Nutrient Cycling in Agroecosystems 62: 73–88, 2002. © 2002 Kluwer Academic Publishers. Printed in the Netherlands.
73
In situ comparisons of ammonia volatilization from N fertilizers in Chinese loess soils M. Roelcke1,∗ , S.X. Li2 , X.H. Tian2 , Y.J. Gao2 & J. Richter1 1 Institute
of Geography and Geoecology, Braunschweig Technical University, Langer Kamp 19c, D-38106 Braunschweig, Germany; 2 Department of Resources and Environmental Sciences, P.O. Box 18, Northwestern Agricultural University, 712100 Yangling, Shaanxi, P.R. China; ∗ Corresponding author. E-mail:
[email protected] Received 24 December 1999; accepted 23 October 2000
Key words: ammonia volatilization, ammonium bicarbonate, China, Dräger-Tubes, in situ method, urea
Abstract Ammonia volatilization loss from mineral N fertilizers was determined on a calcareous Chinese loess soil with a pH (CaCl2 ) of 7.7. An original in situ method that required no electricity or laboratory analyses was used. By means of a bellows pump, ambient air was drawn through four conical cups placed onto the soil (total area 400 cm2 ) and subsequently through an NH3 -specific detector tube with direct colorimetric indication of the ammonia concentration (measuring range, 0.05–700 vol.-ppm NH3 ). Duration of measurement was about 3 min. Following N fertilization to winter wheat in 1990 and to summer maize in 1991, the application methods surface broadcast, uniform incorporation into the 0–15-cm layer, and for maize, a point placement at 10 cm depth were investigated. Ammonium bicarbonate and urea were applied at rates of 100 and 200 kg N ha−1 . In the autumn of 1990, ammonia losses following NH4 HCO3 application were more than twice as large as with urea, fertilizer incorporation reduced NH3 losses 15-fold, and doubling the nitrogen application rate resulted in a 1.7-fold increase in the percentage of nitrogen loss. Cumulative ammonia fluxes were about 2 times higher in the summer of 1991. Comparing application methods in summer, losses were significantly (3 times) lower only with point placement. The above differences were all significant at the P<0.05 level. Due to the very low air exchange rate (0.9 volumes min−1 ), actual volatilization rates were underestimated by this method. Though not yielding absolute amounts, the Dräger-Tube method proved very suitable for comparing relative differences in ammonia fluxes. The measurements clearly reflected the characteristic flux patterns for the different treatments and the effects of environmental factors on their time course.
Introduction The consumption of urea as N fertilizer in developing countries has been steadily increasing (FAO, 2000). Ammonia volatilization is often the main pathway of nitrogen loss under semi-arid conditions (Aggarwal & Praveen-Kumar, 1994; Christianson et al., 1990; Katyal et al., 1987). For example, Zhang et al. (1992) found that 30–32% of the N applied as urea could be lost as NH3 from calcareous soils in N. China. Nevertheless, because of technological constraints, relatively little research on ammonia volatilization has been carried out in developing countries using micro-
meteorological methods (e.g., Cai et al., 1986, 1998; Zhang et al., 1992; Zhu et al., 1989). Instead, a large part of the studies were carried out in the laboratory or have used 15 N-labelled nitrogen fertilizer to indirectly assess the importance of the different pathways of nitrogen loss (Christianson et al., 1990; Pilbeam & Hutchison, 1998; Wang et al., 1991). However, in order to correctly estimate NH3 -volatilization losses by this means, other loss reactions such as denitrification, runoff and leaching must be measured simultaneously (Harper, 1988). Recent studies (Harper & Sharpe, 1998) have shown problems in using 15 N for assessing plant–soil N dynamics due to gaseous NH3 exchange.
74
Figure 1. Bellows pump for NH3 -volatilization measurements with inserted Dräger-Tube (photograph © Drägerwerk AG, Lübeck, Germany).
Micrometeorological approaches using the mass balance (integrated horizontal flux, IHF) method are now commonly accepted as being the most preferable for directly measuring NH3 -volatilization in the field for agronomic purposes (Harper, 1988). A wide range of investigations have been carried out (e.g., Beauchamp et al., 1982; Denmead et al., 1977; McInnes et al., 1986a,b; Wilson et al., 1982, 1983). More recently, modifications using passive flux samplers (e.g., Fox et al., 1996; Leuning et al., 1985; Schjørring et al., 1992) have been developed. Most of these micrometeorological methods require considerable technological input and large circular plots with a radius of 15–40 m, which makes them unsuitable for investigating a large number of fertilization practices. The other main approach uses enclosures covering a certain portion of the soil surface. Most enclosure methods are based on the forced-draft system with air flow rates of 15 to 20 replacement volumes
per minute (Fenn & Kissel, 1973; Kissel et al., 1977; Vlek & Stumpe, 1978). Laboratory measurements using the forced-draft method assess potential maximum NH3 -volatilization rates due to constant temperature, soil moisture and wind speed, and exceed the actual loss rates under field conditions, which are determined by weather conditions and agronomic factors (Vlek, 1994, pers. comm.). Attempts to minimize alterations of the ambient environmental conditions when using an enclosure system have been made by enclosing the plots only for a portion of the time, or by constructing larger wind tunnels and adjusting the air flow to ambient field conditions (Bouwmeester et al., 1985; Huber & Amberger, 1989; Ryden and Lockyer, 1985). In the following study, measurements of ammonia volatilization using the Dräger-Tube method (Richter, 1972; Roelcke, 1994), representing a special case of the forced-draft enclosure method, are presented. Recently, parallel measurements carried out in an alkaline soil in China over several seasons (Pacholski et al., 1999), using both the Dräger-Tube method and two simplified micrometeorological mass balance methods (Leuning et al., 1985; Schjørring et al., 1992) have confirmed the Dräger-Tube method’s validity for providing comparative estimates between treatments. The purpose of our study was to compare relative ammonia volatilization rates under the influence of type of N fertilizer, method of application, application rate, and different seasons at an agricultural site on the semi-arid Chinese Loess Plateau. Experiments on the same site using 15 N-labelled urea, with similar treatments but in different years, resulted in percentages of unaccounted fertilizer N at harvest ranging from 43 to 62% for maize and from 36 to 46% for wheat (Rees et al., 1997). In laboratory measurements and simulations using columns of the same soil, up to 60% of surface applied urea-N had volatilized after 13 days (Roelcke et al., 1996). Other aspects of the N cycle in the area are described in Rees et al. (1997) and Roelcke et al. (2000).
Sites, materials and methods The study was conducted at the southern edge of the Chinese Loess Plateau in the village of Shangzhuang (34◦ 17 N, 108◦ 00 E; altitude 570 m), 5 km W of the Northwestern Agricultural University in Yangling, Shaanxi Province. The soils are developed from quaternary loess and are classified as lou tu (old manured loessial soils) in the Chinese system (Insti-
75 Table 1. Main properties of the topsoil (0–20 cm) on the two experimental field sites I and II in the village of Shangzhuang, Shaanxi, China (from Roelcke, 1994). Particle size fractionation according to the German classification system Site
pH (0.01 M CaCl2 )
CaCO3 (% w/w)
Total N (mg kg−1 )
Org. C (mg kg−1 )
CEC (mmolc kg−1 )
Clay < 2 µm
Silt 2–63 µm (% w/w)
Sand 63–2000 µm
I II
7.8 7.7
9.8 9.5
800 910
6500 7300
164 134
23 23
74 74
3 3
tute of Soil Science, 1990) and as Udic Haplustalf in the USDA (1994) system. The topsoil (0–20 cm) is a highly calcareous silt loam soil (USDA, 1994; Table 1). The climate is subhumid continental with a mean annual temperature of 12.9◦C, a mean precipitation of 632 mm a−1 (Water Economy Bureau Xianyang, 1989) and a reference evapotranspiration (ET0) of 772 mm a−1 (Kang et al., 1992). A longterm rotation of winter wheat (Triticum aestivum L.) or occasionally rapeseed (Brassica napus L.) with summer maize (Zea mays L.) is practiced. Wheat yields are around 5.0–5.25 t ha−1 , maize yields 6.0–7.5 t ha−1 . Nitrogen fertilizer is applied as ammonium bicarbonate (NH4 HCO3 ; 17% N) or urea (46% N), at rates of about 150 kg N ha−1 per crop, usually in a single application. The fertilizer is commonly uniformly incorporated into the 0–15-cm layer for wheat, for maize it is point placed at about 10 cm depth. However, the fertilizer is also applied at shallower depths and urea is sometimes surface broadcast following rainfall. The high soil pH and CaCO3 content lead to NH3 -volatilization as the major pathway for mineral N loss. Measuring device and procedure The method for measuring ammonia volatilization in situ presented here is based on that by Richter (1972) for measuring rates of CO2 evolution from soils, also described in Anderson (1982). The ammonia concentration difference between enclosure air and ambient air is measured without interrupting the quasi-continuous gas flow. However, in contrast to the methods integrating over time which yield average emission rates, this method is based on many direct short-term measurements of NH3 concentrations, resulting in instantaneous values for NH3 fluxes. It requires neither electricity nor laboratory equipment. The gas analysis detector tubes (‘Dräger-Tubes’) for short-term measurements Ammonia 0.25/a, Ammonia 2/a and Ammonia 5/a, supplied by Drägerwerk
AG, Lübeck, Germany or National Draeger Inc., Pittsburgh, PA, USA, are used for ammonia detection (Drägerwerk AG, 1994; Figure 1). These disposable detector tubes contain a solid phase acid compound and bromophenol blue pH indicator and have been calibrated by the provider. A Dräger-Tube is inserted into the corresponding Dräger Gas Detector Pump (a handoperated bellows pump with semi-continuous suction characteristics). A volume of air defined by the number of strokes (1 stroke = 100 cm3 ) required for the specific type of Dräger-Tube and the given measuring range (usually five to 10 strokes) is drawn through the tube. The ammonia is trapped quantitatively, avoiding problems with handling and analysis of the gas. The total length of the colour change is a measure of the NH3 concentration, which is read directly from a scale printed onto the tube, in volume-ppm (µl l−1 ) NH3 . The three types of tubes measure NH3 in different standard concentration ranges, over a total range of 0.25–70 vol.-ppm. This can be expanded to 0.05–700 vol.-ppm by varying the number of strokes between one and 50. Four cups made of tinplate serve as gas collectors, covering a total surface area of 400 cm2 (Figures 2a, b). A polyethylene funnel is fitted into each cup. After lowering the rim into the soil, the resulting conical enclosure under each cup has a volume of approximately 370 cm3 . Two Cu tubes (5 mm i.d.) serve as air inlet and outlet, respectively. The outlets of all four cups are connected via several 35-cm pieces of Teflon tubing to form one portable, hand-held unit. During measurements, the upper end of the Dräger-Tube is connected to the Teflon tube leading to the quadruplet of gas collectors. A blank measurement is first carried out in order to detect any traces of ammonia in the cups and tubing. For each measurement, the rims of the four volatilization cups are then gently pressed about 2 cm into the soil. First, a larger volume of air (about 2–3 l) is pumped through the system employing a used Dräger-Tube, in order to create a quasi-steady state regarding the NH3 concen-
76
Figure 2. Sectional view of one volatilization cup (a), and quadruplet of gas collectors (b).
tration in the enclosure air. Immediately thereafter, the actual measurement is carried out using a new tube. The average time for 1 l of air to be drawn through the system is 45 s; the total duration of each measurement (including preliminary pumping) is only about 3 min. Therefore, changes in temperature, radiation and moisture under the cups are minimized. Parallel to each series of measurements, atmospheric pressure and air temperature are recorded. In contrast to the original procedure (Richter, 1972) for measuring CO2 , no correction of measured NH3 concentrations in the enclosure air for the NH3 in the ambient air is usually carried out, since the sensitivity of the DrägerTubes is too low (natural NH3 concentrations are 1–10 µg NH3 m−3 = 0.0014–0.014 vol.-ppm NH3 ). The background ammonia concentrations at 10 cm height should, however, be checked during periods of maximum emissions. This method detects minimum NH3 fluxes of about 0.06 mg N m−2 h−1 = 14.5 g N ha−1 h−1 . Method calibration The resulting air flow rate under each cup is approximately 0.9 exchange volumes min−1 , corresponding to a very small wind speed of 0.00056 m s−1 (=2 m h−1 ) at the soil surface. This low air exchange rate was expected to lead to an underestimation of the actual volatilization rates. The increase in NH3 concentration in the enclosure air during measurements alters the NH3 gas concentration difference at the soil surface/atmospheric air boundary layer and possibly the NH3 gradients in the soil air. A higher air exchange rate would lead to a smaller NH3 con-
centration increase in the enclosure and to greater NH3 -volatilization. This effect should be seen separately from that of the wind speed at the soil surface. An increase in wind speed provokes higher NH3 volatilization losses both through an increase in the empirical bulk transfer coefficient Ka as well as indirectly through an increase in evaporation rates (Kirk & Nye, 1991). The wind speed of 2 m h−1 at the soil surface under the enclosures of the Dräger-Tube method is much smaller than wind speeds encountered under normal field conditions and is certain to reduce Ka during the period of measurement. Theoretically, an increased accuracy could be achieved by increasing flow rate, but then the method could itself succumb to problems associated with high-flow chamber techniques. Besides the experiments mentioned in the introduction, several direct and indirect calibrations of this method were carried out. Twelve different air exchange rates between 0.40 and 32.7 exchange volumes per min were tested in the laboratory with volatilization chambers (area, 12.5 × 6 cm; depth of soil columns, 15.5 cm; headspace volume, 255 cm3 ) and forced-draft method with acid traps (Roelcke, 1994). The same treatments as those presented in this study were then compared in the laboratory using humidified air at flow rates of 16.3 exchange volumes per min (wind speed at the soil surface of 0.034 m s−1 ). Experimental setup (A) 1990 winter wheat experiment The experiments were carried out on two adjacent farmers’ field sites, I and II. Mineral nitrogen (NO3 − -
77 Table 2. Treatments and duration of measurements in both field experiments Application date 1990 winter wheat: Oct. 19, 1990
Oct. 24, 1990
1991 summer maize: July 10, 1991
Type of fertilizer and method of application
Appl. rates (kg N ha−1 )
Beginning of measurements
End of measurements
NH4 HCO3 surface NH4 HCO3 mixed 0–15 cm urea surface zero N urea surface (additional) urea mixed 0–15 cm
100 and 200 100 and 200 100 and 200 – 100 and 200 100 and 200
Oct. 19, 1990, 18:00 (4–8 h after NH4 HCO3 appl.) (4–18 h after urea appl.)
Nov. 21, 1990
Oct. 25, 1990, 06:00 (16–17 h after urea appl.)
Nov. 21, 1990
NH4 HCO3 surface NH4 HCO3 mixed 0–15 cm NH4 HCO3 point placed 10 cm urea surface urea mixed 0–15 cm zero N
100 and 200 100 and 200 100 and 200 100 and 200 100 and 200 –
July 10, 1991, 18:30 (4–6 h after NH4 HCO3 appl.) (4–6 h after NH4 HCO3 appl.) July 11, 1991, 12:00 (26–33 h after urea appl.) (26–33 h after urea appl.)
Aug. 10, 1991
N and NH4 + -N) contents in the 0–0.9-m layer were determined at harvest of the preceding maize crop (September 25, 1990). On October 15, 100-m2 plots were marked out on each field site and the usual tillage was carried out. The plots were each subdivided into 16 microplots measuring 2 × 2 m, each separated by a 0.5-m guard area. During the following 3 days, a total of 11.3 mm rain fell. Fertilizer was applied on October 19 and 24, 1990. The fertilizers NH4 HCO3 (powder) and urea (granules) were applied at two rates and two methods: surface broadcast (accurately applied), and uniform incorporation into the 0–15-cm layer using a hand-held hoe. The treatments investigated and the duration of the experiment are listed in Table 2. The NH4 HCO3 -treatments were expected to have greater NH3 emissions and were therefore placed on the downwind sides (E). Six zero N plots received the same tillage treatments as the 10 treated plots. Corresponding treatments on the two adjacent sites were evaluated as replicates. According to local practice, winter wheat was sown on the day of fertilizer application. Wheat plants emerged after 2 weeks and did not affect the course of the measurement. Therefore, no fallow plots were included. Due to logistic reasons, the beginning of the measurements was delayed for a few hours after fertilizer application (Table 2). During every series of measurements, two to three replicate measurements were carried out by positioning the cups at random on different spots of the same microplot. Initially, the measuring frequency was every
6 h; it was then gradually reduced. Due to different patterns of ammonia loss from both fertilizers (see below), measurements were carried out more frequently on the NH4 HCO3 -amended plots at the start of the experiment, while a higher frequency was maintained on the urea plots during the later phase. Each day, one measurement was carried out during the time of highest temperature. Unhydrolyzed urea in soil was not determined. (B) 1991 summer maize experiment After the winter wheat harvest in early June 1991, 16 microplots of 2 × 2 m were marked out on each of the sites I and II as in the preceding season, and the wheat stubble incorporated. On each site, the mineral N content in the 0–0.9-m profile was determined prior to the experiment (July 3). Maize had already been sown by hand into the standing wheat crop at the end of May, with inter-row spacings of 0.60 m and intrarow spacings of 0.40 m (4.17 plants m−2 ). Due to a strong drought persisting in the area since mid-June 1991, a flood irrigation of 60 mm was carried out on the plots two days prior to the experiment. At the beginning of the experiment, the maize plants were 20–30 cm high, and at the end, more than 2 m. Plant densities were quite variable and part of the wheat stubble still lay on the soil surface (5–10% cover). Fertilizer was applied on July 10, 1991. The treatments investigated are listed in Table 2. A point placement of fertilizer was included as a third method of application, where the N-
78 fertilizer is applied to every second maize plant, in 10cm deep holes, about 10 cm beside each row. The hole is covered with soil and slightly compacted. Due to different plant densities, the same amount of fertilizer per 4-m2 microplot was distributed among nine holes on site I and 12 holes on site II. Four zero N plots were included on each site. Beginning of measurements was again delayed (Table 2). Measurements were carried out as described for the wheat experiment, except for the point placement treatments, where generally only two cups were placed onto fertilizer-amended spots due to the relatively large distance between the single spots. Measuring frequency was approximately every 8 h during the first 3 days and then reduced. Calculation of results The NH3 readings (vol.-ppm) displayed on every Dräger-Tube were multiplied with a correction factor for the atmospheric pressure: Correction factor = 1013 [hPa] actual atmospheric pressure [hPa]
Precision of measurements (1)
After subtraction of the blank value, the values for the NH3 concentrations (in vol.-ppm) and the durations of measurement (in s) were converted into NH3 -N fluxes [mg N m−2 h−1 ] as follows: FNg =air volume ·|conc| · 10−6 · ρNH3 14 10000 3600 · , · · 17 400 time
For every series of measurements, mean values of the fluxes detected during two to three replicate measurements on the same microplot were used for calculating NH3 losses over the whole time course of the experiment. Fluxes were assumed to be zero during rainfall events. In the statistical evaluation, the cumulative absolute (kg N ha−1 ) and relative (% of N applied) fluxes in the different treatments were compared by means of multifactorial analysis of variance, separately for each season and by seasons, using the program STATGRAPHICS (Statistical Graphics Corporation, 1993). Corresponding treatments on the two sites were evaluated as replicates. Since the NH3 fluxes from the surface broadcast urea treatments of October 19 and 24, 1990, in the winter wheat experiment were of similar magnitude (see below), all treatments from these two application dates were pooled. Except for the temperature fluctuations in the winter wheat experiment, microclimate variables were not included in the analyses.
(2)
where FNg is the NH3 -N flux in mg N m−2 h−1 ; air volume is the throughput of air during one measurement in l; |conc| is the value of the corrected volume concentration (vol.-ppm) of NH3 ; ρ NH3 is the density of NH3 in mg l−1 (ρ NH3 = 696 mg l−1 ; 25◦ C, 1013 hPa); 14/17 is the molecular weight conversion factor of NH3 to N; 10 000/400 is the area conversion factor from cm2 to m2 in m−2 ; and 3600/time is the time conversion factor from s to h in h−1 (time = duration of measurement in seconds). On the treatments with point placement of fertilizer to summer maize, measured NH3 concentrations were converted into fluxes per unit area [m2 ] in a different way. The measured concentrations were related to the amount of N on each spot amended with fertilizer. Fluxes were then calculated according to the number of amended spots covered by the cups during measurements, and to the number of amended spots per unit area.
The relative standard deviation (coefficient of variation) of the indicated values on the Dräger-Tubes is 10–15% (Drägerwerk AG, 1994). The average (n = 175) temporal variation between two replicate measurements was only 5% (difference between two values related to the smaller value) when the same type of Dräger-Tube was used. This corresponds to a variation in air flow rate of ± 0.05 exchange volumes min−1 . The mean variation in the measured NH3 concentrations of the enclosure air between two replicate measurements was 117% (n = 72) in the autumn of 1990, and 151% (n = 114) in the summer of 1991, decreasing with time. This variability may be attributed to the relatively small surface area of the enclosures and the effects of heterogeneity of the soil surface and plant growth. Losses were calculated from mean fluxes of replicate measurements on one microplot, and treatment differences exceeded variability between replications.
Results and discussion Method calibration The results of the laboratory air flow rate experiment are shown in Figure 3. The dependence of NH3 volatilization on the air exchange rate inside the chambers
79
Figure 3. Cumulative NH3 -volatilization losses measured 40 h after application of 448 kg N ha−1 of (NH4 )2 SO4 at different air exchange rates (laboratory experiment using a forced-draft method with volatilization chambers; headspace volume 255 cm3 ).
was similar to that found in a wide variety of experiments using forced-draft systems (e.g., Fenn & Kissel, 1973; Vlek & Stumpe, 1978). The low air flow rate of the Dräger-Tube method lies in the steepest part of the curve. Based on the relationship in Figure 3, an increase in air flow rates from 0.9 to 16.3 exchange volumes per min results in a 2.6-fold increase in the amounts of NH3 volatilized. However, cumulative NH3 losses in the laboratory experiments (16.3 exchange volumes per min) after 14–21 days were 7 to 21 times greater than those of the corresponding in situ treatments (Roelcke, 1994). Besides being due to the different geometry of the volatilization chambers and the cups, this apparent deviation from the factor 2.6 mentioned above is caused by differences between field and laboratory conditions. The flow rate dependency was confirmed in experiments using different-sized cups. An average 10-fold difference was found between total losses measured directly using the Dräger-Tube method and those determined indirectly using 15 N (Rees et al., 1997; Roelcke, 1994). The parallel measurements over several seasons resulted in cumulative losses detected using the Dräger-Tube method being 6 to 27 times (depending on season) lower than those using the two simplified micrometeorological mass balance methods (Pacholski et al., 1999), thus confirming the one order of magnitude obtained from the laboratory and the 15 N experiments. However, a single defined calibration factor cannot be deducted from these results without taking into account the actual wind speeds in the field. This method therefore cannot be used to determine
Figure 4. Time course of daily maximum and minimum air temperatures (1.5 m height), precipitation and daily water vapour pressure deficits at 14:00 in the period October 19 – November 21, 1990.
absolute flux rates and percentages of N loss if used exclusively. The results presented in this paper show the applicability of the Dräger-Tube method in comparing relative treatment differences. All measured and cumulated fluxes subsequently presented in tables and figures are therefore marked with an ‘X’. This represents a variable scaling factor due to the method’s underestimation of the actual fluxes.
Zero N plots
During both seasons, ammonia concentrations measured on the zero N plots of the two sites (0.15 vol.ppm on average) were near the detection limit of the Dräger-Tubes and mostly of the same magnitude as the mean blank value of the measuring device (0.16 vol.-ppm). Only very low effective NH3 fluxes (mean, 0.07 mg N m−2 h−1 ) were detected on a few occasions. For these reasons, results from the zero N plots were not further evaluated.
80
Figure 5. Four examples of NH3 fluxes measured following mineral N application to winter wheat in 1990. (c) The estimation of the nighttime fluxes (during the measuring campaign in the autumn of 1990 only). Measured daytime fluxes, thick line; estimated nighttime fluxes and fluxes set to zero at rainfall events, thin line. In this figure and Figures 7–10, X represents a scaling factor since the method underestimates the actual fluxes.
Time courses of NH3 fluxes and environmental interactions (A) 1990 winter wheat experiment The mean air temperature during the period of measurement (October 19–November 21) was 10.6 ± 2.5◦C, the mean daily vapour pressure deficit 16.7 hPa (Figure 4). Soil mineral nitrogen contents (0–0.9 m) on site I and II prior to the experiment were 80.2 and 92.2 kg N ha−1 , respectively. Average initial soil moisture contents (0–5 cm) were 20.5% (October 19), final contents were 16.5% (November 22). Remains of NH4 HCO3 fertilizer could be seen for up to 50 h after application, while urea granules had dissolved within 24 h. Figure 5c shows an example for the estimation of the nighttime fluxes carried out for all treatments, during the measuring campaign in autumn 1990 only. From the seventh to the 11th day and from the 23rd day onwards, nighttime temperatures were between 0 and 5 ◦ C, while daytime temperatures still reached 20 ◦ C. These large daily air temperature fluctuations
resulted in large day/night variations in NH3 fluxes. To calculate the nighttime fluxes over the whole course of the experiment the day/night flux variations measured during the first few days of the experiment were correlated with the corresponding temperature variations (R2 = 64.55%). The nighttime fluxes were then estimated from the measured daytime fluxes and the night and day (minimum and maximum) temperatures using this regression. During this period, the estimated nighttime fluxes were near 0 mg N m−2 h−1 . McInnes et al. (1985) also assumed a high correlation between fluctuations in temperature and in ammonia concentration in the air above the soil surface. In experiments by McInnes et al. (1986b) no NH3 -volatilization occurred at temperatures around the freezing point. The highest NH3 fluxes were measured following surface broadcasting of fertilizer (Figures 5a–d). On these treatments, NH3 fluxes could be measured for 26 days for NH4 HCO3 and 34 days for urea. The high initial fluxes from surface-applied NH4 HCO3 (20–80 mg N m−2 h−1 ) showed hardly any diurnal variations but
81 decreased rapidly. During the first 21 hours, 50–70% of the total amount of NH3 had volatilized. It is likely that the immediate fluxes were still greater, due to the delay in the beginning of measurements by a few hours. Fluxes on the incorporation treatments were much lower (not shown). During the first 85 h after fertilizer application on October 19, 1990, weather conditions were humid with low evaporation rates and a total of 3.0 mm of rainfall from October 21 to 23 (Figure 4). On the microplots where urea was applied on October 19, 1990 (Figure 5c), fluxes began to increase on the fourth day of the experiment (October 22). From the fifth day onward, high atmospheric pressure, dry westerly winds and increasing evaporation rates prevailed, leading to increasing temperature variations between day and nighttime. Flux variations increased correspondingly, more pronounced with urea than with NH4 HCO3 . Highest fluxes were frequently measured in the late morning hours, during the period of greatest temperature increase and highest drying rate of the soil surface. Fluxes reached a maximum after 7 to 8 days and then decreased rapidly. The high initial soil moisture contents on Oct. 19 created favourable conditions for urea dissolution and hydrolysis, although the process was slowed down by the relatively cool temperatures around 13◦C. The low, scattered rainfall (total 3.0 mm) in the initial phase prevented higher NH3 -volatilization rates during the first 3 days; yet it was not sufficient to cause a major leaching of urea or ammoniacal N (NH3 -N + NH4 + N) into the soil and to notably reduce cumulative losses. In several in situ experiments, light precipitation (< 15 mm) of low intensity generally provoked an increase in total NH3 -volatilization (Craig & Wollum, 1982; McInnes et al., 1986a). In experiments by Bouwmeester et al. (1985), NH3 -volatilization was effectively prevented by 24 mm, but not by 8 mm of rainfall. Mineral N measurements between 1989 and 1991 in the same location showed higher NH4 + -N contents (up to 30 kg N ha−1 ) persisting in the upper 0–20 cm for at least 3 weeks following urea application in autumn (unpublished data). In the present case it can therefore be assumed that hydrolysis and nitrification did not take place so quickly as to notably prevent higher fluxes from occurring after 7–8 days. On the microplots where urea was applied on October 24, 1990 (Figure 5d), the soil surface had already dried. Urea prills dissolved more slowly and fluxes increased less steeply than after October 19, and a delay in maximum NH3 -volatilization rates was observed.
Figure 6. Time course of daily maximum and minimum temperatures (1.5 m height), precipitation and daily water vapour pressure deficits at 14:00 in the period July 10–August 10, 1991, as well as soil temperatures during measurements (5 cm depth; average of the two field sites, July 10–16, 1991).
It can be assumed that urea hydrolysis was incomplete during the first stage. The 3.0 mm of rain on the ninth day after fertilizer application (November 2; Figure 4) led to a temporary decrease in NH3 evolution rates. Following the renewed increase in temperature and evaporation, large increases in NH3 fluxes were observed, reaching a maximum 11 days after fertilization. A dry soil surface of a few mm depth can be sufficient to delay the diffusion of urea to zones of high urease activity and therefore reduce or prevent hydrolysis (Fenn & Hossner, 1985). (B) 1991 summer maize experiment Mean air temperature was 26.8 ± 2.6◦ C, the mean daily vapour pressure deficit at 14:00 was 32 hPa (July 10–August 10), and the average soil temperature at 5 cm depth (July 10–16) was 30.2◦C (Figure 6). Soil mineral nitrogen contents (0–0.9 m) on site I and II prior to the experiment were 91.3 and 136.2 kg N ha−1 , respectively. Average initial soil moisture contents (0–5 cm) were 20.3% (July 10), final contents were 9.1% (August 3). Urea and NH4 HCO3 crystals were totally dissolved after 36 and 96 h, respectively. The high evaporation rates and uneven growth of the
82 maize plants led to increasingly heterogeneous soil surfaces. This led to larger differences between fluxes from identical treatments as compared to the 1990 winter wheat experiment. Time courses of the fluxes following surface application (Figures 7a,c) and incorporation (Figures 7b,d) were similar for both kinds of fertilizers. Fluxes after uniform incorporation were markedly higher than in the autumn of 1990. On the treatments with point placement of fertilizer, the NH3 concentrations detected on the fertilizer-amended spots were extremely high in some cases (maximum 160 vol.-ppm NH3 ), but fluxes were lower than on the surface or incorporation treatments after converting on an area basis (not shown). Maximum initial fluxes with surface applied NH4 HCO3 were a similar order of magnitude as in the winter wheat experiment. On several of the urea plots with surface application and mixed incorporation (Figures 7c,d), the highest fluxes were recorded at the first measuring occasion, greatly exceeding the fluxes of autumn 1990. It is thus very likely that part of the maximum fluxes were missed due to the delayed onset of measurements on the urea-treated plots. Initial temperature and moisture conditions for a very rapid urea hydrolysis and NH3 -volatilization were much more favourable than in the autumn of 1990. Maximum fluxes were detected as soon as one day after urea application. Similarly, in field experiments by Harper et al. (1983) and Hargrove et al. (1987), NH3 losses began just 10 and 16 h, respectively, after urea application in summer. These findings are in sharp contrast with those from laboratory experiments quoted in Fenn & Hossner (1985) where at least 2 days of favourable conditions for urea hydrolysis had to persist for substantial NH3 losses to occur. In our case, it can be assumed that urea was completely hydrolyzed after 2 days on the surface and mixed incorporation treatments. Ammonia volatilization did not cease at nighttime. After 2 days, the soil surface had largely dried. Measurements on shaded, moister and less hardened patches frequently gave higher fluxes than those on patches exposed to direct sunshine. This may be due to ammoniacal N remaining in water solution close to the soil surface on the wetter spots. On July 12 and 13, highest fluxes on the surface treatments were mainly measured in the late morning hours. This diurnal dependency of fluxes is caused by higher atmospheric relative humidity and soil moisture contents at daybreak in combination with the greatest increases in air and soil temperatures and wind speed in the morning
hours, leading to high evaporation rates at the soil surface (Ferguson et al., 1988; Hargrove et al., 1977; McInnes et al., 1986a). Evaporation causes a transport of dissolved ammoniacal N and unhydrolyzed urea with the water stream towards the soil surface. On the mixed incorporation treatments, the highest NH3 emissions on July 12 and 13 were measured in the early evening hours, when soil temperatures at 5 cm depth were 1–4◦ C higher than in the late morning hours. This led to continuing evaporation and upward transport of ammoniacal N from greater depths. Similar findings were reported by McInnes et al. (1986a). Following the drying of the soil surface on July 13 and 14, fluxes on the microplots with uniform incorporation (Figures 7b,d) exceeded those from surface broadcast fertilizer (Figures 7a,c). In case of a dry soil surface, ammonia is no longer in solution and evaporation rates and upward convective transport of ammoniacal N are reduced. This can, however, be compensated by an increased gas phase diffusion of NH3 (Kirk & Nye, 1991), as was confirmed in our soil column experiments and simulations (Roelcke et al., 1996). On the mixed treatments, volatilization at this stage effectively took place within the soil to a greater extent. During the July 13–15 period, there was a strong dew formation at nighttime. Soil moisture contents (0–5 cm) had decreased to 8.8% by July 15. After a short intense rainshower of 15 mm on July 16, NH3 -volatilization strongly increased for 3 days on all microplots, except for the point placement treatments. On one of the urea-amended plots, fluxes exceeded the initial high fluxes. The dewfall and rain may have caused remains of urea to be hydrolysed as well as a desorption of ammoniacal N, making ammonia in the water available for efflux. Comparison of cumulative fluxes (A) 1990 winter wheat experiment The cumulative NH3 fluxes are compared in Figures 8 and 9. Losses measured on the replicate treatments (field sites I and II) agreed very well in the case of urea (Figure 8), but differences were larger in the case of surface-applied NH4 HCO3 (Figure 9). Table 3 gives the fluxes summed over the whole period of measurement as mean values for the different treatments. Mean absolute and relative NH3 losses were 2.6 times higher (P<0.001) with NH4 HCO3 than with urea. However, the effect of the type of fertilizer was entirely due to the surface applications (significant interaction between factors fertilizer and application
83
Figure 7. Four examples of NH3 fluxes measured following fertilization of summer maize in 1991. Table 3. Cumulative relative and absolute NH3 fluxes following N fertilizer application in both experiments. 1990 winter wheat: mean values of two (urea surface applied four) replicates. 1991 summer maize: mean values of two replicates. The X besides the observed figures represents a scaling factor since the method underestimates the actual fluxes Application method depth [cm]
Losses
NH4 HCO3 (100 kg N ha−1 ) (200 kg N ha−1 )
Urea (100 kg N ha−1 ) (200 kg N ha−1 )
1990 winter wheat: Surface 0 Mixed Incorp. 0–15 Zero N
(%) (kg N ha−1 ) (%) (kg N ha−1 )
X 4.46 X 4.46 X 0.10 X 0.10
X 7.13 X 14.26 X 0.18 X 0.36 Traces
X 1.08 X 1.08 X 0.29 X 0.29
X 2.03 X 4.06 X 0.22 X 0.45
1991 summer maize: Surface 0 Mixed Incorp. 0–15 Point placed 10 Zero N
(%) (kg N ha−1 ) (%) (kg N ha−1 ) (%) (kg N ha−1 )
X 4.78 X 4.78 X 5.97 X 5.97 X 0.60 X 0.60
X 5.93 X 11.86 X 7.40 X 14.79 X 2.80 X 5.60 Traces
X 2.30 X 2.30 X 1.44 X 1.44 X 0.76 X 0.76
X 3.52 X 7.05 X 3.68 X 7.37 X 1.29 X 2.58
84
Figure 8. Cumulative fluxes following urea application in 1990. Comparison of two different fertilization dates, two application rates and two methods of application. (c), (d) reprinted with permission from “Soil Erosion and Dryland Farming”. Copyright CRC Press, Boca Raton, Florida, 2000
Figure 9. Cumulative fluxes following NH4 HCO3 application in 1990. Comparison of two application rates and two methods of application. Reprinted with permission from “Soil Erosion and Dryland Farming”. Copyright CRC Press, Boca Raton, Florida, 2000
method, P<0.001). Mean absolute and relative NH3 losses on the incorporation treatments were only 6.6% of those on the surface treatments (P<0.001). Losses after incorporation of NH4 HCO3 were only 1–4% of those after surface broadcasting, but for urea, 7–30% (Table 3). Mean absolute NH3 losses were 3.3 times (P<0.001) and relative losses 1.7 times (P<0.05)
higher on the treatments with 200 kg N ha−1 than with 100 kg N ha−1 (Table 3, Figures 8 and 9). This effect of the application rate was again confined to the surface treatments (significant interaction, P<0.001). A doubling of the N application rate resulted in a higher increase in absolute losses with NH4 HCO3 than with urea (significant interaction, P<0.05).
85
Figure 10. Cumulative fluxes following NH4 HCO3 and urea application in 1991. Comparison of two application rates and three methods of application. Reprinted with permission from “Soil Erosion and Dryland Farming”. Copyright CRC Press, Boca Raton, Florida, 2000
The significantly higher NH3 emissions following application of NH4 HCO3 than that of urea are consistent with results from many Chinese field experiments (e.g., Cai et al., 1986, 1998; Zhu et al., 1989). Commercial NH4 HCO3 is a very unstable and volatile substance with a high vapour pressure, which can directly decompose to NH3 , CO2 and H2 O even before application. In contrast, the processes of urea dissolution and diffusion in soil, hydrolysis and the transport of the hydrolysis products to the soil surface are involved for NH3 -volatilization from urea, and shallower gradients of ammoniacal N and pH build up at the soil surface as compared to NH4 HCO3 . Similarly, a higher N application rate leads to higher resulting NH4 + and pH-gradients and an increase in the percentage of N volatilized (Rachhpal-Singh and Nye, 1986). Incorporation reduces direct NH4 HCO3 -decomposition and the build-up of high NH3 -N concentrations at the immediate soil surface (dilution effect); it may, however, facilitate hydrolysis of urea. Using identical treatments as in situ with the forced-draft system in the laboratory (Roelcke, 1994), the same significant
differences between treatments and similar significant interactions between factors were found as in the winter wheat experiment. (B) 1991 summer maize experiment Since part of the maximum fluxes from urea had been missed in the maize experiment, the multifactorial analyses of variance were carried out separately for the two fertilizers. Differences in cumulative fluxes between replicate treatments on the two sites, though not systematic, were partly greater than the differences between the single treatments themselves (Figure 10). No significant differences were found between surface and uniform incorporation treatments (Table 3). Mean cumulative fluxes following point placement of NH4 HCO3 were about 35% of those following surface broadcasting (P<0.05), and 28% of those following uniform incorporation (P<0.01). For urea, the corresponding portions were 35% (P<0.05 for relative losses) and 39% (n.s.). Mean absolute losses were 2.8 (NH4 HCO3 ) and 3.8 (urea) times higher on the treatments with 200 kg N ha−1 than on those with
86 100 kg N ha−1 (P<0.01), while relative losses were 1.4 times (NH4 HCO3 ) and 1.9 times (urea) higher (n.s.; Figure 10). No significant interactions between factors were found in summer. Comparing both experiments (separately for each fertilizer, surface broadcast and uniform incorporation only), absolute and relative losses were 2.0 times (NH4 HCO3 ; P<0.01) and 2.5 times (urea; P<0.001) higher in summer 1991 than in autumn 1990. In the case of NH4 HCO3 , this was due to the strong increase on the incorporation treatments (significant interaction between season and application method, P<0.01). The increase in urease activity is the major cause for higher NH3 -volatilization from urea at higher temperatures. In the summer of 1991 the high evaporation rates resulted in increased upward convective transport of unhydrolyzed urea and ammoniacal N as well as increased diffusion of gaseous NH3 . This made uniform incorporation of fertilizer much less effective in reducing NH3 -volatilization than for the wheat fertilization. The traditional point placement of N fertilizer to maize was the sole method successful in reducing NH3 losses in summer (Figure 10). These findings are consistent with results of 15 N experiments carried out on the same sites in the summer of 1990, where uniform incorporation did not improve N fertilizer recovery of a maize crop as compared to a surface application, but where deep point placement resulted in the highest fertilizer recovery and significantly higher amounts of fertilizer N taken up by cobs than in the mixing treatment (Rees et al., 1997).
tained with micrometeorological methods for similar treatments. The Dräger-Tube method presented here combines the practicality of enclosure methods for in situ measurements with the advantage of minimizing alterations of natural conditions under the enclosures. It allows for a high mobility, has few technical requirements and comparatively low cost, making it applicable in remote areas. The method’s main limitation is the low air exchange rate, leading to an underestimation of the actual volatilization rates by one order of magnitude.1 It is, however, very useful for relative comparisons between treatments, which resulted in significant differences in cumulative fluxes.
Acknowledgements The volatilization cups were constructed by Mr. W. Walter of the Institute of Soil Science at Hannover University. We thank Mr. Li Tuanmin and Mr. Zhao Junhao in Yangling for technical assistance, and the farmers in the villages of Shangzhuang and Lingjiao for providing us with the experimental sites and their help. The experimental work presented here was carried out with financial support from the German Research Foundation (DFG Ri 269/25), the VolkswagenFoundation (VW I/65 586) and the Chinese Ministry of Agriculture.
References Conclusions The experiments clearly confirmed the potential for ammonia losses under the soil and climatic conditions investigated. The loss-reducing effect of a uniform incorporation of N fertilizer into the topsoil (for winter wheat) and of a point placement (summer maize) was clearly demonstrated. Although split applications would also be preferable, their beneficial effect could easily be outweighed if the fertilizer is not incorporated at sufficient depth (at least 15 cm). The measurements clearly reflected the characteristic patterns of NH3 fluxes for the different fertilization treatments and the effects of environmental factors on the time course of NH3 fluxes. For all treatments and both seasons, cumulative ammonia fluxes ranged between 0.1 and 8.8% of the applied N or 0.1 and 17.6 kg N ha−1 . These amounts were lower than most values ob-
Anderson JPE (1982) Soil respiration. In: Page AL et al. (ed) Methods of Soil Analysis, Part 2, 2nd edn, Agron. Monogr. 9, pp 831–871. Madison, Wisc: American Society of Agronomy, Soil Science Society of America Aggarwal RK & Praveen-Kumar (1994) Availability and management of nitrogen in soils of arid ecosystem. Ann Arid Zone 33: 1–18 Beauchamp EG, Kidd GE & Thurtell G (1982) Ammonia volatilization from liquid dairy cattle manure in the field. Can J Soil Sci 62: 11–19 Bouwmeester RJB, Vlek PLG & Stumpe JM (1985) Effect of environmental factors on ammonia volatilization from a ureafertilized soil. Soil Sci Soc Am J 49: 376–381 Cai GX, Zhu ZL, Trevitt ACF, Freney JR & Simpson JR (1986) Nitrogen loss from ammonium bicarbonate and urea fertilizers applied to flooded rice. Fertil Res 10: 203–215 1 Considering the very low air exchange rate, the values detected could also be interpreted as an indicator for the partial pressure of NH3 in equilibrium with the soil solution at the soil surface, itself a function of ammoniacal N concentration, pH and temperature (Cai GX, 1998, pers. comm.).
87 Cai GX, Fan XH, Yang Z & Zhu ZL (1998) Gaseous loss of nitrogen from fertilizers applied to wheat on a calcareous soil in North China Plain. Pedosphere 8: 45–52 Christianson CB, Bationo A, Henao J & Vlek PLG (1990) Fate and efficiency of N fertilizers applied to pearl millet in Niger. Plant Soil 125: 221–231 Craig JR & Wollum WG II (1982) Ammonia volatilization and soil nitrogen changes after urea and ammonium nitrate fertilization of Pinus taeda L. Soil Sci Soc Am J 46: 409–414 Denmead OT (1983) Micrometeorological methods for measuring gaseous losses of nitrogen in the field. In: Freney JR & Simpson JR (eds) Gaseous Loss of Nitrogen from Plant–Soil Systems, pp 133–157. The Hague: Nijhoff-Junk Denmead OT, Simpson JR & Freney JR (1977) A direct field measurement of ammonia emission after injection of anhydrous ammonia. Soil Sci Soc Am J 41: 1001–1004 Drägerwerk AG (1994) Dräger-Tube Handbook: Soil, Water and Air Investigations as well as Technical Gas Analysis, 9th edn, Lübeck, Germany: Drägerwerk Aktiengesellschaft FAO (2000) Food and Agriculture Organization of the United Nations, FAOSTAT – FAO Statistical Databases, Agriculture Data. http://apps.fao.org Fenn LB & Hossner LR (1985) Ammonia volatilization from ammonium or ammonium-forming nitrogen fertilizers. Adv Soil Sci 1: 123–169 Fenn LB & Kissel DE (1973) Ammonia volatilization from surface applications of ammonium compounds to calcareous soils: I. General theory. Soil Sci Soc Am Proc 37: 855–859 Ferguson RB, McInnes KJ, Kissel DE & Kanemasu ET (1988) A comparison of methods of estimating ammonia volatilization in the field. Fertil Res 15: 55–69 Fox RH, Piekielek WP & Macneal KE (1996) Estimating ammonia volatilization losses from urea fertilizers using a simplified micrometeorological sampler. Soil Sci Soc Am J 60: 596–601 Hargrove WL, Kissel DE & Fenn LB (1977) Field measurements of ammonia volatilization from surface applications of ammonium salts to a calcareous soil. Agronomy J 69: 473–476 Hargrove WL, Bock BR, Raunikar RA & Urban WJ (1987) Comparison of a forced-draft technique to nitrogen-15 recovery for measuring ammonia volatilization under field conditions. Soil Sci Soc Am J 51: 124–128 Harper LA (1988) Comparisons of methods to measure ammonia volatilization in the field. In: Bock BR & Kissel DE (eds) Ammonia volatilization from urea fertilizers, Bulletin Y-206, pp 93–109. Muscle Shoals, AL: National Fertilizer Development Center, TVA Harper LA & Sharpe RR (1998) Atmospheric ammonia: Issues on transport and nitrogen isotope measurement. Atmos Environ 32: 273–277 Harper LA, Catchpoole VR, Davis R & Weir KL (1983) Ammonia volatilization: Soil, plant, and microclimate effects on diurnal and seasonal fluctuations. Agron J 75: 212–218 Huber J & Amberger A (1989) NH3 -Verluste unter verschiedenen Anbaubedingungen. Kongreßband 1989. VDLUFASchriftenreihe 30: 109–115 Institute of Soil Science, Academia Sinica (1990) Soils of China (English edn), Beijing, China: Science Press Kang SZ, Zhong HZ & Zhang X (1992) A model of the crop water requirements and irrigation for different areas of Shaanxi Province. Chinese Water Resources and Electricity Press (in Chinese) Katyal JC, Singh B, Vlek PLG & Buresh RJ (1987) Efficient nitrogen use as affected by urea application and irrigation sequence. Soil Sci Soc Am J 51: 366–370
Kirk GJD & Nye PH (1991) A model of ammonia volatilization from applied urea. V. The effects of steady-state drainage and evaporation. VI. The effects of transient-state water evaporation. J Soil Sci 42: 103–125 Kissel DE, Brewer HL & Arkin GF (1977) Design and test of a field sampler for ammonia volatilization. Soil Sci Soc Am J 41: 1133–1138 Leuning R, Freney JR, Denmead OT & Simpson JR (1985) A sampler for measuring atmospheric ammonia flux. Atmos Environ 19: 1117–1124 McInnes KJ, Kissel DE & Kanemasu ET (1985) Estimating ammonia flux: a comparison between the integrated horizontal flux method and theoretical solutions of the diffusion profile. Agron J 77: 884–889 McInnes KJ, Ferguson RB, Kissel DE & Kanemasu ET (1986a) Field measurements of ammonia loss from surface applications of urea solution to bare soil. Agron J 78: 192–196 McInnes KJ, Ferguson RB, Kissel DE & Kanemasu ET (1986b) Ammonia loss from applications of urea-ammonium nitrate solution to straw residue. Soil Sci Soc Am J 50: 969–974 Pacholski A, Cai GX, Fan XH, Roelcke M, Chen D & Richter J (1999) Eine einfache Methode zur Messung der AmmoniakVolatilisation in situ — Vergleichsmessungen in Henan, V.R. China. Mitteilgn Dtsch Bodenkundl Gesellsch 91 (II): 842–845 Pilbeam CJ & Hutchison D (1998) Fate of nitrogen applied in different fertilizers to the surface of a calcareous soil in Syria. Nutr Cycling Agroecosyst 52: 55–60 Rachhpal-Singh & Nye PH (1986) A model of ammonia volatilization from applied urea. I. Development of the model. II. Experimental testing. III. Sensitivity analysis, mechanisms, and applications. J Soil Sci 37: 9–40 Rachhpal-Singh & Nye PH (1988) A model of ammonia volatilization from applied urea. IV. Effect of method of urea application. J Soil Sci 39: 9–14 Rees RM, Roelcke M, Li SX, Wang XQ, Li SQ, Stockdale EA, McTaggart I, Smith KA & Richter J (1997) The effect of fertilizer placement on nitrogen uptake and yield of wheat and maize in Chinese loess soils. Nutr Cycling Agroecosyst 47: 81–91 Richter J (1972) Zur Methodik des Bodengashaushaltes. II. Ergebnisse und Diskussion. Z Pflanzenern Bodenkde 132: 220–239 Roelcke M (1994) Die Ammoniak-Volatilisation nach Ausbringung von Mineraldünger-Stickstoff in carbonatreichen LößAckerböden. Ph.D. dissertation, Braunschweig Technical University. In: Göttinger Beiträge zur Land- und Forstwirtschaft in den Tropen und Subtropen 92, Göttingen, Germany: Erich Goltze Roelcke M, Han Y, Li SX & Richter J (1996) Laboratory measurements and simulations of ammonia volatilization from urea applied to calcareous Chinese loess soils. Plant Soil 181: 123– 129 Roelcke M, Rees RM, Li SX & Richter J (2000) Studies of the nitrogen cycle on the southern edge of the Chinese Loess Plateau. In: Laflen J, Huang C & Tian J (eds) Proceedings of Conference Soil Erosion and Dryland Farming, Xi’an, China, Sept. 1997, Chapter 12, pp 103–120. New York: CRC Press Ryden JC & Lockyer DR (1985) Evaluation of a system of wind tunnels for field studies of ammonia loss from grassland through volatilization. J Sci Food Agric 36: 781–788 Schjørring JK, Sommer SG & Ferm M (1992) A simple passive sampler for measuring ammonia emission in the field. Wat Air Soil Pollut 61: 13–24 Statistical Graphics Corporation (1993) STATGRAPHICS, Maryland: STSC Inc.
88 USDA (1994) Keys to Soil Taxonomy, 6th edn, Washington, DC: USDA Soil Conservation Service Vlek PLG & Stumpe JM (1978) Effects of solution chemistry and environmental conditions on ammonia volatilization losses from aqueous systems. Soil Sci Soc Am J 42: 416–421 Wang XZ, Zhang SL, Xu YH, Zhu ZL & Freney JR (1991) Fate of fertilizer nitrogen applied to crops grown on fluvo-aquic soil in Huang-Huai-Hai plain. Pedosphere 1: 145–155 Water Economy Bureau Xianyang (1989) Hydrology Manual, Xianyang, Shaanxi, China: Bureau of Water Economy and Soil and Water Conservation of Xianyang City (in Chinese) Wilson JD, Thurtell GW, Kidd GE & Beauchamp EG (1982) Estimation of the rate of gaseous mass transfer from a surface source plot to the atmosphere. Atmos Environ 16: 1861–1867
Wilson JD, Catchpoole VR, Denmead OT & Thurtell GW (1983) Verification of a simple micrometeorological method for estimating the rate of gaseous mass transfer from the ground to the atmosphere. Agric Meteorol 29: 183–189 Zhang SL, Cai GX, Wang XZ, Xu YH, Zhu ZL & Freney JR (1992) Losses of urea-nitrogen applied to maize grown on a calcareous fluvo-aquic soil in North China Plain. Pedosphere 2: 171–178 Zhu ZL, Cai GX, Simpson JR, Zhang SL, Chen DL, Jackson AV & Freney JR (1989) Processes of nitrogen loss from fertilizers applied to flooded rice fields on a calcareous soil in north-central China. Fertil Res 18: 101–115