Effects of controlled release N fertilizers and reduced application rate on nitrous oxide emissions from soybean fields converted from rice paddies

ABSTRACT

This study aimed to investigate the effects of using controlled release fertilizers with reduced application rates on nitrous oxide (N₂O) gas emissions. For this, we monitored the N₂O emissions released from soybean fields converted from rice paddies in Japan for three years, i.e., from 2017 to 2020, under six nitrogen fertilizer treatments: conventional (AC: ammonium chloride, 20 kg N ha⁻¹), controlled release coated urea (CRCU: ammonium chloride, 5 kg N ha⁻¹, coated urea, 15 kg N ha⁻¹, total 20 kg N ha⁻¹), controlled release coated nitrate (CRCN: coated calcium nitrate, 20 kg N ha⁻¹), CRCU at a reduced rate (CRCU-R: 10 kg N ha⁻¹), CRCN at a reduced rate (CRCN-R: 10 kg N ha⁻¹), and no nitrogen fertilizer (NF: 0 kg N ha⁻¹). The soil of the field was classified as Gleyic Fluvisol.

The annual N₂O emissions of the CRCN treatment were significantly lower than those of the AC in the first and second years (p < 0.05) and were not significantly different in the third year, with reductions of 17–32%, while the same yield was maintained. The annual N₂O emissions of the CRCU treatment tended to be lower than those of the AC for the three years, with reductions of 14–19%, although not statistically significant. This suggests that coated nitrate fertilizers were more effective in reducing N₂O emissions, since nitrate produces this gas via denitrification only. In addition, the N₂O emissions of the CRCU-R and CRCN-R treatments were 22–37% and 24–41% lower than that of the AC treatment, respectively. Although not statistically significant, these reductions were slightly greater than those obtained for the N₂O emissions of CRCU and CRCN, suggesting the effect of further mitigation of N₂O emissions through the reduction of application rate. Furthermore, the N₂O emissions per yield tended to decrease due to the use of controlled release fertilizers for the three years, and further decrease due to the reduction of application rate. In conclusion, the use of controlled release fertilizers with reduced application rates can be regarded as an improved climate-smart soil management in soybean fields converted from rice paddies.

KEY WORDS:

• Application rate
• climate-smart soil management
• nitrate-coated fertilizer
• nitrous oxide (N₂O)
• soybean

1. Introduction

Nitrous oxide (N₂O) is a greenhouse gas whose effect is approximately 300 times that of carbon dioxide and has an atmospheric lifetime of approximately 120 years (IPCC, 2007). Soils are by far the largest overall source of N₂O. Agricultural soils particularly are the largest anthropogenic source of this gas (Intergovernmental Panel on Climate Change (IPCC) 2007; Van Groenigen et al. 2010), owing to the direct or indirect (through recycling as manure) application of synthetic nitrogen fertilizers (Van Groenigen et al. 2010).

The world’s consumption of nitrogen fertilizer in 1961 was 11.8 Tg N year⁻¹ and increased approximately nine times to 104.5 Tg N year⁻¹ until 2010 (IFA 2016). N₂O emissions are expected to increase further in the future owing to the growing consumption of nitrogen fertilizer accompanying the increase of food production. For this reason, improving the yields of major grains and stabilizing productions are important international issues. Although the yield of soybean, one of the main upland crops, has increased over the last century, the average worldwide yield remains at a low level. For instance, Japan achieved an average yield of 1.52 Mg ha⁻¹ in 2019, while the world’s average yield was 2.77 Mg ha⁻¹ (FAO 2020). More than 80% of soybean products in Japan are cultivated in paddy-upland fields where the groundwater level is relatively high; therefore, soil moisture is an issue affecting these plantations (Kokubun 2013). These reports indicate that maintaining or increasing the yield of soybean in fields converted from rice paddies is an important issue in Japan.

Fertilizer management, drainage improvement, and cultivation abandonment are effective measures for reducing N₂O emissions in agricultural lands (Intergovernmental Panel on Climate Change (IPCC) 2007). Fertilizer management can be practical, and many studies have reported the effects of controlled release fertilizers (coated fertilizers) on reducing N₂O emissions (Akiyama, Yan, and Yagi 2010; Noda 2001; Hasukawa et al. 2017).

Nitrous oxide is produced in the soil by the microbial processes of nitrification and denitrification after the application of nitrogen fertilizers (Akiyama and Tsuruta 2003). As such, N₂O emissions are reduced by using coated calcium nitrate fertilizers that do not generate N₂O derived from the nitrification process (Akiyama and Tsuruta 2002). In addition, N₂O emissions are generally positively correlated with nitrogen application rates, with emission levels decreasing when the nitrogen application rate is low (Van Groenigen et al. 2010). However, a reduced nitrogen fertilizer application rate generally leads to reduced crop yield. Mitigating greenhouse gas emissions while maintaining productivity is, therefore, important for achieving sustainable agricultural productions. By focusing on maintaining or improving grain yield and reducing N₂O emissions simultaneously with the appropriate levels of nitrogen fertilizer, it is possible to maximize economic viability and environmental conservation (Mosier et al. 2006). For instance, Van Groenigen et al. (2010) proposed a method for evaluating N₂O emissions per yield as agronomic productivity to environmental sustainability.

Hasukawa et al. (2017) revealed that coated fertilizer application reduced the annual N₂O emissions in a wheat-soybean cropping system converted from rice paddy fields, while the wheat and soybean production was equivalent to that of plantations using conventional fertilizers. However, the effect of reducing N₂O emissions by using coated fertilizers in soybean cultivation was low and unstable compared to wheat cultivation and depended on rainfall. In this study, to develop a mitigation strategy aimed at reducing N₂O emissions from soybean fields converted from rice paddies, we evaluated the effects of different types and application rates of nitrogen fertilizers on N₂O emissions from soybean fields. Furthermore, we evaluated these effects on the optimal balance between soybean productivity and the mitigation of N₂O emissions per yield.

2. Materials and Methods

2.1. Experimental plots and design

The experiment was performed for 3 years from 2017 to 2020 in the Shiga Agricultural Technology Promotion Center (Shiga Agricultural Technology Center) in Omi-Hachiman City, Shiga Prefecture, Japan (35°18ʹN, 136°12ʹE; mean annual air temperature: 14.5°C, annual precipitation: 1,731 mm). The experiment was carried out in soybean cultivation fields within a paddy-upland rotation system (paddy–wheat–soybean); therefore, the experimental field changed every year to study the soybean cultivated fields. In the crop rotation system, wheat was sown at the end of October following the rice harvest, and soybean was sown after the wheat harvest at the end of June. Each field was monitored for approximately one year from the soybean sowing period in June to the next paddy rice transplanting period in May of the following year. The soil of the experimental fields was classified as Gleyic Fluvisol according to the World Reference Base Classification (FAO, ISRIC, and ISSS, 2006), or as fine-textured gray lowland soil according to the classification of cultivated soil in Japan, third approximation (Nishina, Sudo, and Yagi et al. 2015). Table 1 lists the soil chemical properties selected before the experiment.

Table 1. Selected physicochemical properties of the experimental fields soils

Year	pH	T-C	T-N	Available P2O5	CEC	Exchangeable base					Three phases distribution^1			Bulk density^1
						Ca	Mg			K	gaseous phase	liquid phase	solid phase
	(H2O)	(g kg^−1)	(g kg^−1)	(mg kg^−1)	(cmolc kg^−1)	(cmolc kg^−1)					(%)			(g cm^−3)
2017–2018 (First year)	6.5	19.7	1.71	170.8	17.1	13.63		3.11	0.45		27.3	35.0	37.7	1.06
2018–2019 (Second year)	6.8	17.8	1.55	149.2	17.6	15.08		4.47	0.54		29.3	33.1	37.6	1.03
2019–2020 (Third year)	6.1	20.7	1.73	83.3	18.5	13.39		3.88	0.56		33.0	31.6	35.4	0.97

Before starting the experiment, five soil samples were collected from the overall experimental field, mixed and air-dried, 2.0 mm sieved, and subjected to soil chemical analysis.

¹The soil core samples (100 ml) were collected at two locations from each treatment after soybean harvest.

Table 2 presents the experimental design. Six nitrogen fertilizer treatments were established: conventional (AC), a compound fertilizer based on ammonium chloride and an application rate of 20 kg N ha⁻¹; controlled release fertilizer 1 (CRCU), a mixture of compound fertilizer containing ammonium chloride at 5 kg N ha⁻¹ and coated urea (Ceracoat R50/70/90) at 15 kg N ha⁻¹, with a total application rate of 20 kg N ha⁻¹; controlled release fertilizer 2 (CRCN), a coated calcium (23%) nitrate fertilizer 70 days type of elution period, with a total application rate of 20 kg N ha⁻¹; CRCU at a reduced rate (CRCU-R: 10 kg N ha⁻¹), CRCN at a reduced rate (CRCN-R: 10 kg N ha⁻¹), and a fertilizer without nitrogen (NF: 0 kg N ha⁻¹). Each treatment had three replicate plots selected with randomized block design in each year. The area of each plot was approximately 20 m².

Table 2. Fertilizer management of treatments

Treatment	Basal fertilizer
	Type of nitrogen fertilizer	Nitrogen application rate			Fertilizer Method
		Total	fast-acting
		(kgN ha^−1)
AC	conventional synthetic fertilizer ^a	20.0	20.0	(100)	broadcasting and incorporation
CRCU	compound fertilizer containing coated fertilizer ^b	20.0	5.0	(25)
CRCN	coated fertilizer ^c	20.0	0.0	0
CRCU-R	compound fertilizer containing coated fertilizer ^b	10.0	2.5	(25)
CRCN-R	coated fertilizer ^c	10.0	0.0	0
NF	-	-	-	-

Three replicates per treatment. Values in parentheses show the percentage of fast-acting nitrogen applied.

^a:compound fertilizer based on ammonium chloride (N-P₂O₅-K₂O(%):5–15-20) .

^b:mixture of compound fertilizer containing ammonium chloride and coated urea (Ceracoat R)

(N-P₂O₅-K₂O(%):16–14-14,fast-acting nitrogen 3.5%,Ceracoat R 10.5%).

^c:coated nitrate calcium fertilizer (N-P₂O₅-K₂O(%):12-0-0,nitrogen fertilization 70 days type of elusion 12%, Calcium 23%).

In NF, nitrogen fertilizer was not applied and only phosphorus and potassium were applied in the same amount as in AC.

Soybeans were grown according to the Cultivation Technology Guideline of Shiga Prefecture (Shiga Prefectural Government 2012). The total amount of wheat residue (straw: 433–618 g m⁻²) of each year was incorporated in soil after harvest in mid-June. The soybean cultivar ‘Kotoyutaka’ (2017 and 2018) and ‘Kotoyutaka A1ʹ (2019) were sown from late-June to mid-July, using a narrow-line non-tillage cultivation method. Nitrogen fertilizer was applied to the entire area (5–10 cm deep) before sowing. Phosphorus and potassium were applied as PK compound fertilizers, following the Shiga Prefecture cultivation technology guidelines. The planting density was about 25 plants m⁻² (length between rows 30 cm, spacing between plants 14 cm). Soybeans were harvested in mid- to late-November. The total amount of soybean residue (stem and pod) was incorporated in soil after harvest from mid-November to early December.

2.2. Measurement of N₂O emissions

Nitrous oxide gas fluxes were measured using the closed-chamber method (Yagi 1997). Gas was collected once per week between 9:00 AM and 12:00 AM. The frequency of measurement was approximately two to three times per week immediately after basal fertilizer application, and biweekly from January to March, the post-soybean fallow period.

An acrylic chamber (60 cm long, 30 cm wide, and 50 cm high) was used for the measurements. To facilitate covering the thick soybean plants inside the chamber during the latter soybean growth periods, the number of the soybean plants inside the chamber was reduced to two in the area of 30 cm × 14 cm. The chamber base was inserted into the soil up to a depth of 8 cm. The height of the chamber was set to 100 cm from mid-August until harvesting period according to the soybean growing height. For gas sampling, the gas in the chamber was mixed several times with a 50 mL syringe, and then 30 mL were sampled three times at 10-min intervals. The gas concentration was analyzed based on the method of Sudo (2012) using a gas chromatograph (GC; GC-14A, Shimadzu, Kyoto, Japan) with an electron capture detector. Nitrous oxide flux was calculated as follows based on Hasukawa et al. (2013).

F = ρ · V/A · ΔC/Δt · 273/T

Here, F is N₂O flux (μg N m⁻² h⁻¹), ρ is the gas density in the standard state (N₂O-N: 1.25 kg m⁻³), V is the volume of air in the chamber (m³), A is the bottom area of the chamber (m²), and ΔC/Δt is the average rate of increase of gas concentration in the chamber (10⁻⁹ m³ m⁻³ h⁻¹), T is the temperature (K) inside the chamber. Cumulative N₂O emissions were calculated using the trapezoidal integration method.

The fertilizer-induced N₂O emission factor (EF) was also evaluated in this research. It is a robust predictor of N₂O fluxes and has been used to estimate national greenhouse gas emissions (IPCC, 2006).

The EF was calculated as follows:

EF (%) = (N₂O emissions in plots where nitrogen fertilizer was applied – N₂O emissions in NF)/(nitrogen fertilizer application rate) × 100

2.3. Measurement of the selected field properties

The air and soil (5 cm deep) temperatures inside the chamber at the time of collection were measured using a temperature data logger (Ondotori, T & D, Matsumoto, Japan). The volumetric moisture content (0–12 cm deep) was measured at five locations around the chamber using a portable soil moisture meter (HydroSense, Campbell, UT, USA) at the time of gas sampling. The solid phase ratio between the time of soybean harvest and the end of the experimental period was 35.6%–40.4% in the first year, 36.4%–39.9% in the second year, and 30.5%–36.8% in the third year. Using this for the volumetric moisture content, the water-filled pore space (WFPS; soil moisture), defined as the ratio of water volume to the soil pore volume, was calculated and the average value for each treatment was determined. To obtain the cumulative precipitation values, meteorological observation data from the Shiga Agricultural Technology Center were used.

The yield of soybean was measured at maturity in each plot with a survey area of 1.2 m² (1.0 m × 4 lines). The grain weight of soybeans with a grain diameter of at least 5.5 mm was determined and converted to the value at 15.0% moisture. The selected soil properties were investigated based on the Soil Nutrient Measurement Method Committee (1983) and the Soil Environment Analysis Method Editing Committee (1997). Before starting the experiment, five soil samples were collected from the overall experimental field, mixed and air-dried, 2.0 mm sieved, and subjected to soil chemical analysis. For the three-phase distribution, the soil core samples (100 ml) were collected at two locations from each treatment after soybean harvest. The available phosphate was measured using the Truog method (Nanzyo 1997). The inorganic nitrogen concentrations of ammonium nitrogen (NH₄-N), nitrite nitrogen (NO₂-N), and nitrate nitrogen (NO₃-N) in the soil were determined with an automatic nitrogen analyzer (TRAACS 2000, Bran+Luebbe, Norderstedt, Germany) for all collected soil samples. The soil samples in all plots were collected from a depth of 0–10 cm around the chamber when gas fluxes were measured. The nitrogen content of soybeans collected at maturity in each plot was analyzed using the Kjeldahl method.

2.4. N₂O emissions per yield

A balance between crop productivity and mitigation of greenhouse gases is important for sustainable agriculture. Mosier et al. (2006) introduced the concept of greenhouse gas intensity by dividing Global Warming Potential by crop yield. Expressing N₂O emissions per unit of yield can account for the management impacts on both N₂O emissions and yield, and might provide a useful metric for greenhouse gas inventories (Venterea, Maharjan, and Dolanet 2011).

In this study, the N₂O emissions per yield were calculated as follows:

N₂O emission per yield (kg N₂O-N Mg⁻¹) = (N₂O emission in each treatment)/(soybean yield in each treatment).

2.5. Statistical analysis

A one-way analysis of variance with the type of treatment was performed for N₂O emissions, emissions factor, soybean yield (grain weight), nitrogen accumulation of the aboveground plant parts, and N₂O emission per yield, followed by a Tukey’s test. In this study, the level of statistical significance was set at 0.05 (R Core Team 2017).

3. Results

3.1. N₂O emissions

The N₂O flux during the 3 years is shown in Figures 1–3 (a), and the soil temperature, precipitation, WFPS, and inorganic nitrogen (NH₄-N and NO₃-N) concentrations during the 3 years are shown in Figures 1–3 (b–e). The annual N₂O emissions and the annual EFs for the 3 years are listed in Table 3.

Table 3. N₂O emissions, yield, and N₂O emission per yield of soybean during the experimental period

Year	Treat -ment	N2O emission				Reduction rate compared to AC		Emission Factor		Yield (Grain Weight)		N accumulation of the above-ground parts		N2O emission per yield
		Total		early stage of growth		Total	early stage of growth
		(kg-N2O-N ha^−1)				(%)		(%)		(Mg ha^−1)		(kgN ha^−1)		(kgN2O-N Mg^−1)
2017–2018 (first year)	AC	2.13 ± 0.23	a	1.41 ± 0.55	a			4.97 ± 2.07	a	3.91 ± 0.34	a	325 ± 31	a	0.55 ± 0.08	a
	CRCU	1.74 ± 0.12	ab	0.67 ± 0.06	ab	(18)	(52)	3.01 ± 1.40	a	3.85 ± 0.80	a	328 ± 60	a	0.47 ± 0.11	ab
	CRCN	1.45 ± 0.14	bc	0.58 ± 0.05	b	(32)	(59)	1.57 ± 1.52	a	3.81 ± 0.19	a	318 ± 21	a	0.38 ± 0.02	ab
	CRCU-R	1.33 ± 0.09	bc	0.79 ± 0.21	ab	(37)	(44)	1.95 ± 1.54	a	3.03 ± 0.82	a	249 ± 63	a	0.46 ± 0.09	ab
	CRCN-R	1.27 ± 0.34	bc	0.64 ± 0.34	ab	(41)	(55)	1.30 ± 1.76	a	3.64 ± 0.87	a	298 ± 72	a	0.35 ± 0.09	ab
	NF	1.14 ± 0.20	c	0.72 ± 0.11	ab	(47)	(49)	-	-	3.89 ± 0.85	a	317 ± 74	a	0.30 ± 0.06	b
	Treatment	**		*				n.s		n.s.		n.s.		*
2018–2019 (second year)	AC	2.85 ± 0.13	a	0.31 ± 0.10	a			4.59 ± 1.32	a	2.93 ± 0.34	a	213 ± 24	a	0.98 ± 0.12	a
	CRCU	2.31 ± 0.45	ab	0.13 ± 0.06	b	(19)	(57)	1.89 ± 1.41	a	2.86 ± 0.09	a	214 ± 7	a	0.81 ± 0.18	a
	CRCN	2.07 ± 0.23	b	0.07 ± 0.03	b	(27)	(77)	0.70 ± 1.12	a	2.53 ± 0.20	a	195 ± 16	a	0.82 ± 0.06	a
	CRCU-R	2.00 ± 0.09	b	0.14 ± 0.04	b	(30)	(56)	0.71 ± 2.65	a	2.96 ± 0.73	a	219 ± 51	a	0.70 ± 0.17	a
	CRCN-R	1.96 ± 0.32	b	0.08 ± 0.03	b	(31)	(73)	0.29 ± 4.87	a	2.76 ± 0.24	a	209 ± 15	a	0.71 ± 0.10	a
	NF	1.93 ± 0.26	b	0.11 ± 0.05	b	(32)	(66)	-	-	2.39 ± 0.56	a	183 ± 47	a	0.85 ± 0.29	a
	Treatment	*		**				n.s		n.s		n.s		n.s
2019–2020 (third year)	AC	5.55 ± 0.77	a	1.94 ± 0.18	a			7.70 ± 6.70	a	3.67 ± 0.07	a	247 ± 10	a	1.51 ± 0.19	a
	CRCU	4.78 ± 0.30	a	1.19 ± 0.39	a	(14)	(39)	3.86 ± 2.87	a	3.43 ± 0.15	a	233 ± 13	a	1.39 ± 0.03	a
	CRCN	4.59 ± 0.27	a	1.03 ± 0.46	a	(17)	(47)	2.87 ± 4.39	a	3.36 ± 0.23	a	226 ± 19	a	1.37 ± 0.17	a
	CRCU-R	4.31 ± 0.78	a	1.14 ± 0.53	a	(22)	(41)	2.95 ± 14.6	a	3.73 ± 0.04	a	253 ± 8	a	1.15 ± 0.20	a
	CRCN-R	4.22 ± 0.24	a	1.05 ± 0.16	a	(24)	(46)	2.05 ± 5.79	a	3.63 ± 0.14	a	251 ± 9	a	1.16 ± 0.06	a
	NF	4.01 ± 0.71	a	1.11 ± 0.25	a	(28)	(43)	-	-	2.84 ± 0.12	b	194 ± 6	b	1.42 ± 0.26	a
	Treatment	n.s		n.s				n.s		**		**		n.s
2017–2020 (Average)	AC	3.51 ± 1.61		1.22 ± 0.78				5.75 ± 3.86		3.50 ± 0.51		262 ± 54		1.01 ± 0.43
	CRCU	2.94 ± 1.43		0.66 ± 0.50		(16)	(46)	2.92 ± 1.94		3.38 ± 0.59		258 ± 61		0.89 ± 0.42
	CRCN	2.70 ± 1.45		0.56 ± 0.48		(23)	(54)	1.71 ± 2.57		3.23 ± 0.59		246 ± 58		0.86 ± 0.44
	CRCU-R	2.55 ± 1.41		0.69 ± 0.52		(27)	(44)	1.87 ± 7.53		3.24 ± 0.66		240 ± 43		0.77 ± 0.34
	CRCN-R	2.48 ± 1.36		0.59 ± 0.46		(29)	(52)	1.21 ± 3.96		3.34 ± 0.63		253 ± 53		0.74 ± 0.36
	NF	2.36 ± 1.34		0.64 ± 0.46		(33)	(47)	-		3.04 ± 0.84		231 ± 78		0.86 ± 0.52

The values are shown as mean ± standard deviation.

The analysis of variance was performed in the treatment one-way (ANOVA) allocation

(**: significant difference at the 1% level, *: significant difference at the 5% level, n.s.: no significant differences).

Reduction rate compared to AC (%) = 100-(N₂O emission in each treatment/N₂O emission in AC)×100.

Emission Factor (%) = (N₂O emission in N applied plot‒N₂O emission in NF)/(nitrogen application rate)×100.

Numbers within a column followed by different letters differ significantly among the treatment (multiple comparisons by Tukey HSD, p < 0.05).

N₂O emissions in the early stages of growth indicate emissions from before the basal fertilization to 30 days after the basal fertilization  .

Figure 1. Changes in (a) N₂O flux, (b) soil temperature, (c) precipitation and WFPS, and (d, e) soil inorganic nitrogen concentration in the first year. Small figures on the right side of (a), (d), and (e) are enlarged ones for the period immediately after fertilization.

Error bars show standard deviation. Soil temperature and WFPS were the average values of all the plots. NO₂-N levels were low in all treatments throughout the measurement period (data not shown).

Figure 2. Changes in (a) N₂O flux, (b) soil temperature, (c) precipitation and WFPS, and (d, e) soil inorganic nitrogen concentration in the second year. Small figures on the right side of (a), (d), and (e) are enlarged ones for the period immediately after fertilization.

Error bars show standard deviation. Soil temperature and WFPS were the average values of all the plots. NO₂-N levels were low in all treatments throughout the measurement period (data not shown).

Figure 3. Changes in (a) N₂O flux, (b) soil temperature, (c) precipitation and WFPS, and (d, e) soil inorganic nitrogen concentration in the third year. Small figures on the right side of (a), (d), and (e) are enlarged ones for the period immediately after fertilization.

Error bars show standard deviation. Soil temperature and WFPS were the average values of all the plots. NO₂-N levels were low in all treatments throughout the measurement period (data not shown).

3.1.1. N₂O flux

The N₂O flux had two periods with relatively high peaks within a year: just after the base fertilization and after rainfall under high temperature conditions from the end of June to August (Figures 1–3 (a)). The N₂O flux increased immediately after the base fertilization (within one week after application) for 3 years. The highest peak was obtained in the third year (690–1,145 μg N m⁻² h⁻¹). N₂O flux showed a higher peak after rainfall under high temperature conditions than immediately after the application of basal fertilizer (within 40 days after application) for 3 years. The peak after rainfall under high temperature was highest in the third year (2,155–3,307 μg-N m⁻² h⁻¹), followed by the second and first years. Over the 3 years, the N₂O flux during the fallow period after soybean cultivation (post soybean fallow period) remained lower in all plots than that during the soybean cultivation period, except for a temporary increase.

3.1.2. Temperature and precipitation

The soil temperature (5 cm deep) (Figures 1–3 (b)) did not show a clear difference between years and treatments and remained above 25°C from the sowing period to mid-August; however, it gradually decreased after September and was approximately 10°C during the mature period (November). In winter, the temperature was low (5–10°C).

The cumulative precipitation (Figures 1–3 (c)) was 1,482 mm in the first year, 1,040 mm in the second year, and 1,196 mm in the third year. Seasonal variation in all three years, namely, there was 133 mm of precipitation in early August and 289 mm in late October in the first year, 204 mm on July 5–7 and 116 mm in late September in the second year, 147 mm in mid-July 102 mm in late-July, 97 mm in mid-August, and 120 mm in late October in the third year.

3.1.3. WFPS

The WFPS throughout the year tended to increase with precipitation (Figures 1–3 (c)). There was no clear difference in the WFPS values between the treatments and years. The WFPS ranges were relatively constant during the 3 years: 43%–62% (average 53% in AC) in the first year, 42%–65% (average 50% in AC) in the second year, and 41%–72% (average 53% in AC) in the third year. Notably, in the third year, the maximum of WFPS was slightly higher than in the other two years.

3.1.4. Soil inorganic nitrogen concentration

The NH₄-N concentration increased immediately after fertilization in the first and third years, with the NO₃-N concentration subsequently increasing, as typically observed in AC (Figures 1–3 (d), (e)). In the second year, although the NO₃-N concentration increased immediately after fertilization (from late-June to early-July), the NH₄-N concentration did not. After the initial peaks, the NH₄-N concentration remained low. However, the NO₃-N concentration temporarily increased after soybean harvest in the second and third years. There were no clear differences between any of the treatments. The NO₂-N concentration remained almost zero (detector limit: 0.06 mg L⁻¹) in all plots throughout the 3 years.

3.1.5. N₂O emissions

In the first year, the annual N₂O emissions in CRCN, CRCU-R, CRCN-R, and NF were significantly lower than those in AC (CRCN: p < 0.05; CRCU-R, CRCN-R, and NF: p < 0.01), with reductions of 32%, 37%, 41%, and 47%, respectively. In the second year, the annual N₂O emissions in CRCN, CRCU-R, CRCN-R, and NF were significantly lower than those in AC (CRCN, CRCU-R, CRCN-R, and NF: p < 0.05), with reductions of 27%, 30%, 31%, and 32%, respectively. The annual N₂O emissions in CRCU tended to be lower than those in AC in the first and second year, with reductions of 18%, 19%, although the effects were not statistically significant. In the third year, the annual N₂O emissions were not significantly different among the treatments, with reductions of 14%, 17%, 22%, 24%, and 28%, respectively. The annual N₂O emissions were highest in the third year, followed by the second and first years. Both the use of controlled release fertilizer and reduced application rates actually decreased the annual N₂O emissions, although the effects were not always statistically significant.

3.1.6. The fertilizer-induced N₂O emission factor (EF)

In the first year, the annual EFs ranged from 1.30% in CRCN-R to 4.97% in AC, but there were no significant differences among the treatments. In the second year, the annual EFs ranged from 0.29% in CRCN-R to 4.59% in AC, but there were no significant differences among the treatments. In the third year, the annual EFs ranged from 2.05% in CRCN-R to 7.70% in AC, but there were no significant differences among the treatments. The annual EFs tended to be highest in AC and lowest in CRCN-R for all three years. Both the use of controlled release fertilizer and reduced application rates actually decreased the annual EFs, although the effects were not statistically significant.

3.2. Yield and nitrogen accumulation of soybean plantations

The annual yield and nitrogen accumulation for the 3 years are listed in Table 3. In the first year, the yield of soybean ranged from 3.03 Mg ha⁻¹ in CRCU-R to 3.91 Mg ha⁻¹ in AC, but there were no significant differences among the treatments. In the second year, the yield ranged from 2.39 Mg ha⁻¹ in NF to 2.96 Mg ha⁻¹ in CRCU-R, but there were no significant differences among the treatments. In the third year, the yield ranged from 2.84 Mg ha⁻¹ in NF to 3.73 Mg ha⁻¹ in CRCU-R, and the yield in NF was significantly lower than that in other treatments (p < 0.01). The soybean yields were similar in the first and third years and slightly lower in the second year. The nitrogen accumulation of the aboveground plant parts showed the same tendency as the yield, and there were no significant differences among the treatments, except for NF in the third year (p < 0.05). In conclusion, differences in yield and nitrogen accumulation were not statistically significant among the nitrogen fertilizer treatments.

3.3. N₂O emissions per yield

The annual N₂O emissions per yield for the 3 years are listed in Table 3. In the first year, the N₂O emissions per yield ranged from 0.30 kg N₂O-N Mg⁻¹ in NF to 0.55 kg N₂O-N Mg⁻¹ in AC, but there were no significant differences among the treatments, except for the NF and AC (p < 0.05). In the second year, the annual N₂O emissions per yield ranged from 0.70 kg N₂O-N Mg⁻¹ in CRCU-R to 0.98 kg N₂O-N Mg⁻¹ in AC, but there were no significant differences among the treatments. In the third year, the annual N₂O emissions per yield ranged from 1.15 kg N₂O-N Mg⁻¹ in CRCU-R to 1.51 kg N₂O-N Mg⁻¹ in AC, but there were no significant differences among the treatments. The annual N₂O emissions per yield were highest in AC for all three years among the treatments. Both the use of controlled release fertilizer and reduced application rates actually decreased the annual N₂O emissions per yield, although the effects were not statistically significant.

4. Discussion

4.1. Temporal changes in the N₂O emissions and their influencing factors

4.1.1. N₂O flux

In our study, two large peaks of N₂O flux were observed during the early stages of growth of soybean plants throughout 3 years (Figures 1–3 (a)), namely from the end of July to August. The two peaks may be influenced not only by soil moisture but also by temperature, as temperature was positively correlated with N₂O flux (Aguilera et al., 2013). The first large peak occurred immediately after the application of basal fertilizer, and the N₂O flux increased most in the AC treatment, followed by the CRCU, and the lowest in the CRCN treatment (Figures 1–3 (a)). The soil NH₄-N and NO₃-N concentrations in the CRCU treatment tended to have a slower increase than in the AC soil (Figures 1–3 (d), (e)), which was attributed to the slower release of NH₄⁺ from the controlled release fertilizer that was applied. In addition, the NH₄-N concentration in the CRCN treatment was consistently very low, while the NO₃^–N concentration was high because coated NO₃⁻ was applied; we suppose these trends caused the reduced N₂O flux, due to minimal nitrification. In the CRCU-R and CRCN-R treatments, the peak N₂O flux tended to decrease with lower application rates, but the concentrations of NH₄-N and NO₃-N were almost the same. There was no difference in WFPS at the first peak among treatments (Figures 1–3 (c)). Namely, the WFPS was approximately 50% in the first and second years, and approximately 60% in the third year when the peak was large. As previously reported by Smith et al. (2003), the N₂O flux tended to increase with the WFPS to approximately 60%.

The second large peak occurred from the end of July to the middle of August, when there was no significant difference in N₂O flux among the treatments. The soil NH₄-N and NO₃-N concentrations were consistently relatively low, while the WFPS increased up to approximately 60%. In addition, Kusa, Sawamoto, and Hu et al. (2010) reported that N₂O production due to denitrification can occur in a local anaerobic environment such as inside soil aggregates even under relatively aerobic conditions. Judging from these points, it seemed that the second large peak of N₂O flux mainly originated from the denitrification process rather than nitrification.

4.1.2. N₂O emissions

The annual N₂O emissions of the CRCN treatment were significantly lower than those of the AC in the first and second years (p < 0.05, Table 3) and were not significantly different in the third year. The annual N₂O emissions of the CRCU treatment tended to be lower than those of the AC for the three years, although not statistically significant. The annual N₂O emissions tended to be reduced by the use of controlled release fertilizers.

The cumulative N₂O emissions in the early stages of growth are shown in Table 3. N₂O emissions in the CRCN treatment were significantly lower than those in the AC treatment in the first and second years, while those in the CRCU treatment were significantly lower than those of the AC treatments in the second year. These results indicate that coated calcium nitrate is more effective in reducing N₂O emissions than coated urea, probably because N₂O is produced only from denitrification when coated calcium nitrate is used, while N₂O is produced from both denitrification and nitrification when using coated urea. Hasukawa et al. (2013) reported that the effect of reducing N₂O emissions in soybean cultivation by coated urea was not stable, but our results suggested that the problem could be solved by utilizing coated calcium nitrate. In addition, the results of our study performed in soybean fields converted from paddy fields with poor drainage (Gleyic Fluvisols) were similar to those reported by Akiyama and Tsuruta (2002) for the upland of Andosols, which had good drainage. Accordingly, the effect of reducing N₂O emissions by coated calcium nitrate can be observed regardless of the level of soil drainage. In the CRCU-R and CRCN-R treatments, N₂O emissions tended to decrease when lower application rates were used, as previously reported by Shcherbak, Millar, and Philip Robertson (2014).

The NF treatment had the highest average reduction rate of N₂O emissions, followed by the CRCN-R, CRCU-R, CRCN, and CRCU. The reduction rate was greater during the early stages of growth than during the entire growth period. Namely, N₂O emissions decreased with the increase in the rate of controlled release fertilizer and with reduced nitrogen application rate in addition to the use of nitrate-N as the source of N in coated fertilizer. Akiyama, Yan, and Yagi (2010) found out, in their meta-analysis, that coated fertilizers reduced N₂O emissions by an average of 35% (95% confidential interval: 14% to 58%). They also reported that the effectiveness of coated fertilizers in reducing N₂O emissions differed with soil type; although N₂O emissions had a 77% reduction in the grasslands of Gleysol, their effectiveness was lower in the upland of Andosol. In our study with Gleyic Fluvisols, a similar reduction rate was obtained (14%–32%).

4.2. Relationship between fertilizer application rate and N₂O emissions

The annual EFs for the 3 years were 4.59–7.70% in AC, 1.89–3.86% in CRCU, 0.70–2.87% in CRCN, 0.71–2.95% in CRCU-R, and 0.29–2.05% in CRCN-R (Table 3). These values were higher than the EF of 0.62% ± 0.48% obtained for upland fields in Japan (Akiyama, Yan, and Yagi 2006), and EF of 1% reported by the IPCC (Intergovernmental Panel on Climate Change (IPCC) 2006). In our survey, the high EF was considered to be due to the low fertilizer application rate of 10–20 kg N ha⁻¹ for nitrogen-fixing soybeans, as supported by Shcherbak, Millar, and Philip Robertson (2014).

The annual EFs in CRCU, CRCN, CRCU-R, and CRCN-R for the 3 years were lower by 39–59%, 63–85%, 61–85%, and 73–94% than those in AC, respectively. Both the use of controlled release fertilizers and reduced application rates decreased the EF, although this effect was not statistically significant. Shcherbak, Millar, and Philip Robertson (2014) reported that EF was significantly lower for controlled release urea treatments than for treatments using conventional synthetic fertilizers.

Fertilized nitrogen is required for soybean production; it is highly effective, especially when the yield is low. For example, the nitrogen fixation function of rhizobia is small due to environmental factors such as drought and shortage of O₂ due to submergence (Shimada 2013) or continuous cropping (Tamura 2000). Therefore, it is common to apply a small amount of nitrogen (20 kg N ha⁻¹) as basal fertilizer at the time of sowing to support the early stages of growth before active nodule development and nitrogen fixation. In addition, Takakai, Takeda, and Kon et al. (2010) showed that the nitrogen balance in paddy-converted soybean fields was negative, and appropriate nitrogen fertilization and organic matter application were required to achieve sustainable productions. The low EF obtained in treatments where controlled release fertilizers were used suggests that the nitrogen release rate from these fertilizers matched the nitrogen requirements of soybean plants. Although reduced nitrogen application rate is also effective in terms of nitrogen utilization efficiency, the balance between yield and N₂O emissions should be considered.

4.3. Relationship between soybean yield and N₂O emissions

The annual N₂O emissions per yield for the 3 years were 0.55–1.51 kg N₂O-N Mg⁻¹ in AC, 0.47–1.39 kg N₂O-N Mg⁻¹ in CRCU, 0.38–1.37 kg N₂O-N Mg⁻¹ in CRCN, 0.46–1.15 kg N₂O-N Mg⁻¹ in CRCU-R, 0.35–1.16 kg N₂O-N Mg⁻¹ in CRCN-R, and 0.30–1.42 kg N₂O-N Mg⁻¹ in NF (Table 3). In all treatments, the third year had the highest N₂O emissions per yield and the first year had the lowest, based on the trends of N₂O emissions and yield. The annual N₂O emissions per yield in the CRCU-R and CRCN-R treatments tended to be reduced steadily throughout the three years compared to those in the AC, although not statistically significant. The combination of controlled release fertilizers with reduced nitrogen application rates was effective in maintaining the yield while reducing N₂O emissions. Soybeans produced by this combination are certified as environmentally friendly agricultural products in Shiga Prefecture. Therefore, this combination can reduce the costs and increase the value of the products. Van Groenigen et al. (2010) reported that yield-scaled N₂O emissions have a significant negative relationship with nitrogen fertilizer use efficiency (defined as the percentage of nitrogen fertilizer taken up by the aboveground parts of the plant), indicating that increasing nitrogen fertilizer use efficiency is directly linked to minimizing N₂O fluxes.

Relatively high soil T-C and T-N concentrations can ensure the growth of soybean plants during the early stages, which leads to stable soybean production. However, the yield level can be further improved by using controlled release fertilizers that can supply nitrogen after the flowering period, because nitrogen absorption of soybean generally increases after the flowering period (Furuhata, Adachi, and Ohono 2011). In our survey, a certain level of yield could be secured even in NF. This was considered to be because the soil was relatively fertile and soybean was a nitrogen-fixing plant. Venterea, Coulter, and Dolanet (2016) further reported that optimizing both the application timing and source of nitrogen can allow for a moderate reduction in the nitrogen rate that does not affect grain yield but decreases N₂O. In such cases, N₂O emissions per yield could be reduced by decreasing the basal nitrogen application rate and increasing the nitrogen utilization efficiency.

4.4. Crop field rotation system and greenhouse gas emissions

Snyder, Bruulsema, and Jensen (2007) reported that management practices such as balanced nitrogen fertilizer management and crop rotation lead to reduced N₂O emissions. Adviento-Borbe et al. (2007) reported that CO₂ equivalent emissions could be reduced by switching continuous maize cropping to maize-soybean crop rotation. Hasukawa et al. (2021) reported that the net greenhouse gas balance (CH₄, N₂O and CO₂) in the paddy-upland rotation (four cultivations (wheat–soybean–rice–rice) over 3 years) and continuous paddy rice fields were estimated to be 2.38 and 7.51 Mg CO₂ ha⁻¹ year⁻¹, respectively, highlighting a 68% reduction in the paddy-upland rotation system. To achieve a good balance between the mitigation of greenhouse gas emissions and crop productivity, it is necessary to verify the effectiveness of comprehensive measures that combine land use management, such as crop and field rotation, with fertilization management.

5. Conclusions

The use of controlled release fertilizers with reduced application rates was able to significantly reduce N₂O emissions. Moreover, it reduced N₂O emission factors while maintaining soybean yield, thereby reducing N₂O emissions per yield. As such, this practice can be regarded as a climate-smart soil management (Paustian et al. 2016) for soybean fields converted from rice paddies.

Acknowledgments

This study was supported by the ‘‘Basic Survey on Farmland Soil Greenhouse Gas Emissions’’ project of the Ministry of Agriculture, Forestry and Fisheries from 2017 to 2020. We thank Prof. Kazuyuki Inubushi, Chiba University for acting as external experts on this project and for providing valuable comments throughout the study, and Dr. Yasuhito Shirato of the Institute for Agro-Environmental Sciences, National Agricultural and Food Research Organization (NARO), for their helpful advice and support. Furthermore, we thank the staff of the Agricultural Technology Promotion Center of Shiga Prefecture for their support.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study was supported by the ‘Basic Survey on Farmland Soil Greenhouse Gas Emissions’ project of the Ministry of Agriculture, Forestry and Fisheries from 2017 to 2020. We thank Prof. Kazuyuki Inubushi, Chiba University for acting as external experts on this project and for providing valuable comments throughout the study, and Dr. Yasuhito Shirato of the Institute for Agro-Environmental Sciences, National Agricultural and Food Research Organization (NARO), for their helpful advice and support. Furthermore, we thank the staff of the Agricultural Technology Promotion Center of Shiga Prefecture for their support.