Introduction

Transpiration in mixed forests can be spatially heterogenous due to size-related and interspecific differences in tree water use. Recently conducted sap flux studies in mixed stands have revealed substantial differences in whole-tree water turnover among co-occurring tree species (Granier et al. 1996; Pataki et al. 2000; Oren and Pataki 2001; Dünisch and Morais 2002). Interspecific variation in tree water use may thus complicate the prediction of stand-level transpiration from tree-level sap flux data if species differences were relevant such as stem density, tree size and age (Köstner et al. 2002). On the other hand, a comparative study on sap wood area and sap flux in more than 20 tree species in a Panamanian rain forest revealed that variation in diameter at breast height accounted for 98% of the variation in sapwood area, and for 91% of the variation in total daily sap flux density in the outermost 2 cm of sapwood if all species are pooled (Meinzer et al. 2001). This result suggests that scaling water use from tree to stand level could be relatively straightforward even in species-rich forest communities. Unfortunately, the number of studies on species differences in water use in mature mixed forests is still too limited to draw more general conclusions on the hydrological relevance of interspecific differences in water use and for relating water use to tree species attributes.

In Central Europe, forestry is fostering the establishment of mixed-species stands which, in a few decades, shall replace large parts of the monospecific forest plantations dominating the landscape today. Additionally, with the prospect of a warmer and perhaps more summer-dry climate, tree species with a putatively higher drought tolerance such as Tilia cordata Scop., Carpinus betulus L., Fraxinus excelsior L. and Acer pseudoplatanus L. may spread in natural woodlands, and are likely to be favoured by foresters in managed stands (Lasch et al. 2002). However, only little information on water use and drought sensitivity of these species exists so far. This gap of knowledge contrasts with the large body of information available on the water relations of Fagus sylvatica L., Picea abies Karst. and Pinus sylvestris L. (e.g. Köstner et al. 1996; Köstner et al. 1998a; Granier et al. 2000) which represent the three commercially most important tree species in Central Europe. Therefore, a conversion of monospecific to mixed stands may have unknown consequences with regard to water use and drought response at the tree level, and for stand transpiration.

In this study, we analysed seasonal patterns of water use at the tree level in five co-occurring species of a mixed broad-leaved forest in the Hainich National Park, Thuringia, Germany. The species were assumed to show a contrasting behaviour with respect to drought sensitivity and regulation of water use. F. sylvatica was found to be drought-sensitive with respect to stem increment, fine root growth and water use at the tree level (Cochard et al. 1999; Leuschner et al. 2001; Schipka 2002). A. pseudoplatanus is found on sites with moderate to high soil moisture and air humidities indicating a moderate drought sensitivity of this species (Ellenberg 1996). F. excelsior thrives on moist alluvial soils but is also found on shallow drought-affected limestone soils (Marigo et al. 2000). C. betulus and T. cordata are most abundant in Central European sub-continental and continental regions with higher drought frequencies. However, precise data on water use and drought sensitivity are limited in these two species. From this information, drought sensitivity of the species should follow a pattern F. sylvatica > A. pseudoplatanus > F. excelsior > C. betulus and T. cordata. We tested this hypothesis by recording sap flux in mature individuals for each species during wet and dry periods of a growing season. In accordance with Pataki et al. (2000) drought tolerance is defined here as the extent of sustained sap flux during the dry period. We used continuous measurements of sap flux density and a characterisation of the conducting sapwood area to compare the water use patterns of the five species at the tree level. To eliminate tree size effects on water use, sap flux data were normalised by stem diameter.

Methods

Study site

The study was conducted in two old-growth broad-leaved forest stands in the Hainich National Park (Thuringia, Central Germany). The stands are situated in the southeastern part of the National Park at 290 m a.s.l., close to the village of Weberstedt (51°05′28″N, 10°31′24″E). One stand is a species-rich mixed forest with the following species present as mature trees (in order of abundance): F. excelsior, A. pseudoplatanus, T. platyphyllos Scop., T. cordata, C. betulus, Ulmus glabra Huds., A. platanoides L., Quercus robur, A. campestre L., F. sylvatica and Prunus avium L. Most trees in the upper canopy layer were about 80 years old with the exception of Q. robur L. and F. sylvatica (about 180 years old), which is a consequence of traditional management (coppice with standard). The stand has not been modified by forestry for the last 40 years. Tree height ranged between 26 and 29 m, stem density was 260 stems ha−1 (≥7 cm DBH). The basal area of trees was 46.7 m−2 ha−1. More details on stand structure are given in Frech et al. (2003). In spring and early summer, a dense field layer with dominant Allium ursinum L., Leucojum vernum L. and Aconitum vulparia RCHB is present. The potential natural vegetation in the area is putatively dominated by F. sylvatica forests (Hordelymo-Fagetum). F. sylvatica (beech) is a rare species in this stand; therefore, a monospecific beech stand in about 800 m away, on comparable soil was also investigated. The beech trees in this stand were ca. 140-year old, stem density was 209 ha−1 and basal area was 44.6 m−2 ha−1. Average leaf area index as determined with an optical analysis system (LAI 2000, LiCor, Lincoln, USA) was 5.1 in the mixed stand (2004) and 4.9 in the beech forest.

Average annual temperature in Weberstedt is 7.5°C and the average annual rainfall is 669 mm (Deutscher 2001). In both stands, the fertile and clay-rich soil with a depth of 50 cm is derived from triassic limestone (Lower Muschelkalk) mixed with a small quantity of quaternary loess. The loess content in the upper soil horizon in the beech forest is slightly greater than in the mixed forest. In both stands the soil type is Cambisol with a thin litter layer. The pH (measured in KCl) in 0–50 cm depth ranged between 5.7 and 6.0, the C/N ratio was 13–14 mol mol−1. The exchangeable cations consisted to about 99% of Ca, K, Mg and Na.

Sap flux measurements

Sap flux measurements were conducted simultaneously in five tree species (A. pseudoplatanus, C. betulus, F. excelsior, F. sylvatica) during 6 July to 31 August 2001. The first four species were studied in the mixed stand, the latter in the nearby beech forest. Five trees per species were selected randomly among the trees that reached the upper canopy level. The studied trees were located in an area of 6600 m2 in the mixed forest and 2400 m2 in the beech forest. Both plots were located in the centre of larger forest stands with similar structure. Structural characteristics of the 25 trees studied are listed in Table 1. Each tree was equipped with four constant heat sap flux gauges constructed after Granier (1987) which were inserted at breast height in four directions of the trunk (N, E, S, W). The sensors used for the diffuse-porous species (A. pseudoplatanus, C. betulus, F. sylvatica and T. cordata) consisted of a copper–constantan thermojunction inserted 1 cm into a 2 cm long and 2 mm wide aluminium probe. In the case of the ring-porous F. excelsior, we used probes of only 1 cm length (thermojunction at 0.5 cm) to avoid sensor length exceeding sapwood depth. At each measuring point two probes were inserted radially into the stem, with one located 15 cm above the other. The upper probe was equipped with a constantan heater supplied with a constant power source. The two thermojunctions were coupled and the temperature difference between the two probes was recorded continuously to obtain sap flux density (J s, in g s−1 m−2) according to an empirically derived formula of Granier (1987). Sap flux data were sampled in 15-s intervals with CR10X dataloggers (Campbell, Cambridge, UK) and averaged every 15 min.

Table 1 Characteristics of the studied trees. Each tree was equipped with four pairs of sap flux sensors
Full size table

Calculation of sap flux per tree is based upon radial changes in J s and the sapwood area of the trunk. We applied two approaches to estimate the size of the conducting sapwood area and to analyse the water transport patterns in the trunk cross section. First, we analysed the relationship between sap flux density and xylem depth in the five species with two sets of four mobile sensors (one set in the mixed forest, one in the beech forest) that were inserted consecutively into the xylem at depths of 2–4, 4–6 and 6–8 cm. These sensors were left for 5 days at a given depth and were moved to another tree when the three depths had been measured. Flux density as a function of depth is expressed as percentage of the synchronously measured flux density at 0–2 cm depth.

Second, we conducted dye experiments to identify the size of the conducting sapwood area in the trunks of 7 trees per species and 12 in the case of F. sylvatica trees. At noon on days with bright weather, we drilled deep radial holes (diameter 12 mm) into the trunk while water coloured with Indigo Carmin infiltrated into the hole. After 1 h, a growth core was extracted from the trunk 5 cm above the first hole and the length of the coloured core section was then taken as the depth of conducting sapwood (Andrade et al. 1998; Meinzer et al. 2001). Based on these data, we calculated equations for the relation between stem diameter in 1.3 m height (DBH) and sapwood area for the five species (see results and Table 3). They were used to estimate the sapwood area of a model tree of each species with a diameter of 49 cm.

Micrometeorological and soil moisture measurements

Air humidity and air temperature were measured in a clearing adjacent to the mixed forest (Rotronic MP 100 A, Ettlingen, Germany). Data were logged synchronously with the sap flux values. In the centre of the study plot, the volumetric soil water content was continuously measured each with two TDR probes (Soil Moisture, Santa Barbara, USA) at 5, 15 and 40 cm of soil depth. A second TDR station was run in the nearby beech stand taking measurements at the same depths. Data were recorded twice a day at 12 and 24 h. Soil water storage was calculated by integration of soil moisture measurements at different depths.

Statistical analyses

The relationship of diameter breast height to sapwood area and sap flux density were tested with linear regressions. The response of sap flux density to changes in environmental parameters were evaluated with linear, non-linear and multiple regression models. The residuals of the regression analyses were tested for homoscedascity and normal distribution. The level of significance was usually P≤0.05. The statistical analyses were conducted with SAS software (version 8.1).

Results

Weather conditions and soil water content in the study period

The average daily air temperature during the study period (6 July–31 August 2001) was between 13 and 25°C and the vapour pressure deficit (D) ranged between 0.2 and 1.8 kPa (Fig. 1). There was considerable rainfall in the beginning of July, but there was almost no rainfall between mid-July and the end of August 2001. Soil moisture content (Θ) increased from the beginning of July to mid-July, and thereafter decreased until the end of August. The calculated soil water storage in 0–50 cm of depth showed comparable values and time courses in the mixed stand and the beech forest (Fig. 2).

Fig. 1
figure 1

Course of air temperature (A) and vapour pressure deficit D (B) on a clearing adjacent to the mixed forest during the study period (6 July to 31 August 2001)

Full size image
Fig. 2
figure 2

Course of soil water storage in 0–50 cm of soil depth in the mixed forest and the neighbouring beech forest (6 July to 31 August 2001)

Full size image

Sap flux

Under moist soil conditions (July 2001), sap flux density (J s) varied substantially among the five studied tree species (Fig. 3). In general, J s decreased in the sequence F. sylvatica > A. pseudoplatanus > T. cordata > C. betulus > F. excelsior. Twenty-one-day means of J s with moist soil (soil water storage > 170 mm) were: 152 g cm−2 per day for F. sylvatica, 102 g cm−2 per day for A. pseudoplatanus, 86 g cm−2 per day for T. cordata, 77 g cm−2 per day for C. betulus and 67 g cm−2 per day for F. excelsior. In two of the species, J s showed a significant linear decrease with DBH (F. sylvatica: r 2=0.39; F. excelsior: r 2=0.24); whereas J s increased with DBH in A. pseudoplatanus (r 2=0.41), T. cordata (r 2=0.99) and C. betulus (r 2=0.31).

Fig. 3
figure 3

Course of average sap flux density (J s) in the five studied tree species on a bright day under moist soil conditions (21 July 2001). J s was measured in five trees per species

Full size image

During the study period, daily sap flux densities tended to decrease in all species except F. excelsior (Fig. 4). In four species (A. pseudoplatanus, C. betulus, T. cordata and F. sylvatica) we found a significant correlation between J s and in 5 cm of soil depth (Fig. 5). The steepest slope and the strongest correlation between and J s were found in F. sylvatica. There was no significant correlation between J s and in 5 cm in F. excelsior. Similarly, soil moisture dynamics in greater soil depths explained little of the observed variance of J s in this species (at 15 cm depth: r 2=0.11, P<0.05).

Fig. 4
figure 4

Daily sap flux density (J s; mean and standard error of five trees per species) of the five studied species (6 July to 31 August 2001)

Full size image
Fig. 5
figure 5

Daily sap flux density (J s) as a function of volumetric soil water content (5 cm of soil depth) in the five studied species. For species with a significant correlation (P<0.05) linear regression lines are drawn. Note different scale of x- and y-axis in the figure of F. sylvatica

Full size image

For an analysis of the response of J s to changes in D under contrasting soil water conditions we selected a wet (21 consecutive days with soil water storage >170 mm) and a dry period (21 consecutive days with soil water storage <170 mm). In both periods, the response of J s to changes in D could be described for all species except F. excelsior by logarithmic functions of the type J s=a+bln D (Fig. 6). Parabolic functions with a decrease of J s at high D values, as were used by, e.g., Pataki et al. (2000), may give better results for datasets with D values >1.8 or 2.0 kPa, which never occurred in our study. In the species studied, the variance in J s explained by D was less under dry than under moist soil conditions in all species except F. excelsior. From the functions, that relate J s to D, the reduction in J s as a consequence of soil drought can be estimated for different values of D. At a D of 0.9 kPa, which is an average value for the study period, T. cordata reduced J s by 44%, F. sylvatica by 39%, A. pseudoplatanus by 37%, and C. betulus by 31% (Fig. 7). The degree of reduction in J s was independent of D in F. sylvatica, C. betulus and T. cordata, but tended to increase with D in A. pseudoplatanus. Multiple regressions revealed that a large proportion of the observed variation in J s was explained by D and in F. sylvatica and A. pseudoplatanus (71 and 66%, respectively). J s of T. cordata, C. betulus and, in particular, F. excelsior showed a lower influence of D and Θ (Table 2). In F. sylvatica, soil moisture contributed more to the explained variance (57%), whereas in A. pseudoplatanus a stronger influence was exerted by D (37% of variance explained). An addition of air temperature to the multivariate regression analysis did not increase the explained variance significantly and was therefore rejected.

Table 2 Relationships between daily sap flux density (J s, in g cm−2 per day) in the five studied tree species and daily means of vapour pressure deficit (D), and soil moisture in 5 cm of soil depth according to a multiple regression analysis for 46 consecutive days. D and soil moisture were not correlated to each other (r 2=0.02, P>0.05)
Full size table
Fig. 6
figure 6

Daily sap flux density (J s) as a function of vapour pressure deficit (D) for periods with higher and lower soil moisture contents (open and filled circles, respectively). For species with a significant relationship (P<0.05) logarithmic functions of the type y=a+b ln x are drawn. Note different scale of y-axis in the figure of F. sylvatica

Full size image
Fig. 7
figure 7

Relative reduction in sap flux density J s between periods with high and relatively low soil water reserves in four tree species as a function of D. J s values at high soil moisture are taken as 1.0. The data were computed with species-specific models that relate J s to D under contrasting soil moisture contents (Fig. 6). There were no significant effects of D and soil moisture in 5 cm of depth on J s of F. excelsior

Full size image

In order to extrapolate sap flux density at 0–2 cm depth to sap flux at the tree level, J s was monitored at different xylem depths. In the five F. sylvatica trees, we found a reduction in J s by 62% between 0–2 and 2–4 cm depth and a further less pronounced reduction (when compared to 0–2 cm: 78% in 4–6 cm and 85% in 6–8 cm; Fig. 8). Our measurements on the J s/depth relationship in the other diffuse-porous species A. pseudoplatanus, C. betulus and T. cordata point to comparable patterns of decrease in J s, but the dataset was limited. Therefore, the results of F. sylvatica were used for calculating the total water use per tree in these species as well. The sap wood area (A s) in the diffuse-porous trees reached depths between 11 and 22 cm in the xylem without marked differences among the species. Linear regression equations for estimating A s from DBH were established for these four species. If all trees of the diffuse-porous species are pooled, a single equation explained 94% of the observed variance in A s by the variable DBH (Table 3). For model trees of 49 cm DBH, daily sap flux was computed in all five species for the period of 6 July to 31 August 2001 (Fig. 9). With 168 kg per day the model tree of F. sylvatica greatly exceeded the other three diffuse-porous species in their daily maxima of sap flux (A. pseudoplatanus 95 kg per day; C. betulus 74 kg per day; T. cordata 75 kg per day). For the ring-porous F. excelsior, three to six times smaller daily fluxes were computed (27 kg per day). In the period of 10–31 August with reduced soil moisture, daily flux rates were strongly reduced in T. cordata and F. sylvatica, and to a lesser extent in the other three species. However, F. sylvatica still exceeded the other studied species with maxima at 88 kg per day (A. pseudoplatanus 62 kg per day, C. betulus 54 kg per day, T. cordata 49 kg per day, F. excelsior 27 kg per day).

Table 3 Sap conducting xylem area of the five studied tree species, and of a bulked sample of all studied trees of diffuse-porous species (A. pseudoplatanus, C. betulus, T. cordata, F. sylvatica). Parameters of linear equations (y=a+b x) used to estimate sapwood area from stem diameter at breast height (including bark) are given
Full size table
Fig. 8
figure 8

Radial decrease of sap flux density (J s) with xylem depth in stems of F. sylvatica. The flux density at 0–2 cm was taken as 100% (y=99.29x −0.92, r 2=0.98, P<0.01; n=5)

Full size image
Fig. 9
figure 9

Computed sap flux of model trees with 49-cm diameter in the five tree species studied (6 July to 31 August 2001). To extrapolate from sap flux density at 0–2 cm to whole tree sap flux, the J s–depth relationship established for F. sylvatica was used in A. pseudoplatanus, C. betulus and T. cordata as well

Full size image

Discussion

This study compares sap flux in mature trees of five broad-leaved species that co-occur in close proximity under similar edaphic and climatic conditions. Comparison among the five species was conducted on two levels, (i) for flux density (J s) in the outermost xylem and (ii) for daily flux per tree. A crucial step in computing whole-tree water fluxes is the extrapolation of measured flux density to the total conducting area in the trunk. Similar to our results, studies on F. sylvatica stands in Southern Germany and Eastern France found an exponential decrease of J s with xylem depth (Köstner et al. 1998b; Granier et al. 2000). However, the magnitude of change in J s with xylem depth varied among these studies. We found a drop by 62% from 0–2 cm to 2–4 cm, whereas Granier et al. (2000) measured a reduction by 45% using a comparable technique. At greater depth the reductions in flux density were more similar in the two studies. The differences in spatial J s patterns could result from a contrasting tree age (32 vs. 140 years in the French and the Hainich stand, respectively), or from site differences (e.g. contrasts in annual precipitation: 820 vs. 669 mm per year in the French and the Hainich stand, respectively). For the other tree species of this study, data on the J s/xylem depth relationship are much more limited. The existing evidence, however, points at a similarly rapid decrease of J s from the outer to the inner xylem in the other three diffuse-porous species as was found for F. sylvatica. Therefore, we assume that our estimates of tree level water fluxes are a useful approximation for these species as well. In the ring-porous F. excelsior, we found that only the outer 1 cm of the xylem was sap conducting and, thus, the total sap wood area of this species was about 11 times smaller than that of the diffuse-porous species.

Dyeing experiments in the trunks showed a close linear relation between stem diameter (DBH) and size of sapwood area (A s) in the four diffuse-porous species. Our data on sapwood area in F. sylvatica correlate well with the results of Falge et al. (2001) from Southern Germany who measured the sapwood area in this species by computer tomography (r 2=0.96, P<0.01 for the two datasets). These results indicate that trunk dimension can serve as a reliable estimator of sap-conducting area in F. sylvatica, and presumably in the other diffuse-porous species as well. There were only small differences in the relationship between DBH and A s between F. sylvatica, A. pseudoplatanus, T. cordata and C. betulus. In a bulk dataset of these species DBH accounted for 94% of the observed variation in A s. However, A s and DBH are strongly autocorrelated because DBH is used to convert sapwood depth to A s (Oren et al. 1998). In contrast to A s, sap flux density in the outermost xylem was found to be influenced significantly by both, the identity of the species and DBH.

Under conditions of a moist soil (>170 mm of water in the 0–50 cm topsoil profile), we obtained a decrease of J s in the sequence F. sylvatica > A. pseudoplatanus > T. cordata > C. betulus > F. excelsior. These sap flux densities showed a significant positive correlation with DBH in three of the species (A. pseudoplatanus, T. cordata, C. betulus) but significantly decreased with DBH in two other species (F. sylvatica, F. excelsior). The studied trees of F. sylvatica exceeded the trees of the other species in DBH. Thus one may expect that if trees of the same diameter were studied, the difference in J s of F. sylvatica and the other studied species would be even greater. A decrease of J s with DBH was also reported by Meinzer et al. (2001), whereas both Granier et al. (1996) and Oren et al. (1999) found that J s increased with DBH, and Pillips et al. (1999) found no consistent relationship between J s and trunk diameter. Thus, evidence exists that the sapwood area of diffuse porous-species can be reliably estimated from trunk diameter irrespective of the species, but, at least at our study site, species-specific differences seem to be important when attempting to predict sap flux density.

Decreasing soil water reserves (from >200 to 150 mm in a profile of 50 cm) reduced J s significantly in four of the studied species (A. pseudoplatanus, C. betulus, F. sylvatica, T. cordata). A multiple regression analysis revealed a greater influence of soil moisture on J s in F. sylvatica than in the other species which were more effectively controlled by D in their sap flux. The reduction of J s under contrasting soil moisture contents was calculated by use of a model (J s vs. D) that allows comparison of J s under a given D and reduces the influence of extreme values. These calculations suggest that the reduction in J s of T. cordata (44%) was larger than of F. sylvatica (39%). A. pseudoplatanus (37%), and C. betulus (31%) showed somewhat smaller relative reductions. No effect of soil moisture effect in 0–5 cm of depth was visible in the ring-porous species F. excelsior. If the relative reduction in J s under soil drought is taken as a measure, a tendency of a decrease in drought sensitivity in the sequence T. cordata > F. sylvatica > A. pseudoplatanus > C. betulus > F. excelsior would appear. This ranking is not in line with observational data on the drought sensitivity of these species in Central German forests (cf. Ellenberg 1996). It can be argued that the parameter ‘relative reduction in J s’ alone is not a good measure of drought sensitivity.

The only ring-porous species in our sample, F. excelsior, differed from the other species not only in wood anatomy, but also in terms of maximal sap flux density, size of conducting sapwood area, and the sensitivity of J s to atmospheric vapour pressure deficit and soil moisture. The low J s is difficult to explain and might indicate inaccuracies in sap flux measurement in this species. For example, the insertion of the sap flux gauges could have provoked embolisms in the large vessels of this species. However, Granier-type sensors have successfully been used in quite a number of other ring-porous species (such as oaks) without obvious embolism-related problems (Bréda et al. 1995; Oren and Pataki 2001). It appears that the low sensitivity of J s to D and in F. excelsior is not an artefact because similar response patterns were observed with non-intrusive heat balance sap flux systems (Coners and Leuschner 2002) mounted on the roots of F. excelsior in the Hainich stand (Sandra Korn, unpublished data). Moreover, when comparing different tree species in a North American forest, Oren and Pataki (2001) found a lower sensitivity of J s to variation in D in ring-porous species compared to diffuse-porous species. The causes of this apparent partial decoupling of sap flux in F. excelsior from atmospheric demand remain unclear and deserve further analyses. Due to the uncertainty in the sap flux measurements of F. excelsior, we did not calculate canopy conductance at the stand level.

An explanation for the unusual behaviour of sap flux in F. excelsior may perhaps be found when studying some allometric characteristics of this species, which distinguish this species from many other temperate tree species. Total leaf area of mature F. excelsior trees was found to be smaller than in other temperate broad-leaved species (Roberts and Rosier 1994; Hölscher et al. 2002; 2004). In the Hainich stand, F. excelsior contributed 45% of the total number of stems and 41% of the total projected crown area in the stand, but only 28% of the total leaf area. Thus, the relatively small sap conducting area can partly be explained by a small leaf area. On the other hand, the F. excelsior trees in the Hainich stand developed greater densities of fine roots in the soil profile than one would expect from the abundance of F. excelsior stems in this stand: 58% of total fine root biomass, but only 45% of the stems belonged to F. excelsior. This species also had a comparably high abundance of coarse roots in 30–50 cm soil depth. The high abundance of F. excelsior roots may facilitate soil water uptake and could reduce the risk of drought-induced xylem embolism in the large vessels of roots and stems due to better access to soil water. A high density of fine roots could thus enhance the drought tolerance of this species.

The observed differences in J s among the species result in considerable differences in whole-tree water flux even if all trees had similar trunk dimensions. This implies a heterogeneous pattern of soil water extraction and canopy transpiration in a mixed stand of the five tree species. Remarkably, the species with the greatest water flux per tree (F. sylvatica and A. pseudoplatanus) are known to be confined to oceanic/sub-oceanic climates with both species showing highest abundance in submontane and montane forest communities where soil drought is rare (Ellenberg 1996). In contrast, T. cordata, C. betulus and F. excelsior with smaller tree-level water fluxes are not restricted to sites with ample water supply but also occur in Central European sub-continental regions with regular or episodic summer drought. From our data on tree water consumption under comparable soil moisture conditions, one may speculate that inherently high water flux rates are one factor that could restrict the distribution of drought-sensitive tree species in regions with periodic water shortage. Other factors, that may also be involved in the limitation of drought-sensitive tree species, include a greater sensitivity to cavitation in the conducting pathway, reduced wood increment rates and a lesser leaf area, reduced fine root growth and/or increased fine root mortality, and elevated seedling and sapling mortalities.