2022 Volume 91 Issue 3 Pages 356-365
Vegetable soybean, cultivated in the Nakakawachi region in Osaka Prefecture, is highly regarded as a special vegetable soybean in Japan. Understanding yield variability, and environmental factors that influence yield, are important for stably producing high quality vegetable soybeans. The objective of this study was to determine the respective effects of soil properties, including physicochemical and biological parameters, on agronomic traits of vegetable soybean cultivated under greenhouse conditions in the Nakakawachi region targeting 18 fields in the 2017 to 2018 season. Substantial differences in vegetable soybean yield were observed among fields, and the yield ranged from 299 to 968 kg/10 a. Regarding agronomic traits, pod production was significantly correlated with total node number, the number of branches and pods, while cultivation duration was also positively correlated with the market base pod yield. Although the nodule dry weight (DW) had no significant correlations with any agronomic traits among farmer’s fields, correlation analysis within each field frequently showed nodule DW had a positive correlation with vegetable soybean production. Regarding soil physicochemical factors, the factors influencing yield included total carbon, total nitrogen (N), and available N. Bacterial diversity in soil was evaluated at harvest of vegetable soybean in 2017. Principal coordinates analysis with 16S rRNA gene amplicon sequencing data showed fields could be divided into three groups. We assessed the correlation between vegetable soybean yield and the relative abundance of each bacterial family, resulting in significant correlations of some bacterial families with pod production. This study provides information on soil chemical factors that led to stable production of vegetable soybean cultivated in the Nakakawachi region.
Vegetable soybean is an important crop in Asian countries and, recently it has become increasingly popular worldwide, due to its high protein content, essential amino acids, and other nutritional compounds such as isoflavones (Zhang and Kyei-Boahen, 2007; Jiang et al., 2018). In Japan, vegetable soybeans are frequently consumed as a snack with alcoholic beverages. The production area of vegetable soybean in Japan is 13,000 ha, and approximately 66,100 t are produced per year (MAFF, 2020). Imports of frozen vegetable soybean amounted to 34,000 t in 1989, predominantly from Taiwan. Because seed freshness is important for taste, vegetable soybeans are domestically produced, close to processing and consumption areas. In some areas, however, attempts have been made to produce special vegetable soybeans, such as Dadacha-mame (Yamagata Prefecture), Kurosaki-chamame (Niigata Prefecture) and Murasaki-zukin (Kyoto Prefecture). To improve the yield and seed quality of vegetable soybean in Japan, cultural practices such as sowing time, cropping pattern, fertilization and harvesting time have been studied by many researchers (Takahashi and Hoshino, 2003; Nishioka and Okumura, 2008; Katayama et al., 2009; Yoshida et al., 2011; Sugimoto et al., 2018).
The vegetable soybean production area of Osaka Prefecture is comparatively small. Even so, this area produces a higher yield of vegetable soybean (8.6 t/ha) compared to other areas in Japan (MAFF, 2020). The Nakakawachi region, which is one of the most predominant production areas of vegetable soybean in Osaka Prefecture, is currently encouraging urban agriculture growers to produce special vegetable soybean, “Yao-edamame”. In this area, vegetable soybeans are frequently cultivated during the cool season (February) in greenhouses. Due to the high population and limited arable land in the Nakakawachi region, agricultural production systems have been intensified to produce various vegetables, and vegetable soybean production must be incorporated into these cropping systems. Therefore, residual soil nutrients from previous crops can influence vegetable soybean production.
To improve seed quality, Takai et al. (2010) evaluated varietal differences in free amino acid and sugar concentrations in vegetable soybean seeds. However, information on cropping management is limited, and research on vegetable soybean, particularly comparisons of agronomic traits among fields, has received little attention. Further, understanding the variabilities of farmer-based agronomic traits, including yield between fields, is important to achieve high yield and satisfactory quality.
Differences in management by individual farmers result in different levels of soil fertility and provide valuable information on the factors influencing yield variability. In the present study, agronomic traits of vegetable soybean grown at various sites in Nakakawachi region, Osaka, were assessed, and their relationships with soil physicochemical/biological properties were examined to determine soil environmental factors influencing vegetable soybean yield.
Field surveys were conducted at farmer’s fields, located in the Nakakawachi region, mainly in Yao city, Osaka, Japan. A total of 18 vegetable soybean fields were selected, including eight greenhouses (Akegawahigashi [Ak1], Fukuei-cho [Fu], Kashimura-cho [Ka1], Kyokouji [Ky1], Miyakoduka [Mi1], Okubo [Ok], Yaogikita [Ya1], and Yaogihigashi [Ya2]) in 2017 and 10 greenhouses (Akegawa-higashi-kita, [Ak2], Akegawa-higashi-naka [Ak3], Akegawa-higashi-minami [Ak4], Kyokouji [Ky2], Miyakoduka-higashi [Mi2], Miyakoduka-nishi [Mi3], Miyakoduka-higashi [Mi4], Miyakoduka-higashi [Mi5], Onjikita-machi-nishi [On1], and Onjikita-machi-higashi [On2]) in 2018. Management of cultivation and fertilization were different depending on the site and individual farmers. Detailed information on the surveyed fields is shown in Table 1. Briefly, all surveyed fields had the same number of rows (2), and planting density ranged from 8,000 to 11,765 plants/10 a. In most cases, plants were transplanted after being raised in a greenhouse. In the present survey, organic fertilizers were applied in all fields. Cottonseed meal was applied as a base fertilizer in two fields (Ak1 and Mi1) in 2017, and seven fields (On1, On2, Ak2, Ak3, Ak4, Mi4, and Mi5) in 2018. For topdressing, different organic fertilizers were applied in all fields, depending on the farmers using them. The timing of topdressing was different among the various fields, but topdressing was often conducted around flowering time. Total inputs of N, P2O5, and K2O (kg/10 a) ranged from 1.6–16.8 N, 1.4–19.6 P2O5, and 0.9–14 K2O, respectively. The earliest transplanting took place on March 2 (On1 and On2), and the latest on April 17 (Ky1).
Crop management of vegetable soybean in each field.
The soybean variety, Taisetsu-midori (Otaniseed Corp., Japan), was cultivated in all fields. In each field, the dates of flowering and harvest were recorded. Ten complete plants, including roots, were sampled in each field on the day before or after harvest by farmers, except for the Mi3 field, where nine plants were sampled. To assess agronomic traits, the numbers of total nodes and branches, total pod number, market-base pod number, total pod weight, and market-base pod weight were investigated at harvest. Pod weight was measured immediately after counting the pod number. Pods containing more than two seeds were considered market-base pods. Average pod weight was determined by dividing total pod weight by total pod number. Market-base pod yield (kg/10 a) was determined by multiplying the market-base pod yield per plant and planting density. The roots of each plant were washed and carefully rinsed with water. Roots were separated from nodules, and the respective dry weight (DW) per plant was determined after drying at 70°C for two days.
Soil physical and chemical propertiesBefore cultivation, soil samples, at an approximate distance of 0–10 cm from the plant base, were randomly collected at five points using a small trowel in each field. Then, soil samples were combined to provide a composite sample for each field. Soil samples were thoroughly mixed and air-dried before nutrient analyses. Available nitrogen (N) concentrations were determined using previously published methods (Tanaka et al., 1998; Sakaguchi et al., 2010). Available phosphate (P) was extracted using the Truog–P method, and concentrations were measured using a molybdenum blue colorimetric method. To assess concentrations of soil exchangeable cations, potassium (K), calcium (Ca), and magnesium (Mg) were extracted using ammonium acetates (pH 7.0) and measured using a polarized Zeeman atomic absorption spectrophotometer (ZA3000; Hitachi High-Tech Science Corp., Tokyo, Japan). Total carbon (C) and total N were determined by the dry combustion method (CN coder MT-700; Yanagimoto Co., Ltd., Japan). Nitrate-N (NO3−-N) and NH4+-N were extracted by 2M KCl solution from the soils. Extracted NO3−-N was determined by a colorimetric method after NO3−-N was reduced to NO2-N using a flow injection analyzer (Auto Analyzer QuAAtro 39; BL TEC K. K., Tokyo, Japan). Extracted NH4+-N was also determined by the modified indophenol blue method using a flow injection analyzer (Auto Analyzer QuAAtro 39; BL TEC).
Soil penetration resistance at the flowering growth stage was assessed using a digital cone penetrometer (model SR-2; Daiki Rika Kogyo Co., Ltd., Kounosu, Japan). Average soil depth at soil penetration resistance, up to 1,500 kPa, which restricts root growth of most plants (Coelho et al., 2000), was determined at five randomly selected sites.
Soil biological propertiesSoil bacterial communities were identified using 16S rRNA-based high-throughput sequencing of DNA extracted from fresh soil collected at harvest. Soils were preserved at −20°C until DNA extraction. Because we could not get −20°C preserved soils in 2018, NGS analysis was conducted using only the soils collected in 2017. Four grams of soil were used for genomic soil DNA extraction with commercially available kits (Kyokuto Pharmaceutical Industrial Co., Ltd., Tokyo, Japan). The extracted soil genomic DNA was used as a template to amplify 16S rRNA genes. The V3–V4 regions of the 16S rRNA gene were amplified using 2 × KAPA HotStart Ready mix (KAPA Biosystems Inc., Wilmington, MA, USA) with forward (5'-TGCCAGCMGCCGCGGTAA-3') and reverse (5'-GGACTACHVGGGTWTCTAAT-3') primers (Klindworth et al., 2013) using a PCR Thermal Cycler Dice (Takara Bio Inc., Shiga, Japan). PCR reaction mixtures consisted of 12.5 μL of KAPA HotStart Ready mix, 5 μL of 1 μM primers and 2.5 μL of genomic DNA, for a final volume of 25 μL. The reaction conditions were as follows: initial denaturation for 3 min at 95°C, followed by 25 cycles of 30 s at 95°C, 30 s at 55°C, and 30 s at 72°C, with a final extension of 5 min at 72°C. For Illumina library preparation, the barcodes were added using 2 × KAPA HiFi HotStart Ready mix (KAPA Biosystems), and Nextera XT v2 Index Primers 1 and2 (Illumina, Inc., San Diego, CA, USA) with eight cycles. Before sequencing, the amplicons were purified with AMPure XP beads (Beckman Coulter Inc., Brea, CA, USA). Amplicons were sequenced for 300 bp at both ends on the Illumina MiSeq platform (Illumina).
Data analyses and statisticsAgronomic trait data was analyzed using one-way analysis of variance (ANOVA). Correlations were determined using Pearson’s correlation coefficient analysis. These statistical analyses were conducted using SPSS software (IBM SPSS Statistics for Windows, Version 26.0; IBM Corp., Armonk, NY, USA). In soil bacterial genome sequence data analysis, raw sequence reads were processed using the QIIME2 pipeline. Forward and reverse reads were joined, denoised, and checked for chimeras using the software package deblur, and low-quality reads were removed. For microbial diversity analysis, QIIME2 software was also used to assess the relative abundance of operational taxonomic units (OTUs). For taxonomy assignment, SILVA was used as a reference database. OTU clusters were defined by a 97% identity threshold. The relative abundance of each OTU was normalized by dividing the readings of individual OTUs by the total readings in a sample. Data from a total of 188 OTUs was analyzed by principal coordinates analysis (PCoA) to examine differences in soil bacterial communities between fields with the R package. Only the bacterial families that had significant correlations with market-base yield were shown in this study.
Average durations from transplanting or sowing to flowering, and from flowering to harvest were 35 days and 41 days, respectively (Table 2). Although the same soybean variety, Taisetsu midori, was cultivated in all fields, relatively large differences in cultivation durations were observed. Durations from transplanting, or direct sowing, to harvest ranged from 62 to 90 days.
Duration from transplanting or sowing to flowering, from flowering to harvest, and from transplanting or sowing to harvest in each field.
Significant differences in all agronomic traits and yield parameters, except for market pod yield, were identified by ANOVA among fields (Table 3). Average numbers of total nodes and branches were 22.3 and 4.2 per plant, respectively. Ky2 and Mi5 fields showed lower values for numbers of total nodes and branches. The average DWs of roots and nodules were 2.2 g and 1.1 g, respectively. The highest nodule DW was observed in On1 and On2 fields, and the minimum was in the Ka1 field. Total pod number ranged from 17.3 to 48.1 per plant. The average number of market-base pods (24.2 pods/plant) decreased by 25% of the average total pod number (32.2 pods/plant), particularly in the Fu, Mi2, and Ak4 fields, which showed low values for the market-base pods ratio. Average pod weight ranged from 2.4 to 3.4 g. Average weights of total pods and market-base pods were 91.1 g and 56.3 g, respectively. The average market-base pod yield was 595 kg/10 a, and ranged from 299 to 968 kg/10 a.
Agronomic traits of vegetable soybeans cultivated in different greenhouses in 2017 and 2018.
Positive correlations were observed among most agronomic traits, total node number, branch number, total pod number, market-base pod number, total pod weight, market-base pod weight and market-base pod yield (Table 4). Average pod weight showed negative correlations with branch number, total pod number, and market-base pod number. Root DW showed a positive correlation only with total node number. However, nodule DW did not show significant correlations with any agronomic traits. Respective correlation coefficients between market-base pod weight and nodule DW within each field are shown in Figure 1. Although no significant correlations between nodule DW and agronomic traits were observed among fields, all fields showed positive correlations with respect to nodule DW. Significant positive correlations were observed in seven fields (Ak1, Mi1, Ok, Ky2, Mi5, On1, and On2).
Correlation coefficients among agronomic traits of green vegetable soybeans cultivated in 2017 and 2018.
Correlation coefficients between market-base pod weight and nodule dry weights in green vegetable soybeans cultivated in greenhouses in 2017 and 2018. Asterisks (*) indicate significance at P < 0.05.
Duration from transplanting, or direct sowing, to flowering was positively correlated with total node number, while duration from flowering to harvest was positively correlated with market-base pod yield (Fig. 2). Duration from transplanting, or direct sowing, to harvest was positively correlated with both total node number and market-base yield.
Relationships among duration from transplanting or sowing to flowering (A, D), duration from flowering to harvest (B, E), and duration from transplanting or sowing to harvest (C, F) to total node number and market-base yield of vegetable soybean. Asterisks (**, *) indicate significance at P < 0.01 and P < 0.05, respectively.
Soil chemical properties of each field are shown in Table 5. The average pH was 6.76, ranging from 5.61 to 7.32, while the average total C and total N contents were 1.16% and 0.12%, respectively. Ky1 and Ky2 fields showed lower values for total C and total N. A very large variation was observed in NO3-N content, which ranged from 0.73 to 48.67 mg N/100 g soil. NH4-N and available N contents ranged from 2.18 to 11.44 mg N/100 g soil, and 37.64 to 372.8 mg N/100 g soil, respectively. Considerable P was accumulated in most fields, with the average available P content being 271.71 mg P2O5/100 g soil. However, Ky1 and Ky2 fields showed lower values of available P content. The average K content was 11.17 mg K2O/100 g soil, ranging from 3.74 to 29.71 mg K2O/100 g soil. The average Ca and Mg contents were 275.17 mg CaO/100 g soil, and 33.48 mg MgO/100 g soil, respectively. In contrast to available P content, Ky1 and Ky2 showed higher values for CaO and MgO compared to other fields. The average soil depth at up to 1,500 kPa penetration resistance was 25.7 cm. The lowest value was 17.8 cm, which was observed in the Ok field, and the highest value was 47.1 cm, observed in the Mi4 field.
Soil physico-chemical properties of each field.
Correlation analysis showed that total C and total N content were positively correlated with branch number, total pod weight, and market-base pod yield (Table 6). NH4-N showed positive correlations with many agronomic traits such as branch number, total node number, total pod number, total pod weight and root DW. Available N also showed positive correlations with branch number, total node number, market-base pod weight, and market-base pod yield. However, negative correlations in nodule DW were observed for NO3-N and available N. Partial correlation analysis showed that duration from transplanting, or sowing, to harvest was positively correlated with available N, total N and total C content in soil (Fig. 3).
Correlation coefficients of soil properties and agronomic traits of vegetable soybean.
Relationship between cultivation duration and (A) total C, (B) total N and (C) available N in soils of vegetable soybean fields. Asterisks (**, *) indicate significance at P < 0.01 and P < 0.05, respectively. Cultivation duration: duration from transplanting or sowing to harvest.
In total, 188 bacteria families were used for statistical analyses. The PCoA distinguished fields as shown in Figure 4. Fields were roughly divided into three groups: the Fu Field was positioned on the right side, fields Ok, Ak1, Ky1, Ya1, and Mi1 were in approximately the same position, and Ya2 and Ka1 were positioned on the left side. Significant correlations between market-base pod yield and some bacterial families were observed (Fig. 5). Bacterial families such as Dehalobacteriaceae, Lachnospiraceae, Paraprevotellaceae, Peptococcaceae, and Pseudomonadaceae showed positive correlations. In contrast, Microbacteriaceae showed negative correlations.
Results of principal coordinates analysis using soil bacterial genome sequence data from greenhouse soils in the 2017 season.
Relationship between market-base yield and relative abundance of bacteria identified from greenhouse soils in the 2017 season. Asterisks (**, *) indicate significance at P < 0.01 and P < 0.05, respectively.
The Nakakawachi region is one of the most important areas for vegetable soybean production in Osaka Prefecture, but variations in vegetable soybean yield and soil properties have not been reported. The present study compared vegetable soybean yield and soil properties of 18 fields in the Nakakawachi region and examined the soil environmental factors influencing the vegetable soybean yield. We found considerable differences in the agronomic traits of vegetable soybeans among the surveyed fields.
Soybean yield consists of several components including planting density, pod number, seeds per pod and seed size. In this study, a high planting density did not always produce a higher yield. However, the Ka1 field, which had the highest planting density, showed the highest yield (Table 3). The pod number is strongly correlated with total node number and branch number (Machado et al., 2017). Our data also showed positive correlations between pod number and total node number or branch number, and these agronomic traits positively correlated with market-base pod yield (Table 4). Meanwhile the average pod weight did not show a positive correlation with market-base yield. This is due to the trade-off between pod number and seed weight.
A longer duration of the vegetative and reproductive phases produces a higher soybean yield (Chen and Wiatrak, 2010), and delaying sowing resulted in a shortening of the length of vegetative development (Calvino et al., 2003). Our survey also showed that the duration from transplanting, or sowing, to flowering was positively correlated with the total node number, which is one of the parameters of vegetative growth, and the duration from flowering to harvest was positively correlated with market-base pod yield (Fig. 2). However, the difference in cultivation duration was not explained solely by only the transplanting or sowing date in the present study, that is, the cultivation duration was different even for fields with around the same time of transplanting, such as the Mi1 and Ak4 fields (Tables 1 and 2). Therefore, other factors must also influence the cultivation duration.
Correlation analysis between agronomic traits and soil properties showed that the factors influencing yield were available N, total N and total C. Soil organic matter, including total C, plays an important role in carbon storage, aggregate formation, plant nutrient supply, and nutrient retention (Aoyama, 2015). Soybean plants require substantial amounts of N for growth and seed production. Thus, available N and total N content are important to determine vegetable soybean yield. Cultivation duration from transplanting, or sowing, to harvest was also positively correlated with available N, total N, and total C (Fig. 3). Therefore, these soil factors likely influence the cultivation duration and yield of vegetable soybean. In the present study, the soils before vegetable soybean cultivation were used for correlation analysis considering the residual soil nutrients from previous crop production. Correlation analysis using soils collected during cultivation, or at harvest, may produce different results in terms of vegetable soybean agronomic traits. However, we confirmed that there were strong positive correlations in most soil nutrients between soils collected before and after vegetable soybean cultivation (data not shown). Therefore, we believe that correlation analysis even using soils collected before cultivation can provide useful information on vegetable soybean production.
N2 fixation by nodules contributes to vegetable soybean N acquisition. Various agronomic traits such as shoot (plant) DW, total seed number, total seed weight, leaf chlorophyll content, and shoot total N were found to be positively correlated with nodule DW in soybean (Mirza et al., 1990; Hungria and Bohrer, 2000; Hamawaki and Kantartzi, 2018). High N levels inhibit nodule formation (Gan et al., 2002; Xia et al., 2017). In the present study, a negative correlation between nodule DW and available N was observed (Table 6). Judging from our results, nodule DW is not the crucial factor influencing soybean yield. However, correlation analysis within the fields suggested different mechanisms; numerous fields showed positive correlations with market-base pod weight and nodule DW (Fig. 1). No identification of native rhizobia was conducted, and the effects of rhizobia were evaluated using only nodule DW. Therefore, further studies to identify native rhizobia, and to evaluate N2 fixation under minimized effects of environmental factors, are required to elucidate the contribution of rhizobia to vegetative soybean production.
PCoA showed that fields could be roughly divided into three groups. The abundance of some bacterial families showed positive correlations with market-base pod yield (Fig. 5). Bacterial community structures associated with soybean cultivation have been shown to be affected by geographical zones (Zhang et al., 2018) and soil management regimes including organic and conventional agriculture (Longley et al., 2020). However, little information about microorganisms associated with crop productivity using high-throughput sequencing data is available. In the present study, we identified specific bacterial families which appeared to significantly affect market-base pod yield. Pseudomonadaceae is known to promote plant growth and exert biocontrol activity against various plant pathogens (Compant et al., 2005). However, there is little information about other bacterial families, such as Dehalobacteriaceae, Paraprevotellaceae, Lachnospiraceae, Peptococcaceae, and Microbacteriaceae, or Paraprevotellaceae on plant growth. Dehalobacteriaceae is reported to be a dechlorinator of pentachlorophenol (PCP), one of the most recalcitrant pollutants (Zhu et al., 2018). Paraprevotellaceae is known as a rumen bacterium in cattle (Abbas et al., 2020). Lachnospiraceae has abundant, diverse carbohydrate active enzymes and is superior at producing short chain fatty acids (such as acetic and butyric acids) from organic matter (Biddle et al., 2013). Peptococcaceae are organic matter decomposers, plausible candidates for decomposing protocatechuate derived from lignin (Kato et al., 2015). Microbacteriaceae are ligninolytic bacteria, degrading lignin to soluble phenolic compounds (Taylor et al., 2012). Most of these bacterial families are relevant to decomposition of soil organic matter, but further studies are needed to reveal their effects on vegetable soybean growth.
ConclusionsThe present study showed differences in agronomic traits of vegetable soybeans cultivated in the Nakakawachi region, which is one of the most productive vegetable soybean cultivation areas in Japan. Even though the number of surveyed fields was limited, this is the first study on the relationships among vegetable soybean yield and physicochemical/biological soil properties targeting farmers’ fields. Some soil environmental factors were selected as influencers on market-base yields of vegetable soybean. Our results showed that low levels of total C, total N, and available N in soil likely reduced the vegetative or reproductive growth period of vegetable soybean, resulting in a lower market-base yield. In the present study, all farmers applied organic fertilizer to the soil, but the qualities and/or amounts of applied fertilizer were different among farmers. Certainly, fertilization is frequently determined empirically, and soil nutrient content may be affected by residuals of preceding crops, but this study showed that keeping higher levels of total C and total N content in soil is essential to improve the market-base yield of vegetable soybean.
We would like to thank the vegetable soybean farmers in Nakakawachi region for their cooperation in this research.
