N₂ fixation of nodules and N absorption by soybean roots associated with ridge tillage on poorly drained upland fields converted from rice paddy fields

Abstract

Sowing on elevated ridges is effective in reducing wet injury of soybean plants cultivated in upland fields converted from rice paddy fields. Therefore, we investigated the effect of ridge tillage (RT) on soybean N accumulation properties. We compared the amounts of plant N associated with N₂ fixation of nodules and from soil and fertilizer in the RT treatment with amounts in conventional tillage (CT) in two fields in 2002–2003. Both fields were upland fields converted from rice paddy fields (Typic Hydraquents). The main difference between the fields was the presence of a field underdrain. The amounts of Rb and K accumulated in the shoots were also determined to estimate soybean root distribution. The grain yields with RT increased in both fields from 106% to 129% compared with CT. Increased pod number and seed weight were the major factors responsible for the yield increase. anova indicated that RT significantly increased the activities of both N₂ fixation of nodules and N absorption by roots until R1 (flowering stage). The ratio of Rb and K accumulated in the shoots indicated that with RT, the root distribution was more abundant in the superficial layers compared with CT. Thus, RT reduced wet injury during the rainy season that overlapped the flowering stage. Nitrogen accumulation from N₂ fixation until the R7 stage with RT was significantly higher than that with CT. We concluded that RT was effective in increasing N₂ fixation of nodules in poorly drained upland fields converted from rice paddy fields.

Key words:

• soybean
• tillage
• wet injury
• nitrogen uptake
• upland field converted from rice paddy field

INTRODUCTION

Wet injury is a major constraint on soybean, Glycine max (L.) Merr., cultivation in upland fields converted from rice paddy fields (UFCPs) in Japan. Soil structure is poorly developed in many of these fields. In addition, these fields display a low permeable layer under the plow layer as a result of puddling for the preceding rice cultivation. In most cases, wet injury occurs during the initial stages of soybean growth, which overlaps the rainy season (from June to July) in Japan. According to Sugimoto (1994), the earlier the stage at which soybean experiences wet injury, the greater the decrease in yield. Declining soybean yield mainly results from a decrease in pod number by N deficiency. Takahashi et al. (2005) confirmed Sugimoto's (1994) conclusion based on a survey of 33 farmers’ UFCPs that had experienced wet injury. They concluded that a decrease in the pod number, which resulted from poor drainage, was the major reason for yield decrease.

Sowing soybean seeds on elevated ridges increased the yield in poorly drained UFCPs (Hosokawa 2004, Hosokawa et al. 2005). In contrast to ridge tillage (RT) in the Corn Belt of the USA (Hatfield 1998), the major objective of raised ridges in Hosokawa's case was to alleviate wet injury and not soil erosion. According to the studies conducted by Hosokawa, RT, whereby the seedbed is raised to 15 cm, increased the yield to 110% or 120% of that in conventional cultivation (i.e. tillage without raised seedbeds). Hosokawa concluded that the apparent decrease in the water table by RT increased the oxygen concentration around the rhizosphere, which resulted in increases in branch and pod numbers per

Table 1 Soil classification, texture, pH, soil N, amount of mineralized N and height of water table in the two fields

	Site A	Site B
Classification of cultivated soils in Japan^†	Mottled Gley Lowland soils	Strong Gley Lowland soils
Soil Taxonomy^‡	Typic Hydraquents	Typic Hydraquents
Clay content (%)	27	45
Silt content (%)	30	41
Sand content (%)	44	19
pH (H2O)	5.5	5.0
Soil N (%)	0.15	0.28
Mineralized N (mg N kg^−1)	50.3	69.3
Height of water table (m)^§	0.7	0.4
^† Cultivated Soil Classification Committee (1995);
^‡ Soil Survey Staff (1994).
^§Measurement was conducted in October 2003.

plant. However, the characteristics of both N₂ fixation of nodules and N absorption by roots associated with RT have not been fully elucidated. These N accumulation characteristics should be clarified because soybean yield is affected by soil N fertility when wet injury is not severe (Takahashi et al. 2005). After clarification, the RT method could be applied to various fields.

Nitrogen accumulation properties of soybean are more complex than those of other crops because half or more of the accumulated N in soybean is derived from N₂ fixation of nodules. Several methods have been developed to estimate the ratio of N derived from N₂ fixation (Herridge and Peoples 1990), including the N accumulation difference between nodulating soybeans and non-nodulating isoline, ¹⁵N dilution method, acetylene reduction method and the relative ureide method. Among these methods, the relative ureide method offers numerous advantages for the study of wet injury because: (1) it is reliable for field experiments, (2) it is applicable for estimation through a wide range of growth stages, (3) it does not require special apparatus for estimation (Herridge et al. 1990; Takahashi et al. 1992).

The objective of the present study was to evaluate the effect of RT on both N₂ fixation of nodules and N absorption by roots using the relative ureide method.

MATERIALS AND METHODS

Fields and soils

To evaluate the effects of RT on N₂ fixation of nodules and N absorption by roots of soybean, we used two UFCPs with different soil moisture conditions, sites A and B, located in Joetsu City, Niigata prefecture, Japan. Both sites had been used as paddy fields prior to 2001 and were converted to upland fields in the spring of 2002. There were no climatic differences between the two fields because the fields were separated by 100 m. The investigations were conducted from 2002 to 2003. Table 1 shows the soil classification, pH, soil texture and soil N

Figure 1  Difference in the volumetric water content between two cultivation treatments (ridge tillage [RT] and conventional tillage [CT]) at 5 cm depth from the soil surface at sites A and B in 2003.

properties, and the height of the water table in both fields. Figure 1 shows the time-course of the volumetric water content of the soil at both sites. Severe wet injury was observed at site B with conventional tillage (CT), where an underdrain was not installed and the soil exhibited higher clay content. The color of the leaves had turned yellow and leaves from the lower nodes had fallen during the rainy season at site B. These were typical symptoms of wet injury, as described by Sugimoto (1994). Site A had an underdrain and was drier than site B. Wet injury at site A during the rainy season was not as severe as that at site B.

Cultivation methods

In 2002 and 2003, the soybean cultivar “Enrei” was sown on 2 June and 3 June and harvested on 1 October and 2 October, respectively. Tillage, making of ridges, sowing and basal application of fertilizers were conducted in one process using a machine developed by Hosokawa (2004). The spacing of each row was 75 cm long and two seeds were sown at 18 cm intervals. Height of the seed locations was 15 cm above the inter-row field surface. In CT the process was identical to that for RT, except for the lack of ridges. Hence, the only difference between the treatments was the vertical location of the seeds. Basal application of fertilizers was conducted by side-dressing at a distance of 5 cm from the seeds. The respective amounts of nitrogen, phosphorus and potassium used for basal application were 1.6, 6.0 (as P₂O₅) and 8.0 (as K₂O) g m⁻². The nitrogen fertilizer applied consisted of ammonium sulfate. We did not use organic fertilizers or apply any topdressing.

Analytical methods for soils

Soil samples to measure the amount of mineralized N were taken in March 2002 from the plow layer (approximately 13 cm depth). The amount of mineralized N was determined by incubation at 30°C under field moisture conditions (Methods of Soil Environment Analysis Committee 1997) for 4 weeks. The amount of mineralized N was calculated from the sum of nitrate, nitrite and ammonium N. Soils used for the N measurement displayed fresh moisture conditions without drying.

We used the method referred to as “ratio of similar elements” (Takahashi 1996) to evaluate the root system of soybean. This method is based on the fact that the absorption behavior of rubidium is similar to potassium for plants (Takahashi 1996) and on the assumption that K is located in more superficial layers than Rb because of the application of fertilizers. The tendency of plant roots to be distributed in deeper layers was evaluated by using the Rb/K ratio in plants. We verified the assumption by analyzing the Rb/K ratio of exchangeable cations in the soil profiles. Soil samples from 0–10, 10–20 and 20–30 cm depths were taken in October 2002 for estimation of the Rb and K distribution toward the vertical soil profiles. Exchangeable Rb and K of soils were extracted two times with a 1 mol L⁻¹ pH 7 ammonium acetate solution for 1 h with shaking. The soil to solution ratio was 1/50 g mL⁻¹.

In 2003, the field moisture content of each plot was measured by inserting time domain reflectometry (TDR) probes at a 5 cm depth from the surface.

Sampling procedures and analytical methods for soybeans

Soybean plants in each plot were sampled (0.75 m²) at the V4 (only in 2003), V5, R1, R3, R5 and R7 stages to measure dry weight and amounts of accumulated N, K and Rb in shoots. The terminology of the growth stages of soybean followed that proposed by Fehr et al. (1971). At each sampling time, four plants were selected and the xylem sap was collected from a stem cut close to the ground using cotton for 2 h. The sampled xylem sap was frozen immediately until an analysis of the composition of the xylem sap solution was carried out. Sampling from the V4 to R7 stages was duplicated. Sampled plants were dried at 80°C immediately, the dry weight was measured and the plants were ground for N, Rb and K measurements. Total plant N content at each stage was determined using the combustion method (Rapid-N III, Elementar Inc., Hanau, Germany). Rb and K in the shoots of the soybean plants were extracted using 50 mL of 0.1 mol L⁻¹ HCl for 0.5 h with shaking; the sample to solution ratio was 1/50 g mL⁻¹ (Takahashi et al. 1991).

Estimation of the N₂ fixation ratio was carried out according to the method of Takahashi et al. (1992). In brief, nitrate, amide and ureide concentrations in the xylem sap solution were determined by colorimetry using the methods of Cataldo (1974), Herridge (1984) and Young and Conway (1942), respectively. The N of the nitrate and amide forms was considered to correspond to the N absorbed from soil or fertilizer by the roots. Ureide form N was considered to be derived from N₂ fixation of nodules. The amount of N derived from N₂ fixation of the nodules, soil or fertilizer was estimated as follows:

In these equations, NS, NF and TN, are the amount of N derived from soil or fertilizer, N derived from N₂ fixation and total N, respectively. FF is the ratio of the amount of N derived from N₂ fixation per total N in the xylem sap solution and i indicates the sampling step. It must be noted that the above estimation is likely to be ambiguous if TN in the plant is lost by, for example, falling leaves. Hereafter, we use the terms plant N to designate the total amount of accumulated N in a plant, fixed N as the amount of N derived by N₂ fixation of nodules and absorbed N as the amount of N derived from soil and fertilizer by root absorption.

At the harvest stage, 1.5 m² of soybean was sampled and used to measure grain yield, dry weight, pod number, seed number and seed weight. Sampling at harvest was conducted in four duplications in 2002 and three duplications in 2003. Grain yield was expressed as 15% of the seed moisture content.

RESULTS AND DISCUSSION

Growth and yield of soybean

The average temperature, solar radiation and precipitation for 2002 and 2003 are shown in Fig. 2. The weather conditions were different in 2002 and 2003. The rainy season lasted from 11 June and 12 June and ended on 23 July and 1 August in 2002 and 2003, respectively.

Figure 2  Average temperature, solar radiation and precipitation for each 5-day period during the cropping season of soybean at sites A and B. Arrows indicate the rainy season in a normal year.

Table 2 Yield and related properties of soybean at harvest

			Grain yield (g m^−2)	Dry matter (g m^−2)	Pod number (m^−2)	Seed number (m^−2)	Seed weight (g)
2002	Site A	CT	353	539	755	1,260	0.280
		RT	378	577	786	1,270	0.296
	Site B	CT	348	515	734	1,210	0.286
		RT	375	586	800	1,310	0.287
2003	Site A	CT	317	506	668	1,090	0.289
		RT	398	601	742	1,250	0.320
	Site B	CT	292	435	519	941	0.311
		RT	331	481	537	943	0.351
RT, ridge tillage; CT, conventional tillage.

Precipitation during the rainy season was relatively low in 2003. Abundant precipitation was recorded from late August to early September in 2003. Figure 1 shows changes in the water content of soils in 2003. Water content of the soil at site B was always higher than that at site A. In contrast to site B, the difference in the water content of the soil between the two croppings appeared only during the rainy season at site A. Such poor drainage at site B was caused by the absence of an underdrain and higher soil clay content. The days after sowing (DAS) for each growth stage were 31 (V5), 50 (R1), 66 (R3), 79 (R5), 98 (R7) and 121 (harvest) in 2002, and 22 (V4), 32 (V5), 51 (R1), 70 (R3), 85 (R5), 108 (R7) and 121 (harvest) in 2003. The flowering stage (R1) occurred late or at the end of the rainy season in both years.

Yield and related properties of soybean are shown in Table 2. anova showed that although RT cultivation increased the soybean yield in both years (Table 3), the average yield in each year was quite different, reflecting the difference in the weather conditions. The growth of the soybean plants was relatively less active at site B than at site A. The level of significance was less than 5% for the dry weight, pod number and seed weight between the two sites, but 7.4% for the grain yield. This lower significance of the grain yield might result from a larger deviation of the values of grain yield. Increased yield with RT was observed at both sites A and B. Although site B experienced more severe moisture conditions (Fig. 2), the yield at site B with RT was higher than that at site A without RT. Hosokawa et al. (2005) revealed that the height of the water table from the surface in the case of RT was 10–15 cm lower than that for CT. This apparent decrease in the water table was consistent with the height of the raised ridges. Furthermore, O₂ concentration within the plow layer did not decrease below 13% for RT, although these values decreased to 3% for CT. The results indicated that RT ameliorated the soil air conditions during the rainy season, which resulted in an increase in yield with RT. The same mechanisms underlying the reduction of wet injury might have operated in our experiments because we used the same machine and fields as Hosokawa et al. (2005).

In previous studies on the effects of RT on soybean yield, a significant yield increase has not been confirmed (Eckert 1987; Fausey 1990; Vyn et al. 1990). The difference between our results and those of Hosokawa et al. (2005) and the results from other investigators may be ascribed to differences in the field moisture conditions. For example, Hatfield et al. (1998) showed that the upper limit of the water table of the field where they conducted RT experiments ranged from 0.7 m to 6 m, with an average value of 3 m. The corresponding values

Table 3 Sources of variation, degrees of freedom and mean squares for soybean properties at harvest in an analysis across years, sites and methods of cultivation

Source	df	Mean squares
		Grain yield	Dry weight	Pod number	Seed weight
Year (Y)	1	5,687**	16,114**	159,198****	63.075****
Site (S)	1	3,278*	14,197**	42,101***	7.253**
Y×S	1	3,060*	13,144**	51,482***	13.872***
Cultivation (C)	1	11,618***	25,980***	16,032*	27.195****
Y×C	1	1,992	468	11	12.826***
S×C	1	493	11	38	0.373
Y×S×C	1	886	2,918	3,504	2.751
Error	20	936	2,025	4,813	1.333
*P < 0.1;
**P < 0.05;
***P < 0.01;
****P < 0.001.

of sites A and B in our study were lower, 0.7 m and 0.4 m, respectively, in October (Table 1). It is reasonable to assume that the water table during the summer season was higher than that in October because the fields that surrounded both sites were flooded for paddy rice cultivation and RT treatment might be more effective under such conditions.

N₂ fixation of nodules and N absorption by roots

The ratio of N derived from N₂ fixation in the xylem sap solution, FF in Eq. 1, is shown in Fig. 3. Although the peak occurred relatively earlier in 2003, the time-course of the ratio was almost the same throughout all plots; that is, the ratio was low at the V5 stage, increased and touched a peak from the R1 to R5 stages, and decreased after the R5 stage. There was no clear difference between RT and CT. Although it appears that the ratio of fixed N was too high at the V4 stage or too low at the V5 stage in 2003, the reasons for such phenomena have not been determined.

The RT treatment increased the amount of plant N in soybean plants during almost all stages aside from the very early stage (Table 4). We could not evaluate the net N accumulation at harvest because soybean leaves fell after the R7 stage. Thus, the amount of plant N at harvest was the apparent amount of plant N. This amount could not be factored into N₂ fixation and the amount derived from root absorption.

anova results of plant N at the R1 stage (end of rainy season) and R7 stage (the last stage at which we were able to evaluate the net N accumulation) are shown in Table 5. At the end of the rainy season (R1), the amount of plant N was significantly lower at site B than site A (Table 4). Even under such conditions, RT treatment increased significantly both the amount of fixed N and absorbed N (Table 5). Hence, the absence of significant differences between FF for RT and CT (Fig. 3) does not

Figure 3  Comparison of the time-course of the ratio of N derived from N2 fixation of nodules per plant N (ratio of fixed N) with the differences in years, cultivation treatments and sites. Letters above the x-axis denote the growth stage of soybeans. RT, ridge tillage; CT, conventional tillage.

imply that RT did not increase the amount of fixed N, but indicates that the activities of both fixed N and absorbed N increased. There are two possibilities of mechanisms for promoting early growth of soybean, namely reduction of wet injury and warming of the seedbed. According to Radke (1982) because RT warms the seedbed, establishment and plant growth are promoted in early spring. Although we cannot deny that the effects reported by Radke (1982) could have been effective in our study, we consider that the reduction in wet injury was more likely to have increased N accumulation for the following reason. By specifically examining the results for site A, the beneficial effect of RT on the increase in the amount of plant N was unclear at the V5 stage in 2002 and the V4 stage in 2003, but became more evident at the V5 and R1 stages, respectively. This pattern suggested that RT did not always increase the amount of plant N before, or at the onset of, the rainy season. In other words, the effect of RT started to appear from the rainy season onward. Therefore, this effect cannot be explained by warming of the seedbed. According to Torigoe et al. (1981), lower branches of soybean plant that differentiate at an earlier growth stage have

Table 4 Estimated amount of N derived from N2 fixation of nodules (fixed N) and of N absorbed by roots (absorbed N)

	Site A						Site B
	Total		Fixed N		Absorbed N		Total		Fixed N		Absorbed N
	CT	RT	CT	RT	CT	RT	CT	RT	CT	RT	CT	RT
2002
V5	1.3	1.3	0.25	0.26	1.1	1.0	0.41	0.83	0.11	0.19	0.30	0.64
R1	4.6	5.7	3.2	4.2	1.4	1.5	4.5	5.0	3.3	3.6	1.1	1.4
R3	7.3	7.6	5.6	6.0	1.8	1.6	9.6	11	8.0	9.3	1.6	2.1
R5	16	14	14	13	2.6	1.9	19	21	16	17	2.7	3.2
R7	18	23	14	18	3.1	5.2	26	26	20	20	5.6	6.0
2003
V4	0.29	0.29	0.18	0.11	0.11	0.18	0.21	0.25	0.07	0.13	0.14	0.12
V5	1.2	1.7	0.31	0.42	0.9	1.3	0.65	0.76	0.15	0.18	0.50	0.59
R1	4.8	5.9	3.4	4.1	1.3	1.8	2.8	2.9	2.0	2.0	0.84	0.90
R3	11	11	8.5	8.5	2.6	2.8	8.2	9.7	6.0	7.5	2.2	2.2
R5	14	19	11	15	3.4	4.5	12	15	8.8	11	3.1	3.8
R7	19	21	12	15	6.0	5.6	18	26	12	16	5.7	9.8
Values are g N m^−2.

Table 5 Sources of variation, degrees of freedom and mean squares for soybean N fixation and absorption of R1 and R7 stages in an analysis across years, sites and methods of cultivation

Source	df	Mean Squares
		R1			R7
		Total	Fixed N	Absorbed N	Total	Fixed N	Absorbed N
Year (Y)	1	0.061*	2.04***	0.061**	25.2	74.61****	13.09*
Site (S)	1	2.052****	3.94***	0.824****	54.1**	14.46****	12.69*
Y×S	1	0.001	2.45***	0.273****	12.8	15.31**	0.111
Cultivation (C)	1	0.273***	0.980**	0.17***	75.9**	31.22***	9.78
Y×C	1	0.015	0.102	0.019	4.87	2.63	0.334
S×C	1	0	0.456	0.015	0.098	1.23	2
Y×S×C	1	0.166	0.009	0.101**	30.2	5.7	9.63
Error	8	0.114	0.165	0.009	8.27	1.98	3.08
*P < 0.1;
**P < 0.05;
***P < 0.01;
****P < 0.001.

the potential to bear many more nodes and pods, and the differentiation of these branches is sensitive to plant conditions. Furthermore, they concluded that II, III and IV branches, which corresponded to lower branches in their nomenclature, tended to bear many nodes and pods. The period when these branches differentiated corresponded to the V5 to V7 stages, which coincided with the rainy season in our study. Torigoe et al.'s (1981) study implied that plant N accumulation, which could be improved by RT in the rainy season, resulted in an increase in pod number (Table 2) through an increase in the number of branches at earlier stages.

At the R7 stage, RT significantly increased the amount of N derived from N₂ fixation, while the amount of absorbed N did not increase significantly (Table 5), presumably because of the lower amount of N absorbed with RT at site A in 2003. The amount of N absorbed with RT was higher at site A in 2002 and at site B in 2003. In other words, RT increased the amount of N derived from N₂ fixation unlike that of N derived from absorption by roots through sites and years. It is well known that O₂ concentration in soil is an important factor facilitating N₂ fixation of root nodules (Ae et al. 1983). It is reasonable to conclude that the effect of RT was mainly associated with an increase in N₂ fixation. High N₂ fixation can alleviate N deficiency during the rainy season and induce a high value for the seed weight (Table 2) at the maturity stage. Takahashi et al. (1992) considered the possibility that a high leaf area index (LAI) at the R1 or R2 stage resulted in efficient seed growth through an increase in the amount of photosynthates. Takahashi et al.'s (1992) study pointed out that N₂ fixation decreased after the R5 stage and senescence was attributed to competition for photosynthates between seed growth and N₂ fixation of nodules. Recently, Takahashi et al. (2005) have supported this hypothesis by indicating that the amount of soil N mineralized until the R1 stage was significantly correlated with seed weight, if wet injury did not decrease the pod number severely. These results are fully consistent with our findings. Therefore, it can be considered that reducing wet injury in the rainy season increased LAI, which led to a high N₂ fixation activity of the nodules and seed weight at the maturity stage.

We must also consider the possibility that the contribution of absorbed N was underestimated. The fact that the amount of plant N increased with RT (Table 4), while the ratio of N₂ fixation of nodules did not change with CT (Fig. 3), indicated that the amount of N absorbed also increased with RT. RT increased the amount of N absorbed except at site A in 2003. The contribution of N absorption by roots increased gradually after the R5 stage because the ratio of N₂ fixation was at a maximum at the R1–R5 stages, after which it decreased (Fig. 3), whereas the amount of total plant N of soybean increased uniformly (Table 4). Although we could not estimate the amount of N absorbed after the R7 stage, if it could be estimated the effect of RT on the increased amount of N absorbed could be clarified. In contrast, RT did not always increase the amount of N absorbed at site A in 2003. We will discuss the relationship between root extension and the amount of N absorbed in this plot in the next section.

Effect of RT on the development of the root system

The profile of the ratio of exchangeable Rb to K in soil changed with soil depth (Fig. 4), which supports the use of the Rb/K ratio in soybean shoots as an indicator of root distribution (Takahashi 1996). Lower Rb/K ratios in plants indicate that the root system of soybean is distributed in the superficial layers.

The Rb/K rate of soybean was significantly lower with RT at the R1 stage (Tables 6,7), which indicates that most of the developing root system was located within the plow layer with RT. RT raised the seedbed by using soil from the plow layer. Thus, the thickness from the soil surface to the subsoil was higher in RT than CT and the depth of the surface of the subsoil was the same as that with CT. The pattern of the Rb/K ratio may indicate that the root system with RT was located in a more superficial area than that with CT, but it would not necessarily indicate that the root system with RT was thicker than that with CT. However, this result does imply that most of the developing root system did not reach the subsoil: this root system was suitable for

Figure 4  Ratio of exchangeable Rb and K in vertical soil profile at sites A and B. The ratio is expressed on a weight basis.

Table 6 Ratio of Rb and K in soybean at R1 and R7 stages

	Site A		Site B
	CT	RT	CT	RT
2002
R1	14	8.8	11	12
R7	13	13	19	25
2003
R1	15	13	16	13
R7	20	17	24	29
Values are mole ratio ×100,000.

reducing wet injury in the rainy season. At the R7 stage, because the value of the Rb/K ratio increased, the root system was distributed in deeper layers than at the R1 stage. The significant difference in the Rb/K ratio between sites A and B at the R7 stage (Table 7) may reflect the difference in the gradient of the Rb/K ratio between the two fields (Fig. 2), but not the soil moisture conditions. No significant differences in the Rb/K ratio were observed between RT and CT at the R7 stage, indicating that the root distribution was oriented toward the vertical direction with RT and that there was no difference with CT at the R7 stage. The fact that the root system with RT was distributed in the superficial zone in the rainy season implies that after the rainy season the soybean roots with RT extended more vigorously than with CT. This conclusion and the fact that RT increased the uptake of

Table 7 Sources of variation, degrees of freedom and mean squares for Rb/K at R1 and R7 growth stages in an analysis across years, sites and methods of cultivation

Source	df	R1	R7
Year (Y)	1	36.6****	81.9***
Site (S)	1	0.123	274****
Y×S	1	0.122	0.004
Cultivation (C)	1	19.8****	15.6
Y×C	1	0.203	0.666
S×C	1	8.12	8.45
Y×S×C	1	15.6	0.193
Error	8	0.125	5.72
***P < 0.01;
****P < 0.001.

N derived from soil at the R5 stage or later indicated that the soybean roots with RT displayed high activity at the maturity stage of soybean growth. The lower Rb/K ratio with RT at site A in 2003 than with CT is consistent with the fact that soybean plants in this plot absorbed a lower amount of N than plants with CT during the R5–R7 stages (Table 4). While the reason for this remains to be determined, it appears that soybean roots with RT did not necessarily extend vigorously throughout the plot. As Takahashi et al. (2003) emphasized, N uptake after the R5 stage is an important strategy to increase soybean yield. Further studies should be carried out to elucidate the relationships among RT, root systems and N accumulation during the soybean maturity stage.

Conclusions

1.	RT significantly increased pod number, seed weight and grain yield in two UFCPs during a 2-year period.
2.	RT significantly increased the amount of plant N. Clear differences in the ratio of the amount of N derived from N2 fixation of nodules per total plant N were not observed between CT and RT.
3.	RT increased both N2 fixation of nodules and N absorption by roots at the end of the rainy season (R1). Roots were distributed within more superficial layers than with CT at this stage. Reduction of wet injury at this stage is considered to increase the number of soybean pods.
4.	Throughout the growth stages until R7, RT significantly increased the amount of N derived from N2 fixation. The increase in the amount of N absorbed with RT was observed at many stages of the experiments, but the effect was neither stable nor significant.

ACKNOWLEDGMENTS

We deeply appreciate the technical assistance provided by Mr J. Asaoka and Mr K. Yazaki. We would also like to thank Mrs A. Okuda for her assistance with the chemical analysis.