Plant material
Fagus sylvatica seeds from Central Germany (Neuhaus, Solling) and South-east Poland (Lutowiska, Podkarpackie) were chosen for study in order to compare autochthonous provenances from the center and the eastern margin of the species’ distribution range (Fig. 1). Because we intended to simulate the natural situation in the stands, we decided not to pre-select the seeds of the two provenances for equal seed weight or seed quality but conducted a random selection of seeds in the two locations. The seeds originated from several tree individuals of each one stand per region (forestry district Neuhaus, No. 81009; forestry district Lutowiska). We focused on a single stand per region because the genetic diversity of F. sylvatica is typically higher within a given stand than the diversity between different stands. For example, in a sample covering six beech forests in Central Germany (Hesse), Sander et al. (2000) found 99% of the diversity within the stands and only 1% of the diversity between the stands. Climatic data of the two regions are given in Table 1.
Table 1 Climatic data of the marginal and central provenances (after Lorenc 2005 and Schipka 2002) Cultivation
For germination, the seeds were placed in regularly watered pots filled with loamy sand (Einheitserde B) in a climate chamber at 15/20°C (night/day), after weighing ten seeds per provenance for subsequent determination of the relative growth rate (RGR) (see below). On April 19, 2006, the seedlings were planted in the center of circular plastic containers (2 l) filled with a mixture of one part loamy sand, one part Perlite (Perligran G, Deutsche Perlite GmbH, Dortmund, Germany) and one part humus material (v:v:v). A commercial NPK-fertilizer (Triabon, COMPO GmbH & Co. KG, Münster, Germany; 16-8-12/N-P-K) was added.
The experiment took place in the Experimental Botanical Garden of the University of Göttingen between May 10, 2006, and September 21, 2006, under a mobile plexiglass roof equipped with a rain sensor, which automatically covered the plants when it rained. The roof was removed automatically a few minutes after the rain stopped. Thus, the beeches grew under local temperature and light conditions, but with complete control of soil water supply. To minimize potential influences of environmental gradients at the experimental site, the provenances and treatments were randomly positioned in alternating order and the positions were changed randomly four times during the experiment.
The pots were well-watered until the drought treatment (DT) was initiated after 14 weeks (July 25, 2006). In total, 36 plants per provenance were cultivated with each 12 plants treated with a different moisture regime, i.e., a control (40%), a moderate stress (20%), and a severe stress (10% soil water content) treatment. These soil moisture levels are roughly equivalent 5, 10 and 20 vol%. The limited volume of the pots made it necessary to add water every 2 days after water loss had been determined by weighing the pots.
Harvesting
At the end of the experiment (September 21, 2006), all leaves were removed from the stem, and the remaining shoot was cut off at the root collar after measuring shoot length and diameter of the stem and counting the number of leaves. All leaves were scanned with a flatbed graphics scanner, and the images were analyzed with the software WinFolia (WinFolia 2005b, Régent Instruments Inc., Québec, QC, Canada) to determine leaf area and calculate specific leaf area (SLA, in cm2 g−1 DM). The roots of the trees were harvested by carefully sifting the root-containing soil material of each pot through a sieve and washing the roots to clean them of soil residues. They were sorted by diameter (fine roots < 2 mm, coarse roots > 2 mm). The roots were spread out in water, scanned and the digitized images processed using the software WinRhizo (WinRhizo 2005c, Régent Instruments Inc., Québec, QC, Canada) which calculates the surface area of each root.
All plant organs were dried (70°C, approximately 80 h) and weighed. Specific root area (SRA, in cm2 g−1 DM), total fine root surface area, root dry weight and fine root/leaf area ratio were calculated from these data for each tree. The RGR (in g g−1 day−1) was calculated for the whole seedling by subtracting seed biomass from total harvested biomass and relating the difference to the duration of the experiment.
One day before the harvest, predawn water potential (Ψpre) of the leaves was measured at 4:00 a.m. using a Scholander pressure chamber (Scholander et al. 1965). The relative water content of the leaves (θ
l) harvested around noon was determined by drying (fresh weight − dry weight/fresh weight).
Chemical analyses
The dried plant material of each organ of a plant was pooled and ground. The leaf δ13C signature and N concentration were determined by mass spectroscopy (Delta Plus, Finnigan MAT, Bremen, Germany) in the Stable Isotope Laboratory (KOSI) of the University of Göttingen. For analyzing plant cation concentration, 100 mg of plant powder were digested with 3 ml HNO3 at 185°C for 5 h and the concentrations of Ca and K measured by atomic absorption spectrometry (AAS Vario 6, Analytic Jena, Jena, Germany).
Statistical analyses
All statistical analyses were performed with SAS Version 8.02 (SAS Institute Inc., Cary, NC, USA) and JMP (JMPIN Version 4.0.4, SAS Institute 2001). Significance was determined at P < 0.05 throughout. Before statistical analyses, all data were tested for normal distribution (Shapiro–Wilk test) and homogeneity of variances (Bartlett test). To achieve normal distribution and homogeneity of variances, the data of fine root biomass and leaf calcium concentration were logarithmically transformed. Two-way analyses of variance with the sources treatment, provenance and their interaction were performed by the ANOVA procedure for balanced data of the variables maximum shoot length, number of leaves per plant, seed weight, total biomass, leaf biomass, root/shoot ratio, RGR and leaf N concentration. In the case of unbalanced data (fine root biomass, SLA, leaf calcium concentration, leaf potassium concentration), general linear models were calculated. Differences between two treatments were analyzed with a Scheffé test, except for root/shoot ratio and RGR which were analyzed with a post hoc Tukey test.
For non-normally distributed data, the influences of provenance and treatment were investigated with a Kruskal–Wallis test (leaf water content, predawn leaf water potential, root collar diameter, shoot biomass, SRA, FR/LA ratio, δ13C, leaf magnesium concentration). Differences between two treatments were analyzed with a U-test after Mann and Whitney. A summary of the results of the different tests comparing the plant morphological, physiological and chemical variables between different DTs and different provenances is given in Table 2.
Table 2 Summary of results of three different statistical tests comparing various plant morphological, physiological and chemical variables between different drought treatments (DT) and different provenances (Pro)