Plant Soil (2009) 323:125–141 DOI 10.1007/s11104-009-9993-1

REGULAR ARTICLE

Analysing the role of soil properties, initial biomass and ozone on observed plant growth variability in a lysimeter study S. Gayler · C. Klier · C. W. Mueller · W. Weis · J. B. Winkler · E. Priesack

Received: 17 November 2008 / Accepted: 6 April 2009 / Published online: 13 May 2009 © Springer Science + Business Media B.V. 2009

Abstract This simulation study is based on a lysimeter experiment with juvenile beech trees (Fagus sylvatica L.) which were grown under ambient or doubled ambient atmospheric ozone concentrations. The aim of the study was to analyze the role of differences in soil properties, differences in initial biomass and ozone impacts on observed plant growth variability at the eight lysimeters of this experiment. For this purpose, we established a new simulation model based on the

Responsible Editor: Johan Six. S. Gayler (B) · C. Klier · E. Priesack Institute of Soil Ecology, Helmholtz Center Munich, German Research Center for Environmental Health, Ingolstädter Landstr. 1, 85764 Neuherberg, Germany e-mail: [email protected] J. B. Winkler Department of Environmental Engineering, Helmholtz Center Munich, German Research Center for Environmental Health, Ingolstädter Landstr. 1, 85764 Neuherberg, Germany C. W. Mueller Lehrstuhl für Bodenkunde, Technische Universität München, Am Hochanger 2, 85354 Freising, Germany W. Weis Fachgebiet Waldernährung und Wasserhaushalt, Technische Universität München, Am Hochanger 13, 85354 Freising, Germany

model system Expert-N by coupling soil water and nitrogen transport models with the plant growth model PLATHO, which was already tested and applied for juvenile beech. In order to parameterize the soil model, for all lysimeters soil hydraulic parameters as well as carbon and nitrogen stocks were measured. Simulation results reveal that the observed decreased growth rates under elevated ozone are due to ozone impacts on plant growth, whereas the high plant growth variability between lysimeters is to a major part the consequence of differences in soil hydraulic properties. Differences in initial biomass are of minor importance to explain plant growth variability in this experiment. Keywords Simulation model · O3 · Free air ozone fumigation · Water availability · Beech · Fagus sylvatica

Introduction The availability of water and mineral nitrogen at the right moment and in appropriate soil depth are crucial factors determining plant growth. Conversely, plants determine the upper boundary condition of soil water transport by transpiration and interception of precipitation, at least during the vegetation period. Due to the common difficulties in filling large scale lysimeters or cutting soil

126

monoliths and due to natural variability in sizes of even-aged beech trees, it is nearly impossible to establish identical growth conditions in an experiment with beech trees growing in different lysimeters as replicates. Consequently, it is not easy to distinguish in an experiment dealing with environmental effects on plant growth whether observed differences in plant growth are due to the experimental factors (e.g. elevated ozone concentrations) or if possible effects are covered by differences between the single lysimeters in soil water dynamics and nutrient availability as a consequence of variability in soil physical properties and soil organic matter quality. In addition, too high variability in initial biomass of investigated trees can hide real effects on biomass growth of experimental treatments, if observed effects are small compared to variance per treatment. An appropriate method to analyze the relevance of the different impact factors in such a study is to use a simulation model that is able to describe environmental impacts on plant growth as well as water and nutrient fluxes in the investigated system. The present study is based on a lysimeter experiment in which juvenile beech trees were fumigated for four vegetation periods with elevated atmospheric ozone concentrations (Schloter et al. 2005; Pritsch et al. 2008; Winkler et al. 2009b). A modelling approach was performed to investigate the question if the observed variability of tree growth between lysimeters and the decrease of growth rates at fumigated lysimeters was actually due to elevated ozone impacts. The modelling approach was also applied to analyze both the role of differences in soil properties and in initial biomass of trees on plant growth in this experiment. The needed simulation model of the soil-plantatmosphere system was established within the modelling system EXPERT-N (Priesack 2006) by coupling models of soil water and nitrogen transport, soil nitrogen turnover and soil heat flow with the plant growth model PLATHO (Gayler and Priesack 2005, 2007). As increased levels of the phytotoxic ozone may have detrimental effects on leaf physiology and plant growth (Dizengremel 2001; Matyssek et al. 2006), for the present study PLATHO had

Plant Soil (2009) 323:125–141

to be extended by a sub-model that considers these effects. O3 diffuses through the stomata into the leaf interior where it reacts in different ways with the leaf tissue. The impact of chronic ozone exposure on trees have been extensively studied at different scales from gene expression to total plant growth rates. Ozone can cause a reduction of transcripts of genes coding for proteins involved in primary metabolism and photosynthesis (Olbrich et al. 2009), it can induce localized cell death in seedlings (Nunn et al. 2005) or produce reactive oxygen species, which are the probable source for signal chains (Dizengremel 2001). Ozone decreases assimilation rates and increases respiration (Grams et al. 1999; Löw et al. 2007). It causes premature leaf senescence (Pritsch et al. 2008) and can shift resource partitioning between shoot and root (Dizengremel 2001). Decreased plant growth rates caused by ozone were observed as well for juvenile trees as in adult tree stands (King et al. 2005; Kozovits et al. 2005; Wipfler et al. 2005). A detailed mathematical model for the uptake and detoxification of ozone based upon the direct reaction of the pollutant with ascorbate localized within the apoplast is presented by Plöchl et al. (2000). This model considers the regeneration of dehydroascorbic acid in the cytosol, the rate of replenishment of cell wall ascorbate and the distribution of ascorbate between sub-cellular compartments. This approach describes primary effects of ozone at the leaf biochemical level, but it is difficult to extrapolate their consequences to the plant or stand level. Consequently, for the purpose of plant growth models simpler approaches are needed that address ozone effects on light use efficiency as well as assimilate partitioning and overall rates and costs for detoxification and repair of enzymes (van Oijen et al. 2004). Until now, only a few plant growth models consider impacts of ozone on plant growth (Weinstein et al. 1991; Chen et al. 1998; van Oijen et al. 2004; Deckmyn et al. 2007). Many studies have shown that leaf damage correlates better with the fluxes into the leaves than with exterior ozone dose (Musselman and Massman 1999). Therefore the first step in modelling ozone effects is the estimation of the effective influx into leaves. However, the most important source of uncertainty in stomatal ozone flux modelling is the stomatal conductance factor

Plant Soil (2009) 323:125–141

(Op de Beeck et al. 2007). Modelling approaches for the calculation of ozone influx into leaves are presented for example by Emberson et al. (2000) and Dewar (2002). Ozone impacts on plant growth are considered in plant growth models in different ways. Weinstein et al. (1991) simulate growth reductions of Red spruce (Picea rubens) under ozone if a critical level of exterior ozone concentration is exceeded. Chen et al. (1998) calculate needle mortality and decrease in photosynthetic capacity as a function of cumulative ozone dose for loblolly pine (Pinus taeda L.). Metabolic costs of detoxification and repair, which reduce the amount of assimilates that are available for partitioning to plant organs are considered in modelling approaches for potato (cv. Bintje) (Wolf and Oijen 2003), for spring wheat (Triticum aestivum L.) (van Oijen et al. 2004) and for adult beech trees (Fagus sylvatica L.) (Deckmyn et al. 2007). In these modelling approaches also reduction of photosynthetic capacity is considered, if metabolic costs for repair cannot be met by the amount of available assimilates (Deckmyn et al. 2007). For our modelling study, we adopted the approach presented by van Oijen et al. (2004) due to its simplicity in model parameterization. For the calculation of stomatal conductance we use the approach presented by Falge et al. (1996). In this paper we first give an overview over the applied simulation model focussing on the used approach to simulate ozone impacts on plant growth. We present the data needed for the parameterization of the soil model and for running the plant growth model for the lysimeter study and then compare simulated and measured percolation and evapotranspiration rates. Finally, we use the model to analyze the impacts of differences in soil properties between the lysimeters, of variability in initial biomass of trees and of the experimental factor, the ozone fumigation, on growth of trees. Experimental data and simulation model

127

of the Helmholtz-Zentrum München, Germany, is described in detail in Winkler et al. (2009b). The authors give also a description of the lysimeter station technique. Therefore we outline only the aspects of the experiment relevant for this simulation study. Eight lysimeters each with a surface of 1 m2 and a depth of 2 m, were filled in 1999 with soil (Haplic Cambisol dystric) from Höglwald, a forest site in South Germany. The lysimeters were filled according to the original horizons at the Höglwald site to obtain equal conditions in the lysimeters and were left untreated for three years. After this, three year old beech trees were planted with a density of four plants per lysimeter in November 2002. Additional trees were planted with the same density in between and around the lysimeters in order to get an homogeneous canopy. Since June of 2003, the plants on four lysimeters (N1, N2, N3 and N4) were fumigated during four vegetation periods with a doubled ozone concentration, compared to the ambient O3 concentration. Lysimeters S1, S2, S3 and S4 represent the control. At the begin of the last vegetation period in 2006, the trees in four of the lysimeters were inoculated with the root pathogen Phytophthora citricola (Fleischmann et al. 2009). This was not further considered in the simulation study, because this infection had no significant effects on tree growth rates in 2006 (Winkler et al. 2009a). Air temperature, precipitation, relative humidity and global radiation were measured at a nearby meteorological station and were recorded every 10 min, ozone concentrations at each lysimeter every 50 min. Percolation amounts were measured in 15 min intervals and mean values of lysimeter weight were recorded in hourly intervals. So evapotranspiration ET [mm] within a given time period could be calculated from measured amounts of precipitation P [mm], percolation L [mm] and the change of total water content ΔW [mm], which can be calculated from lysimeter weight (Winkler et al. 2009b) ETa = P − L − ΔW

(1)

Experimental design The setup of the ozone fumigation experiment which was carried out at the lysimeter station

Stem diameters were measured once a year at the stem basis. At the end of the experiment the trees were harvested completely and total

128

above-ground and below-ground tree biomass was determined Winkler et al. (2009a). The topmost 100 cm of each lysimeter were cut into five discs of 20 cm thickness as described in Seyfarth and Reth (2008) to allow below ground sampling.

Plant Soil (2009) 323:125–141

adjusted these parameters to fit daily measured evapotranspiration rates per lysimeter. Considering ozone impacts on plant growth A list of symbols used in the following presentation of PLATHO subroutines is given in Table 1.

Model overview The modelling system EXPERT-N (Priesack 2006) was used to simulate vertical onedimensional soil water flow, evapotranspiration, solute transport, turnover of soil carbon and nitrogen and plant growth. EXPERT-N consists of several modules for simulating these processes that can be coupled in various combinations to new simulation models. To get a complete description of water fluxes in the soil–plant– atmosphere system, we coupled the plant growth model PLATHO (Gayler and Priesack 2005) with the soil water transport modules following the approach of the model HYDRUS (Simunek et al. 1998), and with modules for nitrogen transport and turnover similar to the approaches of the models LEACHN (Hutson and Wagenet 1992) and SOILN (Johnsson et al. 1987). At the lower boundary of the soil column the seepage water condition was chosen. Run-off and snow processes were not considered. The simulation time step of the soil model is governed by the solver of the Richards equation for soil water transport and varies between 0.001 and 0.1 days, whereas the plant growth model is simulated with a constant time step of 0.1 days. For the calculation of the potential evapotranspiration, the Penman-Monteith dual crop coefficient method was used (Allen 2000; Loos et al. 2007). In this method, the potential evapotranspiration of a hypothetical grass reference crop predicted by the classical Penman-Monteith equation (Monteith 1965) is multiplied by a specific crop coefficient, which is the sum of a basal crop coefficient and the soil water evaporation coefficient. These crop factors can vary during different growth stages of the plants and determine not only potential evapotranspiration but also the partitioning of evapotranspiration between evaporation and transpiration. As crop factors for juvenile beech trees were not available from literature, we

Table 1 List of symbols used in the presented subroutines of PLATHO Symbol Unit

Description

Aav A˜ av

g(CH2 O) g(CH2 O)

AL Aold

m2 g(CH2 O)

Ddetox DM Drepair ETa FO3 ,eff H Ps R Vc,max

g(CH2 O) day−1 g(CH2 O) day−1 g(CH2 O) day−1 mm g(O3 ) day−1 m g(CH2 O) day−1 g(CH2 O) day−1 μmol m−2 s−1

V˜ c,max

μmol m−2 s−1

Wi

g

aL cdetox cO3

m2 m−1 g(CH2 O) g−1 (O3 ) g m−3

crepair

g(CH2 O) g−1 (Rubisco) mm – day−1 –

Available assimilates Reduced amount of assimilates Total plant leaf area Assimilate surplus from time step before Demand for detoxification Demand for maintenance Demand for repair Actual evapotranspiration Effective ozone influx rate Plant height Photosynthesis rate Reserves mobilization rate Maximum carboxylation velocity Decreased carboxylation velocity Biomass of compartment i (= roots, stem, . . . ) Leaf area density Costs of detoxification Atmospheric ozone concentration Cost of repair

dst fdetox flvs frepair gO3 h rgrw rrepair sR sR,crit sR,opt Δt ϕ

Stem diameter Detoxified fraction Partitioning rate to leaves Fraction of assimilates used in repair m day−1 Stomatal conductance for ozone m Height year−1 Biomass growth rate g(Rubisco) day−1 Repair rate of Rubisco g(Rubisco) m−2 Leaf Rubisco concentration g(Rubisco) m−2 Critical leaf Rubisco concentration g(Rubisco) m−2 Optimal leaf Rubisco concentration d Time step g(Rubisco) g−1 (O3 ) Ozone damage coefficient

Plant Soil (2009) 323:125–141

129

PLATHO simulates phenological development, photosynthesis, water and nitrogen uptake by roots, biomass growth, respiration and senescence. The gain and consumption of carbohydrates and nitrogen for growth- and maintenance-related processes are calculated. The amount of assimilates which are available for supplying the demand for growth, Aav [g(CH2 O) d−1 ] is calculated from the actual plant photosynthesis rate, Ps [g(CH2 O) d−1 ], potential reserve remobilization R [g(CH2 O) d−1 ] and the demand for maintenance processes, DM [g(CH2 O) d−1 ] during the actual time step Δt [d] and the surplus assimilates remaining from the time step before, Aold [g(CH2 O)]: Aav = (Ps + R − DM )Δt + Aold

internal nitrogen, or if actual growth rates of plant organs are limited by one of these factors. A detailed description of the procedure that simulates allocation of carbohydrates to the different compartments of the model is given in Gayler and Priesack (2005). For the present study, PLATHO was enhanced by the modelling approach developed by van Oijen et al. (2004) to simulate additional demands of assimilates for ozone detoxification, Ddetox [g(CH2 O) d−1 ], and repair of Rubisco, Drepair [g(CH2 O) d−1 ], and to simulate the impact of ozone on photosynthesis rate. Consequently Eq. 2 was changed to   A˜ av = Ps + R − DM − Ddetox − Drepair Δt + Aold

(2)

Ps is calculated by integrating the leaf gross photosynthesis per unit leaf area over plant height. The light distribution profile in the canopy is simulated by an enhancement of the method of Kropff and Laar (1993) accounting for shading by next neighbours (Gayler et al. 2006). Leaf internal CO2 concentration is calculated according to the approach of Falge et al. (1996), which was enhanced by a water shortage term that considers stomatal closure if actual transpiration is lower than potential transpiration. The response of leaf gross photosynthesis to irradiance and leaf internal CO2 concentration is calculated following the approach of von Caemmerer and Farquhar (1981). Glucose consumption for maintenance processes is assumed to be proportional to organ biomass and to depend on temperature following a Q10 relationship. In each simulation time step, the allocation of carbohydrates to the different biomass compartments of the model (fine and coarse roots, stem, branches and leaves) is simulated in two steps. First, the demands for carbohydrates and nitrogen of the single organs are estimated from the potential total plant growth rate taking into account a functional balance approach between leaves and fine roots, allometric relationships between the compartments and empirical seasonal factors, to consider the different phenological stages of the plant. In a second step, it is checked if these potential growth rates can be fulfilled by Aav and the availability of plant

(3) where A˜ av [g(CH2 O)] is the reduced amount of assimilates that are available for growth taking into account the consumption of assimilates due to ozone impacts. The rate of effective ozone uptake per simulation layer at height h [m] is calculated from the atmospheric ozone concentration cO3 [g m−3 ], the stomatal conductance for ozone in this layer gO3 (h) [m d−1 ], the leaf area density aL (h) [m2 m−1 ] in this layer and a factor fdetox [-] that describes the detoxified fraction of the ozone flux entering the stomata. The total effective ozone influx rate into the plant, FO3 ,eff [g(O3 ) d−1 ], is then calculated by integration of the influx rates over the plant height H [m]:  H FO3 ,eff = gO3 (h) aL (h) cO3 (1 − fdetox ) dh (4) 0

where stomatal conductance gO3 (h) is calculated following the method of Falge et al. (1996) from the net assimilation rates in the single leaf layers. In the model of van Oijen et al. (2004) it is assumed that the overall cost for the detoxification process is proportional to the detoxification rate itself:  H Ddetox = gO3 (h) aL (h) cO3 fdetox cdetox dh (5) 0

where Ddetox [g(CH2 O) d−1 ] is the amount of assimilates used for detoxification and cdetox [g(CH2 O) g−1 (detoxified O3 )] is the cost coefficient of detoxification.

130

Plant Soil (2009) 323:125–141

Assimilation rate per leaf area is not affected by ozone as long as the Rubisco content may be restored by repair processes. This is the case if rrepair > ϕ FO3

(6)

where rrepair [g(Rubisco) d−1 ] is the repair rate of Rubisco and ϕ [g(Rubisco) g−1 (O3 )] the ozone damage coefficient. Based on observations it is assumed that the repair rate is directly proportional to the amount of assimilates partitioned to leaves: rrepair = Aav flvs

frepair crepair

(7)

where Aav [g(CH2 O)] is the amount of carbohydrates that are available for growth and maintenance (Eq. 2), flvs [d−1 ] is the rate of carbohydrates partitioned to leaves, frepair [-] is the fraction of assimilates that is used in repair and crepair [g(CH2 O) g−1 (Rubisco repaired)] is the repair cost coefficient. Direct proportionality between repair rate and the amount of assimilates used for repair, Drepair [g(CH2 O) d−1 ], is assumed: Drepair = rrepair crepair

(8)

Thus, the concentration of Rubisco sR [g(Rubisco) m−2 (leaf)] in leaves changes with dsR 1 = dt AL

 sR,opt

dAL + rrepair − ϕ FO3 dt

 (9)

where AL [m−2 ] denotes the total plant leaf area and sR,opt [g(Rubisco) m−2 (leaf)] the optimal leaf Rubisco concentration. In our model we assume that maximal carboxylation velocity, Vc,max [μmol m−2 s−1 ] is decreased as soon as the Rubisco concentration falls below a critical level sR,crit [g(Rubisco) m−2 (leaf)]   sR ˜ Vc,max = Vc,max min 1, sR,crit

(10)

Model input data PLATHO was already parameterized and tested for simulating growth of juvenile beech trees in previous studies (Gayler et al. 2006, 2008). The

parameters needed for the new ozone module impact were adopted from van Oijen et al. (2004), who estimate fdetox = 0.9 [–], frepair = 0.05 [–], cdetox = 0.375 [g(CH2 O) g−1 (O3 )] and crepair = 18 [g(CH2 O) g−1 (Rubisco)]. For ϕ van Oijen et al. estimate a value between 21 and 34 [g(Rubisco) g−1 (O3 )] but use a value of 10 in their simulation study. For lack of own measurements we decided in this study to use a mean value of 24. From measured leaf nitrogen concentration and the assumption that 50% of leaf N are Rubisco-N, sR,opt was estimated to be 0.035 [g(Rubisco) m−2 (leaf)]. sR,crit was set to 0.75 sR,opt according to measured relation between leaf Rubisco content and photosynthesis rate as described in Müller et al. (2005). For running the whole model in the present study, we needed additional data for soil model parameterization as well as start values of biomass compartments for the plant growth model. Data for soil model parameterization Putative differences in soil physical and soil chemical properties between the single lysimeters were reassessed after finishing the experiment by measuring the following parameters in different soil layers of each lysimeters down to a soil depth of 1 meter: soil bulk densities, water retention and unsaturated hydraulic conductivity curves, saturated hydraulic conductivity as well as total carbon and nitrogen stocks. Undisturbed soil cores (68.2 cm3 ) were taken in triplicate for bulk density determination at all eight lysimeters at 0–5, 10–15, 20–25, 30–35, 45– 50, 65–70 and 90–95 cm soil depths. The samples were dried at 105◦ C for 3 days and the density was calculated from the weight of the dried samples. The C and N contents were measured in duplicate by dry combustion (Elementar, vario MAX CNS Analyzer). Since all samples were free of carbonates, the measured C concentrations equal the organic C concentrations. The C and N stocks were calculated by multiplying the C and N contents with the bulk density. Nitrogen influx from atmosphere at the lysimeter facility was assumed to be 4.2 [g(N) m−2 y−1 ] which is similar to the influx rate at the original site Höglwald (Rennenberg et al. 1998).

Plant Soil (2009) 323:125–141

Litter degradation rates were deduced from a litterbag experiment that was conducted in the same experiment. Litter degradation was not significantly influenced by the ozone treatment (Pritsch et al. 2008). To simulate water transport is simulated by the Richards equation, we use hydraulic functions according to the van Genuchten-Mualem approach (van Genuchten 1980). For the determination of the van Genuchten parameters α and n and the unsaturated hydraulic conductivity curve, 16 undisturbed soil cores of 10 cm height and a diameter of 15 cm were dug from two depth of all lysimeters and sealed at the bottom in the laboratory. The cores were completely saturated with water and left to desiccate at room temperature, until both miniature pressure-transducer tensiometers (T5, UMS GmbH, Munich, Germany) lost the hydraulic connection between ceramic cup and pressure transducer. Gravimetric water content was measured with a ECH2O-probe (ECH2O EC10, Decagon Devices, Inc., USA). In addition to the unsaturated hydraulic conductivity, the hydraulic conductivity at saturation was estimated by means of the constant head method as described in Klute and Dirksen (1986). Start values for running the plant growth model For running the plant growth model PLATHO, start values of biomass compartments are needed. These start values were estimated from stem diameters, which were measured at the beginning of the experiment, by means of allometric relations, a common method to estimate biomass of tree components from stem diameters (Zianis et al. 2005). These allometric relations were deduced from a data set that comes from different other experiments with juvenile beech trees (Fleischmann et al. 2002, 2005; Koch 2006; Kozovits et al. 2005; Luedemann et al. 2005). High correlations were found between stem diameter dSt [mm] and aboveground woody biomass WSt+Br [g] (Eq. 11, r2 = 0.93, n = 451) as well as between WSt+Br and total root biomass WR [g] (Eq. 12, r2 = 0.94, n = 438). Partitioning of WSt+Br between stem biomass WSt [g] and branch biomass WBr [g] (Eq. 13, r2 = 0.95, n = 244) as well as partitioning of WRoot between fine root biomass

131

WFR [g] and coarse root biomass WCR [g] (r2 = 0.93, n = 364) were also deduced from the same data set. WSt+Br = 0.0152 dSt 3.058

(11)

WRoot = 1.796 WSt+Br 0.774

(12)

WSt = 0.59 WSt+Br ,

WBr = 0.41 WSt+Br

(13)

WCR = 0.80 WRoot ,

WFR = 0.20 WRoot

(14)

Biomass growth rates rgrw [y−1 ] of the trees were calculated from the estimated initial stem+branches biomass, Wini [g], (Eq. 11), and stem+branches biomass that was measured at the final harvest, Wend [g]: rgrw = (log Wend − log Wini )/ΔT

(15)

were ΔT [y] represents the time period of the experiment.

Simulation scenarios To analyze the role of soil properties, initial biomass and ozone on simulated plant growth, we defined eight scenarios considering these factors in various combinations. We applied our simulation model to see the consequences of these scenarios. In naming the simulation scenarios, a ‘+’ denotes inclusion of a putative impact factor into the simulation, whereas a ‘−’ denotes neglection of this factor. ‘S’ denotes ‘soil properties effect’, ‘I’ the ‘initial biomass effect’ and ‘O’ the ‘elevated ozone effect. Thus, for example ‘H(−S+I+O)’ means that differences in initial biomass and effect of elevated ozone are considered, but differences in soil properties are neglected; or ‘H(+S−I−O)’ means that only differences in soil properties are considered.

Statistical criteria Two statistical criteria were used to assess the adequateness of plant growth simulations. The

132

Plant Soil (2009) 323:125–141

root mean square error RMSE as defined by Wallach and Goffinet (1989) was used to assess over the whole period of the experiment for each lysimeter the deviation between predicted stem diameters, Pi , and observed stem diameters, Oi , ¯ proportioned against the mean observed value O:

Results Plant growth

where Pj and Oj are the predicted and observed ¯ values of aboveground woody biomass, and O ¯ and P are the corresponding mean values over all lysimeters. A value < 0 of ME indicates that ¯ would be a better estimation for observed Pj = O variability than the simulation results. If simulated values are completely in accordance with observed values, ME equals 1.

The growth of beech trees in this experiment is discussed in detail by Winkler et al. (2009a). We therefore present only the results relevant for this study. At the start of the experiment, stem diameters of plants at fumigated lysimeters were significantly higher compared to stem diameters at control lysimeters ( p = 0.006 at individual plant level). During the time course of the experiment, this advantage was more and more lost until after three vegetation periods mean stem diameters of trees exposed to elevated ozone equals mean stem diameters of control trees (Winkler et al. 2009a). Also after four vegetation periods, no significant differences of stem diameters as well as of plant biomasses were found. However, mean stem diameter growth rates of control trees were significantly higher than mean growth rates of fumigated trees (0.36 ± 0.05 [y−1 ] vs. 0.31 ± 0.05 [y−1 ], p = 0.008, n = 16). The correlation between initial stem diameters and stem diameters at the respective time point of observation decreased during the experiment to a coefficient of determination of r2 = 0.18. This low value shows that initial plant sizes had only a minor part on the observed plant growth variability. Mean values per lysimeter of stem diameters [millimeter] at the beginning of the experiment as well as stem diameters and dry mass [g tree−1 ] of biomass compartments at the end of the experiment are shown in Table 2. A high variability of tree

Table 2 Mean values ± standard deviations per lysimeter of stem diameters dst,i at the beginning of the experiment (23 April 2003) and dst,e at the end of the exper-

iment (8 August 2006); below ground biomass, WRoot , stem+branches biomass, WSt+Br and leaf biomass, WLv , at final harvest (22–23 August 2006)

1 RMSE = ¯ O

n i=1

(Oi − Pi )2 n

(16)

where n is the number of observations during the experiment times the number of lysimeters (n = 40). The lower the value of RMSE, the better is the agreement between simulation and observation. The Modelling efficiency ME (Willmott 1982) was used to assess the variability in plant biomass at the time point of final harvest between all lysimeters. 2 Oj − Pj ME = 1 −   

8  8 ¯ 2 ¯ 2 j=1 Oj − O + j=1 Pj − P

8



j=1

(17)

Lysimeter

dst,i

dst,e

S1 S2 S3 S4 N1 N2 N3 N4

7.5 ± 1.7 9.1 ± 1.4 8.0 ± 0.6 7.2 ± 0.2 9.2 ± 1.2 9.0 ± 0.6 8.7 ± 0.9 9.3 ± 1.6

WRoot

WSt+Br

WLv

300 ± 84 370 ± 54 341 ± 82 563 ± 58 321 ± 122 360 ± 164 372 ± 64 437 ± 257

83.8 ± 4.2 116.8 ± 7.5 85.1 ± 5.9 82.7 ± 11.7 86.6 ± 19.2 90.9 ± 5.0 77.4 ± 14.6 75.0 ± 8.7

g tree−1

mm 24.8 ± 2.5 27.6 ± 2.6 24.2 ± 1.2 28.6 ± 1.6 25.2 ± 3.4 24.3 ± 4.7 25.5 ± 1.1 27.4 ± 6.1

166 ± 35 222 ± 49 210 ± 34 291 ± 29 165 ± 34 183 ± 69 196 ± 23 264 ± 113

S1, S2, S3 and S4: ambient ozone; N1, N2, N3 and N4: elevated ozone

Plant Soil (2009) 323:125–141

133

biomass was found both within the group of control lysimeters (S1–S4) and within the group of fumigated lysimeters (N1–N4). Data for soil model parameterization Measured soil parameters reflect the situation within the lysimeters at the end of the experiment. Measuring these parameters during the time course of the experiment was not possible, because the sampling would have influenced soil water dynamics and plant growth. In our simulations, we assume that soil parameters were constant for the period of the experiment. Figure 1 shows soil bulk densities in different soil depths for the single lysimeters together with the soil bulk densities which were measured at the original site at Höglwald (Kreutzer and Bittersohl 1986). The slight variability that was found between the single lysimeters reveals no significant differences between fumigated and control lysimeters. A similar picture was found for total carbon and nitrogen stocks in different soil depths (Figs. 2 and 3). The values vary slightly between the single lysimeters

Fig. 1 Vertical soil bulk density profiles in single lysimeters and at the original site Höglwald. Höglwald values after Kreutzer and Bittersohl (1986)

Fig. 2 Vertical profiles of total carbon stocks in lysimeters measured at final harvest of the experiment. Values are normalized to soil layers of 1 cm thickness

Fig. 3 Vertical profiles of total nitrogen stocks in lysimeters measured at final harvest of the experiment. Values are normalized to soil layers of 1 cm thickness

134

Plant Soil (2009) 323:125–141

but do not significantly differ between ozone treatment and control trees within the single soil horizons. However, pronounced differences in soil hydraulic parameters were found between the lysimeters at the time point of the final harvest of the lysimeters. The samples for the determination of soil hydraulic parameters covered a depth gradient (0–5, 5–15, 25–35, 45–55, 65–75, 85–95 cm) with two samples taken from different lysimeters at each depth. At depth 5–15 cm samples were taken from six lysimeters (lysimeters S1, S2, S4, N1, N2, N3), as this horizon showed the highest root densities and thus impacts of soil physical properties on transpiration were supposed to be most significant. The most pronounced deviations from lysimeter means were found for lysimeter S4, with the highest saturation water content and lysimeter N1 with a hydraulic conductivity curve that differs considerably from curves of the other lysimeters. In Table 3 the soil hydraulic parameters of this soil horizon are presented for each lysimeter and for the original site Höglwald, where θres as well as van Genuchten α and n were fitted to measured water retention curves. The data for Höglwald were estimated from data presented by Kreutzer and Bittersohl (1986). For our simulations, the missing values for lysimeters S3 and N4 in this horizon were replaced by the

Table 3 Hydraulic parameters of the horizon from 5 to 30 cm soil depths in each lysimeter and at the original site Höglwald: saturated volumetric water content θsat , residual water content θres , saturated hydraulic conductivity Ksat , van Genuchten α and n. Höglwald values after Kreutzer and Bittersohl (1986) Lysimeter

θsat –

θres –

α cm−1

n –

Ksat cm day−1

S1 S2 S3a S4 N1 N2 N3 N4a Höglwald

0.55 0.46 0.51 0.53 0.46 0.53 0.46 0.51 0.49

0.00 0.05 0.00 0.00 0.00 0.05 0.00 0.00 0.00

0.347 0.060 0.116 0.029 0.106 0.017 0.301 0.116 0.125

1.103 1.088 1.135 1.132 1.102 1.229 1.079 1.135 1.105

29.5 64.0 51.9 24.9 24.6 80.3 98.4 51.9 –

a No values were measured for this horizon at these lysimeters. Mean values of other lysimeters were used for simulations

mean of the values that were measured at the other lysimeters. Simulation of soil water dynamics Simulated percolation, evapotranspiration and changes in water storage were compared with measured data for all lysimeters. Simulation results show that the model is able to reproduce the dates and the size of most of the percolations peaks quite well. Figure 4 shows for lysimeter S4 simulated and measured daily evapotranspiration rates, changes in lysimeter weight (converted to millimeter water) and percolation amounts, together with daily precipitation amounts. Evapotranspiration rates were calculated as described in Eq. 1 from precipitation, percolation and lysimeter weight measurements. Similar peaks of percolation amounts could also be identified for all the other lysimeters, both, in measured and in simulated percolation rates. Only small differences in the dates as well as height and width of the peaks were found. Changes in soil water storage are met by the model quite well for the main part of the simulation period. However, the strong depletion of the soil in July 2006 is not adequately described by the model and consequently the model fails in simulating the extreme high measured transpiration rates in this month. In the model water uptake of plants from the loam horizons (5–30 cm and 30–90 cm soil depth) is underestimated, if water tension reaches values greater than 103 hPa. Extrapolation of the water retention curve to the permanent wilting point, where modelled water uptake is completely stopped, results in a volumetric water content of still 22%. Apparently, this value is overestimated. Evapotranspiration simulations fit daily measurements not as good as percolation simulations and simulations of changes in water storage. Plant growth simulation Plant growth was simulated for each lysimeter. The impact of elevated ozone concentrations on plant growth was estimated from effective ozone influx into plants, which depends directly on stomatal conductance in the applied modelling

Plant Soil (2009) 323:125–141

135

Fig. 4 Daily precipitation rates, P, together with simulated (solid lines) and measured (grey bars) daily evapotranspiration ET, changes in water storage ΔW and percolation, L, of lysimeter S4

approach according to van Oijen et al. (2004). Figure 5 shows simulated effective daily ozone influx per plant together with simulated mean values of stomatal conductance to water vapour and daily means of atmospheric ozone concentrations for lysimeter S4 (ambient ozone) and lysimeter N4 (elevated ozone). The model reacts as expected on climatic variations and soil water availability. The high stomatal conductance in the mid of 2006 correlates to the high water uptake rates in this period (Fig. 4). Simulated stomatal conductance to water vapor is hardly affected by the ozone treatment over a wide part of the simulation period. However, at the beginning of the vegetation period 2005 a markedly higher stomatal conductance is simulated for trees under elevated ozone due to a higher soil water availability in this lysimeter.

Based on the simulation of enhanced cost for ozone detoxification and repair of Rubisco under elevated ozone and by considering the differences of soil hydraulic properties and initial plant biomass, the growth model is able to represent the different growth conditions at the lysimeters with a sufficient accuracy. Simulated mean values per lysimeter of stem diameters [mm] and biomass [g tree−1 ] compartments at final harvest are presented in Table 4. Comparison with Table 2 shows good correlations between simulated and measured root biomass (r2 = 0.54) as well as between simulated and measured stem+branches biomass (r2 = 0.60), whereas simulation of leaf biomass correlates with measurements only with r2 = 0.24. A considerable part of the observed variability in plant growth and of the growth rate decrease at lysimeters N1–N4

136

Plant Soil (2009) 323:125–141

Fig. 5 Course of A daily means of atmospheric ozone concentration, B simulated daytime means of stomatal conductance to water vapor and C simulated effective ozone

influx into plants at lysimeter S4 (ambient ozone—solid lines, black bars) and at lysimeter N4 (doubled ozone— dotted lines, grey bars)

can be explained by the model. This is pointed out in Table 5, where measured and simulated mean values per ozone treatment of above-ground woody biomass, stem diameters and mean annual biomass growth rates were compared for this scenario. The model meets measured mean values well, however simulated standard deviations are much lower than in measurements. This indicates that the pronounced growth variability that was

observed between the single lysimeters is not completely explained by the model. Starting from this best simulation run, we analyzed our model to assess the role of soil properties, initial biomass and ozone on observed tree growth variability and growth rate decrease at lysimeters N1–N4. For this purpose, we applied our simulation model to see the consequences of the eight hypotheses considering these factors in various combinations as described above. The simulated effect on mean growth rates per ozone treatment, Δrgrw , as well as values of the root mean square error RMSE and the modelling efficiency ME for all scenarios are presented in Table 6 where Δrgrw is the percentaged deviation

Table 4 Simulated mean values per lysimeter of stem diameters (8 August 2006), dst,e , below ground biomass, WRoot , stem+branches biomass, WSt+Br and leaf biomass, WLv , (22–23 August 2006) Lysimeter S1 S2 S3 S4 N1 N2 N3 N4

dst,e

WRoot

mm

g tree−1

25.7 26.1 26.1 27.5 25.0 25.4 26.1 27.1

188 200 197 218 167 180 184 201

WSt+Br

WLv

362 385 380 446 323 351 371 416

84 82 82 78 80 77 80 77

S1, S2, S3 and S4: ambient ozone; N1, N2, N3 and N4: elevated ozone

Table 5 Measured and simulated mean values per ozone treatment of stem diameters (8 August 2006), dst,e , stem+branches biomass, WSt+Br (22–23 August 2006), and biomass growth rates, rgrw Measured amb. O3

Simulated elev. O3

amb. O3

elev. O3

WSt+Br (g) 394±116 372±48 393±36 365 ± 39 dst,e (mm) 26.3±2.1 25.6±4.0 26.9±0.8 26.3±1.0 rgrw (year−1 ) 1.02±0.08 0.92±0.04 1.03±0.05 0.91±0.04

Plant Soil (2009) 323:125–141

137

Table 6 Simulated mean growth rate decrease of trees at lysimeter N1–N4 compared to Lysimeters S1–S4, Δrgrw , as well as root mean square errors RMSE and modelling efficiencies ME of the different scenarios H(+S+I+O) H(+S+I−O) H(+S−I+O) H(+S−I−O) H(−S−I+O) H(−S+I+O) H(−S+I−O) H(−S−I−O)

Δrgrw (%)

RMSE

ME

−11.5 −2.0 −9.2 +0.5 −10.6 −12.5 −2.2 ±0

0.13 0.15 0.13 0.15 0.15 0.15 0.17 0.15

0.66 0.50 0.56 0.51 0.13 0.13 −0.40 −0.26

between the mean of growth rates at lysimeters S1–S4 and lysimeters N1–N4. The starting configuration of the model, H(+S+I+O), reached values for ME of 0.66 and for RMSE of 0.13. This indicates that 66% of the observed variability between all lysimeters is explained by the model and that measured stem diameters are missed by simulated values for 13% on average. From the scenarios in which only one impact factor is considered, H(+S−I−O), H(−S+I−O) and H(−S−I+O), it can be seen, that the proportion of soil properties in simulated growth variability dominates the other factors (ME = 0.51 compared to ME = 0.13 and ME = −0.40). On the other hand, the major part of the decrease in growth rates at lysimeters N1–4 is explained by considering additional ozone impacts under elevated ozone. Changing from ‘−O’ to ‘+O’ in one of the scenarios lowers always simulated growth rates at these lysimeters, irrespective of the consideration of other factors. The effect of ‘+S’ on growth rates is ambiguous, whereas changing from ‘−I’ to ‘+I’ decreases growth rates at lysimeters N1–4 to a smaller extend than the ozone effect. The simulation of hypothesis H(−S+I−O) reduces rgrw at lysimeters N1–4 to ca. 2%, but produces a negative ME Thus, a small part of the reduced growth rates at these lysimeters is explained by initial biomass effect, however nothing of the variability between the lysimeters, indicated by the negative value of ME. RMSE is hardly affected by the different scenarios. Best values are reached for H(+S+I+O) and H(+S−I+O). The highest value was simulated for the scenario in which only differences in

initial tree biomass was considered and for which also the worst ME was calculated.

Discussion A simulation model was applied to investigate growth variability in a lysimeter study with juvenile beech trees which were grown for 4 vegetation periods under ambient (control, lysimeters S1–S4) or doubled ambient atmospheric ozone concentrations (lysimeters N1–N4). In this experiment trees at lysimeters N1–N4 started to grow with significant higher stem diameters compared to control trees but had significantly decreased growth rates over the period of the experiment. However, at the time point of final harvest of trees, no significant difference of mean biomass was measured, due to a high variability of biomass per lysimeter. Besides the differences in initial tree biomass and ozone impacts, differences in soil properties were conceived to be putatively responsible for observed growth variability and decreased growth rates at lysimeters N1–4. To investigate the role of the latter factor, a considerable effort was put into soil model parameterization. At several soil depths soil bulk densities, carbon and nitrogen stocks, water retention curves and soil hydraulic conductivity curves were measured after finishing the experiment. No significant differences between elevated ozone lysimeters and control lysimeters were found for bulk densities. The high agreement between the single lysimeters and the original soil profile at the Höglwald site (Kreutzer and Bittersohl 1986) indicates how carefully the lysimeter filling procedure was done. Also carbon and nitrogen stocks showed no significant ozone impact, which is supported by the more detailed study of soil organic matter distribution in the same experiment by Mueller et al. (2009). In a lysimeter refilling procedure it is more difficult to reproduce water retention functions and hydraulic conductance of the soil horizons as to reproduce vertical soil density profiles. This can be seen as a disadvantage of the soil refilling procedure compared to the method where soil monoliths were cut. However, if eight monoliths would be cut at a forest site, the variability

138

in hydraulic properties between the lysimeters would probably be even higher, because of the typical high soil heterogeneity in forest soils compared to agricultural soils where ploughing levels out heterogeneities of the top soil. Additional to the difficulties that are inherent to a filling procedure, root growth and dying of coarse roots can influence the hydraulic parameters due to preferential flow paths. Consequently some markedly differences in water retention curves and soil hydraulic conductivity curves of the lysimeters were present at the end of the experiment. Simulations of soil water balances show that the measured soil hydraulic parameters provide an adequate soil model parameterization. Evapotranspiration simulations fit daily measurements not as well as percolation simulations or simulations of water storage changes. At least for the period until mid of 2006 we explain this by the uncertainty, that is inherent to daily and direct evapotranspiration measurement by water balance calculation based on Eq. 1. The principal difficulties of these measurement are discussed by Loos et al. (2007). Water balance estimations at high time resolution are difficult, because lysimeter weight measurements are often affected by spatial variability of precipitation or wind pressure. As most of the peaks are reproduced by the model and also differences in simulated percolation amounts occur between single lysimeters in accordance with observations, we conclude that the parameterization of the soil water model is suitable to describe soil water dynamics in the lysimeters. Consequently, we can use our model to investigate possible effects of differences in soil water dynamics on plant growth. In our simulation model only indirect impacts of ozone on stomatal aperture are considered, as far as enhanced effective ozone influx FO3 ,eff results in a decreased Vc,max and, following the Falge approach (Falge et al. 1996), in decreased stomatal conductance values. Nevertheless, plausible interactions of stomatal conductance with climatic factors were simulated. For example in the second half of the exceptional warm and dry year 2003 (Winkler et al. 2009b), effective ozone influx is hardly affected by high atmospheric ozone concentrations due to stomatal closure in consequence of soil water shortage.

Plant Soil (2009) 323:125–141

The model postulates a clear feedback mechanism between ozone impact on plant growth and soil water budget (see Fig. 5). Higher growth rates of control trees in 2004 and at the beginning of 2005 result in an enhanced soil water depletion and hence in stomata closure as a consequence of water shortage. For a more detailed modelling approach of ozone influx and ozone detoxification, which simulates also direct ozone impacts on stomatal aperture, knowledge about biochemical reactions between ozone and relevant leaf metabolites as well as knowledge about environmental impacts on these metabolites are required. A feasible starting-point for such a modelling approach could be coupling the detoxification model based on leaf ascorbate of Plöchl et al. (2000) with the model of stomata regulation by abscisic acid according to Dewar (2002). However, neither ascorbate nor abscisic acid were measured in the present lysimeter study. Starting point of the impact analysis was a simulation run, which considers all the available information about soil properties of the lysimeters, about initial biomass of trees, and includes the ozone impact model by van Oijen et al. (2004). It was tested how model performance changes if one, two or all of these three factors was omitted from the model configuration. The simulation results suggest that the slight differences between lysimeters in soil hydraulic properties are responsible for the main part of the observed variability of tree biomass at the end of the experiment, but do not contribute to the decreased growth rates at the fumigated lysimeters N1–N4. The large effect of small differences in soil properties demonstrates, how important it is during the juvenile phase of tree growth that water and nutrients are available at the right moment and in the right soil depth, were fine roots are present. Simulations also show that the higher initial tree biomass at lysimeters N1–N4 explains a small part of the lower growth rates at these lysimeters. This is in accordance with Schwinning (1996) who show that growth rates of plants are negatively correlated to initial plant biomass under not too high competition pressure. However, the major part of the significant decrease of growth rates at lysimeters N1–N4 was explained by simulated ozone impacts on carboxylation rate as well

Plant Soil (2009) 323:125–141

as by additional carbohydrate consumption for ozone detoxification and repair of Rubisco under elevated ozone concentrations. Conclusions In this study, it was demonstrated how superposing impacts on plant growth in a lysimeter study could be separated using a simulation model. By the model it was possible to demonstrate that putative ozone effects in this experiment were neither covered by the high tree growth variability in this experiment nor by variability in initial plant biomass. Simulation results revealed that the observed decreased growth rates under elevated ozone are due to ozone impacts on plant growth, whereas the high plant growth variability between lysimeters is to a major part the consequence of differences in soil hydraulic properties. Differences in initial biomass are of minor importance to explain plant growth variability in this experiment. The model could explain 66% of the observed plant growth variability. Thus, apparently additional factors that are not considered in the simulation model are also responsible for unequal plant growth in the lysimeters. The most likely factors are genetic variability between beech individuals and oversimplifications in plant growth modelling and modelling of water flows and nitrogen cycle. Acknowledgements We are grateful to the Deutsche Forschungsgemeinschaft which funded this study within the frame of Sonderforschungsbereich 607 Growth and Parasite Defence—Competition for Resources in Economic Plants from Forestry and Agronomy. We further thank Dr. Thorsten Grams for providing above ground biomass and stem diameter data, Wolfgang Graf, Hans Lang and Oliver Gefke for providing climate and soil water balance data as well as Dr. Frank Fleischmann, Dr. Felix Haesler, Dr. Jürgen Esperschütz, Gunda Stoelken, Gudrun Hufnagel, Monika Kugelmann, Wolfgang Eigner, Marcus Lehmann, Josef Heckmair and Tina Schmidt for digging soil cores and roots out of approximately 30,000 kg loam.

References Allen RG (2000) Using the FAO-56 dual crop coefficient method over an irrigated region as part of an evapotranspiration intercomparison study. J Hydrol 229: 27–41

139 Chen CW, Tsai WT, Lucier AA (1998) A model of airtree-soil system for ozone impact analysis. Ecol Model 111:207–222 Deckmyn G, Op de Beeck M, Löw M, Then C, Verbeeck H, Wipfler P, Ceulemans R (2007) Modelling ozone effects on adult beech trees through simulation of defence, damage, and repair costs: implementation of the CASIROZ ozone model in the ANAFORE forest model. Plant Biol 9:320–330 Dewar RC (2002) The Ball-Berry-Leuning and TardieuDavies stomatal models: synthesis and extension within a spatially aggregated picture of guard cell function. Plant Cell Environ 25(11):1383–1398 Dizengremel P (2001) Effects of ozone on the carbon metabolism of forest trees. Plant Physiol Biochem 39(9):729–742 Emberson LD, Ashmore MR, Cambridge HM, Simpson D, Tuovinen JP (2000) Modelling stomatal ozone flux across Europe. Environ Pollut 109(3):403–413 Falge E, Graber W, Siegwolf R, Tenhunen JD (1996) A model of the gas exchange response of Picea abies to habitat conditions. Trees 10:277–287 Fleischmann F, Schneider D, Matyssek R, Oßwald WF (2002) Investigations on net CO2 assimilation, transpiration and root growth of Fagus sylvatica infested with four different Phytophthora species. Plant Biol 4(2):144–152 Fleischmann F, Koehl J, Portz R, Beltrame AB, Oßwald W (2005) Physiological changes of Fagus sylvatica seedlings infected with Phytophthora citricola and the contribution of its elicitin Citricolin to pathogenesis. Plant Biol 7:650–658 Fleischmann F, Winkler JB, Oßwald W (2009) Effects of ozone and Phytophthora citricola root infection on carbon partitioning in European beech (Fagus sylvatica). Plant Soil (this issue). doi:10.1007/s11104-009-9927-y Gayler S, Priesack E (2005) PLATHO-documentation. http://www.sfb607.de/english/projects/c2/platho.pdf Gayler S, Priesack E (2007) PLATHO - a dynamic plant growth model considering competition between individuals and allocation to carbon-based secondary compounds. In: Fourcaud T, Zhang X (eds) Proceedings of PMA06—plant growth modeling and applications, Beijing, China, 13–17 November 2006. IEEE Comp Soc, Los Alamitos, California, pp 85–92 Gayler S, Grams TEE, Kozovits A, Luedemann G, Winkler JB, Priesack E (2006) Analysis of competition effects in mono- and mixed cultures of juvenile beech and spruce by means of the plant growth simulation model PLATHO. Plant Biol 8(4):503–514 Gayler S, Grams TEE, Heller W, Treutter D, Priesack E (2008) A dynamical model of environmental effects on allocation to carbon-based secondary compounds in juvenile trees. Ann Bot 101(8):1089–1098 Grams TEE, Anegg S, Häberle KH, Langebartels C, Matyssek R (1999) Interactions of chronic exposure to elevated CO2 and O3 levels in the photosynthetic light and dark reactions of European Beech (Fagus sylvatica). New Phytol 144(1):95–107 Hutson JL, Wagenet RJ (1992) LEACHM: leaching estimation and chemistry model: a process-based model

140 of water and solute movement, transformations, plant uptake and chemical reactions in the unsaturated zone. Version 3.0. Research Series No. 93-3. Cornell University, Ithaca, NY Johnsson H, Bergström L, Jansson PE, Paustian K (1987) Simulated nitrogen dynamics and losses in a layered agricultural soil. Agric Ecosyst Environ 18:333–356 King JS, Kubiske ME, Pregitzer KS, Hendrey GR, McDonald EP, Giardina CP, Quinn VS, Karnosky DF (2005) Tropospheric O3 compromises net primary production in young stands of trembling aspen, paper birch and sugar maple in response to elevated atmospheric CO2 . New Phytol 168(3):623–636 Klute A, Dirksen C (1986) Hydraulic conductivity and diffusivity: laboratory methods. In: Klute A (ed) Methods of soil analysis part I: physical and mineralogical methods. Agronomy 9, A. Soc. Of Agronomy, Inc., Soil Sci. Soc. Am., Inc., Madison, Wisconsin USA, pp 687–733 Koch N (2006) Scaling within ontogenetic stages and growth scenarios of beech (Fagus sylvatica) and spruce (Picea abies) under ozone exposure. Thesis (in German). http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb: 91-diss20060210-1615467692. TU München, Fakultät Wissenschaftszentrum Weihenstephan Kozovits AR, Matyssek R, Blaschke H, Göttlein A, Grams TEE (2005) Competition increasingly dominates the responsiveness of juvenile beech and spruce to elevated CO2 and O3 levels throughout two subsequent growing seasons. Glob Chang Biol 11:1387–1401 Kreutzer K, Bittersohl J (1986) Untersuchungen über die Auswirkungen des sauren Regens und der kompensatorischen Kalkung im Wald. Forstw Cbl 105: 273–282 Kropff MJ, van Laar HH (eds) (1993) Modelling cropweed interactions. CAB International, Wallingford, UK Loos C, Gayler S, Priesack E (2007) Assessment of water balance simulations for large-scale weighing lysimeters. J Hydrol 335:259–270 Löw M, Häberle KH, Warren CR, Matyssek R (2007) O3 flux-related responsiveness of photosynthesis, respiration, and stomatal conductance of adult Fagus sylvatica to experimentally enhanced free-air O3 exposure. Plant Biol 9:197–206 Luedemann G, Matyssek R, Fleischmann F, Grams TEE (2005) Acclimation to ozone affects, host/pathogen interaction, and competitiveness for nitrogen in juvenile Fagus sylvatica and Picea abies trees infested with Phytophthora citricola. Plant Biol 7:640–649 Matyssek R, Thiec DL, Löw M, Dizengremel P, Nunn AJ, Häberle KH (2006) Interactions between drought and O3 stress in forest trees. Plant Biol 8:11–17 Monteith J (1965) Evaporation and the environment. Proc Sympos Soc Exp Biol 19:205–234 Mueller CW, Brüggemann N, Pritsch K, Stoelken G, Gayler S, Winkler JB, Kögel-Knabner I (2009) Initial differentiation of vertical soil organic matter distribution and composition under juvenile beech trees. Plant Soil (this issue). doi:10.1007/s11104-009-9932-1

Plant Soil (2009) 323:125–141 Müller J, Wernecke P, Diepenbrock W (2005) LEAFC3-N: a nitrogen-sensitive extension of the CO2 and H2 O gas exchange model LEAFC3 parameterised and tested for winter wheat (Triticum aestivum L.). Ecol Model 183:183–210 Musselman RC, Massman WJ (1999) Ozone flux to vegetation and its relationship to plant response and ambient air quality standards. Atmos Environ 33(1):65–73 Nunn AJ, Reiter I, Häberle KH, Langebartels C, Bahnweg G, Pretzsch H, Sandermann H, Matyssek R (2005) Response patterns in adult forest trees to chronic ozone stress: identification of variations and consistencies. Environ Pollut 136:365–369 Olbrich M, Gerstner E, Welzl G, Winkler JB, Ernst D (2009) Transcript analysis in leaves of ozone-treated beech saplings at an outdoor free air fumigation site. Plant Soil (this issue) Op de Beeck M, Löw M, Verbeeck H, Deckmyn G (2007) Suitability of a combined stomatal conductance and photosynthesis model for calculation of leaf-level ozone fluxes. Plant Biol 9:331–341 Plöchl M, Lyons T, Ollerenshaw J, Barnes J (2000) Simulating ozone detoxification in the leaf apoplast through the direct reaction with ascorbate. Planta 210: 454–467 Priesack E (2006) Expert-N - dokumentation der modellbibliothek. FAM-Bericht 60, Hieronymus Verlag, München Pritsch K, Ernst D, Fleischmann F, Gayler S, Grams TEE, Göttlein A, Heller W, Koch N, Lang H, Matyssek R, Munch JC, Olbrich M, Scherb H, Stich S, Winkler JB, Schloter M (2008) Plant and soil system responses to ozone after 3 years in a lysimeter study with juvenile beech (Fagus sylvatica L.). Water Air Soil Pollut: Focus 8(2):139–154 Rennenberg H, Kreutzer K, Papen H, Weber P (1998) Consequences of high loads of nitrogen for spruce (Picea abies) and beech (Fagus sylvatica) forests. New Phytol 139(1):71–86 Schloter M, Winkler JB, Aneja M, Koch N, Fleischmann F, Pritsch K, Heller W, Stich S, Grams TEE, Göttlein A, Matyssek R, Munch JC (2005) Short term effects of ozone on the plant–rhizosphere–bulk soil system of young Beech trees. Plant Biol 7:728–736 Schwinning S (1996) Decomposition analysis of competitive symmetry and size structure dynamics. Ann Bot 77:47–58 Seyfarth M, Reth S (2008) Lysimeter soil retriever (LSR)— an application of a new technique for retrieving soils from lysimeters. Water Air Soil Pollut: Focus 8: 227–231 Simunek J, Huang K, Van Genuchten MT (1998) The HYDRUS code for simulating the one-dimensional movement of water, heat, and multiple solutes in variably-saturated media. Version 6.0. Technical report 144. Salinity Laboratory, USDA, Riverside, CA van Genuchten MT (1980) A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci Soc Am J 44:892–898

Plant Soil (2009) 323:125–141 van Oijen M, Dreccer MF, Firsching KH, Schnieders BJ (2004) Simple equations for dynamic models of the effect of CO2 and O3 on light-use efficiency and crop growth. Ecol Model 179:39–60 von Caemmerer S, Farquhar GD (1981) Some relationships between the biochemistry of photosynthesis and the gas exchange of leaves. Planta 153:376–387 Wallach D, Goffinet B (1989) Mean squared error of prediction as a criterion for evaluating and comparing system models. Ecol Model 44:299–306 Weinstein DA, Beloin RM, Yanai RD (1991) Modeling changes in red spruce carbon balance and allocation in response to interacting ozone and nutrient stresses. Tree Physiol 9:127–146 Willmott CJ (1982) Some comments on the evaluation of model performance. Bull Am Meteorol Soc 64: 1309–1313 Winkler JB, Fleischmann F, Gayler S, Grams TEE, Scherb H, Matyssek R (2009a) Do chronic aboveground O3 exposure and below-ground pathogen stress affect

141 growth of young beech trees (Fagus sylvatica L.)? Plant Soil (this issue). doi:10.1007/s11104-009-9968-2 Winkler JB, Lang H, Graf W, Munch JC (2009b) Experimental setup on field lysimeters for studying effects of elevated ozone and below-ground pathogen infection on a plant–soil-system of juvenile beech (Fagus sylvatica L.). Plant Soil (this issue). doi:10.1007/s11104009-9936-x Wipfler P, Seifert T, Heerdt C, Werner H, Pretzsch H (2005) Growth of adult Norway Spruce (Picea abies [L.] Karst.) and European Beech (Fagus sylvatica L.) under free-air ozone fumigation. Plant Biol 7:611–618 Wolf J, Oijen MV (2003) Model simulation of effects of changes in climateand atmospheric CO2 and O3 on tuber yield potential of potato (cv. Bintje) in the European Union. Agric Ecosyst Environ 94: 141–157 Zianis D, Muukkonen P, Mäkipää R, Mencuccini M (2005) Biomass and stem volume equations for tree species in Europe. Silva Fenn Monogr 4:63