Effects of water table management and row width on the growth and yield of three soybean cultivars in southwestern Japan

Agricultural Water Management 192 (2017) 85–97

Contents lists available at ScienceDirect

Agricultural Water Management journal homepage: www.elsevier.com/locate/agwat

Effects of water table management and row width on the growth and yield of three soybean cultivars in southwestern Japan Naoki Matsuo a,∗ , Masakazu Takahashi b , Tetsuya Yamada c , Motoki Takahashi c , Makita Hajika c , Koichiro Fukami a , Shinori Tsuchiya a a

National Agriculture and Food Research Organization, Kyushu Okinawa Agricultural Research Center, 496 Izumi, Chikugo, Fukuoka 833-0041, Japan National Agriculture and Food Research Organization, Kyushu Okinawa Agricultural Research Center, 2421 Suya, Koushi, Kumamoto 861-1192, Japan c National Agriculture and Food Research Organization, National Institute of Crop Science, 2-1-18 Kannondai, Tsukuba, Ibaraki 305-8518, Japan b

a r t i c l e

i n f o

Article history: Received 3 November 2016 Received in revised form 25 May 2017 Accepted 26 June 2017 Keywords: Growth Row width Soybean Water table management Yield

a b s t r a c t In southwestern Japan, soil water fluctuations from flooding to drought cause unstable soybean yields. Water table management (WTM) with sub-irrigation/drainage systems will overcome the soybean yield instability by inhibiting these fluctuations. Narrow row cultivation is expected to increase soybean yields. The effects of WTM and row width on soybean growth and yield in this region are not clear. We evaluated the effects of WTM with sub-irrigation/drainage systems and row widths (35 or 70 cm) on the growth and yield of one conventional (tall main stem) and two newly developed (short main stem) soybean cultivars. The WTM consisted of (1) fluctuation of the water table between the natural water table depth and that at 30 cm depth according to the growth stage and weather conditions, especially rainfall events (newly developed); (2) maintaining the water table at a 30 cm constant depth throughout the growth period (recommended in Japan); and (3) the natural water table with an underdrain (control). No significant interaction was observed between the WTM and cultivar or row width treatment, indicating that cultivars and row width treatments responded similarly to WTM. WTM 1 and 2 decreased the soybean yield by approx. 5% when the natural water table depth in control existed at 50–60 cm depths throughout the growing period, indicating that the natural water table depth in control was near optimum for soybean growth and yield. Before performing WTM, therefore, the natural water table depth should be measured and considered. The combination of newly developed cultivars with narrow rows had similar or greater yields than conventional cultivation (cultivar and row width), due mainly to an increase in pods m−2 and a decrease in yield loss without severe lodging. Thus, yield potential in southwestern Japan could be increased by narrow row cultivation, but cultivars with short main stem lengths should be cultivated. © 2017 Elsevier B.V. All rights reserved.

1. Introduction Soybean (Glycine max [L.] Merr.) is one of the important crops in the world, because of its abundant protein and oil contents. In Japan, more than 80% of soybean crops are cultivated in fields that are converted from paddy fields (MAFF, 2015). Those paddy fields were originally designed to hold irrigation water, and thus the drainage speed in the fields is generally low. In southwestern Japan, the optimum planting date is considered to be early July to mid-July (Uchikawa et al., 2003), but the rainy season (early June to mid- to late July) sometimes overlaps with this period. Thus, grow-

∗ Corresponding author. E-mail address: [email protected] (N. Matsuo). http://dx.doi.org/10.1016/j.agwat.2017.06.024 0378-3774/© 2017 Elsevier B.V. All rights reserved.

ers sometimes cannot plant soybean seeds at the optimum planting period. In addition, if a rainfall occurs before planting, planting will be delayed because the agricultural machinery cannot be operated at such times. If a heavy rainfall occurs after planting, the soil surface will be flooded, causing the reduction of seedling establishment or early plant growth (Hamada et al., 2007; Nakayama et al., 2004). The rainy season in southwestern Japan sometimes continues until late July, resulting in late (late July to early August) planting which often causes yield reduction due to insufficient vegetative growth (Egli et al., 1987; Fatichin et al., 2013). Therefore, the speed of the drainage of paddy soil should be improved. In addition to flooding stress during the rainy season in southwestern Japan, the amount of rainfall is relatively small after the rainy season (late July to September). This imposes drought stress on soybean plants, decreasing the seedling establishment or early

86

N. Matsuo et al. / Agricultural Water Management 192 (2017) 85–97

vegetative growth. This may cause a reduction of the seed yield. In this way, the soil moisture fluctuation from flooding to drought inhibits the stability of the soybean yield in southwestern Japan. Lysimeter and greenhouse experiments demonstrated the existence of an optimum water table for obtaining high soybean yields (Nathanson et al., 1984; Madramootoo et al., 1995; Shimada et al., 1995, 1997), but it may be difficult to control the groundwater level precisely in actual crop fields. To solve these problems, field equipment with a sub-irrigation/drainage system, which is called a farm-oriented enhancing aquatic system (FOEAS), was developed in 2005 (Shimada et al., 2012; Wakasugi and Fujimori, 2009) to manage the water table depth even under the field conditions. When the drainage mode is used, extra surface water that is present during the rainy season is quickly drained, facilitating the field operation with agricultural machinery soon after rainfall. When the sub-irrigation mode is used, the water table can be set at the desired depth between +10 and −35 cm from the soil surface. Thus, paddy rice and upland crops, such as soybean, wheat and barley, can be cultivated in the same fields. For the upland crops, if drought spell continues for several days, the sub-irrigation system can alleviate drought stress. The effects of water table management (WTM) with a subirrigation/drainage system on soybean growth and yield under field conditions were examined in the U.S. and Canada (Cooper et al., 1991; Fisher et al., 1999; Mejia et al., 2000; Nelson et al., 2011). In most of those studies, WTM with a sub-irrigation/drainage system increased the soybean yield compared to that under freedrainage conditions. In Japan, however, there are few reports on the effect of WTM on soybean growth and yield under field conditions. Shimada et al. (2012) reported that WTM with a FOEAS at the constant depth of 20 or 32 cm throughout the growing period significantly increased the soybean (cv. Tachinagaha) yield compared to the yield obtained without WTM (i.e., only open-ditch drainage) in mid-Japan. In contrast, Matsuo et al. (2013) reported that in southwestern Japan, WTM with a FOEAS at a constant depth of 20 cm throughout the growing period tended to decrease the yields of two soybean cultivars compared to the yields obtained under the control conditions (i.e., only an underdrain was used), and the effect of WTM at the constant depth of 35 cm throughout the growing period on the yield was not significant for either cultivar. Thus, the effect of WTM might differ between cultivation areas and cultivars. However, in the above-cited studies in Japan, the water tables were maintained at constant depths throughout the growth period, even though the FOEAS used could freely change the water table from +10 to −35 cm depth at any time. Ayars et al. (2006) proposed the ideal WTM in which the water table is maintained at the position where the root systems exist at an early growth stage and is then lowered as the root systems grow. It is possible that the use of a FOEAS will help develop new techniques for optimum WTM, as proposed by Ayars et al. (2006). To increase agricultural income, low-cost and labor-saving cultivation techniques are required. To realize low-cost cultivation, the yield should be increased and the usage of agricultural materials, such as agrichemicals, should be minimized. To realize labor-saving cultivation, some agricultural practices should also be minimized. To achieve these purposes, narrow-row cultivation may be one of the desirable techniques. Several studies demonstrated that narrow-row cultivation increased soybean yields (Bullock et al., 1998; Cooper, 1997; Cox and Cherney, 2011; Robinson and Wilcox, 1998). In addition, weed emergence and/or weed growth can be inhibited by narrow-row cultivation, because the aboveground parts of plants grown in narrow-row cultivation cover the ground more quickly than those in wider row cultivation (Matsuo et al., 2015). Thus, the costs for herbicide or the labor for intertillage and ridging, which are conventional agricultural practices used in Japan to control weed population, can be minimized.

The soybean cultivar Fukuyutaka, which is a leading cultivar in southwestern Japan and has long been cultivated for more than 30 yr, has a long main stem. Thus, lodging often occurs with this cultivar under narrow-row cultivation, because intertillage and ridging cannot be performed under this cultivation (Matsunaga et al., 2003). Lodging sometimes makes harvesting by agricultural machinery difficult, and it reduces soybean yields. Thus, new soybean cultivars with a short main stem and lodging resistance even under narrow-row cultivation are needed in order to shift from conventional wide-row cultivation to narrow-row cultivation in southwestern Japan. Indeed, some cultivars with a short main stem and lodging resistance were recently bred in Japan. The objective of the present study was to evaluate the effects of three types of WTM on the growth, lodging, yield and yield components of three soybean cultivars under narrow (35 cm) and conventional normal row (70 cm) cultivation. We hypothesized that: (1) WTM depending on the growth stage and/or weather conditions (especially rainfall events) may increase soybean growth and yield compared to soybean plants grown under conditions in which the water table is maintained at a constant depth of approx. 30 cm throughout the growing season (recommended in Japan) or not managed under natural water table conditions (i.e., only an underdrain was used), and (2) narrow-row cultivation is suitable for newly developed soybean cultivars with short main stems. 2. Materials and methods 2.1. Site description and plant materials The field experiments were conducted in 2011, 2012 and 2013 at the Kyushu Okinawa Agricultural Research Center (KARC), Chikugo, Fukuoka, Japan (33◦ 12 N, 130◦ 30 E, 10 m elevation). There were three FOEASs at the KARC during those years, next to paddy rice fields. Before conducting the experiment, a field was divided into three plots and FOEASs were constructed in each plot. Thus, three WTM treatments could be conducted in a single year. The distance between the WTM treatments examined in the present study was approx. 3 m. The water movement between the treatments and seepage were inhibited by compacting the soil and by burying vinyl sheets down to a depth of >60 cm. The soil was lowland paddy soil (Typic Endoaquept) and contained 23.0% sand, 36.1% silt and 41.9% clay (light clay). The previous crop was wheat for all three years. The soybean cultivars used were Fukuyutaka (maturity group VI), Sachiyutaka A1 (maturity group V) and Hatsunagaha (maturity group IV). Fukuyutaka is a conventional cultivar, released in 1980 and cultivated widely in southwestern Japan and Sachiyutaka A1 was bred by the Institute of Crop Science, Tsukuba, Ibaraki, Japan and released in 2012. Hatsunagaha was bred by the KARC and released in 2014. Fukuyutaka has a long main stem, whereas Sachiyutaka A1 and Hatsunagaha have short main stems. 2.2. Crop management Soybeans were grown in the same field for all three years. The planting date was July 11, July 18 and July 16 in 2011, 2012 and 2013, respectively (Table 1). After seedling establishment, the following three WTM treatments were conducted. (1) The water table was manipulated during crop season depending on the growth stage and weather conditions (especially rainfall events) by using both sub-irrigation and drainage modes (FL plot). (2) The water table was set at −30 cm depth throughout the crop season by using only the sub-irrigation mode (30-cm plot). (3) An artificial water table was not set throughout the crop season, by using only the drainage mode with the natural water table (Control plot).

N. Matsuo et al. / Agricultural Water Management 192 (2017) 85–97 Table 1 Planting dates and days to reach specific reproductive stages during the 2011–2013 growing season. Year

Cultivar

Planting

R2a

R5a

R8a

2011

Fukuyutaka Sachiyutaka A1 Hatsunagaha

7/11 7/11 7/11

8/25 8/19 8/17

9/16 9/14 9/9

11/4 10/30 10/28

2012

Fukuyutaka Sachiyutaka A1 Hatsunagaha

7/18 7/18 7/18

8/30 8/27 8/24

9/21 9/18 9/13

11/6 10/29 10/26

2013

Fukuyutaka Sachiyutaka A1 Hatsunagaha

7/16 7/16 7/16

8/27 8/23 8/21

9/20 9/18 9/13

11/14 11/4 11/1

a R2: Open flower at one of the two uppermost nodes. R5: The seed in the pod at one of the four uppermost nodes is 3 mm long. R8: 95% of the pods have reached their mature color (Fehr et al., 1971).

The basic concept of the FL plot was that the water table was set at approx. −20 cm depth after seedling establishment to promote early plant growth, and then the water table was lowered to the −30 cm depth. Thereafter, the water table was not set (as in the Control plot), in order to promote root growth. During the reproductive stage, the water table was raised, in order not to impose drought stress on the soybean plants. In 2011, the water table of the FL plot was set at −20 cm depth from 15 days after sowing (DAS) to 29 DAS. The groundwater was then drained until mid-September, at which point the groundwater was raised to −30 cm depth until mid-October. In 2012, water table was set at −20 cm depth from 10 to 21 DAS and then set at −30 cm for next 2 weeks. Thereafter, groundwater was drained. Because >10 mm rainfall occurred frequently in 2012, the water table was set at −30 cm depth temporarily only when there was no precipitation for >7 days. In 2013, the water table was set at −20 cm from 10 to 21 DAS and then at −30 cm for the next 2 weeks. Thereafter, groundwater was drained. Because >10 mm rainfall did not occur for 30 days from early September, the water table was again raised to −20 to −30 cm depth from mid-September to early October. Because the WTM treatments could not be replicated within each year, a single year was considered one replication. The WTM treatments were arrange randomly in each year. In each year, each WTM treatment had three sub-sample plots, arranged in a splitplot design: the main plot was the cultivar and the sub-plot was the row width. The area of each FOEAS was ≥360 m2 , and the area of each sub-sample plot was ≥14.4 m2 . Seeds were sown in the normal (70 cm) or narrow-row (35 cm) width. Three seeds of each cultivar were sown at spaces of 70 cm row width × 20 cm intrarow spacing or 35 cm row width × 20 cm intrarow spacing. Irrigation water was applied for the first 10 days using irriga® tion tubes (Sumisansui , Sumika Agrotech Co., Osaka, Japan) to ensure uniform seedling establishment. The seedlings were then thinned to one plant per hill (7.1 and 14.2 plants m−2 for each 70 cm and 35 cm row treatment, respectively). For both row-width treatments, intertillage and ridging, a normal practice in Japan to control lodging and weeds, were not carried out. The fields received 10 t ha−1 of pellet-type manure approx. 1 month before sowing. No additional fertilizer was applied, according to conventional agricultural practice in this region. A combination of insecticide, herbicide and manual weed control was used to maximize the yields. 2.3. Measurement of weather conditions, SWP and shoot growth parameters The daily mean temperature and rainfall amount were measured at the meteorological station of the KARC, located approx. 100 m away from the experimental field. From these meteorological data, the daily reference evapotranspiration (ET0 ) was

87

calculated based on the Food and Agricultural Organization of the United Nations (FAO) Irrigation and Drainage paper No. 56 (FAO56; Allen et al., 1998). The daily soil water potential (SWP) at depths of 10 and 30 cm was measured with tensiometers (HD-001, Sensez, Tokyo). The SWP was measured only for the Fukuyutaka plants that underwent the 70 cm row treatment, due to the limitation of the number of tensiometers. In each of the three WTM treatments, the SWPs of three subsample plots were measured. In total, 18 tensiometers (three WTMs × three sub-plots × two depths) were used in the present study. For all of the WTM treatments, a polyvinyl chloride pipe (inner dia. 10 cm, height 65 cm) per WTM treatment were installed down to approx. 60 cm at the central position, and the daily water table depth was measured with water-level loggers (U20-001-04, Onset Computer, Bourne, MA, USA). The recording intervals were 60 min for both the tensiometers and the water-level loggers. The data were averaged and are presented here on a daily basis. Because the measuring limit of the water table depth was 60 cm, water table depth data below 60 cm are not presented. Because >80% of the soybeans grown in Japan are cultivated in fields converted from rice paddy fields, and since paddy rice is often cultivated near soybean fields in Japan, there was enough irrigation water in the irrigation channels around the soybean field, and growers do not consider how much irrigation water they use. Therefore, the amount of irrigation water was not measured in this study. At the R2 and R5 stages (according to Fehr et al., 1971), crop samples were collected to determine the shoot dry weight (SDW; g m−2 ) and leaf area index (LAI; m2 m−2 ). The R2 and R5 dates for each year are shown in Table 1. The plants in a 1.12 m2 area per subsample plot were collected at each sampling, leaves were separated from whole plants and the leaf area was determined with a leaf area meter (LI-3000C, LI-COR, Lincoln, NE). The leaf area index (LAI; m−2 ) was calculated by dividing measured leaf area by sampling area. The aboveground parts were then dried at 80 ◦ C in a ventilated oven for at least 72 h to determine the shoot dry weight (SDW; g m−2 ). The crop growth rate (CGR; g m−2 d−1 ) during R2 to R5 was calculated by subtracting the SDW at R2 from the SDW at R5 and dividing this value by the number of days from R2 to R5. After the measurement of shoot traits, the root systems of only Fukuyutaka plants in the 70 cm row treatment were collected at two positions per sub-sample plot down to a 30 cm depth from the soil surface with a soil sampler (length 30 cm, inner dia. 5 cm; HS-30S, Fujiwara Scientific, Tokyo). The soil samples were collected between hills and were further divided into sections of 0–15 cm and 15–30 cm and stored in a refrigerator at 4 ◦ C before analysis. The soil was carefully removed from the root samples with tap water, and the nodules were completely removed. The root length density (RLD; cm cm−3 ) was analyzed with the WinRHIZO image analysis software package (Regent Instruments, Montreal, Canada). The root systems were then dried at 80 ◦ C in a ventilated oven for at least 72 h to determine the root weight density (RWD; g cm−3 ). 2.4. Post-harvest measurements of agronomical traits At harvest (the dates of harvest for each year are shown in Table 1), the lodging angle of each plant’s main stem was determined according to the method of Matsuo et al. (2015) for 10 plants per sub-sample plot. Briefly, a 15 cm × 15 cm clear acryl plate on which straight lines were drawn with a marker at 10◦ , 20◦ , 40◦ and 60◦ to the upright was used. If the lodging angle was between 0◦ and 10◦ , 10◦ and 20◦ , 20◦ and 40◦ , 40◦ and 60◦ or >60◦ , the lodging score was defined as 0, 1, 2, 3, or 4, respectively. After the lodging score measurement, the plants in a 2.8 m2 area per sub-sample plot were sampled to determine the plant height, the number of main stem nodes, the lowest node height with pods (LNHP), the yield and the yield components (i.e., pods m−2 , seeds

88

N. Matsuo et al. / Agricultural Water Management 192 (2017) 85–97

pod−1 , and the 100-seed weight). Because harvesters collect seeds above approx. 10 cm from the soil surface on average, the yield loss was also estimated as described by Matsuo et al. (2015). Briefly, before harvest, the plants were marked with a spray at a position of 10 cm above the ground surface with the aid of a metal plate (10 cm high × 20 cm long). Then the pods above 10 cm were separated from those below 10 cm. The yield, pods m−2 , seeds pod−1 , and seed weight >10 cm above the soil surface were measured, but the yield components below 10 cm from the soil surface were not measured (only the seed weight was determined). The seed yields were adjusted to 130 g kg−1 moisture.

2.5. Experimental design and statistical analysis Because the WTM treatments could not be replicated within each year, a single year was regarded as one replication as noted above. The average value of the three sub-sample plots with respect to each year was used for the statistical analysis. The experimental design was a split–split plot on a randomized complete block design with three replications (years). The main plot was the WTM treatment, the sub-plot was the cultivar, and the sub–sub plot was the row width. The statistical analysis was carried out using SPSS v. 23 software (SPSS, Chicago, IL). WTM treatment, cultivar, row width and their interaction were considered fixed effects, and replication (year) was used as a random effect. An analysis of variance (ANOVA) was conducted to test the effects of WTM treatment, cultivar, row width and their interaction on shoot growth parameters (SDW, LAI, CGR) and agronomical traits at harvest (plant height, the number of main stem nodes, LNHP, yield, yield components, yield loss and lodging score). The effect of WTM treatment on the RLD and RWD only for Fukuyutaka was also analyzed. The significance level was set at 0.05 for the shoot and root growth parameters, whereas it was set at 0.05 or 0.10 for the measurements carried out at harvest according to Mejia et al. (2000) who studied the effect of WTM on soybean in the field condition and analyzed yield and yield components data setting threshold for significance at p < 0.10. Significant treatment effects (p < 0.05 or p < 0.10) were further checked by using Fisher’s protected least significant difference (LSD) test.

3. Results 3.1. Weather conditions The mean air temperature during the years 2011–2013 was higher than the 30-yr average, except for November in 2012 and 2013 (Table 2). The mean air temperature in July to October tended to be higher in 2013 than in 2011 and 2012. In 2011, the amount of rainfall in July was lower than the 30-yr average (Table 1), and rainfall events were concentrated in early to mid-July (i.e., before the present experiments) (Fig. 1a). After the rainy season, the rainfall concentrated around mid- to late August (Fig. 1a), but thereafter, no rainfall of >10 mm was observed except for mid-September and mid-October (Fig. 1a). In 2012, the amount of rainfall in July was more than twice the 30-yr average (Table 1), but the rainfall was concentrated before the onset of the experiments (Fig. 1b). Although the total amount of rainfall in August and September was smaller in 2012 than the 30-yr average (Table 1), rainfall of >10 mm occurred frequently (12 d) during August to September (Fig. 1b). In 2013, the rainfall was concentrated between late August and early September (Fig. 1c), because two typhoons came close to the experimental field. Thereafter, rainfall of >10 mm was not observed until early October.

3.2. Water table In 2011, the water table in the FL plot was set at −20 to −30 cm depth until mid-August (Fig. 1a). Because the groundwater was drained thereafter, the water table decreased and it fluctuated depending on the rainfall events. The water table was then set at approx. −30 cm until early October, because the amount of rainfall was low after early September. In 2012, the water table in the FL plot was set at −20 cm for the first 10 d and then lowered to −30 cm for the next 10 d (Fig. 1b). Thereafter, the groundwater was drained until early October, because frequent rainfall of >10 mm occurred in 2012. The water table between late August to early October was approx. −50 cm. In early October, the water table was temporarily raised to approx. −30 cm to prevent drought stress and then the groundwater was drained again. In 2013, the WTM in the FL plot was similar to that in 2012, except for mid- to early October; because there was little rainfall during this period, the water table was kept at −20 to −30 cm (Fig. 1c). Thereafter, the groundwater was drained again. The water table in the 30-cm plot was maintained at approx. −30 cm throughout the growth season, except for late August to early September when two typhoons came close to the experimental field. The water table in the 30-cm plot increased in mid-September to early October when the water table in the FL plot was increased, implying that there might have been a seepage between the FL plot and the 30-cm plot. The water table in the Control plot was −50 to −60 cm until early October, except for late August and early September, due to the occurrence of heavy rainfall. The water table in the 30-cm and the Control plot was maintained at approx. −30 cm and −50 to −60 cm, respectively, throughout the growth season for three years, except when heavy rainfall occurred (Fig. 1). 3.3. Soil water potential Regardless of the WTM, the SWP at −10 cm depth fluctuated largely depending on the rainfall events for all three years (Fig. 1a, c, e). The SWP at −30 cm depth in the 30-cm plot was kept above −10 kPa throughout the growth period for all three years (Fig. 1b, d, f). In the Control plot, the SWP at −30-cm fluctuated greatly depending on rainfall events (Fig. 1b, d, f). In general, the SWP at −30 cm depth in the Control plot decreased to less than −30 kPa between mid-September and mid-October. In the FL plot, however, the decrease in the SWP at −30 cm depth was alleviated in comparison to that in the Control plot (Fig. 2b, d, f), because the water table was raised during this period if the amount of rainfall was small (Fig. 1). Fluctuation of the SWP below −30 kPa among sub-sample plots was relatively large. 3.4. Reference evapotranspiration The daily ET0 values during the crop seasons are shown in Fig. 3. The ET0 exceeded 5 mm day−1 until late August for all three years. The ET0 then decreased to below 5 mm day−1 after September, except for mid-September in 2013; the ET0 exceeded 6 mm day−1 . The ET0 values gradually decreased toward November. 3.5. Plant growth Table 3 shows the effects of WTM, cultivar and row width on the LAI and SDW at R2 and R5, and the CGR during R2 to R5. The main effects of WTM, cultivar and row width on the LAI were significant at R2 and R5, except for the WTM effect at R5. At R2, the 30-cm plot and the Control plot had greater LAI values compared to the FL plot, but there was no significant difference in LAI at R5 among

N. Matsuo et al. / Agricultural Water Management 192 (2017) 85–97

89

Table 2 Monthly mean air temperature and rainfall during the 2011–2013 growing seasons and the 30-yr averages. Mean air temperature (◦ C)

Month

July August September Octorber November

Rainfall (mm)

2011

2012

2013

30-yr

2011

2012

2013

30-yr

27.6 28.0 25.0 19.0 15.2

27.6 28.8 24.4 18.5 11.2

28.5 29.2 25.0 20.3 12.1

27.1 27.9 23.9 18.5 12.7

231 238 82 128 158

733 149 105 106 105

143 576 145 124 74

340 200 157 70 76

Fig. 1. Groundwater level and rainfall amounts in the 2011 (a), 2012 (b) and 2013 (c) crop seasons. Dashed, dotted and solid lines indicate the FL, 30-cm, and Control plots, respectively. Vertical bars indicate the rainfall amount.

90

N. Matsuo et al. / Agricultural Water Management 192 (2017) 85–97

Fig. 2. Rainfall amounts and soil water potential (SWP) at depths of 10 cm (a, c, e) and 30 cm (b, d, f) in the 2011 (a, b), 2012 (c, d) and 2013 (e, f) crop seasons. Grey circles with dashed line, white circles with dotted line and black circles with solid line indicate the FL, 30-cm, and Control plots, respectively. Data are shown as mean ± S.E. (n = 3).

Table 3 Effects of water table management (WTM), cultivar and row width on the leaf area index (LAI) and shoot dry matter weight (SDW) at stages R2 and R5 and the crop growth rate (CGR) during the R2 to R5 stages. Data were averaged over three years. LAI (m−2 m−2 )

SDW (g m−2 )

CGR during R2-R5 (g m−2 d−1 )

R2

R5

R2

R5

WTM FL 30 cm Control

3.2 ba 3.6 a 3.5 a

4.3 a 4.5 a 4.6 a

177 b 193 a 190 a

503 b 523 ab 532 a

13.7 a 13.9 a 14.4 a

Cultivar Fukuyutaka Sachiyutaka A1 Hatsunagaha

4.2 a 3.5 b 2.6 c

5.1 a 4.4 b 4.0 c

241 a 184 b 135 c

558 a 537 b 463 c

13.2 b 13.8 b 15.0 a

Row width 35 cm 70 cm

4.1 a 2.7 b

5.0 a 4.0 b

225 a 148 b

574 a 465 b

14.7 a 13.3 b

WTM (W) Cultivar (C) Row width (R) W×C W×R C×R W×C×R

<0.001 <0.001 <0.001 nsb ns ns ns

ns <0.001 <0.001 ns ns ns ns

0.025 <0.001 <0.001 ns ns 0.028 ns

0.027 <0.001 <0.001 ns ns ns ns

ns 0.003 0.001 ns ns 0.049 ns

a b

Means with the same letter do not differ by Fisher’s protected LSD (p < 0.05). ns: no significant difference.

N. Matsuo et al. / Agricultural Water Management 192 (2017) 85–97

91

Fig. 3. Reference evapotranspiration (ET0 ) in the 2011 (a), 2012 (b) and 2013 (c) crop seasons.

the three types of WTM. At R2 and R5, Fukuyutaka produced the greatest LAI, followed by Sachiyutaka A1 and Hatsunagaha. The LAI values in the 35 cm row treatments were greater than those in the 70 cm row treatments at R2 and R5. There was a significant main effect of WTM on the SDW at R2. The SDW at R2 was greater in the 30-cm plot and the Control plot compared to that in the FL plot. A significant cultivar × row width interaction on the SDW at R2 was revealed (Table 3). For both row widths, Fukuyutaka had the greatest SDW at R2, but the magnitude of increase in the SDW in response to row width differed among the

cultivars: the difference in the SDW at R2 between the 35 cm and 70 cm row treatments was 90, 77, and 60 g m−2 for Fukuyutaka, Sachiyutaka A1 and Hatsunagaha, respectively (Table 4). There were significant main effects of WTM, cultivar and row width on the SDW at R5. The SDW at R5 was the greatest in the Control plot, followed by the 30-cm plot and the FL plot in that order. The SDW of Fukuyutaka was the greatest, followed by Sachiyutaka A1 and Hatsunagaha. The SDW in the 35 cm row treatments was greater than that in the 70 cm row treatments.

92

N. Matsuo et al. / Agricultural Water Management 192 (2017) 85–97

Table 4 Interaction of cultivar with row width for shoot dry matter weight (SDW) at R2 stage and crop growth rate (CGR) during R2 to R5 stages. Data were averaged over three years. Cultivar

Fukuyutaka Sachiyutaka A1 Hatsunagaha LSD0.05

SDW at R2 (g m−2 )

CGR during R2-R5 (g m−2 d−1 )

Row width

Row width

35 cm

70 cm

35 cm

70 cm

288 222 165

194 145 105

13.4 14.2 16.4

13.0 13.4 13.6

17

The main effect of WTM on the CGR during R2 to R5 was not significant (Table 3). Main effects of cultivar and row width on the CGR during R2 to R5 were significant. Hatsunagaha had a greater CGR value than Fukuyutaka and Sachiyutaka A1 and narrower row width resulted in a greater CGR value than wider row width. There was also a significant cultivar × row width interaction on the CGR during R2 to R5. In the 35 cm row treatment, the cultivar Hatsunagaha had the greatest CGR (16.4 g m−2 d−1 ), whereas there was no cultivar difference in the 70 cm row treatment (Table 4). There was no difference in the CGR between row widths for Fukuyutaka and Sachiyutaka A1, whereas for Hatsunagaha, the CGR in the 35 cm rows treatment was greater than that in the 70 cm row treatment by 2.8 g m−2 d−1 (Table 4). Table 5 shows the effect of WTM on the RLD and RWD at R2 and R5 only for the Fukuyutaka soybeans. For the RLD, a significant effect of WTM was found for the depth of 15–30 cm at R5. Among the three types of WTM, the 30-cm plot had the greatest RLD at 15–30 cm depth. For the RWD, there was no significant effect of WTM at any growth stage or depth. 3.6. Plant height, the number of main stem nodes and LNHP at harvest Table 6 shows the effects of WTM, cultivar and row width on plant height, the number of main stem nodes and the LNHP at harvest. The main effect of WTM on plant height was significant (P < 0.10), and the plant height was greater in the 30-cm plot than in the Control plot. There was a significant cultivar × row width interaction on plant height. Regardless of row width, Fukuyutaka had the highest plant height, followed by Sachiyutaka A1 and Hatsunagaha (Table 7). The magnitude of increase in plant height in response to row width differed among the three cultivars; the 35 cm row treatment increased plant height by 8.1, 7.2 and 4.2 cm for Fukuyutaka, Sachiyutaka A1 and Hatsunagaha, respectively, compared to the 70 cm row treatment (Table 7). There were significant (P < 0.10) main effects of WTM and cultivar on the number of main stem nodes. The 30-cm plot had the greatest number of main stem nodes, followed by the FL plot and the Control plot. The number of main stem nodes of Fukuyutaka was the highest among the three cultivars. The main effects of WTM (P < 0.10), cultivar and row width on the LNHP were significant. The LNHP in the FL plot was greater than that in the Control plot. Fukuyutaka had the greatest LNHP, followed by Sachiyutaka A1 and Hatsunagaha. The LNHP in the 35 cm row treatment was greater than that in the 70 cm row treatment. 3.7. Yield, yield components and yield loss The effects of WTM, cultivar and row width on the soybean yield, yield components and yield loss are shown in Table 6. There were significant (P < 0.10) main effects of WTM, cultivar and row width on the yield. The yield in the Control plot was approx. 5% greater than that in the 30-cm plot. Fukuyutaka and Sachiyutaka A1 had

1.4

> 10% greater yields than Hatsunagaha. The yield in the 35 cm row treatment was approx. 14% greater than in the 70 cm row treatment. The main effects of cultivar and row width on the pods m−2 were significant. Fukuyutaka had greater pods m−2 values than the other two cultivars, and the 35 cm row treatment produced greater pods m−2 than the 70 cm row treatment. There were no significant main effects or interactions on seed pod−1 . A significant main effect of cultivar on 100-seed weight was revealed. Sachiyutaka A1 had the greatest 100-seed weight among the three cultivars, followed by Hatsunagaha and Fukuyutaka in that order. A significant (P < 0.10) main effect of WTM on yield loss was also observed. The yield loss in the Control plot was greater than that in the FL plot. There was a significant cultivar × row width interaction on yield loss. With the 70 cm row treatment, the yield loss of Fukuyutaka was significantly less than those of Sachiyutaka A1 and Hatsunagaha, whereas a cultivar difference in yield loss was not observed with the 35 cm row treatment (Table 7). The yield loss of Fukuyutaka did not differ between the 35 cm and 70 cm row treatments, whereas the yield loss in the 70 cm row treatment was greater than that in the 35 cm row treatment for Sachiyutaka A1 and Hatsunagaha. 3.8. Lodging scores The effects of WTM, cultivar and row width on the lodging scores are shown in Table 6. A significant cultivar × row width interaction on the lodging score was found. For the cultivars Fukuyutaka and Hatsunagaha, the lodging score did not differ significantly between the two row treatments, whereas the lodging score in the 35 cm row treatment was smaller than that in the 70 cm row treatment for Sachiyutaka A1 (Table 7). With the 35 cm row treatment, Fukuyutaka had greater lodging scores than Sachiyutaka A1 and Hatsunagaha, but with the 70 cm row treatment, the lodging scores of Fukuyutaka and Sachiyutaka A1 were greater than that of Hatsunagaha. 4. Discussion 4.1. Accuracy of water table control by a FOEAS In the present study, it was observed that the rainfall was concentrated and heavy rainfall occurred at the last stage of the rainy season (i.e., early to mid-July) (Fig. 1), which is characteristic of the rainy season in southwestern Japan. This rainfall pattern often delays soybean planting, because the drainage speed of the fields in that region (which were converted from paddy fields) is generally low. However, field operations such as plowing, harrowing and planting can be performed within a few days after the end of rainy season as in this study, if a FOEAS is constructed in the field. In Canada, Mejia et al. (2000) aimed to keep the water table constant at depths of 50 cm and 75 cm with sub-irrigation/drainage systems under field conditions. Although the water tables in those

N. Matsuo et al. / Agricultural Water Management 192 (2017) 85–97

93

Table 5 Effect of water table management (WTM) on root length density (RLD) and root weight density (RWD) of Fukuyutaka at depths of 0–15 and 15–30 cm measured at R2 and R5 stages. Data were averaged over three years. RLD (cm cm−3 )

WTM R2

FL 30cm Control ANOVA

RWD (g cm−3 ) R5

R2

R5

0–15 cm

15–30 cm

0–15 cm

15–30 cm

0–15 cm

15–30 cm

0–15 cm

15–30 cm

0.267 a 0.270 a 0.286 a 0.828

0.208 a 0.222 a 0.197 a 0.920

0.846 a 0.683 a 0.907 a 0.421

0.265 b 0.520 a 0.181 b 0.022

0.017 a 0.024 a 0.024 a 0.489

0.014 a 0.010 a 0.019 a 0.482

0.050 a 0.054 a 0.056 a 0.822

0.016 a 0.022 a 0.016 a 0.403

Means with the same letter do not differ by Fisher’s protected LSD (p < 0.05).

Table 6 Effects of water table management (WTM), cultivar and row width on plant height, the number of main stem nodes (Node no.), the lowest node height with pods (LNHP), yield, pods m−2 , seeds pod−1 , 100-seed weight, yield loss and lodging score at harvest. Data were averaged over three years. Plant height (cm)

Node no. (no. plant−1 )

LNHP (cm)

Yeild (g m−2 )

Pods m−2

Seeds pod−1

100-seed weight (g)

Yield loss (g m−2 )

Lodging score (0–4 scale)

WTM FL 30 cm Control

47.3 ABa 47.7 A 46.0 B

13.5 B 13.7 A 13.4 B

10.1 A 9.5 AB 9.0 B

326 AB 324 B 342 A

620 a 621 a 640 a

1.63 a 1.63 a 1.67 a

32.1 a 32.2 a 32.3 a

10 B 13 AB 14 A

1.8 a 1.8 a 1.8 a

Cultivar Fukuyutaka Sachiyutaka A1 Fukuhayate

60.8 ab 42.8 b 37.4 c

14.8 a 12.9 b 12.9 b

10.9 a 9.4 b 8.3 c

348 a 338 a 306 b

711 a 590 b 580 b

1.62 a 1.67 a 1.64 a

30.2 c 34.2 a 32.2 b

7b 16 a 15 a

2.1 a 1.7 b 1.5 c

Row width 35 cm 70 cm

50.3 a 43.8 b

13.5 a 13.6 a

10.9 a 8.2 b

352 a 310 b

678 a 582 b

1.63 a 1.65 a

32.1 a 32.3 a

7b 18 a

1.7 b 1.9 a

WTM (W) Cultivar (C) Row width (R) W×C W×R C×R W×C×R

0.074 <0.001 <0.001 nsc ns 0.038 ns

0.058 <0.001 ns ns ns ns ns

0.070 <0.001 <0.001 ns ns ns ns

0.078 <0.001 <0.001 ns ns ns ns

ns <0.001 <0.001 ns ns ns ns

ns ns ns ns ns ns ns

ns <0.001 ns ns ns ns ns

0.058 <0.001 <0.001 ns ns <0.001 ns

ns <0.001 0.024 ns ns 0.082 ns

a b c

Means with the same capital letter do not differ by Fisher’s protected LSD (p < 0.10). Means with the same lower case letter do not differ by Fisher’s protected LSD (p < 0.05). ns: no significant difference.

Table 7 Interaction of cultivar with row width for plant height, yield loss and lodging score at harvest. Data were averaged over three years. Cultivar

Fukuyutaka Sachiyutaka A1 Hatsunagaha LSD a b

Plant height (cm)

Yield loss (g m−2 )

Lodging score (0–4 scale)

Row width

Row width

Row width

35 cm

70 cm

35 cm

70 cm

35 cm

70 cm

64.8 46.5 39.5

56.7 39.2 35.3

5 9 7

8 23 23

2.1 1.5 1.4

2.1 2.0 1.6

2.2a

5a

0.3b

Normal text: LSD at p < 0.05. Bold text: LSD at p < 0.10.

treatments were significantly higher than that obtained with a free drainage system alone throughout the growth period, the water tables could not be kept constant and fluctuated largely in response to the rainfall pattern. Elmi et al. (2000) also reported a large fluctuation of the water table in response to the rainfall pattern when this sub-irrigation/drainage system was used in Canada. Although it seems difficult to control the water table at constant depths under field conditions, a FOEAS could be used to control the water table at designated depths relatively constantly (Fig. 1). Results reported by Shimada et al. (2012) and Matsuo et al. (2013)

also support the accuracy of a FOEAS for maintaining the water table at constant depths. 4.2. Effect of WTM on the growth and yield It was observed that the SWP at −30 cm in the Control plot decreased to less than −40 kPa from mid-September to midOctober for all three years (Fig. 2b, d, f). This reduction in the SWP from mid-September to mid-October seems to be one of the meteorological characteristics in southwestern Japan. Matsuo et al. (2013) also reported that the SWP at −30 cm fell below −80 kPa toward

94

N. Matsuo et al. / Agricultural Water Management 192 (2017) 85–97

mid-October in 2008 and 2010 at the same experimental field, where the water table was not controlled. The period from midSeptember to mid-October coincided with the reproductive stage (i.e., pod set to seed filling) of three cultivars used in the present study (Table 1). It is well known that drought stress during these reproductive stages reduces the soybean yield (Korte et al., 1983a,b). Therefore, we raised the water table up to approx. −30 cm in the FL plot to mitigate drought stress only when the amount of rainfall was small during the reproductive stage (Fig. 1). We expected that the growth and yield in the FL plot would be greater than those in the Control plot. Contrary to our expectation, however, the SDW at stage R5 was significantly lower (6.8%) and the yield tended to be lower (4.4%, not significant) in the FL plot compared to those in the Control plot (Tables 3 and 6). It was reported that a temporary increase in water tables at depths of 45 and 90 cm imposed at the R4–R5 stages for 7 days significantly reduced the soybean yield (Stanley et al., 1980). Those authors concluded that root systems could not tolerate the temporary excessive moisture caused by an increase in the water table at these growth stages. In the present study, the temporary increase in water tables during the pod-set to seed-filling stages in the FL treatment might negatively affect soybean growth and yield. Although the reasons for the yield reduction in the FL plot compared with that in the Control plot was unclear here, it is possible that part of the root system was decomposed by hypoxia stress and new roots did not emerge after this late growth stage. Further studies are needed to clarify the optimum duration of sub-irrigation during reproductive stages. The yield in the 30-cm plot was significantly (5.3%) lower than that in the Control plot (Table 6). In the 30-cm plot, all yield components tended to be lower than those in the Control plot, which might have led to the significant difference in the yield between these two plots. Several studies showed that the optimum groundwater level for soybean yield ranged from 60 to 80 cm deep (Madramootoo et al., 1995; Meijia et al., 2000; Shimada et al., 1995; Williamson and Kriz, 1970). In contrast, Nathanson et al. (1984) demonstrated that a groundwater level of 3 or 15 cm increased the soybean yield compared to the yield obtained without a water table (i.e., overhead irrigation). In Japan, Shimada et al. (2012) reported that WTM at 20 and 32 cm increased the soybean yield compared to the yield obtained without a water table in Mid-Japan, while Matsuo et al. (2013) showed that WTM at 20 cm tended to decrease the soybean yield compared to the yield obtained without water tables in southwestern Japan. These results indicate that the effect of WTM by a FOEAS differs among cultivars or cultivation areas in Japan. In the present study’s 30-cm plot, the SWP at 10 cm depth (i.e., the soil surface) fluctuated largely in response to the rainfall pattern, while the SWP at 30 cm depth was constant around −5 kPa. This result indicates that a constant water table at a depth of 30 cm might cause soil saturation below a depth of 10 cm from the soil surface and saturated soil condition might cause hypoxia stress on root systems. The effects of soil-saturated culture on soybean growth and yield have been reported in tropical and subtropical Australia (Garside et al., 1992; Troedson et al., 1989a,b) and in the temperate U.S. (Purcell et al., 1997). Purcell et al. (1997) concluded that soil saturation culture was effective for tropical or subtropical areas, but not for temperate areas. Although the soil in the present 30-cm plot was not completely saturated, our results partly support their conclusion. Overall, it may be concluded that maintaining a water table at depths above 30 cm throughout the growth stage may not be suitable for southwestern Japan. Because the experimental field was next to a rice paddy field in this study, the water table in the Control plot was at a depth of 50–60 cm until at least late September (depending on the experimental year). Babajimopoulos et al. (2007) also pointed out that seepage from rice paddy fields resulted in a high water table in the

neighboring fields in Greece. In light of the results of these previous studies, the optimum water table for soybeans exists at depths of 60–80 cm from the soil surface. Therefore, the depth of the water table in the present Control plot might be more suitable for soybean growth (Table 3) and yield (Table 6) than FL and 30-cm plots. Because the effect of WTM on root growth parameters did not show a consistent trend at different growth stages (Table 5), these parameters seem not to influence soybean growth and yield greatly under the conditions in which the natural water table is relatively high. The root depth of Fukuyutaka plants was monitored with a minirhizotron camera system in two of the present study’s three years, and this depth reached at least 35 cm until the early reproductive stage in all three types of WTM (data not shown). These results indicate that the root systems could absorb soil water below a depth of 35 cm in the Control plot, even though the SWP at a depth of 30 cm during reproductive stage was lower than those in FL and 30-cm plots (Fig. 2). Thus, severe drought stress might not occur in the Control plot as expected. Although the oxygen concentration in the rhizosphere was not measured in this study, hypoxia stress might occur in the FL and 30-cm plots due to high water tables. It is known that hypoxia stress negatively affects root functions (water and nutrient uptake) (Bramley and Tyerman, 2010; Elzenga and van Veen, 2010) or nodule activity (Maekawa et al., 2011). Thus, it was possible that these root functions rather than root morphological traits might affect soybean growth and yield differently among the WTM treatments. Further studies are necessary to investigate this issue. 4.3. Perspective for water table control with a FOEAS It was observed that the ET0 values during mid-September to mid-October were smaller than those during July to August (Fig. 3) and that the trend of ET0 during this period did not coincide with that of the SWP at −30 cm in the Control plot (Figs. 2, 3). This means that calculating ET0 could not estimate the soil moisture conditions. To address this contradiction, the actual crop evapotranspiration (ETc) must be calculated in a future study to predict exact water balance for soybeans and to schedule sub-irrigation. In this study, the shoot growth parameters (SDW and LAI) differed significantly among the three cultivars and between the two row width treatments, and thus it was difficult to determine which crop coefficient (Kc) should be used, because Kc values are dependent on the biomass level and should thus be considered to be biomassdependent (Mueller et al., 2005). In the U.S., an irrigation support system called ‘Soywater’ (http://hprcc-agron0.unl.edu/soywater/) was developed and has been widely used by soybean producers. With the Soywater system, Kc and ETc values are automatically calculated by inputting data of the location (latitude, longitude and altitude), soil types, meteorological data, planting or emergence date, and the cultivar’s maturity group (cultivation practices, such as plant densities, are not required). When the cumulative ETc decreases below a certain threshold of soil water content (which the producer can set by themselves), Soywater tells the producer that it’s time to irrigate. Now, FOEAS spreads to an area of approx. 10000 ha in commercial farms (including construction schedule), and those areas are gradually increasing (Wakasugi et al., 2016). Although some soybean growers are using sub-irrigation mode when the amount of rainfall is small, they use it empirically. Therefore, the Soywater system may contribute to the creation of sub-irrigation schedules with a FOEAS in the drought-prone areas with a low water table in Japan. The groundwater level was relatively high even in the present study’s Control plot where water table was not controlled (i.e., natural groundwater level). In such a field condition, capillary rise might occur due to shallow groundwater level. Furthermore, root depth reached at least 35 cm until early reproductive stage. There-

N. Matsuo et al. / Agricultural Water Management 192 (2017) 85–97

fore, even though the SWP at a depth of 30 cm was low in the Control plot in September, severe drought stress might not be imposed on soybean plants. To create sub-irrigation schedules, the natural groundwater level, the occurrence of capillary rise, soil characteristics and actual root depth should be taken into account.

4.4. Differences in the cultivars’ agronomical traits in response to row width The insignificant cultivar × row width interaction on yield (Table 6) indicated that the effect of the row width on yield was similar among the three cultivars used in this study. Although averaged across WTM and row width, the cultivar Hatsunagaha produced a significantly lower yield than Fukuyutaka and Sachiyutaka A1 (Table 6), the yields of Fukuyutaka, Sachiyutaka, and Hatsunagaha in the 70 cm row treatment were 336, 314, and 279 g m−2 , and those in the 35 cm row treatment were 361, 361, and 333 g m−2 , respectively (data not shown). The yield differences between the 35 and 70 cm row treatments were 25, 47 and 54 g m−2 for Fukuyutaka, Sachiyutaka A1 and Hatsunagaha, respectively. In southwestern Japan, planting Fukuyutaka in 70–80 cm rows is a conventional cultivation method; our present findings demonstrate that planting cultivars with short main stems in narrow rows can provide similar or greater yields than the conventional cultivation method in this region. In Japan, almost all growers perform intertillage and ridging once or twice during the vegetative stage to control weeds, if they plant soybean seeds at wider (more than 60 cm) rows. If the weed density is high, growers sometimes use herbicide. Narrow row cultivations can reduce weed emergence because the aboveground parts of plants grown in narrow row cultivation cover the ground quickly (Matsuo et al., 2015). Although weed biomass was not measured in this study, this trend was observed. If the weed emergence and growth are inhibited by narrow row cultivations, labor for intertillage and ridging and material costs for weeding (herbicide, fuel for tractors and etc.) can be saved. This will lead to low-cost and labor-saving cultivation. Although the seed costs will increase with narrow row cultivation compared to wide row cultivation, the total production cost can be reduced by narrow row cultivation. In Japan, among total production costs, the rate of seed cost is less than 10% (MAFF, 2016). If the yield is increases by narrow row cultivation, agricultural income may also increase. The difference in the CGR during R2 to R5 between the 35 and 70 cm row treatments was 2.8, 0.8 and 0.4 g m−2 d−1 for Hatsunagaha, Sachiyutaka A1 and Fukuyutaka, respectively (Table 4). This order coincided with that for the yield difference between the 35 and 70 cm row treatments for these cultivars. These results suggest that smaller yields of short-stem cultivars planted in 70 cm rows could be increased by narrow-row cultivation via the enhancement of dry matter accumulation during R2 to R5, resulting in a similar or greater yield than that obtained by the conventional cultivation method (i.e., Fukuyutaka planted in 70 cm rows). In the southern U.S., narrow-row cultivation increased the seed yield compared to wide-row cultivation mainly due to a greater CGR during the vegetative (emergence to R1) and early reproductive stages (R1 to R5), when planting was done in July (Board et al., 1990, 1992). Bullock et al. (1998), who studied the effect of row width (114, 76 and 38 cm) on the growth and yield of indeterminate soybeans in the midwestern U.S., also reported that the seed yield increased with decreasing row width, and the CGR increased with decreasing row width until about R5, but thereafter the CGR did not differ among row widths. From these results, they concluded that the increase in seed yield with narrow-row cultivation was due predominantly to beneficial effects experienced before the grain-filling period. Our results coincided with their report.

95

In the present study, among the yield components, pods m−2 was significantly greater in the 35 cm row treatment compared to the 70 cm row treatment (Table 6) and was significantly and positively correlated with yield (r = 0.826, p < 0.001, n = 18) (data not shown). These results indicate that the increase in pods m−2 led to the yield increase in the 35 cm rows. Our results coincided with previous studies that demonstrated that an increase in seed yield by narrow-row cultivation was due mainly to an increase in pods m−2 (Bullock et al., 1998; Furuhata et al., 2008; Matsuo et al., 2015; Saitoh et al., 2007). In our study, yield components other than pods m−2 were not affected by row width (Table 6), as reported by Board et al. (1990). In the present study, we estimated the yield loss by measuring the yield below 10 cm from the soil surface. It was observed that there was a significant cultivar × row width interaction on yield loss (Table 6). The yield losses of Sachiyutaka A1 and Hatsunagaha were significantly reduced by the narrow row treatment (Table 7). The narrow-row cultivation increased not only the plant height, but also the LNHP (Table 6). The LNHP was significantly and negatively (r = −0.810, p < 0.001, n = 18) correlated with yield loss, and the yield was significantly and negatively (r = −0.742, p < 0.001, n = 18) correlated with yield loss (date not shown). These results indicated that the use of the 35 cm row width decreased the yield loss by increasing the LNHP for the short-stem cultivars (Sachiyutaka A1 and Hatsunagaha), partly resulting in a yield increase for these cultivars in the 35 cm row treatment. The significant cultivar × row width interaction on the lodging score (Table 6) revealed that the response of the lodging score to the row width differed among the cultivars. Hatsunagaha, which has the shortest main stem length among the three cultivars tested, showed a significantly lower lodging score (1.6) than Fukuyutaka and Sachiyutaka A1 (2.1 and 2.0, respectively) in the 70 cm row treatment, whereas in the 35 cm row treatment the short-stem cultivars Sachiyutaka A1 and Hatsunagaha had significantly lower scores (1.5 and 1.4, respectively) than Fukuyutaka (2.1) (Table 7). The short-stem cultivars Sachiyutaka A1 (not significant) and Hatsunagaha (significant) tended to have lower lodging scores in the 35 cm row treatment than in the 70 cm row treatment, indicating that the two short-stem cultivars used in this study are suitable for narrow-row cultivation from the view point of lodging. Unlike the short-stem cultivars, Fukuyutaka had the same lodging score in the 35 and 70 cm row treatments, suggesting that Fukuyutaka is not suitable for narrow-row cultivation and even for a conventional row width if intertillage and ridging are not performed. This implies that intertillage and ridging are essential management practices for conventional cultivation to inhibit lodging. For future labor-saving cultivations, these practice should be minimized. Lodging sometimes makes harvesting with agricultural machinery difficult and leads to yield loss because the number of pods between 0 and 10 cm from the soil surface increases. The effect of artificial lodging treatment at various growth stages on yield loss has been widely studied (Mancuso and Caviness, 1991; Noor and Caviness, 1980; Saitoh et al., 2012; Woods and Swearinginm, 1977). Saitoh et al. (2012) concluded that lodging occurred at vegetative stage, R1 or R6 reduced seed yield by approx. 10%, whereas if lodging occurred between R3 and R5, the seed yield was decreased by 25%–34%. In southwestern Japan, R3 to R5 occur in late August to mid-September (Table 1), and typhoons sometimes hit or come close to this region, causing severe lodging. Therefore, future cultivars should have lodging tolerance to prevent the yield loss caused by lodging even under narrow-row cultivation. The combination of cultivars with lodging tolerance and narrow-row cultivation will lead to future cost- and labor-saving soybean production systems in southwestern Japan.

96

N. Matsuo et al. / Agricultural Water Management 192 (2017) 85–97

5. Conclusion Although the use of a FOEAS enabled us to control the water table at the designed depth, the WTM tested in this study (FL and 30 cm plot) decreased the soybean yield by approx. 5.0% compared with the Control plot (drainage only), where the natural water table existed at depths of approx. 50–60 cm almost throughout the growing period. The high water table in the 30-cm plot or the temporary increase in the water table in the FL plot might have negatively affected the root or nodule function, rather than the root growth. Therefore, the natural water table should be measured and taken into account before WTM is conducted with a FOEAS. Furthermore, future studies will be needed to create the sub-irrigation schedules based on both crop growth and meteorological data (i.e., the ETc). Our findings indicate that newly developed soybean cultivars with short main stem lengths were suitable for narrow-row cultivation. The yields of Sachiyutaka A1 and Hatsunagaha in narrow-row cultivation were similar to or greater than those obtained by conventional cultivation (i.e., planting Fukuyutaka in 70 cm rows) with less lodging. The combination of cultivars with short main stem lengths and narrow-row cultivation will lead to future cost- and labor-saving soybean production systems in southwestern Japan. Acknowledgments We are grateful to Mr. Yukinari Kawahara, Mr. Teruyuki Miike, Mr. Akitoshi Honbu, Mr. Hiroyuki Itoh, Mr. Sadahiro Higashi, Ms Tamiko Shimogawa and Ms. Mari Nishida for their field management and data collection in this study. References Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop Evapotranspiration for Computing Crop Water Requirements. FAO Irrigation and Drainage Paper 56. FAO, Rome. Ayars, J.E., Christen, E.W., Soppe, R.W., Meyer, W.S., 2006. The resource potential of in-situ shallow ground water use in irrigated agriculture: a review. Irrig. Sci. 24, 147–160. Babajimopoulos, C., Panoras, A., Georgoussis, H., Arampatzis, G., Hatzigiannakis, E., Papamichail, D., 2007. Contribution to irrigation from shallow water table under field conditions. Agric. Water Manage. 92, 205–210. Board, J.E., Harville, B.G., Saxton, A.M., 1990. Branch dry weight in relation to yield increase in narrow-row soybean. Agron. J. 82, 540–544. Board, J.E., Kamal, M., Harville, B.G., 1992. Temporal importance of greater light interception to increased yield in narrow-row soybean. Agron. J. 84, 575–579. Bramley, H., Tyerman, S.D., 2010. Root water transport under waterlogged conditions and the roles of aquaporins. In: Mancuso, S., Shabala, S. (Eds.), Waterlogging Signalling and Tolerance in Plants. Springer, Berlin, pp. 151–180. Bullock, D., Khan, S., Rayburn, A., 1998. Soybean yield response to narrow rows in largely due to enhanced early growth. Crop Sci. 38, 1011–1016. Cooper, R.L., Fausey, N.R., Streeter, J.G., 1991. Yield potential of soybean grown under a subirrigation/drainage water management system. Agron. J. 83, 884–887. Cooper, R.L., 1997. Response of soybean cultivars to narrow rows and planting dates under weed-free conditions. Agron. J. 69, 89–92. Cox, W.J., Cherney, J.H., 2011. Growth and yield responses of soybean to row spacing and seeding rate. Agron. J. 103, 123–128. Egli, D.B., Guffy, R.D., Heitholt, J.J., 1987. Factors associated with reduced yields of delayed planting of soybean. J. Agron. Crop Sci. 159, 176–185. Elmi, A.A., Madramootoo, C., Hamel, C., 2000. Influence of water table and nitrogen management on residual soil NO3- and denitrification rate under corn production in sandy loam soil in Quebec. Agric. Ecosyst. Environ. 79, 187–197. Elzenga, J.T.M., van Veen, H., 2010. Waterlogging and plant nutrient uptake. In: Mancuso, S., Shabala, S. (Eds.), Waterlogging Signalling and Tolerance in Plants. Springer, Berlin, pp. 23–35. Fatichin, Zhen, S.H., Narasaki, K., Arima, S., 2013. Genotypic adaptation of soybean to late sowing in southwestern Japan. Plant Prod. Sci. 16, 123–130. Fehr, W.R., Caviness, C.E., Burmood, D.T., Pennington, J.S., 1971. Stage of development descriptions for soybean, Glycine max (L.) Merrill. Crop Sci. 11, 929–931. Fisher, M.J., Fausey, N.R., Subler, S.E., Brown, L.C., Bierman, P.M., 1999. Water table management, nitrogen dynamica, and yields of corn and soybean. Soil Sci. Soc. Am. J. 63, 1786–1795. Furuhata, M., Morita, H., Yamashita, H., 2008. Performance of dry matter and seed production under narrow-row-dense- planting culture of soybean cultivar,

Sachiyutaka, in south-western Japan. Jpn. J. Crop Sci. 77, 409–417 (In Japanese with English abstract). Garside, A.L., Lawn, R.J., Byth, D.E., 1992. Irrigation management of soybean (Glycine max (L.) Merrill) in a semi-arid tropical environment. III Response to saturated soil culture. Aust. J. Agric. Res. 43, 1019–1032. Hamada, Y., Shaku, I., Sawada, Y., Kojima, H., 2007. Factors causing submergence damage on soybean seedling emergence under no-tilled culture conditions. Jpn. J. Crop Sci. 76, 212–218 (in Japanese with English abstract). Korte, L.L., Williams, J.H., Specht, J.E., Sorensen, R.C., 1983a. Irrigation of soybean genotypes during reproductive ontogeny. I. Agronomic responses. Crop Sci. 23, 521–527. Korte, L.L., Specht, J.E., Williams, J.H., Sorensen, R.C., 1983b. Irrigation of soybean genotypes during reproductive ontogeny. Ii. Yield component responses. Crop Sci. 23, 528–533. MAFF, 2015. Crop Statistics. Ministry of Agriculture, Forestry and Fisheries, Japan (Accessed 16 10.103) (in Japanese) http://www.e-stat.go.jp/SG1/estat/List. do?lid=000001129531. MAFF, 2016. Statistics of Production Costs for Agricultural Products. Ministry of Agriculture, Forestry and Fisheries, Japan (Accessed 17.04.07) (in Japanese) http://www.maff.go.jp/j/tokei/kouhyou/noukei/seisanhi nousan/. Madramootoo, C.A., Broughtonm, S.R., Dodds, G.T., 1995. Watertable management strategies for soybean production on a sandy loam soil. Can. Agric. Eng. 37, 1–7. Maekawa, T., Shimamura, S., Shimada, S., 2011. Effects of short-term waterlogging on soybean nodule nitrogen fixation at different soil reductions and temperatures. Plant Prod. Sci. 14, 349–358. Mancuso, N., Caviness, C.E., 1991. Association of selected plant traits with lodging of four determinate soybean cultivars. Crop Sci. 31, 911–914. Matsunaga, R., Takahashi, M., Komatsu, K., 2003. Dense planting and no-intertillage cultivation using new soybean variety ‘Sachiyutaka’ with high lodging torelance. Rep. Kyushu Br. Crop Sci. Soc. Jpn. 69, 53–55 (In Japanese). Matsuo, N., Takahashi, M., Nakano, H., Fukami, K., Tsuchiya, S., Morita, S., Kitagawa, H., Nakano, K., Nakamoto, H., Tasaka, K., 2013. Growth and yield response of two soybean cultivars grown under controlled groundwater level in southwestern Japan. Plant Prod. Sci. 16, 84–94. Matsuo, N., Yamada, T., Hajika, M., Fukami, K., Tsuchiya, S., 2015. Planting date and row width effects on soybean production in southwestern Japan. Agron. J. 107, 415–424. Mejia, M.N., Madramootoo, C.A., Broughton, R.S., 2000. Influence of water table management on corn and soybean yields. Agric. Water Manage. 46, 73–89. Mueller, L., Behrendt, A., Schalitz, G., Schindler, U., 2005. Above ground biomass and water use efficiency of crops at shallow water tables in a temperate climate. Agric. Water Manage. 75, 117–136. Nakayama, N., Hashimoto, S., Shimada, S., Takahashi, M., Kim, Y.H., Oya, T., Arihara, J., 2004. The effect of flooding stress at the germination stage on the growth of soybean in relation to initial seed moisture content. Jpn. J. Crop Sci. 73, 323–329 (in Japanese with English abstract). Nathanson, K., Lawn, R.J., DeJabrun, P.L.M., Byth, D.E., 1984. Growth: nodulation and nitrogen accumulation by soybean in saturated soil culture. Field Crops Res. 8, 73–92. Nelson, K.A., Smoot, R.L., Meinhardt, C.G., 2011. Soybean response to drainage and subirrigation on a claypan soil in Northern Missouri. Agron. J. 103, 12-16-1222. Noor, R.B.M., Caviness, C.E., 1980. Influence of induced lodging on pod distribution and seed yield in soybeans. Agron. J. 72, 904–906. Purcell, L.C., Vories, E.D., Counce, P.A., King, C.A., 1997. Soybean growth and yield response to saturated soil culture in a temperate environment. Field Crops Res. 49, 205–213. Robinson, S.L., Wilcox, J.R., 1998. Comparison of determinate and indeterminate soybean near-isolines and their response to row spacing and planting date. Crop Sci 38, 1554–1557. Saitoh, K., Hirata, K., Kashiwagi, Y., 2007. Effect of row-width and planting density on podding and yield performance of early soybean cultivar ‘Enrei’ with reference to raceme order. Jpn. J. Crop Sci. 76, 204–211 (In Japanese with English abstract). Saitoh, K., Nishimura, K., Kitahara, T., 2012. Effect of lodging on seed yield of field-grown soybean −artificial lodging and lodging preventing treatments. Jpn. J. Crop Sci. 81, 27–32 (In Japanese with English abstract). Shimada, S., Kokubun, M., Matsui, S., 1995. Effects of Effect of water table on physiological traits and yield of soybean: I. Effect of water table and rainfall on leaf chlorophyll content, root growth and yield. Jpn. J. Crop Sci. 64, 294–303. Shimada, S., Kokubun, M., Matsui, S., 1997. Effects of Effect of water table on physiological traits and yield of soybean: II: Effect of water table and rainfall on leaf water potential and photosynthesis. Jpn. J. Crop Sci. 66, 108–117. Shimada, S., Hamaguchi, H., Kim, Y.H., Matsuura, K., Kato, M., Kokuryu, T., Tazawa, J., Fujimori, S., 2012. Effects of water table control by farm-oriented enhancing aquatic system on photosynthesis, nodule nitrogen fixation, and yield of soybeans. Plant Prod. Sci. 15, 132–143. Stanley, C.D., Kaspar, T.C., Taylor, H.M., 1980. Soybean top and root response to temporary water tables imposed at three different stages of growth. Agron. J. 72, 341–346. Troedson, R.J., Lawn, R.J., Byth, D.E., Wilson, G.L., 1989a. Response of field-grown soybean to saturated soil culture. 1. Patterns of biomass and nitrogen accumulation. Field Crops Res. 21, 171–187. Troedson, R.J., Lawn, R.J., Byth, D.E., Wilson, G.L., 1989b. Response of field-grown soybean to saturated soil culture. 2. Effect of treatments to alter photosynthesis and leaf nitrogen supply. Field Crops Res. 21, 189–201.

N. Matsuo et al. / Agricultural Water Management 192 (2017) 85–97 Uchikawa, O., Fukushima, Y., Matsue, Y., 2003. Statistical analysis of soybean yield and meteorological conditions in the northern Kyushu. Jpn. J. Crop Sci. 72, 203–209. Wakasugi, K., Fujimori, S., 2009. Subsurface water level control system FOEAS that promotes the full use of paddy fields. J. Jps. Soc. Irrig. Drain. Rural Eng. 77, 705–708 (in Japanese with English abstract).

97

Wakasugi, K., Haraguchi, N., Zukemura, C., 2016. Quantifying subsoil hydraulic conductivity fir stable subsurface irrigation in FOEAS system. Tech. Rep. Natl. Inst. Rural. Eng. Japan 218, 115–123 (in Japanese with English abstract). Williamson, R.E., Kriz, G.J., 1970. Response of agricultural crops to flooding, depth-of water table and soil gaseous composition. Trans. ASAE 13, 216–220. Woods, S.J., Swearinginm, M.L., 1977. Influence of simulated early lodging upon soybean seed yield and its components. Agron. J. 69, 239–242.