To reach a mechanistic understanding of radial growth depends essentially on linking microclimate with tree water relations and carbon (C) balance. Radial growth depends on C as a source of compounds for the cambial activity, but it also depends on tree water status as a controlling factor for the metabolism of the entire tree. Tree water deficit closes stomata and thus reduces net CO2 assimilation, but it also inhibits cell division and, even more sensitively, cell enlargement (Hsiao and Acevedo, 1974). The two sides (C and water) are closely interrelated and have the potential to limit growth, both individually or in combination (Hinckley and Lassoie, 1981). A growth mechanism needs, therefore, to be able to describe the feedback loops between these two balances and it has to deal with the storage effects mainly on the C-side. One important question is, therefore, how strongly is radial growth determined by the availability of C and to what extent is it driven by water-related conditions?
In the investigation presented here, radial growth rates were continuously measured to collect dynamic information about the biggest C-sink in the stem, sap flow, and water-related stem radius changes. These changes could correlate tree water-related responses with microclimate, and microclimate could link physiology to weather conditions. Calculations of potential transpiration allowed the ratio of actual and potential transpiration (T PET–1) to be obtained and with it an approximation of the degree of stomatal aperture and thus the potential net CO2 assimilation rate. It is not yet possible to provide a complete mechanism for radial growth, but the results at least allow some conclusions to be drawn about physiological causes and effects between climate and radial growth.
Approaching carbon allocation
Radial growth of the above-ground conducting wood was found to be one of the most important C-sinks of a tree beside the input to organs like leaves, roots, and fruits; the oxidation of C for respiration and the C-storage as a resource are further C-sinks (Cannell and Dewar, 1994). According to Hoch et al. (2003), the replacement of the whole canopy would need one-quarter of the total C stored in a deciduous tree, whereas the above-ground xylem increment in an average year would need about one-third of it. On the C-production side, there are well-established methods to measure the net CO2 assimilation rate of tree canopies directly (Schulze, 1981; Häsler, 1982). It is also possible to model the carbon income of different species from microclimatic conditions (Farquhar and Sharkey, 1982; Noe and Giersch, 2004). The most difficult part is simulating stomatal regulation of gas exchange. At the individual-tree level, the newest generation of net CO2 assimilation rate-models includes the hydraulic properties of tree water relations in stomatal regulation. This enables a reasonable simulation of the net CO2 assimilation rate even during drought periods (Leuning et al., 2003; Tuzet et al., 2003).
With respect to C-sinks, important experimental steps towards a mechanistic understanding of C-allocation in trees were made through the investigation of the dynamics of C-pools (Saranpää and Höll, 1989; Fischer and Höll, 1992; Barbaroux and Breda, 2002; Ludovici et al., 2002; Barbaroux et al., 2003; Hoch et al., 2003). Lacointe et al. (2004) demonstrated with potted oak and walnut trees when and under what conditions C is shared and exchanged between branches. They showed that branches are largely autonomous during the summer, but there is a strong dependence on C-reserves at budburst and during initial growth. There seems to be a balance between growth and reserves in tree species such as walnut, oak or beech: the more a tree part grows, the more carbon it stores and utilizes. The dynamics of C-storage in the stem wood of beech and oak trees was quantified by Barbaroux and Breda (2002). They showed how the non-structural carbohydrate (NSC) reserves were used in a ring-porous (oak) and a diffuse-porous species (beech) and compared them with the largely concurrent radial growth during the first part of the season. A marked decrease in NSC concentration was observed in oak trees during budburst and early wood growth, whereas seasonal fluctuations in NSC concentrations of beech trees were small despite a similar timing of radial growth. Corresponding findings were reported by Hoch et al. (2003), who compared ten tree species, including conifers and broadleaved species, in a forest in the midlands of Switzerland in terms of the seasonal fluctuations of their NSC-stores in above- and below-ground organs. They concluded that the NSC-stores were never a limitation for growth since the C-reserves covered more than the total C-demand of a year. Even during times of extensive fruit production (masting) of Fagus sylvatica the C-reserves were never substantially depleted. The C-reserves contributed about 50% of the carbon to the newly formed xylem, in conifers and broadleaved trees. The effort required to track these C-stores is considerable and produces only a rough picture of the dynamics over a season because of the limited number of samples which can be taken and analysed.
Intra-annual radial growth
By contrast with the investigation of C-reserves, the continuous detection of radial growth as one of the C-sinks is relatively easy. With an increment-punching tool, thin samples of the actual tree rings can be taken at regular time intervals on the same naturally grown trees in the field (Forster et al., 2000) and the newly formed cell layers can be counted and measured. Even less labour-intensive is the application of a dendrometer for the continuous detection of radial fluctuations of a tree stem (R) (Zweifel and Häsler, 2001; Zweifel et al., 2001; Wullschleger et al., 2004). However, R consists of several components, of which radial growth and water-related swelling and shrinkage of the bark explain the fluctuation to a large extent (>90%) (Zweifel and Häsler, 2000; Daudet et al., 2005; Zweifel et al., 2005). Zweifel et al. (2005) suggested an algorithm to separate the course of R into growth- and water-related fluctuations which allows intra-annual growth to be directly tracked from stem radius measurements. In addition to the growth rate, the water status of a tree is obtained as the difference between actual and fully hydrated states. This is called the tree water deficit as introduced by Hinckley and Lassoie (1981).
The application of these or similar methods (e.g. the pinning method; Schweingruber, 1996) to quantify intra-annual radial growth has been available for over 30 years (Dougherty et al., 1979). Little interest, however, was paid to questions addressing the determination of radial growth of individual trees in terms of C-utilization and actual microclimatic conditions with a high temporal resolution. Many dendrochronological investigations focused on much longer timescales from decades to thousands of years for data sets with large numbers of trees (Esper et al., 2002). The highest temporal resolution usually ended with the distinction between early and late wood, leading to a half-year resolution at best (Cherubini et al., 1997; Rigling et al., 2001, 2002). One reason for this coarse scale is that the distinction between early and late wood is morphologically preserved in the wood structure and can be analysed at any time, whereas the information on the timing of radial growth needs to be measured in the field at the time of growth. Small variations in wood density within a seasonal structure usually cannot be back dated (Schweingruber, 1996).
Mechanism for radial growth?
Five main factors were found to be involved in a potential mechanism for radial growth: (i) carbon as a source of energy (respiration) and C-compounds to form new layers of cells (Hsiao and Acevedo, 1974; Le Roux et al., 2001); (ii) nutrients and temperature as limiting factors for growth in general (Linder and Axelsson, 1982); (iii) auxin as a plant hormone determining cambial activity (Dengler, 2001; Domec and Gartner, 2002); (iv) mechanical stress as an additional activation factor for cambial activity (Osawa, 1992); and (v) tree water relations as a factor for growth-enabling water pressure conditions in the cambium (Lockhart, 1965; Hsiao and Acevedo, 1974; Lambers et al., 1998; Barbaroux and Breda, 2002; Steppe et al., 2006). Many C-based models of individual tree growth are based on a source and sink approach, on balanced-growth considerations or on optimality principles for shoot and root compartments. They only partially account for the known growth factors and lack a mechanism which allows the prediction of intra-annual radial growth (Cannell and Dewar, 1994; Le Roux et al., 2001). Radial growth is usually proportionally attributed to leaf area which is mainly pre-set by the formation of buds in the previous season (Yang and Midmore, 2005). This follows from the pipe-model theory (Shinozaki et al., 1964) but largely decouples radial growth from the current-year microclimatic conditions. There is empirical evidence that the width of a tree ring depends on the microclimatic conditions in the current year (Breda and Granier, 1996; Corcuera et al., 2004) and on those in the preceding ones (Rigling et al., 2002; Fonti and Garcia-Gonzalez, 2004).
Aims of this work
The investigations reported here offer an insight into the timing of radial growth, budburst, leaf expansion, and physiological processes of three tree species (Quercus pubescens Willd., Pinus sylvestris L., Picea abies L. Karst.) over 3 years at two dry south-exposed sites in the central Wallis, Switzerland (Swiss Alps). The aims of this work were (i) to show simultaneously recorded microclimate conditions, physiological dynamics (T and T PET–1) and radial growth rates of different tree species in a high temporal resolution, (ii) to use these primary results to discover the relative importance of individual climate factors on radial growth, and (iii) to investigate how strongly radial growth determined is by the C-balance and to what extent it is driven by the tree water status. To conclude, the dependence of radial growth on current and past physiological and microclimatic factors are discussed and a suggestion for a model is given. The fact that the data set includes the very dry summer of 2003 is a fortunate coincidence and increased the explanatory power of the results.