the Creative Commons Attribution 3.0 License.

Abstract. The exchange of ammonia between crop canopies and the atmosphere depends on a range of plant parameters and climatic conditions. However, little is known about effects of management factors. We have here investigated the stomatal ammonia compensation point in response to cutting and fertilization of a grass sward dominated by Lolium perenne. Tall grass had a very low NH3 compensation point (around 1 nmol mol−1), reflecting the fact that leaf nitrogen (N) concentration was very low. During re-growth after cutting, leaf tissue concentrations of NO3−, NH4+, soluble N and total N increased along with apoplastic NH4+ concentrations. In contrast, apoplastic pH decreased resulting in largely unaltered NH3 compensation points. Nitrogen fertilization one week after cutting caused the apoplastic NH4+ concentration of the newly emerging leaves to increase dramatically. The NH3 compensation point peaked between 15 and 25 nmol mol−1 the day after the fertiliser was applied and thereafter decreased over the following 10 days until reaching the same level as before fertilisation. Ammonium concentrations in leaf apoplast, bulk tissue and litter were positively correlated (P=0.001) throughout the experimental period. Bulk tissue NH4+ concentrations, total plant N and soil NH4+ concentrations also showed a positive correlation. A very high potential for NH3 emission was shown by the plant litter.


Introduction
Ammonia is emitted from plants when the atmospheric NH 3 concentration is lower than the NH 3 compensation point, the latter being equal to the NH 3 concentration in the substomatal cavity (Farquhar et al., 1980;. In the opposite situation, i.e. when the atmospheric NH 3 concentration exceeds the NH 3 compensation point, deposition of NH 3 occurs. The quantity of NH 3 exchanged between crop canopies and the atmosphere may vary between seasons, depending on climatic conditions Sommer et al., 2004). In particular, temperature is known to have a major effect on the NH 3 exchange under controlled environmental conditions Mattsson et al., 1997) as well as in the field (van Hove et al., 2002;Trebs et al., 2006).
The NH 3 emission potential of grasslands may vary with species composition (Horvath et al., 2005) because grass species differ in NH 3 compensation point (Hanstein et al., 1999;Herrmann et al., 2001;Mattsson and Schjoerring, 2002;Mattsson et al., 2008). In a non-fertilized managed grassland in The Netherlands, NH 3 emission fluxes were frequent, covering about 50% of the time in a warm, dry summer period (Wichink Kruit et al., 2007). In contrast, during a wet, cool autumn period, deposition fluxes dominated (80% of the time) due to small canopy compensation points caused by low temperatures and a generally wet surface (Wichink Kruit et al., 2007). Nitrogen fertilisation is one of the major management factors of grasslands and NH 3 volatilisation can be influenced by the form, timing and dosage of N fertiliser (Riedo et al., 2002). Measurements of NH 3 volatilisation under controlled laboratory conditions have shown that high amounts of N supplied to the roots increase NH 3 emission (Mattsson et al., 1998;Mattson and Schjoerring, 1996) and NH 3 compensation points (Mattsson and Schjoerring, 2002). Increasing the N availability to plant roots leads to elevated steady state levels of different N pools within the plant tissue. In a field experiment over two years, the NH 3 losses from wheat, oilseed rape and barley increased under conditions of high N concentration in the foliage . In a Scottish experiment, a higher NH 3 compensation point of the grass was seen after only one of the two cuttings and fertilisations (Loubet et al., 2002). Little is known about the NH 3 emission potential of grasslands where repeated cuttings and N fertilisations are normal management practice. A better understanding of the component parameters influencing the NH 3 emission potential is needed in order to model NH 3 exchange between grasslands and the atmosphere.
The aim of the present study was to estimate the NH 3 emission potential of grassland in relation to common management practice. In order to do this, the temporal variation in the NH 3 compensation point and its underlying components of grass leaves and soil were followed at a field site , starting with tall grass and spanning subsequent events of cutting, lifting and N-fertilization.

Materials and methods
The investigation took place as part of the GRAMINAE integrated experiment conducted on a field near Braunschweig from 22 May to 15 June 2000. The main field was 600×300 m in size and consisted of a mixed sward dominated by Lolium perenne (around 60% abundance), Phleum pratense (∼15% abundance) and Festuca pratensis (∼12% abundance; Mattsson et al., 2008). The data presented for leaves of tall grass plants are mean values of these 3 most abundant species, weighted by their relative abundance in the field. The grass was cut on 29 May and lifted for silage on 31 May. An area of 10×10 m was left uncut for additional sampling of tall grass. Fertilizer (100 kg N ha −1 in calcium ammonium nitrate) was applied on the main field on the 5 June. A 10×10 m plot was left unfertilized and another plot of the same size received 200 kg N ha −1 in calcium ammo-nium nitrate. Growth and development of the grass were as described in Sutton et al. (2008).

Sampling of plant material
Throughout the entire experiment, plants were sampled almost every day between 12:00 and 03:00 p.m. (GMT). Cut green leaves were immediately taken to the field laboratory where the apoplastic solution was extracted using a vacuum infiltration technique (Husted and Schjoerring, 1995). Whole leaves were infiltrated in isotonic sorbitol solution (280 mM) at a pressure of 16 bar under vacuum for 5 s. The procedure was repeated 5 times in order to ensure full infiltration. Infiltrated leaves were carefully blotted dry and kept in plastic bags to equilibrate for 15 min in daylight. Leaf apoplastic solution was extracted by centrifugation at 800 g for 10 min at 4 • C. After extraction, pH of the apoplastic samples was measured with a micro-combination pH electrode (9810, Orion, Beverly, USA) and samples were frozen at −18 • C. Leaf samples for bulk tissue NH + 4 and NO − 3 analysis were also frozen down at the same time for later extraction. Samples of litter (senescent leaves) and stubble (cut stems) were frozen every day after the grass was cut. For total N concentration, samples of leaves, litter and stubbles were taken daily and immediately dried in an oven (70 • C) over night. Guttation droplets were collected on the main field and the high fertilized plot between 03:00 and 06:00 a.m. and immediately frozen.

Plant analysis
Ammonium in apoplastic extracts was determined by fluorometry on an HPLC system (Waters Corp. Milford, USA) equipped with a pump, a column oven with a 3.3 m stainless steel reaction coil, an autosampler cooled to 2 • C and a scanning fluorescence detector. The reaction between NH + 4 and o-phthaldehyde (OPA) to form an alkylthioisoindole fluorochrome was performed at neutral pH with 2mercaptoethanol as reducing agent. This fluorochrome was detected at an exitation wavelength of 410 nm and an emission wavelength of 470 nm (Husted et al., 2000a).
The plant leaves, litter and stubble were homogenised in 10 mM formic acid in a cooled mortar with a little sand. The homogenate was centrifuged at 25000 g (2 • C) for 10 min and the supernatant was transferred to polysulphone centrifugation filters (Size 500-µl, mesh 0.45 µm; Micro VectraSpin, Whatman Ltd., Maidstone, UK) and spun at 5000 g (2 • C) for 5 min. The filtered solution was used for analysis of NO − 3 and NH + 4 concentrations on a flow injection system (Quick Chem instrument, Lachat Istruments INC, Milwaukee, USA). Tissue extracts were also analysed for total soluble N concentration (so-called substrate N) using an ANCA-SL Elemental Analyser coupled to a 20-20 Tracermass Mass Spectrometer (SerCon Ltd., Crewe, UK). The same equipment was used for analysis of total N and C concentrations in oven dried plant material ground to a fine powder. For bulk tissue pH measurements, 0.2 g sample of leaf material was homogenized in 2 ml of deionized water in a cooled mortar with a little sand. The homogenate was centrifuged at 14 000 g (4 • C) for 10 min and pH in the supernatant measured with a microelectrode (Metrohm, Herisau, Switzerland).

Soil sampling and analysis
Soil samples were taken at least every third to fourth day with a soil auger at random positions over the field. Soil cores were separated into two layers (0-10 cm and 10-30 cm) and frozen at −18 • C. A sub-sample was analysed for moisture content by calculating % weight loss after drying the soil for 24 h at 108 • C. Another sub-sample (10 g) was used for pH measurements after extraction for 1 h in 25 ml 0.01 m CaCl 2 . Plant available NH + 4 and NO − 3 were analysed with flow injection after extraction of 25 g of soil in 50 ml 2M KCl.

Calculation of the NH 3 compensation point
The stomatal NH 3 compensation point (χ NH 3 mol NH 3 mol −1 air) at 25 • C was calculated by use of Eq. (1) derived from  taking into account that K d [H + ] apoplast within the range of apoplastic pH values: is the dimensionless ratio between the apoplastic NH + 4 and H + concentrations, and K H and K d are thermodynamic constants of 10 −1.76 atm l mol −1 and 10 −9.25 mol l −1 at 25 • C, respectively. Equation (1) literally calculates the pressure of NH 3 (unit: atm), which according to Dalton's law of partial pressures is equal to the mol fraction (or volume fraction) at a given atmospheric pressure. The calculated χ NH 3 at 25 • C (T ref ) was adjusted to the actual canopy temperature T c by the following equation derived from : is the requested NH 3 compensation point at the actual canopy temperature T c (K), H 0 dis the enthalpy of NH + 4 dissociation (52.21 kJ mol −1 ), H 0 vap the enthalpy of vaporization (34.18 kJ mol −1 ), and R the gas constant (0.00831 kJ K −1 mol −1 ).

Apoplastic parameters
The apoplastic NH + 4 concentration was below 50 µM in the tall grass both before and after the main field was cut May. An area of 100 m 2 was left uncut for additional sampling of tall grass. Fertilizer (100 kg N ha −1 in calcium ammonium nitrate) was applied on the main field on 5 June. A 100 m 2 plot was left unfertilized and another plot of same size was applied 200 kg N ha −1 in calcium ammonium nitrate. Vertical dotted lines indicate times of cutting and fertilisation, respectively. Values are means of three replicates ±S.E. (Fig. 1a). New grass leaves emerging after cutting showed slightly elevated apoplastic NH + 4 concentrations compared to leaves of the tall grass (Fig. 1a). Following application of 100 kg N ha −1 to the main field 6 days after cutting of the grass, apoplastic NH + peaked at around 500 µM, but thereafter decreased over the next 10 days until reaching almost the same level as before fertilisation (Fig. 1a). Plants on a plot receiving 200 kg N ha −1 attained a maximum apoplastic NH + 4 concentration around 800 µM (Fig. 1a). When no nitrogen was applied (0 N plot) apoplastic NH + 4 concentrations remained below 100 µM throughout the experimental period.
Apoplastic pH was higher in the tall grass compared to the cut grass (Fig. 1b). Fertilisation caused a transient increase in apoplastic pH without showing any difference between plants receiving 100 or 200 kg N ha −1 (Fig. 1b). The ratio between apoplastic NH + 4 and H + concentrations ( apoplast ) ranged from 10 to 150 before fertilisation (Fig. 1c). The slight increase in apoplastic NH + 4 following cutting was counteracted by decreasing pH (Fig. 1b). Accordingly, apoplast hardly changed between cutting and fertilisation (Fig. 1c). During the first 2 days after fertilisation, apoplast increased to above 1000, but thereafter started to decrease in parallel with the NH + 4 concentration (Fig. 1c). In the 0 N plot, apoplast remained below 150 throughout the experiment.

Bulk tissue nitrogen status
Bulk tissue NH + 4 concentrations of the tall grass as well as the cut grass prior to fertilisation were lower than 2 µmol g −1 fresh weight (Fig. 2a). After fertilisation, bulk tissue NH + 4 increased rapidly and substantially, peaking around 14 µmol g −1 FW with little difference between 100 and 200 kg N ha −1 treatments (Fig. 2a). Plants not receiving N fertilizer (0 N treatment) maintained a bulk tissue NH + 4 level below 4 µmol g −1 FW.
Bulk tissue NO − 3 concentrations were extremely low in the tall grass (Fig. 2b), while in the new leaves developing after cutting, NO − 3 increased considerably. Fertilisation caused a dramatic increase (4 to 5 fold) in bulk tissue NO − 3 and the  high level remained until the end of the experiment (Fig. 2b). In unfertilised grass, NO − 3 concentrations decreased towards the end of the experiment to values similar to those of the tall grass.
Total N concentration in the tall grass leaves decreased from 3% (dry weight basis) before cutting to about 2% 9 days later (Fig. 3b). The remaining part of the cut stems (stubble) also had a total N concentration of ca. 2% throughout the rest of the experimental period. In the newly produced leaves of the 100 N treatment (main field), the total N concentration increased from around 3% just after cutting to around 5% at the end of the experiment (Fig. 3b). The final foliar N concentration in the 0 and 200 N treatments were 3.5 % and 5.5 %, respectively (data not shown).
Tissue extracts were also analysed for total soluble N concentration which can be interpreted as a dynamic N pool available for plant growth. This so-called "substrate N" was very high in leaves remaining or developing after cutting (Fig. 3a). Following fertilisation, plants in the 100 and 200 N treatments had significantly higher substrate N than unfertilised grass (0 N treatment). Substrate N constituted between 10 and 40% of total leaf N.      The litter component of the grassland consisting of senescent plant leaves either attached to the lower part of the stems or lying on the ground constituted about 20% of the total above-ground biomass before cutting (data not shown). Prior to cutting, litter concentrations of NH + 4 and NO − 3 were below 50 mM (Fig. 3), while 3 days after cutting the NO − 3 concentration in the litter had increased to about 350 mM, while litter NH + 4 remained below 50 mM (Fig. 4). After fertilization (100 kg N ha −1 ), litter NH + 4 increased to around 500 mM but started to decrease again already after a few days (Fig. 4). Nitrate concentrations were slightly higher than NH + 4 concentrations after fertilisation but followed the same temporal pattern (Fig. 4).  In parallel to apoplastic extracts, NH + 4 and H + concentrations in bulk tissue extracts can be used to derive values for different plant components of the sward in order to compare their potential NH 3 exchange (Table 1). Bulk pH values of green leaves and stems were similar to those in the apoplastic solution ranging between 6.2 and 6.6 (Table 1; Fig. 1b). Due to higher NH + 4 concentrations in the tissue extracts (Table 1) than in the apoplastic solution (Fig. 1a), the resulting value in green leaves was around 40 times higher than apoplast (Fig. 1c). Extracts of plant litter showed both higher pH and, particularly, NH + 4 concentrations compared to leaves and stems which resulted in a litter value of around 173 500, i.e. 45-60 times higher than leaf and stem (Table 1).

Soil parameters
Soil NH + 4 and NO − 3 concentrations (Fig. 5) were low before fertilisation with NH + 4 concentrations being higher than NO − 3 concentrations in both the top soil fraction (0-10 cm) and the deeper soil fraction (10-30 cm, not shown). The top soil concentration of inorganic N increased slightly after cutting of the grass (Fig. 5). Application of 100 kg N ha −1 caused a dramatic increase in both NH + 4 and NO − 3 (Fig. 5), while soil inorganic N increased less in the 200 N treatment. This difference may reflect the fact that the N fertiliser in the 200 N plot was applied by hand a few hours later and under drier conditions compared to that in the 100 N treatment of the main field where rainfall followed within a couple of hours after the application.

Correlation analysis
The NH + 4 concentrations in apoplastic solution, leaf tissue and litter were mutually positively correlated ( Table 2). The NH 3 compensation point derived from the apoplastic measurements was also positively correlated with leaf tissue NH + 4 , but not with total leaf N content. Apoplastic pH and the NH + 4 concentration in guttation droplets were not significantly correlated with any of the other parameters (Table 2).

Discussion
Before cutting, the tall grass had low NH + 4 concentrations in both leaf apoplast and bulk tissue (Fig. 1a, 2a). This resulted in NH 3 compensation points so low that the grass was not likely to emit NH 3 before cutting which is in agreement with atmospheric NH 3 concentration gradients above the canopy showing predominantly deposition fluxes . Extremely low tissue NO − 3 concentrations (Fig. 2b) as well as low soil NH + 4 and NO − 3 levels (Fig. 5) also indicated that the small amounts of inorganic N available to the plants were efficiently taken up and utilised for growth and seed development at this stage. Before fertilisation, the soil content of NH + 4 was 4 times higher than that of NO − 3 (Fig. 5), which is not unusual for grassland soil (Whitehead, 1995).
Between cutting and fertilisation there was a re-growth period of one week in which the grass leaves first showed increased NO − 3 concentrations (Fig. 2b) and 5 days later also increased NH + 4 concentrations (Fig. 2a). Ryegrass has been shown to rapidly accumulate NO − 3 in both leaves and stubble after cutting as NO − 3 is involved in the osmotic adjustment (Ourry et al., 1989). Also the soluble N and total N concentrations of the leaves showed higher values during re-growth after cutting compared to the tall grass (Fig. 3a, b). The increase in these N pools was paralleled by slightly increasing NO − 3 and NH + 4 concentrations in the soil. Several authors have reported that during the first days after cutting, uptake of N is inhibited (Bakken et al., 1998;Ourry et al., 1988). In such case plants respond by allocating N from reserves in root and stubble to the developing leaves. This inhibition of uptake could explain the increasing levels of soil NO − 3 and NH + 4 after cutting. In Lolium perenne and Bromus erectus grown in nutrient solution, tissue NH + 4 concentrations of expanding leaves did not start to increase until 6 days after cutting (Sutton et al., 2001). In the same experiment, apoplastic NH + 4 concentrations increased in the new expanding leaves 3-6 days after cutting. Also in the present study, apoplastic NH + 4 concentrations showed slightly higher values in expanding leaves during re-growth compared to the leaves of the tall grass (Fig. 1a). Increasing tissue concentrations of NH + 4 and NO − 3 can also result from shortening of the leaf growth zone and smaller dilution of the N transported to this zone after defoliation compared to the fully expanded leaves before cutting (Schäufele and Schnyder, 2001). Grass cutting has in several cases been reported to lead to NH 3 volatilization from grassland (Milford et al., 1999Loubet et al., 2001). The emitted NH 3 may originate from the plants as a consequence of increased N pools during the period of leaf expansion. However, the potential for NH 3 emission did not seem to increase after cutting in the present work since apoplast (the ratio between apoplastic NH + 4 and H + ) were unaltered due to counteracting effects of decreased pH and increased NH + 4 concentrations (Fig. 1c). Another source of NH 3 emission could be the litter, i.e. senescent leaves attached to the stems or lying on the ground surface. Ammonium concentrations were considerably higher in the litter material compared to green leaves (Fig. 4) due to the protein degradation processes going on in the litter (Mattsson and Schjoerring, 2003). Senescence-related processes were probably also enhanced after cutting when both the climatic conditions and the proportion of litter out of total biomass were changed at the bottom of the canopy (David et al., 2008). In a tall canopy, NH 3 emitted from the litter can be taken up by leaves positioned higher above the ground (Husted et al., 2000b;Nemitz et al., 2000) while in the absence of a tall canopy, the litter NH 3 may escape to the atmosphere. High NH + 4 concentrations and relatively high pH values in the litter also resulted in an extremely high litter value (Table 1) indicating a strong potential for NH 3 emission. In a non-fertilized grassland in the Netherlands, Wichink Kruit et al. (2007) observed an average canopy value of 2200, which is in line with that recorded for stems and young leaves in the present work (Table 1).
After fertilisation, all plant N pools increased with peak values already on the first day after fertilisation. Micrometeorological measurements also showed high NH 3 emissions after fertilisation with some contribution from the fertiliser itself during the first 2 days . The fertiliser was rapidly dissolved in the soil solution since it was raining the same afternoon as the main field was fertilised . Consequently, the fertiliser contamination was restricted to a very short period. The fertilisation was also reflected in higher NH + 4 concentrations in guttation droplets collected at the leaf tips in the early mornings after fertilisation (not shown).
Ammonium concentrations in both leaf tissue and apoplast started to decrease again already a few days after fertilisation (Fig. 1a, 2a) while leaf tissue NO − 3 concentrations remained high for the rest of the experiment (Fig. 2b). This may partly reflect declining soil NH + 4 levels (Fig. 5) and partly rapid assimilation of NH + 4 in the plant cells, while NO − 3 was stored in the leaf cell vacuoles for later use. It has previously been shown that when NH 4 NO 3 is supplied to plant roots, NH + 4 is absorbed more readily than NO − 3 (Bloom, 1981;Clarkson et al., 1986). Nitrate accumulates in grass herbage when the rate of uptake by the roots exceeds the rate of conversion to organic N (Whitehead, 1995). The NH + 4 concentrations in apoplast and bulk tissue were obviously sensitive parameters responding rapidly to fluctuations in soil nitrogen availability. Also in a laboratory experiment with Lolium perenne and Bromus erectus both leaf tissue and apoplastic NH + 4 concentrations were shown to respond rapidly to changing NH + 4 concentrations in the nutrient solution (Mattsson and Schjoerring, 2002).
A correlation analysis revealed that the stomatal NH 3 compensation point calculated on the basis of apoplastic parameters was positively correlated with the bulk tissue NH + 4 concentration in leaves and litter (Table 2). Some previous investigations have likewise shown good correlation between apoplastic and leaf tissue NH + 4 concentrations (Mattsson et al., 1998;Mattsson and Schjoerring, 2002), suggesting that the tissue NH + 4 concentration may be used as an indicator of the NH 3 compensation point. Other studies have found the correlation to depend on growth conditions (Herrmann et al., 2008) or not to be present .

Conclusions
We conclude that the management practice has a major impact on the potential plant-atmosphere NH 3 exchange in grassland by influencing both plant and soil N parameters. The NH 3 compensation point derived from apoplastic measurements was positively correlated with bulk tissue NH + 4 concentrations in leaves and litter. This suggests that measurements of NH + 4 and pH in bulk extracts of plant material in grassland can be used as a simple indicator of the NH 3 exchange potential.