Measurement of ammonia volatilization loss using a dynamic chamber technique: A case study of surface-incorporated manure and ammonium sulfate in an upland field of light-colored Andosol

Abstract

The present study aimed to elucidate ammonia (NH₃) volatilization loss following surface incorporation (0–15 cm mixing depth) of nitrogen (N) fertilizer in an upland field of light-colored Andosol in central Japan. A dynamic chamber technique was used to measure the NH₃ effluxes. Poultry manure, pelleted poultry manure, cattle manure, pelleted cattle manure and ammonium sulfate were used as N fertilizers for basal fertilization to a bare soil with surface incorporation. All three experiments in summer and autumn 2007 and in summer 2008 showed negligible NH₃ volatilization losses following the application of all N fertilizers with the same application rate of 120 kg N ha⁻¹ as total N; these negligible losses were primarily ascribed to chemical properties of the soil, that is, its high cation exchange capacity (283 mmol_c kg⁻¹ dry soil) and relatively low pH(H₂O) (5.9). In addition, the surface incorporation, the very small ratio of ammoniacal N to total N for the manure, and the decrease in soil pH to ≤5.5 following applications of ammonium sulfate were also advantageous to the inhibition of NH₃ volatilization loss from the field-applied N fertilizers.

Key words:

• ammonium sulfate
• ammonia emission
• composted manure
• surface incorporation
• volcanic ash soil

Introduction

Anthropogenic emission of ammonia (NH₃) is one of the main sources of airborne nitrogen (N). Agriculture is the largest source of atmospheric NH₃ to which livestock waste and field-applied N fertilizer primarily contribute. For example, agriculture accounts for 90% of the total NH₃ emissions in Europe (Erisman et al. 2008). Information regarding NH₃ emission factors is necessary to assess the impacts of airborne NH₃ emission, transportation and deposition. Application methods, weather and soil conditions, and fertilizer types strongly affect NH₃ emission from field-applied N fertilizers (Huijsmans et al. 2003). Therefore, the actual status of NH₃ emission induced by fertilizer application should be elucidated for each region and country to derive representative emission factors of NH₃.

Relevant studies of NH₃ emission from field-applied N fertilizer have been increasing only recently in Japan. For example, Hayashi et al. (2006, 2008) reported NH₃ volatilization losses following applications of urea to a paddy field, and Matsunaka et al. (2008) showed NH₃ volatilization losses following applications of cattle slurry to grassland. Using incubation experiments, Fueki et al. (2007) also reported the potential for NH₃ emission from volcanic ash soils following applications of several N fertilizers. However, little information on the NH₃ emission from Japanese upland fields is available. Volcanic ash soils cover approximately 50% of Japanese upland fields (Takata et al. in press). Furthermore, it is known that physicochemical properties are largely different between volcanic ash soils and non-volcanic ash soils (Shoji et al. 1993), which leads to a possible difference in NH₃ emissions. Thus, concrete data on NH₃ emissions from upland fields of volcanic ash soils should be accumulated to establish NH₃ emission factors.

Small experimental plots, such as lysimeters, are advantageous for simultaneous investigations of NH₃ emissions under various treatments. However, a micrometeorological technique that gives relatively accurate fluxes, but requires a large area with homogeneous conditions (e.g. Fowler et al. 2001) is not applicable to such small plots. Instead, a dynamic chamber technique (e.g. Kissel et al. 1977) seems to be the best alternative. Therefore, the objectives of the present study were to use a dynamic chamber technique to measure NH₃ effluxes from lysimeter plots and to elucidate the NH₃ volatilization loss following basal fertilization in a Japanese upland field of volcanic ash soil. Specifically, NH₃ volatilization loss from surface-incorporated composted manure and ammonium sulfate in an upland field of light-colored Andosol, a typical soil type in upland fields in central Japan, was investigated.

Materials and methods

Experimental field

The study site included lysimeter plots located at the National Institute for Agro-Environmental Sciences (NIAES), central Japan (36°01′N, 140°07′E); the site was originally established to study greenhouse gas emissions. The study site has been used to elucidate and evaluate carbon and nitrogen cycles in agroecosystems, including greenhouse gases and relevant substances. The present study forms part of this overall aim. Six lysimeter plots of upland fields were used in the present study. Each plot, 3 m × 3 m, was filled with light-colored volcanic ash soil, classified as Andosol (Food and Agriculture Organization 1998). Soil from the plowed layer and the subsoil, both Andosols, was collected in March 1994 from depths of 0–15 cm at an upland field in Nishinarado and from depths of 50–100 cm at a construction field in Fukuoka, both in Tsukubamirai City, that is, the former Yawara Village. The subsoil was used to fill the lysimeter plots to a depth of 80 cm, and then 20 cm of plowed layer soil was filled onto the subsoil. Light-colored Andosols, the parent material of which is a mixture of sediment and volcanic ash or resediment of volcanic ash, are common soil types in central Japan. The main properties of the soil in the lysimeter plots are listed in Table 1; the soil texture refers to the plowed layer and the other values refer to the topsoil (depth 0–5 cm). The soil pH values shown in Table 1 are the mean values of six lysimeters just before summer cropping in 2007. The cation exchange capacity (CEC) and exchangeable cation values are the mean values of six lysimeters in November 2008. Soil samples to determine CEC and exchangeable cations were collected from depths of 0–5 cm in the topsoil at five points per lysimeter. A composite sample per lysimeter was then made and air-dried. The CEC and exchangeable cations were determined using the semi-micro Schollenberger method (Schollenberger and Simon 1945).

Table 1 Main properties of the soil in the lysimeter plots

Classification	Texture	Bulk density	Sand	Silt	Clay	T-C	T-N
		(Mg m^−3)	(% dry soil)
Andosols	Clay loam	0.56	37	45	18	3.13	0.26
pH	pH	CEC	Ex. Na^+	Ex. K^+	Ex. Mg^2+	Ex. Ca^2+	BS
(H2O)	(KCl)	(mmolc kg^−1 dry soil)					(%)
5.9	5.1	283	1	11	26	131	59.7
BS, base saturation; CEC, cation exchange capacity; Ex., exchangeable cations; T-C, total carbon; T-N, total nitrogen. Source of T-C and T-N: Akiyama et al. (2000).

The lysimeter plots have been used in experiments since the summer of 1995. Cultivation experiments of carrot, barley, qing-geng-cai and spinach were conducted from summer cropping in 1995 to summer cropping in 2000. Magnesium lime was applied to correct the soil pH at the beginning of summer and autumn cropping in 1997 at a rate of 2,200 kg ha⁻¹, and at the beginning of summer cropping in 1998 at a rate of 2,000 kg ha⁻¹. The lysimeter plots were fallowed from July 2000 to May 2002. The cultivation history after May 2002 is shown in Table 2. The lysimeter plots were also fallowed from September 2002 to September 2005. Leaf vegetables were cultivated in recent years.

Cultivation and fertilizer application

The NH₃ volatilization loss from field-applied N fertilizers was investigated during three experimental periods: summer cropping in 2007 (sum-07), autumn cropping in 2007 (aut-07) and summer cropping in 2008 (sum-08). Komatsuna (Brassica rapa L. var. perviridis) was cultivated during these experiments.

Three treatments of basal fertilizer were applied during each of the experimental periods, that is, applications of manure, pelleted manure and chemical fertilizer. Two lysimeter plots per treatment were used. Poultry manure (PM) and pelleted poultry manure (PP) without additional materials were used in sum-07, whereas cattle manure (CM) and pelleted cattle manure (CP) with sawdust and straw as bedding materials were used in aut-07 and sum-08. Poultry manure and PP were provided by Mie Prefecture Agricultural Research Institute, Japan, and CM and CP were provided by the National Agricultural Research Center for Kyushu Okinawa Region, Japan. The manure was well fermented (composted) and was not fresh farmyard manure. The manure and the pelleted manure had the same origin. The manure and the pelleted manure were special in terms of their low water content, which was a necessary condition in pelletizing. No problems were observed regarding changes in the composition of CM and CP between aut-07 and sum-08 because they were well dried. In contrast, the chemical fertilizer was a combination of ammonium sulfate (AS), potassium chloride or potassium bromide, and superphosphate.

Table 2 Cultivation history of the lysimeter plots

Cropping period^†	Crops	Fertilizer in relation to nitrogen			Application rate	NH3 meas.^§
		Lysimeter 1, 4	Lysimeter 2, 5	Lysimeter 3, 6	(kg N ha^−1)^‡
May–September 2002	Sorghum	No fertilization (cleaning cropping)			0	No
September–November 2005	Chinese cabbage	AS + CM	AS + CM	AS + CM	150 + 120	No
September–November 2006	Komatsuna	PM	PP	AS	120	No
June–July 2007	Komatsuna	PM	PP	AS	120	Yes
September–October 2007	Komatsuna	CM	CP	AS	120	Yes
May–June 2008	Komatsuna	CM	CP	AS	120	Yes
^†Fallow period intervenes between cropping periods. ^‡Application rate as total nitrogen. ^§Measurement of ammonia volatilization following fertilizer application. Sorghum, Sorghum bicolor (L.) Moench; Chinese cabbage, Brassica rapa L. var. glabra; Komatsuna, Brassica rapa L. var. perviridis; AS, ammonium sulfate; CM, cattle manure; CP, pelleted cattle manure; PM, poultry manure; PP, pelleted poultry manure.

The dates of basal fertilization were 11 June 2007, 19 September 2007 and 27 May 2008 for sum-07, aut-07 and sum-08, respectively. These fertilizers were applied to the lysimeter plots by surface incorporation (0–15 cm mixing depth) to bare soil using a mini-tiller. The application rate was 120 kg N ha⁻¹ as total N (T-N). The experiments were designed to apply the fertilizers at the same application rate as T-N. As a result, the application rate in the present study was low for the manures compared with the actual rate used in upland cropping in Japan. The main properties of the manure and the pelleted manure are shown in Table 3. The manure pH was determined from an extract solution obtained by adding 200 mL of deionized water to 50 g of fresh manure. The PM and PP were relatively low in carbon content, but rich in N content. Ammoniacal N (NH₄-N) accounted for 6–11% of T-N. In contrast, CM and CP were rich in carbon content, obtained to an extent from plant materials, but were relatively low in N content. NH₄-N accounted for 4–5% of T-N.

Table 3 Main properties of the manure and pelleted manure used in the present study

Manure type	Condition	Bedding material	pH(H2O)	DM% (w/w)
PM	Fermented	No	8.6	81
PP	Fermented then pelleted	No	8.9	76
CM	Fermented	Sawdust + straw	7.7	82
CP	Fermented then pelleted	Sawdust + straw	6.7	88
Manure type	T-C	T-N	C:N ratio	NH4-N	NO3-N	NH4-N/T-N
	(g kg^−1 dry weight)			(g kg^−1 dry weight)		(%)
PM	289	62.5	4.6	3.78	0.0012	6.0
PP	268	59.8	4.5	6.75	0.0032	11.3
CM	407	28.1	14.5	1.22	1.44	4.3
CP	397	30.4	13.1	1.41	4.28	4.7

Row sowing of Komatsuna (12 rows per lysimeter) with a sufficient number of seeds for thinning was conducted by hand just after the basal fertilization. Immediate sowing after fertilization resulted in no failure to germinate because the manure was well fermented. Seedlings were thinned out two to three times until harvest. The mean yields of Komatsuna for the six plots were 1.7–1.8, 0.8–1.0 and 0.3–0.9 Mg dry matter ha⁻¹ for sum-07, aut-07 and sum-08, respectively; the CM and CP plots in sum-08 had particularly low yields of 0.3–0.4 Mg dry matter ha⁻¹. The low application rate of manure in the present study compared with the amount generally used in upland cropping seemed to trigger the low yields observed in sum-08. However, the growth stages of Komatsuna during the measurements of the NH₃ efflux were seeds to small seedlings. Furthermore, the chambers were set at the lysimeter plots to avoid the rows of Komatsuna seeds. It is, therefore, hypothesized that the cultivation conditions of Komatsuna had little impact on the observed NH₃ volatilization losses.

Field measurements

A dynamic chamber method, that is, an enclosure technique with forced ventilation, was adopted in the present study as the most appropriate method applicable to lysimeter plots with a limited area. A schematic view of the dynamic chamber system is shown in Fig. 1. The system was composed of two sampling lines, one to collect NH₃ in the ambient air and the other to collect NH₃ in the exhaust air of a dynamic chamber that contained both ambient and volatilized NH₃. The two sampling lines were separated in sum-07 (Fig. 1a); however, they had the same inlet to homogenize the ambient NH₃ between the two sampling lines in aut-07 and sum-08 (Fig. 1b). The rectangular, stainless steel chamber had two vents on each of the narrow sides, a base area of 400 cm² and a volume of 4 L (excluding anterior and posterior spaces). The chamber used in aut-07 and sum-08 had an open–close top. The reason for the change in the designs of the dynamic chamber system, that is, the same inlet for the two sampling lines and the open–close top of the chamber, in and after aut-07 was to further reduce errors in measurement. A two-stage filter holder (NL-I or NL-O; NILU, Kjeller, Norway) was used as the dry NH₃ collector. The first (upstream) stage removed dust and particles, including particulate ammonium, from the air using a polytetrafluoroethylene (PTFE) filter with a pore size of 0.8 μm (T080A047A; Advantec, Tokyo, Japan). The second (downstream) stage collected NH₃ in the air using a cellulose filter (51A; Advantec) impregnated with phosphoric acid. The preparation method and the advantages of the dry NH₃ collector were stated in Hayashi et al. (2006). Six chamber system sets were prepared and one set per lysimeter was used.

Figure 1 Schematic view of the dynamic chamber system used in the present study. (a) The chamber used in the experiment in summer 2007 (sum-07) and (b) the chamber used in the experiments in autumn 2007 (aut-07) and summer 2008 (sum-08). PTFE, polytetrafluoroethylene.

The NH₃ emission was measured for 7 or 8 days following each fertilizer application. The exact time periods were 11–17 June 2007, 19–26 September 2007 and 27 May–2 June 2008 for sum-07, aut-07 and sum-08, respectively. The measurement frequency in each experiment was as follows: sum-07, three or four times per day with sampling times of 8.00–11.00 hours, 12.00–15.00 hours and 16.00–19.00 hours (and 20.00–23.00 hours when measurements were four times per day); aut-07, three times per day with sampling times of 8.00–11.00 hours, 12.00–15.00 hours and 16.00–19.00 hours; and sum-08, three times per day with sampling times of 8.30–10.30 hours, 12.00–14.00 hours and 15.30–17.30 hours.

The chamber was inserted into the soil to a depth of half its height (approximately 2 L internal volume) just after fertilization, avoiding the rows of Komatsuna seeds. In sum-07, the chamber was removed during the intervening periods and reinserted just before the next measurement. In contrast, in aut-07 and sum-08, an improved chamber with an open–close top was left in the soil after the installation. The top was closed during the measurements and left open during the intervening periods. The airflow rate in each sampling line was approximately 17 L min⁻¹, which corresponded to an air exchange rate of approximately 8.5 exchange volume min⁻¹ for the chamber, with a wind velocity within the chamber of approximately 0.06 m s⁻¹. Except for the measuring times, rainwater was allowed to reach the soil surface where the chambers were located because the chambers were removed from the plots (sum-07) or the chamber tops were opened (aut-07 and sum-08). Although the chamber tops intercepted the rainwater during the measurements, the measuring times were slightly staggered to avoid heavy rain when forecasted.

After the samples were collected, the NH₃ trapped on the phosphoric acid impregnated filter was extracted by adding deionized water. The concentration in the extracted solution was determined using a flow injection analyzer (AQLA-1000; Aqualab, Tokyo, Japan). The NH₃ emission was expressed by.

(1)

where F _vol denotes the NH₃ flux (μg N m⁻² s⁻¹), C _cham and C _amb denote the NH₃ concentration in the chamber exhaust and the ambient air, respectively (μg N m⁻³), Q is the cumulative airflow (m³), A is the chamber base area (m²) and t is the sampling time (s). A linear interpolation was adopted to connect the respective measured fluxes. Then, the positive areas were integrated to derive the cumulative fluxes as volatilization loss.

The air temperature and relative humidity at a height of 0.1 m above the soil surface of lysimeter plot No. 3, corresponding to AS1 plot, were measured every 10 min using a sensor (HMP45A; Vaisala, Helsinki, Finland) and recorded on a datalogger (CR10X; Campbell Scientific, North Logan, UT, USA). Hourly rainfall data were obtained from the weather observatory of NIAES. Water-filled pore spaces (WFPS) of the soil at depths of 5 and 10 cm were measured at each lysimeter plot using time domain reflectometry (TDR) sensors (CS615; Campbell Scientific).

Soil sampling for chemical analysis was conducted every 2–3 days in principle. Five replicate surface soil (0–5 cm depth) samples were collected from each lysimeter and then merged into a composite sample. Fifteen grams of fresh soil was extracted by adding 100 mL of a 10% (w/v) potassium chloride solution under 1 h of shaking. The extracted solution was used to determine the NH₄-N and nitrate N (NO₃-N) contents of the soil using an auto analyzer (TRAACS2000; Bran + Luebbe, Norderstedt, Germany). Another portion of fresh soil was extracted by adding deionized water with a dry soil : water ratio of 1:5 (w/v) under 1 h of shaking. The pH of both solutions, the former as pH(KCl) and the latter as pH(H₂O), was also measured.

Results

NH₃ emissions

Overall, the measured NH₃ fluxes showed very weak volatilization, indicating a rather negative value, that is, dry deposition, in some cases (Fig. 2). In sum-07, NH₃ volatilization was clear only on the day of fertilization at one of the PM plots and at one of the PP plots and at 4–5 days after fertilization at the PP plots (Fig. 2). The maximum efflux was 0.46 μg N m⁻² s⁻¹ on the day of fertilization at one of the PM plots. The mean air temperature during sum-07 was 22.5°C. The soil received 15 mm of rainfall on the day before fertilization. The soil was moist during sum-07 and the WFPS was approximately 50%, excluding the direct effect of rain (Fig. 3). The rainfall 3–4 days after fertilization perhaps induced the delayed peak of NH₃ volatilization at the PP plots. More specifically, chapping of the pellets as a result of swelling by the absorption of rainwater might have resulted in the volatilization of some of the NH₄-N in the inner pellets.

In aut-07, NH₃ volatilization was entirely negligible at the CM and CP plots. In contrast, very weak volatilization was found at the AS plots for several days following fertilization (Fig. 2). The mean air temperature during aut-07 was 24.0°C. The soil was dry during aut-07 and the WFPS was stable at approximately 30% (Fig. 3); there was no rain during aut-07 and for 1 week prior to the fertilization.

In sum-08, NH₃ volatilization was also weak, although the emissions were slightly larger than those under the same treatments in aut-07 (Fig. 2). Weak volatilization was found at the CP and AS plots; however, no volatilization was observed at any of the plots on rainy days, that is, 2 and 4 days after fertilization (Figs 2, 3). In addition, the rain in the intervening time between measurements could fall onto and moisten the soil inside the chambers. The mean air temperature during aut-07 was 16.4°C. The temperature was particularly low 2–4 days after fertilization (Fig. 3). The soil was moist during sum-08 and the WFPS was approximately 40%, excluding the direct effect of rain (Fig. 3).

Figure 2 Ammonia emission following surface incorporation of poultry manure (PM), pelleted poultry manure (PP), cattle manure (CM), pelleted cattle manure (CP) and ammonium sulfate (AS). Arrows denote the fertilizer applications at a rate of 120 kg N ha−1 as total nitrogen. The numbers 1 and 2 refer to the respective lysimeter plots.

Figure 3 Air temperature and relative humidity at a height of 0.1 m, water-filled pore space (WFPS) at a soil depth of 5 cm, and rainfall. Arrows denote the fertilizer applications at a rate of 120 kg N ha−1 as total nitrogen.

Cumulative NH₃ volatilization loss

Reflecting the weak NH₃ volatilization, the NH₃ volatilization losses were very small for all of the treatments and experiments (Table 4). Losses to the applied NH₄-N were only 0.7, 1.5, 0.0–0.1 and 0.0–0.3% for the PM, PP, CM and CP plots, respectively. In contrast, the AS plots showed no volatilization loss to the applied NH₄-N. Therefore, the losses to the applied T-N were 0% for all treatments excluding the PP plots (0.2% of the applied T-N) (Table 4).

Table 4 Ammonia volatilization loss following surface incorporation (0–15 cm mixing depth) of the manure, pelleted manure and ammonium sulfate in the lysimeter plots of light-colored Andosol

Experiment	Fertilizer	Application rate (kg N ha^−1)		NH3 volatilization loss (%)^†
		T-N	NH4-N	To applied T-N	To applied NH4-N
sum-07 (June 2007)	PM	120	7.3	0.0	0.7
	PP	120	13.5	0.2	1.5
	AS	120	120	0.0	0.0
aut-07 (September 2007)	CM	120	5.2	0.0	0.0
	CP	120	5.6	0.0	0.0
	AS	120	120	0.0	0.0
sum-08 (May–June 2008)	CM	120	5.2	0.0	0.1
	CP	120	5.6	0.0	0.3
	AS	120	120	0.0	0.0
^†Average values of two lysimeter plots. AS, ammonium sulfate; CM, cattle manure; CP, pelleted cattle manure; NH4-N, ammoniacal nitrogen; PM, poultry manure; PP, pelleted poultry manure; T-N, total nitrogen.

Figure 4 Soil pH(H2O) and contents of ammoniacal nitrogen and nitrate nitrogen of the surface soil (depth 0–5 cm). Arrows denote the fertilizer applications at a rate of 120 kg N ha−1 as total nitrogen and the crossbars indicate the period of ammonia flux measurement. PM, poultry manure; PP, pelleted poultry manure; CM, cattle manure; CP, pelleted cattle manure; AS, ammonium sulfate.

Changes in soil chemistry

Changes in the pH(H₂O) and the contents of NH₄-N and NO₃-N of the surface soil (0–5 cm depth) around each of the experimental periods are shown in Fig. 4. In sum-07, the soil pH tended to decrease following fertilization at all plots; this tendency was particularly marked at the AS plots, with a decrease to ≤5.5 (Fig. 4). The increases in the NH₄-N contents were less marked at the PM and PP plots, where the application rates of NH₄-N were less than one-tenth of those at the AS plots (Table 4). However, the NH₄-N contents at the PP plots slightly increased 5 days after fertilization, perhaps because of the release of NH₄-N from the swollen pellets. In contrast, large increases in the NH₄-N contents were found at the AS plots (Fig. 4), which could be ascribed to the rapid dissolution of AS owing to the moist soil and the relatively high temperature (Fig. 3). The decreases in pH were perhaps a result of nitrification in conjunction with a tendency toward large increases in the NO₃-N contents at all plots (Fig. 4).

In aut-07, the decreases in pH were weak at the CM and CP plots, dropping to near 6 and then increasing. In contrast, the decreases in pH were marked at the AS plots, similar to those in sum-07. The NH₄-N contents at the CM and CP plots remained flat because of the small application rates of NH₄-N (Table 4). The relatively small increases in the NH₄-N contents at the AS plots compared with those in sum-07 and sum-08 (Fig. 4) perhaps resulted from the inhibition of the dissolution of AS owing to the dry conditions (Fig. 3). The NO₃-N contents increased for 1 week following fertilization and then began to decrease (Fig. 4).

In sum-08, when the same treatments as those used in aut-07 were adopted, changes in the soil pH and soil inorganic N content showed similar tendencies to those observed in aut-07. However, increases in the NH₄-N contents at the AS plots in sum-08 were larger than those in aut-07, which reflected the facilitation of the dissolution of AS as a result of the moist soil conditions in sum-08. The rainfall during sum-08 likely resulted in the small increases in the NO₃-N content owing to NO₃-N leaching from the surface soil by rainwater.

Discussion

Causes of the negligible volatilization loss

Adsorption of released NH₄-N from N fertilizers onto soil particles inhibits NH₃ volatilization. The present study supported this particularly clearly at the AS plots where no marked NH₃ volatilization took place (Fig. 3), despite marked increases in NH₄-N content following the application of AS (Fig. 4). A large CEC results in further adsorption of ammonium ions (NH₄ ⁺) onto soil particles. Incubation experiments using 22 types of soil in China with CEC values ranging from 63.6 to 254 mmol_c kg⁻¹ dry soil (Duan and Xiao 2000) showed a negative correlation between CEC and NH₃ volatilization (R = −0.568, P < 0.01). Furthermore, this relationship was well explained by an exponential function (R ² = 0.564). According to the results of Duan and Xiao (2000), a CEC above 130 mmol_c kg⁻¹ dry soil strongly inhibited NH₃ volatilization. The CEC of light-colored Andosol in the present study, 283 mmol_c kg⁻¹ dry soil (Table 1), was larger than the maximum value of CEC, 254 mmol_c kg⁻¹ dry soil, in the study of Duan and Xiao (2000). It is, therefore, concluded that the high CEC of the light-colored Andosol was the primary cause of the negligible NH₃ volatilization losses observed in the present study. The low soil pH levels, which dropped to 5.0–5.5 following fertilization at the AS plots (Fig. 4), were also effective in inhibiting NH₃ volatilization because a low pH moves the dissociation equilibrium of NH₄ ⁺ in the liquid phase to a decrease in the fraction of aqueous NH₃ (e.g. Hayashi et al. 2008). In addition, the marked decreases in pH at the AS plots could be ascribed to nitrification in conjunction with increases in soil NO₃-N content (Fig. 4). Furthermore, dry conditions inhibit NH₃ volatilization through a reduction in NH₄-N release from N fertilizers (e.g. Kresge and Satchell 1960), which might, in part, have caused the result obtained in aut-07. In contrast, excessively moist conditions caused by rainfall or irrigation also inhibit NH₃ volatilization through the effects of dilution and leaching (e.g. Whitehead and Raistrick 1991), which might, in part, have caused the result obtained in sum-08. According to Kresge and Satchell (1960), an initial water content of 20%, that is, the approximate field water capacities of the test soils, resulted in the most enhanced NH₃ volatilization loss following surface applications of urea.

NH₃ volatilization originates directly from the NH₄-N in the soil. Fermented manure and its pellets lose most of their NH₄-N as a result of volatilization loss during fermentation. Therefore, fermented manure has a smaller potential for NH₃ volatilization than fresh manure. For example, McGinn and Sommer (2007) reported that the NH₃ volatilization losses following surface applications of fresh manure and compost of beef cattle were 5.7% and 0.2% of the applied T-N, respectively. According to McGinn and Sommer (2007), the very small loss of 0.2% of applied T-N resulted from the application of compost with a rate of 27.7 Mg fresh weight ha⁻¹. Thus, it is inferred that changes in NH₃ volatilization loss would be small if the low application rates of manure in the present study, corresponding to 2.4–5.2 Mg fresh weight ha⁻¹, were increased to those of actual upland cropping (e.g. 10–20 Mg fresh weight ha⁻¹).

Regarding the potential for NH₃ emission from AS, the European Environment Agency (2007) suggested a range of emission factors, depending on climate, of 1.5–2.5% of applied T-N. However, these emission factors include emissions from both surface-applied fertilizer and crop foliage. The European Environment Agency (2007) further noted that a multiplier of 10 should be applied to the emission factors in soils with high pH levels, such as calcareous soils. In the present study, the soil pH was relatively low at 5.9 (Table 1). Surface incorporation was conducted instead of surface application, and foliage emission was not considered. It is, therefore, hypothesized that the potential for NH₃ volatilization loss from the AS plots would be less than that stated by the European Environment Agency (2007).

The surface incorporation method of application has a strong effect on the inhibition of NH₃ volatilization, although no direct comparison of NH₃ volatilization loss between surface incorporation and surface application was conducted in the present study. For example, Huijsmans et al. (2003) reported that the weighted means of NH₃ volatilization loss from field-applied pig slurry by surface application (top-dressing), surface incorporation (mixing with the plowed layer) and deep placement (15–20 cm depth) were 68, 17 and 2% of the applied NH₄-N, respectively. Thus, in the case of slurry, surface incorporation can reduce the NH₃ volatilization loss to a quarter of that experienced with surface application. In contrast, manure has a lower potential for NH₃ volatilization loss compared with slurry because a certain amount of NH₃ has already been lost during the storage and/or fermentation of manure. Furthermore, deep placement is unpractical for solid manure. It is, however, inferred that surface incorporation is more effective to reduce NH₃ volatilization loss than surface application (Huijsmans et al. 2003), although its effect is weaker than the effect recorded for slurry.

A negative NH₃ flux was occasionally found, particularly at the CM and CP plots (Fig. 3). Negative fluxes resulted from higher NH₃ concentrations in the ambient air than in the soil air, which was ruled by the partition equilibrium of NH₃ between the gas–liquid phases in the soil. As a result, the transfer of NH₃ from the atmosphere to the soil, that is, dry deposition (absorption), took place instead of the inverse transfer, that is, volatilization. Dry deposition is strongly affected by micrometeorological factors (e.g. Andersen and Hovmand 1999). The present study aimed to elucidate NH₃ volatilization loss. Moreover, the dynamic chamber technique used in the present study was not suitable for measuring dry deposition, including micrometeorological effects, because the chamber covered the soil surface. Hence, negative fluxes were excluded from discussion in the present study.

The pH levels of upland soils, particularly those from volcanic ash, occasionally dip from the range suitable for cropping. In such cases, alkaline materials, such as lime and dolomite, are applied to adjust the pH level. However, an excessive application of these materials results in an unnecessarily high pH level, which could enhance NH₃ volatilization loss. Furthermore, the applied N fertilizer, in the intervening time between application and incorporation, corresponds to the case of surface-applied fertilizer even if the surface incorporation method is conducted. The longer the time between application and incorporation, the more likely that NH₃ volatilization will occur. However, the potential for NH₃ volatilization loss with the use of manure is smaller than that with the use of slurries. Thereby, the difference in NH₃ volatilization loss would be small for manure using the surface application and surface incorporation methods. In addition, the effects of successive applications of manure on NH₃ volatilization loss were out of the scope of the present study because, in the experimental design, the application rates of manure as fresh weight were smaller than those used in actual upland cropping.

The atmosphere contains NH₃. For example, the mean NH₃ air concentration in the warm season on turf grassland near the study site was 2.7 μg N m⁻³ (0°C and 1013 hPa) (Hayashi et al. 2007). The NH₃ concentrations in the ambient air in the present study ranged from 0.40 to 10.3 μg N m⁻³. The concentrations were high in the daytime and just after fertilization and low in the nighttime and on rainy days (data not shown). The partition equilibrium of NH₃ between the gas–liquid phases rules the NH₃ volatilization. Hence, the effect of ambient NH₃ on NH₃ efflux should be considered when evaluating NH₃ volatilization loss in the environment. In contrast, Fueki et al. (2007) reported a small NH₃ volatilization loss from surface-applied AS in incubation experiments using Andosols in Hokkaido, northern Japan. This relatively large volatilization loss, in contrast to the negligible loss obtained in the present study, could be ascribed to the surface application method. In addition, the NH₃-free air that they used to ventilate the vessels appeared to increase volatilization loss to some extent because the NH₃-free gas phase enhances NH₃ efflux from the liquid phase. It is interpreted that incubation experiments using NH₃-free air, such as Fueki et al. (2007), are used to elucidate potential NH₃ volatilization loss.

Uncertainty of the dynamic chamber method

It is known that the ventilation rate within the chamber affects the NH₃ emission. Kissel et al. (1977) reported that a ventilation rate of less than 15 exchange volume min⁻¹ resulted in an underestimation of NH₃ emission. According to Kissel et al. (1977), the ventilation rate of 8.5 exchange volume min⁻¹ in the present study corresponded to a 14% underestimation of the fluxes. However, Huijsmans et al. (2003) pointed out that the wind velocity in the actual environment affected the temporal patterns of NH₃ emission, but rarely affected the cumulative NH₃ flux, that is, the loss to the applied N. The present study also focused on the volatilization loss rather than on the respective fluxes. In addition, the clear reductions in the soil NH₄-N content during the experimental periods (Fig. 4) indicated that marked NH₃ volatilization was unlikely to occur beyond the experimental periods. Hence, the derived negligible losses in the present study can be considered valid.

Conclusions

The NH₃ volatilization loss from surface-incorporated manure and ammonium sulfate was investigated at an upland field of light-colored Andosol in central Japan. The three experimental periods, summer and autumn 2007 and summer 2008, showed negligible NH₃ volatilization losses following applications of poultry manure, pelleted poultry manure, cattle manure, pelleted cattle manure and ammonium sulfate at the same application rate of 120 kg N ha⁻¹ as T-N. Although the pelleted manure showed a delayed release of NH₄-N to some extent, this delay in release did not affect the NH₃ volatilization loss.

The high CEC of the light-colored Andosol used in the present study, 283 mmol_c kg⁻¹ dry soil, was primarily effective in inhibiting NH₃ volatilization loss. The relatively low soil pH, 5.9 as pH(H₂O), also reduced the NH₃ volatilization loss. In particular, the marked decreases in soil pH to ≤5.5 following applications of ammonium sulfate effectively inhibited NH₃ volatilization loss. The fermented manure (compost) used in the present study had a small fraction of NH₄-N to T-N, which also contributed to the negligible NH₃ volatilization loss. The NH₃ efflux is positively and linearly proportional to the NH₄-N concentration in soil solutions (Hayashi et al. 2008). Therefore, when the absolute amount of NH₄-N applied to upland fields was small owing to the small fraction of NH₄-N to T-N in manure, the NH₃ volatilization loss was also small. It is also known that the application method used in the present study, that is, surface incorporation, is highly effective in inhibiting NH₃ volatilization loss, although no direct comparison was conducted in the present study with the surface application method. Thus, it is necessary to steadily accumulate concrete data based on case studies, such as the present study, and then establish the NH₃ emission factors of Japanese agriculture.

Acknowledgments

We would like to express our appreciation to Messrs Hiroshi Kamimura, Hironori Wakabayashi and Toshiyuki Okada, NIAES, for their practical help in managing the experimental field. We thank Dr Masaki Takeuchi, Mie Prefectural Agricultural Research Institute, for his provision of poultry manure and pelleted poultry manure, and Dr Yusuke Arakawa, National Agricultural Research Center for Kyushu Okinawa Region, for his provision of cattle manure and pelleted cattle manure. We also thank Dr Yasuhito Shirato, NIAES, for his assistance with soil analysis. Rainfall data were obtained from the Weather Data Acquisition System of NIAES. This study was supported in part by a Grant-in-Aid for Scientific Research (A), No.19201008, provided by the Ministry of Education, Culture, Sports, Science and Technology, Japan.