Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Drought-induced forest decline, like the Scots pine mortality in inner-Alpine valleys, will gain in importance as the frequency and severity of drought events are expected to increase. To understand how chronic drought affects tree growth and tree-ring δ13C values, we studied mature Scots pine in an irrigation experiment in an inner-Alpine valley. Tree growth and isotope analyses were carried out at the annual and seasonal scale. At the seasonal scale, maximum δ13C values were measured after the hottest and driest period of the year, and were associated with decreasing growth rates. Inter-annual δ13C values in early- and latewood showed a strong correlation with annual climatic conditions and an immediate decrease as a response to irrigation. This indicates a tight coupling between wood formation and the freshly produced assimilates for trees exposed to chronic drought. This rapid appearance of the isotopic signal is a strong indication for an immediate and direct transfer of newly synthesized assimilates for biomass production. The fast appearance and the distinct isotopic signal suggest a low availability of old stored carbohydrates. If this was a sign for C-storage depletion, an increasing mortality could be expected when stressors increase the need for carbohydrate for defence, repair or regeneration.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Ecosystems such as the dry forests at low elevation in the inner Alpine valleys in Europe are excellent field localities to analyse the role of drought on tree performance. Here, precipitation is generally low because of the rain shadow of the surrounding high mountain ranges and frequent drought years (Rebetez & Dobbertin 2004). During the last three decades, climatic conditions have continually shifted towards increasing summer droughts caused by a changing seasonality of precipitation (Weber, Bugmann & Rigling 2007) and increasing evapotranspiration (Rebetez & Dobbertin 2004), leading temporarily to a more negative water balance of the trees (Zweifel et al. 2006).
Recently, the sub-boreal Scots pine (Pinus sylvestris L.), naturally dominating the dry forest at low elevation, shows high mortality rates in the main inner-Alpine dry valleys [e.g. in the Aosta Valley (Vertui & Tagliaferro 1998), in the Inn Valley (Oberhuber 2001) or in the Swiss Rhone Valley (Dobbertin et al. 2005)]. Locally in the Swiss Rhone Valley, almost half of the Scots pine population has died since 1995 (Rebetez & Dobbertin 2004), with the highest mortality rates on dry sites following drought years (Dobbertin et al. 2005).
Continuous depletion of stored carbohydrates was discussed as one mechanism leading to drought-induced tree mortality (Bréda et al. 2006; McDowell et al. 2008). The theory is based on the fact that drought reduces photosynthetic carbon assimilation, which is reflected in a decreased primary production of forests (Ciais et al. 2005; Granier et al. 2007; Reichstein et al. 2007), but also leads to a diminished amount of carbohydrate storage (Bréda et al. 2006). Reduced photosynthesis under drought may lead to a negative carbon balance when the demand for carbohydrates for the maintenance of metabolic or repair processes exceeds the assimilatory supply. Thus, to meet the tree's carbon needs, an increased amount of stored carbohydrates are used. Increasing frequency and/or severity of drought events may lead to tree death as neither the C-assimilation nor the stored carbohydrates suffice to cover the metabolic demands for carbohydrates.
Tree-ring widths (TRWs) yield direct information on tree productivity, which is used as a proxy for climatic impacts on tree growth (e.g. Jacoby, D'Arrigo & Davaajamts 1996; Fritts 2001; Briffa, Osborn & Schweingruber 2004). The combination with stable isotopes in tree-ring analysis allows the distinction between drought situations or very cold growth seasons (e.g. Francey & Farquhar 1982; McCarroll & Loader 2004) because the carbon isotope ratio in tree rings is more directly linked to the plant carbon water relations. During photosynthesis, plants discriminate against the heavier 13CO2 in favour of the lighter 12CO2, leaving plant biomass depleted in 13C. During warm weather conditions, the photosynthetic rate is usually higher than when it is cold. Consequently, higher isotope ratios (less negative δ13C values) are a strong indicator for warm and dry (leading to a reduction in stomatal conductance) weather conditions, and even more expressed for drought periods, with limited water supply.
A further aspect to consider is the isotopic difference in early- and latewood. Carbon isotope values in the latewood show a clear response to current climate condition (Porte & Loustau 2001; Jäggi et al. 2002). Generally, 13C/12C isotope ratio in earlywood is not well correlated with the current climatic parameters, because earlywood is expected to be predominantly formed from old stored carbohydrates, which were assimilated in the previous years (Smith & Paul 1988; Kagawa, Sugimoto & Maximov 2006). Therefore, 13C/12C isotope values in earlywood are poorly related to the ongoing climate because of the mixing of old and new assimilates (von Felten et al. 2007; Keel et al. 2007). Whether this is also true for trees on dry sites with potentially low and eventually exhausted carbohydrate reserves because of chronic drought (Bréda et al. 2006; McDowell et al. 2008) must be examined.
Isotope studies at the seasonal scale provide higher temporal resolution between the impact of environmental conditions and the 13C/12C ratio (e.g. Barbour, Walcroft & Farquhar 2002). A prerequisite is, however, that wood can be dated within the season. A precise technique to determine the seasonal xylem increment is the pinning method (Wolter 1968). However, until now no combination of seasonal 13C/12C ratio and the pinning method has been undertaken.
We installed an irrigation experiment in a site where the Scots pine trees were exposed to drought since their germination. These pine trees growing under contrasting water supply treatments were compared by analyzing tree growth and 13C/12C ratio values on the inter- and seasonal scale using the pinning method.
The goal of this study was to answer the following questions: (1) Is the current water availability [calculated as a seasonal drought index (DI)] recorded in early- and latewood δ13C in equal measure? (2) How fast do trees respond to an improved water supply after a chronic drought stress, and to what degree is it reflected in the width and the carbon isotope ratio of tree rings?
MATERIALS AND METHODS
The experimental site (46°18′ N, 7°36′ E, 615 m a.s.l.) is located in a xeric mature Scots pine forest (average age 90 years) at the valley floor in the driest central part of the Swiss Rhone valley (Valais). The climate is continental with a mean annual temperature of 9.2 °C and mean annual precipitation of 599 mm (standard period 1961–1990, Meteo Swiss, weather station Sion, at 20 km distance from the experimental site).
The forest of the experimental site can be described as Erico-Pinetum sylvestris. The soil type is a pararendzina, which is shallow and skeletal (soil depth 60 cm) characterized by low water retention. The experimental site (about 1 ha) was split up into eight plots of 1000 m2 each, separated by a 5 m buffering strip. Four randomly selected plots were irrigated since June 2003, using water from an adjacent water channel, fed by the river Rhone. Irrigation, corresponding to a supplementary rainfall of about 700 mm per year, was applied at night during the vegetation period (April to October) using sprinklers of about 1 m height. Irrigation earlier than April was not possible because of the risk of equipment damages by late frost. Site conditions differ only in soil moisture because other factors, such as radiation or soil properties, were similar on all plots. Because of the small plot size and the constantly high wind speeds, vapour pressure deficit (VPD) is the same between the plots.
Pinning technique and sample preparation
Three dominant trees per treatment (irrigated and control) were included in this study. To allow a pairwise statistical analysis, the selected trees were classified by crown transparency estimated by visual rating using reference photographs [range: 0% = a fully foliated tree, to 100% = a dead tree; for details on the method of crown transparency estimation, see Dobbertin, Hug & Mizoue (2004)]. The first estimation for the rating of the sample trees took place in March 2003 before the new shoots emerged. To assess the development of foliage density caused by irrigation, the rating was repeated in March 2006. In this study, we concentrated on trees with low defoliation rates with very low to mid crown transparency (see Table 1) to assure that the tree rings were wide enough for seasonal analysis of δ13C.
Table 1. Principal characteristics of the six analysed pine trees
Crown transparency (%)
Tree-ring width (TRW) (mm)
Tree height (m)
Stem diameter (cm)
Crown length (m)
Crown diameter (m)
TRW was calculated for the 10 years before irrigation started (1993–2002). Crown transparency value shows the estimates of March 2003, before the start of irrigation. Tree height, stem diameter, crown length and crown diameter were measured during the experiment (spring 2006).
The pinning method was used for dating the seasonal tree ring formation. Pinning is a technique using the wound reaction resulting from regularly marking the cambium by inflicting micro-injury to date wood formation (e.g. Wolter 1968; Yoshimura et al. 1981; Nobuchi, Fujisawa & Saiki 1993). Trees were wounded at breast height according to Schmitt, Jalkanen & Eckstein (2004) on a biweekly basis during the growth period 2005 (13 April to 13 October) using a small needle (diameter 0.6 mm).
In spring 2006, the trees were cut, and the stem sections with the pinning wounds and two additional stem discs from above the pinning stem sections were taken and stored until further processing at low temperature (−20 °C) to avoid fungal infestation of the samples.
Intra- and inter-annual increments
To assess the variation in annual increment, one stem disc per tree was sanded (35 µm particle diameter), and TRW was measured on two radii using a Lintab digital positioning table and the software TSAP (both from RINNTECH, Heidelberg, Germany). The two measurements per disc were averaged to obtain a mean value per tree.
To study the seasonal increment, cores (19 mm) including the pinning wound were extracted from the stem section and stored in Strasburger solution [one-third ethanol (95%), one-third glycerol (99%) and one-third distilled water] at −20 °C. Using a sliding microtome (Reichert, Vienna, Austria), thin sections (thickness of 10 µm) were cut from the cores including the pinning wound. The thin sections were stained with safranin (1% solution) and astrablue (2% solution) for a better contrast between cell wall and cell lumen. Subsequently, the sections were dehydrated with an alcohol concentration gradient (70, 95% and absolute ethanol), then in xylol (>98%) and permanently embedded in Canada balsam. Pictures of the thin sections were taken (100× magnification, microscope: Olympus BX41, Tokyo, Japan; camera: ColorView III, Soft Imaging system, Muenster, Germany), and the seasonal increments as well as the corresponding TRW were measured on the same thin section using the software IMAGE PRO PLUS (Media Cybernetics, Silver Spring, MD, USA). The seasonal increment was measured as the distance between the last tree-ring border (between year 2004 and 2005) and the last mature cell below the pinning wound (= a cell with a secondary wall, determined by polarized light).
The timing of wood formation was analysed by identifying three phases of tree ring formation: (1) the first mature tracheid found below the wound tissue was regarded as a sign for the beginning of wood formation; (2) early-/latewood transition was reached if the last mature cell was surrounded by latewood cells; and (3) the tree ring was regarded as finalized, if a continuous band of flattened latewood cells was visible, completing the current tree ring.
Sample preparation for isotope analysis
The second stem disc from each tree was dried in the laboratory, and a radial bar, including the last 10 tree rings, was cut from each disc using a band saw. Early- and latewood between the years 1996 and 2005 were separated with a scalpel under a stereomicroscope (Leica Wild M3B, Wetzlar, Germany; 40× magnification), and homogenized using a mill (Ultra Centrifugal Mill ZM 200, Retsch, Haan, Germany).
To analyse the seasonal variation of δ13C values, an additional bar was cut from the disc, and the last tree ring (2005) was sliced by cutting tangential thin sections (thickness of 15 µm) of the specimen using a sliding microtome. Depending on the TRW (0.7–2.1 mm), between 43 and 138 thin sections per tree ring were obtained. To combine the tissue built in the same period of time, we used the relative increment, giving the percentage of the total annual increment built between the pinning dates. It was calculated by relating the seasonal increment (incrementseasonal) to the corresponding TRW, measured near the wound (incrementrelative = incrementseasonal/TRW). To assign the thin section to the time scale, the total number of thin sections per tree ring was regarded as 100%, and all sections representing the tissue built between two pinning dates were pooled. For example, if 5% of the annual increment was developed since the date of the first pinning, then the first 5% of thin sections were pooled as the first sample. If 20% of the tree ring developed until the second pinning, then the next 15% of thin sections were pooled, and so on. All thin sections belonging to the same pinning date were crushed with an ultrasonic processor (UP100S, Hielscher, Teltow, Germany) as milling was not possible because of the small sample size.
The wood powder (about 0.6 mg) of the early- and latewood samples and the thin sections was weighed into tin cups, combusted to CO2 in an elemental analyser (EA-1110, Carlo Erba Thermoquest, Milan, Italy). The elemental analyser was linked to an isotope ratio mass spectrometer (Delta-S Finnigan MAT, Bremen, Germany), which was operating in the continuous flow mode for the determination of the C-isotope ratio. All δ13C values were expressed relative to the international standard Vienna Pee Dee Belemnite (VPDB): δ13C = (Rsample/RVPDB − 1), with R being the 13C/12C ratio of the sample or the VPDB standard. The measured isotope ratio was expressed in the δ notation in ‰. To refer tree-ring δ13C data relative to the pre-industrial standard, data were corrected by adding the difference between the current atmospheric δ13C value at the time when the respective tree ring was formed and the standard (pre-industrial) value to the current δ13C value of each tree ring (McCarroll & Loader 2004).
To compare inter-annual δ13C values with climatic conditions, a DI was calculated as estimate for water availability. The DI was computed as DI = P − PET using monthly data of precipitation sums (P) and potential evapotranspiration (PET) according to Thornthwaite (1948), with climate data coming from the weather station in Sion. For the climate correlation with δ13C values, the monthly DI was inverted and summed up for the 2 months before and the month of early- or latewood formation (earlywood formation = mid-April to mid-June; therefore, climate data were summed up for the period February to June; latewood formation = mid-June to the end of September; therefore, climate data were summed up for the period April to September). For the seasonal comparison between δ13C values and climatic conditions, daily data of temperature (weather station in Sion) and soil moisture [measured on the experimental site at 10 cm depth using Time Domain Reflectometry (1502B cable tester; Tektronix, Beaverton, OR, USA)] were smoothed by a moving window of 15 d.
Inter-annual δ13C signal
Inter-annual variation of TRW showed a rather high conformity to the seasonal DI in the pre-irrigation period until 2002 (Fig. 1). However, because of the strong variation among the trees, the mean correlation averaged over the individual correlation between the tree ring series and the seasonal DI was not significant (r = 0.44). Regarding δ13C values, the conformity with water availability was more pronounced (Fig. 2). The correlation between δ13C and the inverse seasonal DI was highly significant for early- and latewood. In contrast to the correlation between earlywood δ13C and the previous earlywood (see autocorrelation in Table 2) or the previous latewood δ13C (see ‘latewoodprevious’ in Table 2) was not significant at all. As δ13C values in early- and latewood both strongly responded to the water availability (high correlation with the seasonal DI) under naturally dry conditions, the linear regression between δ13C in both tissues was highly significant (see ‘non-irrigated’ in Fig. 3; P < 0.001, R2 = 0.81).
Table 2. Autocorrelation coefficient (Auto) and correlation coefficients between the tissues (earlywood versus current or the previous latewood) and with climate conditions [drought index (DI)] of δ13C values in earlywood and latewood
δ13C in earlywood was correlated with the DI summed up for the months February to June. δ13C in latewood was correlated with the DI summed up for the months April to September.
The irrigation experiment started late in June 2003, which was an unusual hot and dry year. As the bulk of the tree ring (especially earlywood) was already formed at this date, TRW in all analysed trees, even in the irrigated ones, decreased in this dry year (Fig. 1). But, in 2004, a clear differentiation according to the treatment was observed, as TRW clearly increased in all irrigated trees, while it remained low in the control, non-irrigated trees.
For the δ13C values, we found an immediate response to irrigation as δ13C values in latewood decreased already in 2003, the very first year of the experiment (Fig. 2). Earlywood δ13C values also responded immediately to irrigation, as detected by a decrease of the isotopic signature in 2004, the first earlywood formation under improved water supply. Only in tree2 the response to irrigation in earlywood and latewood was delayed by 1 year. However, in control trees, δ13C values increased and remained constantly high in and after the drought year 2003.
Seasonal variation in δ13C
The 2005 seasonal δ13C values showed a common pattern with increasing isotope ratios to a maximum in mid-summer and decreasing values until the end of the growth period (Fig. 4a). The shape of the seasonal δ13C curves is rather smooth with small deviations at the sub-seasonal (bi-weekly) scale.
The irrigated trees showed significantly lower δ13C values over the entire growth period compared to the control trees (Wilcoxon signed rank test, P < 0.001) because of clearly higher soil moisture (Fig. 4c). In addition, the maximum of the seasonal δ13C curve was shifted because of irrigation and occurred 2 weeks earlier compared to the control trees.
For the control trees, maximal δ13C values were observed after the hottest and driest days of the year (Fig. 4a versus c) concurrently with early/latewood transition and a drop in the growth rate (see slope in Fig. 4b).
In irrigated trees, the highest δ13C values were reached in the week prior to the hottest and driest period of the year. The maximum in δ13C values were not associated with the early/latewood transition or an early decrease of the growth rate. In contrast to control trees, no growth depression or stop was observed in irrigated trees before the end of the growing season. In contrast to the control trees, no clear correlation with growth or climatic conditions was observed in the occurrence of the highest δ13C values in irrigated trees.
Previous studies showed that earlywood formation strongly depends on previously stored carbohydrates (Smith & Paul 1988; Kagawa et al. 2006). Thus, δ13C in earlywood is expected to reflect a climate impact, which is more related to the previous year(s) than to the weather conditions of the current year (see Porte & Loustau 2001 for Pinus pinaster). In contrast, our data showed a high sensitivity of early- and latewood δ13C to instantaneous (current year) water availability (see Fig. 2 and Table 2 correlation with the seasonal DI, P < 0.01) and low correlations with the climate conditions of the previous year (see Table 2, autocorrelation and correlation between early- and the previous latewood). As δ13C in early- and latewood reflects the current climate conditions, the relationship between δ13C in the two tissues is very high (P < 0.001; Fig. 3). These results suggest a tight coupling of wood formation and currently produced photosynthates. The assumption of a rapid and direct use of currently produced assimilates for biomass production was also supported by the immediate response to irrigation observed in early- and latewood δ13C (Fig. 2). Hence, an intensive mixing of old and new carbohydrates masking the current year's climate signal in earlywood is very unlikely for the trees we studied. This is in contrast to the other observations for deciduous trees in a temperate forest (Keel, Siegwolf & Körner 2006; Keel et al. 2007) and for conifer trees at the upper tree line (von Felten et al. 2007). We assume that the low mixing of new assimilates with stored photosynthates indicated by our data is caused by a low amount of stored carbohydrates. During the preceding years of chronically dry conditions in the Valais, the plants could not synthesize sufficient carbohydrates to sustain growth and the ongoing metabolic processes, which resulted in a depletion of C-storage. The deviating response pattern of tree2 could be caused by the following: compared with the other irrigated trees, tree2 showed a premature switch to latewood formation and an earlier cessation of wood formation in 2005 (Fig. 4). The year 2003, the first year of irrigation, was particularly dry. As earlywood latewood transition and the cessation of wood formation are expected to occur even earlier under drought (Dünisch & Bauch 1994), it is very likely that tree2 was not able to response to the start of irrigation as the bulk of latewood was already formed. This could have caused the increase in δ13C for the latewood in 2003 leading to a carry-over effect in the δ13C of the earlywood of the following year (2004), which rather increased instead of decreased, contrary to trees 1 and 3. As for the reason, we can only assume that the irrigation was not as effective for tree2, either because the irrigation water did not reach the roots as abundantly and fast as for the other two trees. The reason for the decreasing TRW in 2005 in tree2 might be a result of natural variation in radial increment or because of the generally high increment rate in tree2, allowing no strong increase in the radial increment after the maximal TRW in 2004.
However, our results suggest that trees growing under constantly dry conditions are source limited. According to the source limitation theory (see Berman & DeJong 1996), carbon availability is reduced under drought if photosynthetic activity is reduced, but carbon demand remains constantly high. Most recent modelling results support this theory: photosynthesis was shown to be more sensitive than respiration to the influence of drought at the stand level (Schwalm et al. 2010). The resulting storage depletion might explain the long-lasting growth depression in Scots pine after severe drought events (Eilmann et al., unpublished results), which can also be observed in tree2 and tree3 of the control plot after the drought year 2003 (Fig. 1).
However, according to the sink limitation theory, drought would reduce tree growth first. This lack of sink activity would lead to carbohydrate accumulation in the leaves inhibiting photosynthesis. The direct effects of drought on photosynthesis are expected to be rather low (Körner 2003; Bacelar et al. 2006). Following this theory, carbohydrate supply should never be limited. In case of a sink limitation, we would not have seen such an immediate response to irrigation in δ13C. The new assimilates would have been mixed with the large pool of stored carbohydrates, and the change in signal would have been delayed and dampened, which is in contrast to the results of our experiment. The sink limitation theory is based on studies with species that are well-adapted to drought, like Olea europaea or Pistacia lentiscus. In contrast, we analysed the sub-boreal Scots pine growing at the edge of its distribution, which have to cope with meteorological conditions that are changing towards a sub-Mediterranean climate regime. Therefore, the conclusion lays at hand that if a species is growing beyond their ecological optimum, an ongoing drought situation can lead to storage depletion as a result of a persistent carbon imbalance. Low carbon pools reduce trees' resistance against stressors (Bréda et al. 2006). Hence, increasing mortality must be expected, when a stressor either increases the need (parasitism) or decreases the production (drought) of carbon reserves.
The strong impact of drought on tree growth in the Valais is underlined by the high δ13C values observed in trees growing under the prevailing dry conditions. The mean values of −23.6‰ for earlywood and −23.3‰ for latewood (calculated for the non-irrigated period in irrigated trees and for the whole period in control trees) were among the highest ever found in the wood of pine species (e.g. the following values for different trees from different regions were found: −23.9‰ for wood in Pinus halepensis in the east Iberian Peninsula (Ferrio & Voltas 2005); <−24‰ analysed for wood in P. sylvestris from Southern Germany (Gessler et al. 2009); −25.77‰ analysed for wood in P. sylvestris from Norway to Sibiria (Saurer, Siegwolf & Schweingruber 2004). These high δ13C values reflect highly reduced stomatal conductance as a consequence of a chronic drought situation for Scots pine in the Valais.
In this study, we present exactly dated seasonal δ13C values referring to the timing and the quantity of wood formation. Seasonal variations in micrometeorology and especially altering soil moisture were discussed as main factors responsible for seasonal changes in δ13C (e.g. Leavitt 1993; Barbour et al. 2002; Schulze et al. 2004). This is in agreement with our data, as maximal δ13C values in control trees were observed after the hottest and driest period of the year (Fig. 4a versus c). The continuous decrease in δ13C values until the end of the vegetation period might be caused by decreasing VPD and increasing relative humidity in the second part of the vegetation period. At the same time, as δ13C was at its maximum, noticeable reductions in wood formation occurred. Trees strongly reduced the growth rate or even stopped growing (see slope in Fig. 4b). This underlines the strong limitation of tree growth under the prevailing dry conditions. Besides the amount of produced wood, its structural constitution changed. Wood formation turned towards latewood production, forming smaller cells with thicker cell walls. This earlier switch to latewood formation is again a reaction to the very dry conditions (Dünisch & Bauch 1994).
In irrigated trees, maximal δ13C values occurred previous to the hottest and driest period of the year (Fig. 4a versus c). In contrast to the early maximum in the seasonal δ13C curve of the irrigated trees, the transition from early- to latewood formation was delayed by improved water supply (Dünisch & Bauch 1994; Rigling et al. 2003; Eilmann et al., unpublished results). The increment curve showed a late flattening and constantly high growth rates throughout the growing season, a rather common pattern for conifers (Rossi, Deslauriers & Morin 2003). Thus, a substantial limitation of tree growth by climatic conditions could not be observed in irrigated trees. Regarding the level of seasonal δ13C values, we observed significantly lower δ13C values throughout the entire growing season in irrigated trees (Fig. 4a). This is in agreement with the results on the inter-annual level (Fig. 2) and is in line with previous findings (e.g. Livingston & Spittlehouse 1993; Warren, McGrath & Adams 2001; Ferrio & Voltas 2005).
To summarize: in irrigated trees, the point of maximal δ13C values seemed to be decoupled from changes in wood formation or varying climatic conditions, while trees, growing under prevailing dry conditions, show a strong coupling between climate, tree growth and maximal δ13C values.
We found a strong relationship between early- and latewood δ13C as both mainly reflect the current climate conditions. Accordingly, the irrigation signal was immediately recorded in the δ13C values. These data, indicating a rapid and direct incorporation of current photosynthates, support the source limitation theory with chronic storage depletion as important mechanism leading to drought-induced mortality. This phenomenon of restricted tree growth as a result of source limitation might only be valid for trees at the edge of their distribution growing under changing climate conditions. However, an ongoing climate change will substantially alter the growth conditions for various species and in different ecosystems. Thus, tree mortality because of preceding storage depletion as a result of continuous drought-induced source limitation is expected to become increasingly relevant.
We would like to thank M Dobbertin for valuable discussions and assistance in the field; E. Graf Pannatier for surveying and providing soil moisture data; R. Köchli, U. Wasem and M. Burger for their help in sample preparation. S. Paget for English corrections to the manuscript, and A. Kress for support in the lab. We are grateful to the forest services of the canton Valais, Burgergemeinde Leuk and HYDRO Exploitation SA for infrastructure, to W. Landolt und P. Bleuler for the managing of the irrigation system, and to K. Egger and his team from the forest service Leuk for their help in the field and their support on the experimental site. This study was conducted in the frame of the CCES-project MOUNTLAND funded by the Velux Foundation and the Canton Valais.