Rice yield and soil fertility of an organic paddy system with winter flooding

ABSTRACT

Organic farming is promoted as a way to produce food while minimizing harm to ecosystems. We examined the feasibility of organic farming using applications of common organic materials to produce standard yields of rice (Oryza sativa L.). From 2012 to 2016, we investigated the effects of rice bran, leftover grains, and soybean (Glycine max (L.) Merr.) curd residue in place of inorganic fertilizer and the effects of winter flooding on brown rice yield and on the availability of nitrogen (N) and phosphorus (P) under rice-rice-soybean rotation in Ibaraki Prefecture, Japan. The yield was greatest at 581 g m⁻² with winter flooding in 2013 and smallest at 347 g m⁻² without winter flooding in 2016. The organic materials applied 7.3–7.5 g N m⁻² and 3.2–4.7 g P m⁻², which supplied 142% ± 37% of the N and 429% ± 125% of the P present in the ear of rice at harvest in plots without winter flooding. The rates of organic materials applied would maintain soil N and P content. Winter flooding increased the ammonium concentration in the soil solution, straw dry matter at harvest, and available N in the plow layer during winter. Our practice, which improved N supply in later growth stages and P availability, supported an adequate rice yield and a conserved available N.

KEY WORDS:

• Organic farming
• winter flooding
• paddy rice
• available N
• ammonium-N

1. Introduction

Ecologically friendly management strategies and agroecology, whose core principles include recycling nutrients and energy on the farm, are promoted as ways to produce food while reducing off-site harms (Matson et al. 1997; De Schutter 2010). To establish organic farming as a viable tool for these strategies, it is necessary to understand adequately the factors causing lower yields in organic crops (Seufert, Ramankutty, and Foley 2012). In 2014, the Japanese government stated an aim to increase the area of organic farming from 0.4% to 1.0% of the cultivated area (MAFF, 2014). Organic farming recycles waste materials as organic fertilizer (Barker 2010). In Japanese rice-growing regions, common materials include rice bran, leftover wheat and soybean grains, and bean curd residue (Inaba 2007), but these materials are little used as fertilizer substitute in conventional farming, and their nutrient supply patterns are not well understood for achieving adequate rice yields. Kinetic analysis of the N supply pattern revealed 27 g N kg⁻¹ dry matter (DM) in rice bran (Carbon/N ratio (C/N) = 19) and 55 g N kg⁻¹ DM in soybean curd residue (C/N = 10) (Nira 2010a). Following the incorporation of rice bran into paddy soil, N was immobilized for 9–25 days and 50% mineralization took 32–63 days. Following the incorporation of soybean curd residue, N was not immobilized, and 50% mineralization took 19–41 days. These N mineralization patterns indicate the effectiveness of rice bran and soybean curd leftover as fertilizer, but the effects of nutrient release on organic rice production are not yet known.

Since 1998, rice has been replaced by other crops on more than 400,000 ha, or 17% to 20% of the area of paddy fields cultivated annually in Japan (Nira 2010b). Soybean is the main alternative. Several studies have confirmed that soybean yields and the amount of available N in soil decreased with increasing duration of soybean cultivation in drained paddy soil (Hirokawa, Inahara, and Koike 2011; Odahara et al. 2012; Sumida, Kato, and Nishida 2005). Nira and colleagues have investigated how to maintain available N through soil organic matter (OM) content in such fields, and the results suggest that manure compost application increases soil N availability (Nira 2010b; Nira and Hamaguchi 2012). Organic rice farming could help since OM as a fertilizer substitute is converted to humus in soil (Barker 2010).

It is well known that winter flooding provides habitat for waterfowl (Pernollet et al. 2015). It also has benefits in controlling weeds and helping rice straw decomposition (Kurechi 2007) and is commonly used to enhance straw decomposition (Eagle et al. 2001). Linquist, Brouder, and Hill (2006) reported that winter flooding increased early-season soil N availability, allowing a reduction in N fertilizer application. On Japanese organic rice farms, winter flooding and the application of organic materials were effective in increasing earthworm abundance (Ito et al. 2015). Tubificid worms increased available N and P in submerged paddy soils in proportion to their density, since their activity stimulated soil OM decomposition and the development of reduced conditions (Ito and Hara 2010). However, the effects of winter flooding on N dynamics in temperate rice systems have not been studied extensively (Pernollet et al. 2015), and the effects on organic rice production are as yet inadequately understood.

The aims of this study were to evaluate the productivity of organic rice using rice bran, leftover grains, and soybean curd residue under rice-rice-soybean rotation with winter flooding, and to determine the effect of organic farming with winter flooding on N and P availability.

2. Materials and methods

2.1. Site description and management

The experiment was conducted in an experimental paddy field of the National Agriculture and Food Research Organization’s Agricultural Research Center in Yawara (140.01ʹE, 36.00ʹN). We used eight 100-m² plots bounded with concrete walls. The plots had been filled with 1.4 m of soil about 30 years previously. Paddy rice and soybean have been grown using agrichemicals and inorganic fertilizer since 1999. In 2002, the soil was characterized as having a plow layer of Grey Lowland soil at 0–0.17 m, a grey-brown soil at 0.17–0.23 m, and a brown soil at 0.23–0.30 m (Kaneko, unpublished report in our laboratory). In April 2009, the plow layer soil had 23.5 ± 1.2 g C kg⁻¹ (mean ± SD, n = 8), 2.0 ± 0.1 g N kg⁻¹, C/N = 11.6 ± 0.2, available P (Truog 1930) = 32.8 ± 2.5 mg P kg⁻¹, and pH = 6.3 ± 0.2.

Winter flooding was started in 2006–07. Four of the eight plots were flooded with irrigation water in late November or early December of each year and drained in late March of the next year (Table 1). The same four plots were flooded in 2006–07 to 2009–10 and from 2011–12 to 2012–13. The water looked clear, and after filtration through a 0.20-µm-pore filter, water sampled on 28 March 2013 had 1.04 mg N L⁻¹. Hairy vetch (Vicia villosa Roth) was sown in all plots in winter in 2011 and 2014 before soybean was sown. The other four plots were flooded in 2014–15 and 2015–16.

Table 1. Details of winter flooding and application of cattle manure (CM), fermented organic fertilizer 1–3 (OF1-3), weed suppression material (WS), rice bran (RB), and hairy vetch in every plot and of crops from 2007 to 2016

		Plot No.
Year	Season	1 and 2	3 and 4	5 and 6	7 and 8	Crop (conventional or organic)
2007	winter		flooding		flooding
	spring			CM	CM	Rice (conventional)
	summer
	fall
2008	winter		flooding		flooding
	spring					Soybean (conventional)
	summer
	fall				OF1
2009	winter		flooding		flooding
	spring	OF2, WS (excl. 1plot)	OF2, WS	OF2, WS (excl. 1plot)	OF2, WS	Rice (organic)
	summer	OF3	OF3	OF3	OF3
	fall			RB	RB
2010	winter		flooding		flooding
	spring	OF2, WS	OF2, WS	OF2, WS	OF2, WS	Rice (organic)
	summer	OF3	OF3	OF3	OF3
	fall
2011	winter	hairy vetch	hairy vetch	hairy vetch	hairy vetch
	spring					Soybean (organic)
	summer
	fall			RB	RB
2012	winter		flooding		flooding
	spring	OF2, WS	OF2, WS	OF2, WS	OF2, WS	Rice (organic)
	summer	OF3	OF3	OF3	OF3
	fall			RB	RB
2013	winter		flooding		flooding
	spring	OF2, WS	OF2, WS	OF2, WS	OF2, WS	Rice (organic)
	summer	OF3	OF3	OF3	OF3
	fall
2014	winter	hairy vetch	hairy vetch	hairy vetch	hairy vetch
	spring					Soybean (organic)
	summer
	fall			RB	RB
2015	winter	flooding		flooding
	spring	OF2, WS	OF2, WS	OF2, WS	OF2, WS	Rice (organic)
	summer	OF3	OF3	OF3	OF3
	fall			RB	RB
2016	winter	flooding		flooding
	spring	OF2, WS	OF2, WS	OF2, WS	OF2, WS	Rice (organic)
	summer	OF3	OF3	OF3	OF3
	fall

Application rates: CM, 4 kg m⁻²; OF1, OF2, RB, 100 g m⁻²; WS, 80 g m⁻²; OF3, 17 g m⁻² in 2009, 18 g m⁻² in 2010, and 20 g m⁻² in 2012, 2013, 2015, and 2016.

Rice was planted with cattle manure application at 4 or 0 kg m⁻² in 2007, and soybean was grown with agrichemicals and inorganic fertilizer in all eight plots in 2008; and organic production was started after harvest in 2008 (Table 1). Inorganic fertilizer and agrichemicals were not applied, and weeds were controlled mechanically from the fall of 2008 onward. From 2009, a rice-rice-soybean rotation was begun, using ‘Koshihikari’ rice and ‘Nattosyoryu’ soybean. All plots were flooded again after the fermented organic fertilizer was broadcast in mid April, and rice was transplanted in early June and harvested in late September. After the incorporation of hairy vetch into the soil in spring 2011 and 2014, soybean was sown in late June and harvested in late October. Organic rice management followed a method proposed by Inaba (2007). All rows were 0.3 m apart, and plants were spaced at 0.20 or 0.22 m within the rows.

In organic rice production, just before winter flooding, we broadcast fermented organic fertilizer 1 (OF1), made from rice bran, soybean curd residue, and rice straw in two plots in 2008 and rice bran in four plots in 2009, 2011, 2012, 2014, and 2015 at 100 g m⁻² (Table 1). After drainage, OF2, made from rice bran, wheat seeds, crab and shrimp shells, and salt, was broadcast at 100 g m⁻² in all eight plots in mid April. At the time of rice transplanting, pellets made from rice bran, soybean curd residue, and crushed soybean seed were broadcast as weed suppression material (WS) at 80 g m⁻² in six plots in 2009 and in all eight plots in 2010, 2012, 2013, 2015, and 2016. In late June, OF3, made from rice bran, soybean curd residue, guano, and rapeseed oil cake was broadcast at 17–20 g m⁻² in all plots.

Air temperature and precipitation data during organic production were taken from the weather data acquisition system of the Institute for Agro-Environmental Sciences, NARO (mserver.narcb.affrc.go.jp/disp.php). The minimum and maximum monthly mean temperatures were 1.9°C in January 2012 and 28.1°C in August 2010 (Figure 1). The mean temperature during the winter flooding period ranged from 4.1°C to 6.3°C. Annual rainfall ranged from 1240 to 1525 mm. Total rainfall during the winter flooding period ranged from 171 to 311 mm.

Figure 1. Time course of monthly mean air temperature and monthly cumulative precipitation after start of organic production

2.2. Experimental design

Organic rice was planted in a fixed field layout each year from 2010 (Table 1). After soybean was sown in all eight plots in 2011, we started to investigate organic rice production. In 2012 and 2013, four treatments were used, each with two replications: treatment 1 (control), with no rice bran in fall and no winter flooding; treatment 2 (+RB), with rice bran in fall and no winter flooding; treatment 3 (+WF), with no rice bran in fall and with winter flooding; and treatment 4 (+RB+WF), with rice bran in fall and winter flooding. Rice bran in fall was used to analyze the effect of additional organic material application on soil fertility. Plots were assigned using a two-factor randomized block design (n= 2). To confirm the effect of winter flooding, the four plots with winter flooding before 2014 were not flooded and the other four plots were flooded in 2014–15 and 2015–16 (Table 1).

2.3. Sampling and analysis of organic materials and plants

The organic materials were sampled during application and were oven-dried at 105°C for chemical analysis. The aboveground parts of rice plants at harvest were separated into stem + leaves and ears, oven-dried at 80°C, and weighed. All dried samples were ground and passed through a 2-mm screen.

The N concentration in each sample was determined by the dry combustion method using an NC analyzer (Sumigraph NCH-22F; Sumica Chemical Analysis Service, Tokyo, Japan). Following digestion in nitric and perchloric acids, the P concentration was determined by the molybdenum blue colorimetric method.

In late September, a total of 32 plants were harvested in four rows per plot for yield analysis.

2.4. Soil solution and soil sampling and analysis

After the surface soil was puddled in mid April, we inserted the resinous porous part (100 mm × 2.5 mm wide) of a soil solution sampling tool (DIK-301A; Daiki Rika Kogyo Co., Kounosu, Japan) to 0–0.1 m depth vertically in the surface soil of each plot. Every week, about 5 mL of soil solution was drawn into a 10-mL syringe. After rice planting, the sampling tool was reset centrally between plants. The concentration of ammonium-N in the soil solution was determined by the indophenol method (Scheiner 1976) with flow injection equipment (FL-200; Aqualab Co., Tokyo, Japan).

For analysis of available N, the plow layer soil (about 0.1 m thickness) was sampled with a garden trowel at six or nine positions in each plot in fall before the broadcasting of rice bran and in spring after winter flooding. The bulked soil samples were air-dried and passed through a 2-mm sieve. A 10-g sample of air-dried soil with pure water to the top was put into a glass cylinder (27 mm wide, 120 mm deep), which was then sealed with a rubber stopper. In the cylinder, the soil was 23 mm deep, the water was 79 mm deep, the stopper was 18 mm thick, and there was no air. The samples were incubated for 28 days at 30°C. Both before and after incubation, the ammonium-N was extracted with 10% KCl solution, and its concentration was determined as above by the indophenol method. Available N was determined as the difference in ammonium-N between before and after incubation.

To measure available P in the surface 0.1 m, we sampled soil from between rice plants in a 55-mm-diameter cylinder on 18 July in 2012 and 2013 (maximum-tiller-number stage). Available P was extracted with Bray 2 solution (0.03 N in NH₄F + 0.1 N HCl) (Shoji, Miyake, and Takeuchi 1964). The concentration of P in the extract was determined by the molybdenum blue colorimetric method.

2.5. Statistical analyses

Means of duplicates were tested by analysis of variance (ANOVA) using the GLM procedure of SAS v. 9.4 software (SAS Institute Inc., Cary, NC, USA). Data analyses for brown rice yield, some rice plant properties at harvest, concentrations of N and P in rice organs, and available N and available P in the plow layer in 2012 and 2013 were performed with treatments arranged in a 2 × 2 × 2 factorial (rice bran application × winter flooding × year) randomized block design with two replications, with year as a fixed effect. Ammonium-N was rarely detected in the soil solution in most plots in August, so concentrations from April to July were tested. Data analyses for 2012 and 2013 were performed with treatments arranged in a 2 × 2 × 14–15 factorial (rice bran application × winter flooding × sampling time) randomized block design with two replications, with sampling time as a fixed effect by year.

3. Results

3.1. Amount of N and P incorporation

Each rice-cropping year, control plots, which received no organic materials in the previous fall, received 4.8–5.0 g N m⁻² and 2.1–3.0 g P m⁻² from the application of organic materials (Table 2). Application of rice bran added 2.3–2.5 g N m⁻² and 1.1–2.1 g P m⁻². In total, up to 7.5 g N m⁻² and 4.7 g P m⁻² was supplied. Compared with recommended amounts of N (3–7 g N m⁻²) and P (3.1–4.4 g P m⁻²) in inorganic fertilizer in the main production area in Japan (Niigata Prefectural Agriculture, Forestry and Fishery Division 2005), control plots received recommended amounts of N and 48–97% of P.

Table 2. Rates (g m⁻²) of N and P applied by organic materials in plots with control treatment, with winter flooding (+WF), with rice bran (+RB), and with rice bran and winter flooding (+RB+WF) in 2012–2016

	2012		2013		2015		2016
Materials	N	P	N	P	N	P	N	P
Rice bran	2.4	1.1	2.5	1.1	2.3	1.6	2.5	2.1
OF2^a	1.8	0.8	1.7	0.7	1.7	1.0	1.7	0.7
WS^b	2.6	1.0	2.8	0.8	2.8	1.0	2.6	1.4
OF3^c	0.5	0.6	0.6	0.7	0.6	0.9	0.7	0.6
Total in Control and +WF	4.8	2.4	5.0	2.1	5.0	3.0	5.0	2.7
Total in +RB and +RB+WF	7.3	3.4	7.5	3.2	7.3	4.6	7.4	4.7

^aFermented organic fertilizers made from rice bran, wheat seeds, crab and shrimp shells, and salt.

^bWeed suppression material.

^cFermented organic fertilizer made from rice bran, soybean curd residue, guano, and rapeseed oil cake.

3.2. Rice growth, yield, and plant nutrition

In the control (no organic material in fall, no winter flooding), straw DM, brown rice yield, stem length, number of ears, and N and P contents in rice plants were larger in 2012 and 2015, following soybean crops, than in 2013 and 2016 (Tables 3, 4). The maximum brown rice yield was 581 g m⁻² in +WF in 2013; the minimum was 347 g m⁻² in the control in 2016.

Table 3. Brown rice yield, plant properties, and N and P contents of aboveground parts of rice plants at harvest in plots with control treatment, with rice bran (+RB), with winter flooding (+WF), and with rice bran and winter flooding (+RB+WF) in 2012 and 2013

		Straw dry matter	Brown rice yield^a	Thousand-kernel-weight^a	Stem length	Ear length	Number of ears	Percentage of ripened grains	Aboveground part
Year	Treatment	(g m^−2)	(g m^−2)	(g)	(m)	(m)	(n m^−2)	(%)	N(g m^−2)	P(g m^−2)
2012	Control	676	448	21.1	0.82	0.19	302	69	7.5	1.4
	+RB	721	473	21.3	0.84	0.19	302	79	10.6	1.9
	+WF	807	532	20.7	0.89	0.19	369	72	10.5	1.6
	+RB+WF	817	493	20.6	0.94	0.19	382	55	14.4	2.1
2013	Control	594	397	20.6	0.76	0.18	255	76	5.1	0.9
	+RB	716	507	20.7	0.83	0.18	306	72	7.3	1.4
	+WF	803	581	20.4	0.89	0.19	352	79	9.5	1.4
	+RB+WF	945	556	19.9	0.94	0.19	450	62	12.1	1.5
ANOVA
	RB	**^b	ns^c	ns	**	ns	**	**	**	*
	WF	***	***	***	***	ns	***	**	***	ns
	Year	ns	ns	***	ns	ns	ns	ns	**	*
	RB×WF×Year	ns	ns	ns	ns	ns	ns	ns	ns	ns
	RB×WF	ns	**	*	ns	ns	ns	***	ns	ns
	RB×Year	*	ns	ns	ns	ns	**	ns	ns	ns
	WF×Year	*	*	ns	ns	ns	*	ns	ns	ns

^a > 1.8 mm thickness grain, at 15% moisture.

^b* P < 0.05; ** P < 0.01; *** P < 0.001.

^c Not significant.

Table 4. Brown rice yield, plant properties, and N and P contents of aboveground parts of rice plants at harvest in plots with control treatment, with rice bran (+RB), with winter flooding (+WF), and with rice bran and winter flooding (+RB+WF) in 2015 and 2016 after switching winter flooding position

		Straw dry matter	Brown rice yield^a	Thousand-kernel-weight^a	Stem length	Ear length	Number of ears	Percentage of ripened grains	Aboveground part
Year	Treatment	(g m^−2)	(g m^−2)	(g)	(m)	(m)	(n m^−2)	(%)	N(g m^−2)	P(g m^−2)
2015	A: Control (after +WF)	689	422	22.0	0.89	0.19	312	68	10.3	2.5
	B: +RB (after +RB+WF)	750	417	22.0	0.93	0.20	313	64	10.3	2.2
	C: +WF (after Control)	743	426	22.0	0.93	0.20	315	70	11.2	2.6
	D: +RB+WF (after +RB)	756	418	22.0	0.95	0.21	285	68	11.6	2.7
2016	A: Control (after +WF)	459	347	21.8	0.80	0.19	210	77	5.9	1.5
	B: +RB (after +RB+WF)	608	451	22.2	0.83	0.19	261	78	5.9	1.4
	C: +WF (after Control)	615	463	22.0	0.84	0.19	270	70	6.5	1.6
	D: +RB+WF (after +RB)	683	487	22.1	0.90	0.20	288	71	10.9	2.1
Difference between average in 2015–16 and in 2012–13
A from +WF in 2012–13		−231	−172	1.37	−0.05	0.00	−99	−3	−1.9	0.5
B from +RB+WF in 2012–13		−202	−91	1.83	−0.06	0.00	−129	13	−5.2	0.0
C from Control in 2012–13		44	22	1.17	0.09	0.01	14	−3	2.6	1.0
D from +RB in 2012–13		1	−37	1.05	0.09	0.02	−18	−6	2.3	0.8

^a > 1.8 mm thickness grain, at 15% moisture.

Winter flooding significantly increased straw DM, stem length, number of ears, and N content (all p < 0.001) in 2012 and 2013 (Table 3). Rice bran broadcasting in fall significantly increased straw DM, stem length, number of ears, and N and P content (p < 0.05 or p < 0.01). The interaction of flooding × rice bran broadcasting in fall significantly affected brown rice yield (p < 0.01), 1000-kernel-weight (p < 0.05), and percentage of ripened grains (p < 0.001). Year significantly affected 1000-kernel-weight (p < 0.001), N content (p < 0.01), and P content (p < 0.05). The interaction of winter flooding × year significantly affected straw DM, brown rice yield, and number of ears (all p < 0.05). The interaction of rice bran broadcasting in fall × year significantly affected straw DM (p < 0.05) and number of ears (p < 0.01). Reanalysis of statistics as a 2 (winter flooding) × 2 (year) factorial design by rice bran broadcasting treatment showed significant effects of winter flooding on 1000-kernel-weight only in treatments with rice bran broadcasting (Bonferroni method: p < 0.01). Winter flooding and the interaction of winter flooding × year significantly affected brown rice yield only in treatments without rice bran broadcasting and percentage of ripened grains only in treatments with rice bran broadcasting (Bonferroni method: p < 0.01 or p < 0.05).

Switching winter flooding position in 2014–15 and 2015–16 affected straw DM, stem length, and aboveground N (Table 4): all three were larger in +WF than in the control and were larger in +RB+WF than in +RB in 2016. Winter flooding increased the means in the control and +RB in 2012 and 2013, and its omission in 2014–15 and 2015–16 decreased the means in +WF and +RB+WF in 2012 and 2013.

The application of materials supplied 118% ± 33% of the N and 328% ± 147% of the P present in the ears of rice at harvest in the control (mean ± SD of 4 years), and 142% ± 37% of N and 429% ± 125% of P in plots that received rice bran in fall and no winter flooding (Table 5). Winter flooding substantially decreased all ratios.

Table 5. Ratios (%) of applied N and P to ear N and P at harvest in plots with control treatment, with rice bran (+RB), with winter flooding (+WF), and with rice bran and winter flooding (+RB+WF) in 2012–2016

	2012		2013		2015		2016		Average ± SD
Treatment	N	P	N	P	N	P	N	P	N	P
Control	101	285	157	541	82	206	131	279	118 ± 33	328 ± 147
+RB	107	292	159	491	116	363	186	571	142 ± 37	429 ± 125
+WF	78	226	84	325	71	174	115	245	87 ± 20	242 ± 62
+RB+WF	92	282	107	461	108	257	105	361	103 ± 7	340 ± 92

3.3. Ammonium-N in soil solution

The ammonium-N concentration of the soil solution was <0.3 mg L⁻¹ in all plots in mid to late April; it gradually increased to near maximum in mid June just after rice planting time, decreased in July, and was hardly detected after late July (Figure 2). The significance of the winter flooding × sampling time interaction effects in 2012 (p < 0.05) and 2013 (p < 0.001) indicates a greater seasonal increase in concentration in plots with winter flooding (Table 6). The maximum values in the plots with winter flooding in 2014–15 and 2015–16 (+WF after control, +RB+WF after +RB) were higher than those in the plots without winter flooding in 2012 and 2013 (Figure 2.: ○ and × in each year). On the other hand, the maximum values in the plots without winter flooding after 2014 (control after +WF, +RB after +RB+WF) were lower in 2015 but not lower in 2016 than those in the plots with winter flooding in 2012 and 2013 (▲ and ■ in each year).

Table 6. Analysis of variance of ammonium-N concentrations in the soil solution from April to July of 2012 and 2013

Souce of variation	2012	2013
Rice bran	ns^a	ns
Winter flood	***^b	***
Time	***	***
Rice bran×winter flood×time	ns	ns
Rice bran×winter flood	ns	ns
Rice bran×time	ns	ns
Winter flood×time	*	***

^aNot significant.

^b* P < 0.05; *** P < 0.001.

Figure 2. Seasonal changes in ammonium-N concentration in the soil solution in plots with control treatment, with rice bran (+RB), with winter flooding (+WF), and with rice bran and winter flooding (+RB+WF) in 2012, 2013, 2015, and 2016. Vertical bars indicate SD (n = 2)

3.4. Available N in soil

The available N in the plow layer soil increased from fall to spring in plots with winter flooding, but it decreased or did not change in plots without winter flooding (Table 7). The increase was evident also in 2016 after the flooding treatment was switched between plots (Table 8). Winter flooding significantly increased available N in spring 2012 and 2013 (P < 0.001). Rice bran broadcasting in fall significantly increased it in spring 2012 and 2013 (P < 0.01). Year effects in the previous fall and spring of both years were significant (P < 0.001).

Table 7. Available N in the plow layer in the previous November and in April after winter flooding and available P in the plow layer at maximum tillering stage in plots with control treatment, with rice bran (+RB), with winter flooding (+WF), and with rice bran and winter flooding (+RB+WF) in 2012 and 2013

		Available N (mg kg^−1)		Available P
Year	Treatment	Previous Nov.	April	(mg kg^−1)
2012	Control	91	80	222
	+RB	109	87	204
	+WF	106	106	182
	+RB+WF	106	126	247
2013	Control	156	145	188
	+RB	157	159	237
	+WF	155	188	223
	+RB+WF	169	200	258
ANOVA
	RB	ns^a	**^b	ns
	WF	ns	***	ns
	Year	***	***	ns
	RB×WF×Year	ns	ns	ns
	RB×WF	ns	ns	ns
	RB×Year	ns	ns	ns
	WF×Year	ns	ns	ns

^aNot significant

^b* *P < 0.01; *** P < 0.001.

Table 8. Available N in the plow layer in the previous November and in April after winter flooding in plots with control treatment, with rice bran (+RB), with winter flooding (+WF), and with rice bran and winter flooding (+RB+WF) in 2015 and 2016 after switching winter flooding position

		Available N (mg kg^−1)
Year	Treatment	Previous Nov.	April
2015	A: Control (after +WF)	142	151
	B: +RB (after +RB+WF)	141	155
	C: +WF (after Control)	131	147
	D: +RB+WF (after +RB)	144	163
2016	A: Control (after +WF)	140	137
	B: +RB (after +RB+WF)	152	150
	C: +WF (after Control)	147	173
	D: +RB+WF (after +RB)	145	205
Difference between average in 2015–16 and in 2012–13
A from +WF in 2012–13		11	−3
B from +RB+WF in 2012–13		9	−11
C from Control in 2012–13		15	48
D from +RB in 2012–13		12	61

Switching winter flooding position affected available N in spring in 2015 and 2016 (Table 8). Available N was larger in +WF than in the control and was larger in +RB+WF than in +RB in April 2016. Winter flooding increased the average values in the control and +RB in 2012 and 2013, and its omission in 2014–15 and 2015–16 decreased the means in +WF and +RB+WF in 2012 and 2013.

3.5. Available P in soil at maximum-tiller-number stage

Bray 2 available P in soil sampled at the maximum tiller number stage (Table 7) was around the value of 218 mg P kg⁻¹ proposed by Ito (2014) as the lower limit for adequate rice growth without inorganic P fertilizer in alluvial soils with less active iron (oxalate-extractable Fe concentration <1%). Values did not show any significant differences among treatments.

4. Discussion

4.1. Rice growth, yield, and N and P uptake

In Niigata Prefecture, where the quality of ‘Koshihikari’ is high, the recommended production goals are a brown rice yield of 540 g m⁻², 380 ears m⁻², a stem length of 0.9 m, 90% ripened grains, and a 1000-kernel-weight of 22.0 g (Niigata Prefectural Agriculture, Forestry and Fishery Division 2005). In our control treatment, the yield, number of ears, stem length, and percentage of ripened grains in all years were below these values. Winter flooding increased the yield to around the goal in 2012 and 2013 and increased the number of ears to near the goal in 2012. Fall application of rice bran did not increase the yield to the goal. In contrast, winter flooding with fall application of rice bran increased the number of ears above the goal in 2012 and 2013, and stem length above the goal in 2012, 2013, and 2015; and winter flooding without rice bran increased stem length to near to the goal. No treatment increased the percentage of ripened grains to near the goal, and the 1000-kernel-weight of all plots was lower than the goal before 2015. Fall application of rice bran slightly improved growth and yield; winter flooding gave adequate growth and yield. Rice bran plus flooding led to excess stem length, causing lodging and consequent low light use efficiency, and thus lower yields. The increase in growth almost paralleled N uptake.

The N content of rice plants was larger in 2012 and 2015, following soybean crops, than in 2013 and 2016. This result cannot be explained by warmer temperatures, as the air temperature from June to September was higher in 2016 than in 2015. This result supports some reports that N uptake by rice increased in the year following a soybean crop, since paddy drainage for soybean planting increases the availability of easily decomposable organic N in soil (Moroyu 1983; Kitada et al. 1993) and permits deeper root extension and higher root activity of rice (Kaneta, Kodama, and Naganoma 1989). In the absence of readily available N, as in organic farming without inorganic fertilizer, such an increase in N release after soybean planting is beneficial for rice. The P content of rice plants paralleled the N content among years. Stimulation of growth by N may encourage P uptake.

4.2. Evaluation of N and P incorporation through organic materials

The application rates of 4.8–5.0 g N m⁻² and 2.4–3.0 g P m⁻² in the control plots are close to those in the conventional application of inorganic fertilizer in Japan. The release of N from rice bran and soybean curd residue incorporated in the same paddy field in mid April reached 50% of total N content by 63 and 41 days, respectively; following incorporation from June to early July, the time was decreased to 32–34 and 19–20 days, respectively; and the N mineralization rate of both materials reached 83% at 63 days (Nira 2010a). Therefore, following application in mid April, about half of the N in each material is estimated to be released by the rice planting time, and about half of the N in the weed suppression materials and topdressing is estimated to be released during June and July. The rates of organic materials applied here were adequate for plant growth, but not for the percentage of ripened grains. A low percentage of ripened grains is related to a low leaf area index due to low N uptake after heading or to plant lodging (Miyama 1988). So an increased release of N in later growth stages is needed. The rates of organic materials applied in plots with fall application of rice bran would maintain soil N content because the materials supplied 103% ± 7% to 142% ± 37% of the N present in ears of rice at harvest.

Stem number depends on adequate P in the early growth stage; this P is easily retranslocated to other organs later (Sasaki and Hirata 1995). The small number of ears in our results presumably indicates that P availability, which was around the lower limit for adequate rice growth in alluvial soils with less active iron (Ito, 2014), was low for organic rice. Therefore, greater application of P in the early growth stage might be effective, although the P application rate was 2.4–4.3 times the ear P content and could cause accumulation of P in soil. Further study is needed to improve rice growth.

In organic farming, N and P uptake by weeds must be considered. Our fields had more weeds than conventionally managed fields. Nutrient uptake by weeds and release from incorporated weeds must be evaluated to allow more effective application of nutrients from organic materials.

4.3. Relationship between rice growth and soil labile N

The ammonium-N concentration in the soil solution at rice planting time (Figure 2) in plots with winter flooding was related to straw DM, rice stem length, and number of ears at harvest in each year except 2016. The relationships indicate that winter flooding increases N availability and thus controls rice growth. The low concentration in April shows that there was little accumulation of inorganic N by inflow following winter irrigation and by inhibition of nitrification under flooding. The higher ammonium-N concentration in May and June in plots with winter flooding could be derived from N mineralization in warmer conditions. Fall application of rice bran barely increased the ammonium-N concentration in the soil solution during the growing season. Since rice bran has a low C/N ratio and is fine textured, it is quickly mineralized in winter. The N released from it might then be leached out and denitrified. However, some of the rice bran could have remained, because the available N increased (Table 7).

In the early rice growth period, application of inorganic fertilizer and mineralization of soil OM increased the ammonium-N concentration in the soil solution. In a previous study, the concentration stayed at about 2 mg L⁻¹ soil solution in heavy clay soil basally dressed with 4.5 g N m⁻² during the 30 days following rice planting and then was reduced by plant uptake (Toriyama 1994). In our plots with winter flooding, the concentrations were >2 mg L⁻¹ during mid June to early July after planting (Figure 2), and thus our application of organic material without inorganic fertilizer supplied adequate N for rice growth. However, the amounts and timing of application of organic materials that we used may not be suitable for other soil types with different cation exchange properties, because the concentration of ammonium-N in the soil solution is affected by both its quantity and the cation exchange properties of soil (Toriyama 1994). Further study is needed for the development of appropriate techniques to supply N mineralized from organic materials in all growing seasons.

The differences in rice biomass among treatments could be explained by the potential N availability in spring, as higher available N caused better growth. In 2016, the better growth of rice in the plots with winter flooding was related to greater available N, although the soil solution ammonium-N concentrations were higher in the other plots (Figure 2). As the soil solution was sampled at only one position per plot, it might have been sampled from an unrepresentative position in 2016. In contrast, since soil for available N analysis was sampled at nine positions in a plot and bulked, the available N was more representative of the plot.

4.4. Effect of winter flooding on soil available N and P

In California, winter flooding stimulated rice straw decomposition and increased inorganic N and potentially mineralizable N in soil, but the latter increase was not well understood (Linquist, Brouder, and Hill 2006). If winter flooding stimulates the decomposition of all kinds of OM, it would not restore the fertility of paddy soils. The decomposition of rice straw was slower under anaerobic conditions than under aerobic conditions in one incubation experiments (Pal and Broadbent 1975) and was similar between conditions in another (Kanazawa and Yoneyama 1980). The gross N mineralization rate was lower under submerged conditions than under aerobic conditions in another incubation experiment (Nishio et al. 1994). These results indicate the possibility of suppressing OM decomposition and N mineralization by flooding, thus making winter flooding effective at restoring the fertility of paddy soils following the accumulation of soil OM. Stimulation of rice straw decomposition by winter flooding (Linquist, Brouder, and Hill 2006) might be related to the change from air-dried condition to wet condition in fields. Bird et al. (2001) reported that winter flooding did not affect straw ¹⁵N in the soil before planting, indicating that the rice straw provides an easily mineralizable N pool in spring. In spring 2016, after the winter flooding positions were switched, available N increased following winter flooding (Table 8). This result indicates that two winter treatments were adequate to reveal the effects of winter flooding on soil available N.

Other mechanisms also may stimulate the release of N from April to July. In northeastern Japan, Ito et al. (2015) found higher densities of aquatic oligochaetes in organic paddy fields with winter flooding during the rice-growing season. Bioavailable N and P increased in proportion to the densities of Branchiura sowerbyi, a tubificid worm (Ito and Hara 2010). The tubificids stimulated the decomposition of soil OM with a consequent redox reduction, which increased inorganic N and dissolution of P adsorbed by ferric oxide. Winter flooding might allow more tubificid worms to overwinter and lay more eggs in spring. Higher densities of tubificid worms in our organic paddy fields with winter flooding might stimulate N release during the growing season also. The concentrations of available P in our soils were near the lower limit of recommended values, and thus dissolution of P by tubificid activity may promote rice growth after the accumulation of more P.

5. Conclusions

Winter flooding without fall application of rice bran gave the highest rice yield from the use of common organic materials, above the guideline yield in conventional production. The additional application of rice bran in fall increased straw dry matter and stem length at the expense of grain yield. The N application rate in our organic method was near the amounts in conventional inorganic fertilizer use, but more release of N in later growth stages is needed to improve the percentage of ripened grains. Although the P application rate was 429% ± 125% of the P in ears of rice at harvest, greater application of P in the early growth stage might be more effective in soil with low available P. Winter flooding plays a significant role in organic rice production by increasing the supply of N to rice. The absence of a reduction of N availability during fall 2011 to spring 2016 and the increase of available N by additional fall application of rice bran show that our organic practice would be an effective way to restore soil fertility in rice paddies sown to soybean.

The types and amounts of organic materials should be determined according to N availability and the cation exchange properties of soils. Further studies should support the uptake of organic rice production, as our results show the potential.

Acknowledgments

We would like to thank Yoshiaki Tomiita, Makoto Kudo, and Kimihiko Yamazaki for invaluable fieldwork assistance, Youko Yoshiba for laboratory assistance, and Dr. Takayuki Mitsunaga for advice on statistical analysis. We gratefully thank the Kurata Shokuhin shop and the Okano tofu shop for providing soybean curd residue.

Disclosure statement

No potential conflict of interest was reported by the authors.