial row of consecutive tracheids across a tree ring to a relative position, and thus allow a comparison among tree rings. In these cases, at least five radial files of tracheids have to be measured to obtain reliable data on the variability of cell sizes across tree rings. Methods to monitor cambial dynamics, such as repeated pinning (e.g. Dünisch et al., 2002; Seo et al., 2007) or micro-coring (Deslauriers et al., 2003; Rossi et al., 2006; van der Werf et al., 2007), permit the determination of seasonal growth patterns that allow each cell in a radial file of tracheids to be assigned to the time of the season at which it was formed.
Time series of wood-anatomical variables and their environmental signals
However, time series analysis with wood-anatomical variables has primarily been used to explore the potential to obtain high-resolution proxies (1) by identifying which environmental factor mainly influences wood-anatomical variability in a certain species and environmental setting, (2) by defining when in the season the signal is registered, and – to a lesser extent – (3) to determine the physiological mechanisms that cause the variability in wood anatomy. However, wood-anatomical variables have rarely been applied to infer functional adjustments of xylem hydraulic architecture to temporally changing conditions (e.g. Sterck et al., 2008). Year-to-year analyses will permit the establishment of a link between climatic conditions and the anatomical characteristics of the forming wood. The attribution of these results to specific physiological responses and elucidation of the functional costs and benefits of the adjustment (see example in Fig. 1) would contribute to a better understanding of the plasticity in xylem hydraulic architecture and the different strategies adopted by trees when they are exposed to changing environmental conditions.
Conclusion and perspectives
Tree-ring anatomy provides a valuable opportunity to add a time component to the study of plant responses to changing environments. As a consequence of the direct relationship between cell structure (e.g. vessel area and vessel density) and function and their short period of formation, water-conducting cells can record and permanently encode environmental information with a high temporal resolution. With respect to traditional tree-ring variables, chronologies of wood-anatomical variables can thus provide novel information that is not necessarily limited to trees growing under harsh conditions in marginal habitats. Decoding this information, which is strongly related to the characteristics and the position of the cells within the annual ring, requires specific methodological approaches, including the survey of promising wood-anatomical variables, appropriate preparation techniques, and sophisticated statistical tools to build chronologies and to analyse the relation with environmental factors.
Although this multidisciplinary approach is still at an early stage of development and in some cases involves tedious measuring work, it deserves to be further developed as it has the potential to provide new information in global change research. First, relevant relationships between the physical environment and the physiological response of trees can be recognized and analysed retrospectively, as this information is permanently registered within the wood structure. Secondly, the high time resolution of the environmental influence on wood anatomy can be valuable to identify how and when growth processes are sensitive to the environment and therefore might contribute to disentangling the processes that control tree growth. This is important for understanding both physiological mechanisms and the functional meaning of growth responses. This is crucial for evaluating the range of plasticity and the capacity for resilience of trees growing under certain environmental conditions and ultimately to predict plant responses under future climatic scenarios.
For broader application of this approach in global change research, a concerted effort involving diverse disciplines (functional ecology, wood anatomy, plant physiology and dendrochronology) is required to address some methodological and conceptual issues. Methodologically, there is a need for (1) accurate and efficient measuring along series of rings to increase sample size, (2) expansion of the range of possible wood-anatomical variables to be measured (e.g. cell grouping, pit structure and degree of lignification), (3) understanding of how physiological processes and ageing modify wood formation, (4) improvement of the procedures to identify and enhance environmental signals in different frequency domains, and (5) evaluation of the synergistic effect of combining more tree-ring related proxies. In parallel there is a need for a better understanding of the processes that regulate the hydraulic responses across species, space and time and their functional meaning.
We thank three anonymous reviewers for valuable feedback on and improvements to an earlier draft of this article.