Change in past environments – secrets of the tree hydrosystem


Forest fires or falling rocks will mark a tree – mechanical injury will remain permanently as a timestamp, precisely to the year or even season of the event. This is the simplest type of ecological information that can be yielded retrospectively, allowing reconstruction of forest fire frequency or the history of rock avalanches. Another easily visible feature is the width of the annual rings which can be observed on a tree stump and, maybe, even interpreted without complicated statistical methods or high-tech measuring devices. As early as the 16th century, Leonardo da Vinci had already recognized an association between tree-ring width and the prevailing weather. But the majority of environmental events and changes are documented in trees in a varied and complex manner. It is impossible, for example, for the naked eye to notice the annually changing wood density or the chemical composition of the wood. Recently, the hydrosystem of trees has been proposed as a further promising feature for retrospective ecological studies. The paper in this issue by Fonti & Garcia Gonzales (pp. 77–86) is proof of the evidence that can be gained from the conducting tissue for a narrow time slot dated to an exact calendar year, with fascinating implications for dendroecology and dendroclimatology.

Which features to examine?

The classic feature of dendroecology and dendroclimatology is the tree-ring width (Schweingruber, 1997; Box 1). It can be determined easily and largely nondestructively. As early as 1870 (i.e., only 20 yr after the basics of tree thickness growth were understood), tree-ring width was used as an ecological indicator of air pollution around smelting plants. It has proven its validity more recently in forest decline studies. The tree-ring width, however, integrates all the positive and negative influences which affected the growth of a tree during a year (or several previous years). Since the mid-1970s, it has been possible, routinely, to determine the annually changing wood density of the homogenously structured softwoods with the aid of X-ray densitometry. The maximum density of the latewood contains a distinct climate signal (e.g., the summer temperature at the boreal timberline). In a similar manner to tree-ring width, wood density also integrates several wood-anatomical features over time, such as cell wall thickness and lumen diameter, despite the higher temporal resolution.

Table 1. 
Box 1Tree-ring research – a primer
Dendroecology addresses past environmental change – it attempts to identify and date exogenous disturbing influences in the tree-ring series of trees dated to their exact year of formation and to interpret these in the form of indirect evidence as possible causes of changes.
The closely related goal of dendroclimatology is to identify the climate signal in a tree-ring series and to use it to reconstruct past climate changes from several hundred to several thousand year old tree-ring chronologies (Fritts, 1976).
The term ‘tree-ring chronology’ refers to a continuous time series which is created by a tree based on its growth (thickness and height) formed in annual spurts. During this process, environmental information enters the tree and is stored there permanently – as long as the tree lives or is preserved in the ground after its death, or as long as the building timber gained from it still exists. With the aid of dendroecology, this information can be ‘read’, and thus the past environment can be reconstructed.
The information regarding environmental events and changes is documented in the tree, at its simplest, by mechanical injury. Environmental events can leave also traces in a physiological manner, such as in the form of a ‘frost ring’ caused by late frost in spring or early frost in autumn. By comparison, those environmental effects which are actually perceived by the tree via metabolic-physiological processes but are ‘encoded’ in the form of width, density, structure and chemical composition of the annually formed wood are harder to identify.

For hardwoods with a more complex cellular structure, resolution of the intensively integrating tree-ring width was first limited to separation into early- and latewood width of the ring-porous oak tree. Only with the development of semiautomatic image analysis did wood structure measurements in hardwoods become possible (Eckstein et al., 1977), and sufficiently long test series could only be set up in sufficient numbers after the arrival, in the 1990s, of fully automatic image analysis technology (Woodcock, 1989). The size of the vessels (i.e., the water-conducting cells), which varies from year to year and within a tree ring, is used as a feature (Fig. 1).

Figure 1.

Interacting external and internal influences on vessel formation and enlarged, three-dimensional, ring-porous structure of oak wood.

The radial diameters of the water-conducting cells (tracheids) of softwoods have been compared, such as by Vaganov (1990), with simultaneous weather and ground-water conditions. This feature does not generally react to the climate during a vegetation period but rather to narrowly time-limited stress conditions (disturbing influences). Thus it is possible to interpret a feature at the cellular level retrospectively as an ecophysiological characteristic. A summary of the relationship between wood anatomy and the environment is provided by Wimmer (2002); newer approaches are introduced by Deslauriers et al. (2003).

To complete the suite of useful dendroecological features, it should be noted that environmental information is also stored biochemically in the annual rings of trees. Information carriers are the hydrogen, oxygen or carbon stable isotopes which are present in rainfall and in the atmosphere, and whose proportions fluctuate depending on the ambient climate. These isotopes enter the trees via the roots with water uptake, or via the leaves through photosynthetic activity, and are permanently fixed in the cell walls of the wood in the form of cellulose and lignin components in their exact year of formation. These can be analyzed by mass spectrometry (Schleser et al., 1999).

The utility of the hydrosystem

Fonti & Garcia Gonzales (2004) systematically compared three tree-ring width variables and four earlywood vessel features with regard to the type and strength of their climate signals. The variables were taken as correctly dated time series from, in total, 51 contemporaneously living chestnut trees on the southern side of the Swiss Alps. The common interval for all trees covered the 40-yr period from 1956 to 1995. All variables considered, except the earlywood vessel area, reflected the same signal and hence the same type of information. The signal recorded by the earlywood vessel area was air temperature, preferably during the onset of cambial activity (i.e., in early spring) but to a lesser amount also during the previous autumn. It means that an above-average warm previous autumn as well as early spring resulted in smaller earlywood vessels, and vice versa. Although these results initially describe only the statistical relationship between two sets of variables, viz. earlywood vessel size and temperature, the authors offered good reasons for a real cause/effect-relationship. This proof of biological plausibility supports the usefulness of the hydrosystem of hardwoods as an eco-physiological and, possibly, also climatic archive.

The hydrosystem (in this case of hardwoods) is a three-dimensional structure of tubular cells joined in a network. In ring-porous tree species, such as oak or chestnut, the hydrosystem is fully functional only during the year of its formation; in diffused-porous tree species, such as beech or linden, it is fully functional for several decades (Zimmermann, 1983). The axial dimension of the conducting tissue (i.e., the vessel length) is not measured because of the preparatory effort required – rather, data are obtained from cross sections as vessel area for the sake of simplicity.

The vessel area of hardwoods was considered as a time-stable feature for a long time and is therefore used primarily for the anatomical identification and differentiation of wood species. From evolutionary standpoints, however, it is also treated as a variable feature that adapts to varying ecological conditions in the long term (Carlquist, 1988). The vessel area as a quickly varying feature was initially noticed only in climatically extreme years (Knigge & Schulz, 1961), before it was determined as a dated time series in the following years and tentatively interpreted dendroecologically (Eckstein et al., 1977). But it took 15 more years until value recording was developed sufficiently to enable a statistically sufficient number of samples to be taken at an acceptable time and cost expenditure. Some examples of this research can be found in the publications of Woodcock (1989), Sass & Eckstein (1995), Pumijumnong & Park (1999), and Garcia Gonzales & Eckstein (2003).

It is amazing that a cutting of only approx. 2–10 mm2 with 50–800 vessels (depending on vessel size) reliably represents the growth of an entire year. It also turned out to be advantageous that the vessel area time series do not contain a pronounced age trend. Furthermore, the vessel areas depend less on the living conditions of the previous year or previous years than the tree-ring width, which is usually burdened with substantial first and higher autocorrelation. Another important finding is that the vessel areas store an environmental signal different from that in tree-ring width and density.


The encouraging initial success in the retrospective interpretation of the hydrosystem of trees invites further research in different directions. One such direction may be the reconstruction of the palaeoclimate of a specific region, such as the palaeomonsoon in south-east Asia. For climatologists the Asian palaeomonsoon is a key phenomenon whose dynamics remain little known. However, south-east Asia suffers from the limited length of instrumental weather records. Therefore, trees (among other sources of information) can be taken into consideration as biological recorders of climate with a 1-yr time resolution. One promising tropical tree species suitable for the purpose of reconstructing the Asian palaeomonsoon is teak, with an area of natural distribution which includes India, Myanmar, North-Thailand and Laos. However, since we know that climate may have changed differently in different seasons of the year, a time resolution better than 1 yr is desirable. At this point, the vessel size of teak can become involved (Pumijumnong & Park, 1999; A. Bhattacharyya, pers. comm.) in addition to tree-ring width and stable isotopes.

In such seasonally dry tropical forests where rainfall is the primary climatic driving force for phenology, cambial activity and cell expansion, there are further fields in which vessel size could be applied as an eco-physiological characteristic. An interesting aspect for tropical silviculture would be to compare deciduous trees, semideciduous trees and deciduous stem succulents in order to better understand the relationships between climate, site morphology, phenology, water status, cambial activity and finally the periodicity of growth of tropical trees (Borchert, 1999).

The vessel area of European oak – together with tree-ring width and stable isotopes – can also contribute to reconstruction of the climate in Central Europe over hundreds if not thousands of years. This would probably be the most tedious yet most worthwhile approach. A continuous tree-ring width chronology of oak for the entire Holocene is already available (Friedrich et al., 1999). However, the climatic signal of the tree-ring width of European oak is rather weak: the climatic conditions for oak in Central Europe only rarely limit growth; the main factors, either moisture or warmth, do not continuously act as limiting over time, but interchange with each other; and the tree-ring width integrates both the favourable and unfavourable effects on growth over one or more years. By contrast, vessel size is not an integrator. The ‘window’ through which the environmental influences ‘enter’ a tree is open only for 3–4 wk during the ontogeny of, say, an earlywood vessel until its secondary wall layer has been deposited and lignified. Therefore, the signal ‘archived’ by the vessel size should be unequivocal.

Many basic details have to be clarified before synoptic observation will be possible. In addition, further efforts to achieve fast quantitative assessment of wood anatomical structures of hardwoods are necessary (Vansteenkiste, 2002).