Introduction

Soybean (Glycine max L.) is an important source of high-quality, inexpensive protein, oil and vitamin B. In addition, soybean improves soil fertility by adding nitrogen (N) from the atmosphere (Giller 2001). This advantage has a major benefit to African farming systems, where soils have become exhausted because of the need to produce more food to support the increasing populations and where fertilizers are not available or are too expensive for farmers (Bekunda et al. 1997).

The Crop Improvement Program of the International Institute for Tropical Agriculture (IITA), Nigeria contributed significantly to promoting the production of promiscuous soybean, especially in the northern Guinea savanna of Nigeria. But limited efforts have been made to promote the crop in other sub-Saharan African countries such as Benin. However, in the last decade, interest in growing soybean in Benin has increased, possibly because the commonly grown cotton crop drastically exhausts soil fertility while soybean has been observed to replenish it (Houngnandan et al. 2005; Oikeh et al. 2008). The increase in production has been due largely to land expansion but with little improvement in grain yields, which still averaged less than 500 kg dry matter yield (DM) grain per hectare (Annual Report DPP/MAEP 2005). Due to the persistent high cost of chemical fertilizers, the exploitation of soybean biological nitrogen fixation (BNF) could be an attractive strategy for sustainable agricultural production in highly degraded soils of Benin.

There are contrasting views on the role of rhizobial inoculation on promiscuous soybean cultivars. Osunde et al. (2003), studying the response to rhizobial inoculation by promiscuous soybean varieties in Nigeria, suggested that it is pertinent to know prior to the introduction of a promiscuous soybean cultivar if the symbiotic association with indigenous rhizobia satisfies the optimum yield potential of the cultivar at a given location or whether inoculation can better promote nodulation and N2 fixation. Other authors have reported that the “promiscuous” soybean cultivars nodulate freely with indigenous rhizobia, thus eliminating the need to inoculate with Bradyrhizobia japonicum (Kueneman et al. 1984; Pulver et al. 1985). They also observed that the selection for promiscuity was based only on the number of nodules formed. Sanginga et al. (2000), summarizing the results from several studies carried out at IITA, concluded that promiscuous soybeans could be grown without inoculation. However, recent studies have shown positive responses to inoculation by the promiscuous soybean (Okereke and Eaglesham 1993; Sanginga et al. 1996, 2000; Houngnandan 2006).

The role of N2 fixation in legume N nutrition is well known. However, it has been reported that in various varieties/cultivars of grain legume, the estimation of N2 fixation depends on the methods (Hardarson et al. 1993).

In the last decade, isotope ratio mass spectrometry has been used to detect small variations in the abundance of 15N in soil and plant materials (Letolle 1980; Barrie et al. 1995), and these changes can be used for assessing N2 fixation under field conditions without the additional costs of 15N-enriched fertilizer (Unkovich et al. 1994). The natural abundance method utilizes the difference in δ15N natural abundance between N2-fixing and non-fixing plants to estimate the contribution of atmospheric N2 for N acquisition (Shearer and Kohl 1986; Doughton et al. 1995; Herridge et al. 1995).

Calculation of isotopic discrimination (B value) in non-fixing plants is required in order to improve the accuracy of estimation of N2 fixation based on these differences in the natural abundance (Shearer and Kohl 1986; Okito et al., 2004). These authors emphasized that B value has almost always been determined by growing legumes inoculated with rhizobium in N-free hydroponic, sand or sand/vermiculite culture, and they underlined that the B value determined in N-free culture may be considerably different from that of the legume grown in the field. Indeed, N transformations in soil result in an abundance of 15N values higher than that of the atmosphere. These small differences in 15N between the soil and natural atmospheric N are expressed as parts per thousand and are referred to δ15N values (Peoples et al. 1989). The isotopic fractionation factors (β) during physical, chemical and biochemical processes in the nitrogen cycle result in considerable variations in 15N abundance in natural materials and could be determined for estimating the fractional contribution of biologically fixed N of legumes (Shearer and Kohl 1986; Handley and Raven 1992; Nguluu et al. 2001).

Selection of the reference plant for assessing the integrated δ15N of the N absorbed by the N2-fixing plant from soil is therefore very critical for this method (Danso et al. 1992). The non-N2-fixing reference plant should be totally dependent on soil N for growth. In most cases, cereals (non-legumes) are used as reference plants. However, the reference plants are generally significant sources of error in estimating N2 fixation mainly as a result of temporal and spatial variation in δ15N signatures of soil N pools (Danso et al. 1993). However, a major assumption of the 15N natural abundance method for estimating N2 fixation is that both the legume and the reference plant should have the same level of fractionation associated with N uptake from soil N (Spriggs et al. 2003), which is not often the case with cereals as reference plants. The problems associated with the determination of N2 fixation by measuring δ15N have recently been reviewed (Handley and Scrimgeour 1997; Högberg 1997; Boddey et al. 2000). We are not aware of any study in sub-Saharan Africa that has used non-fixing soybean species as the reference plant in estimating N2 fixation from 15N natural abundance technique.

The objective of this study was to estimate the biological N2 fixation of multiple soybean cultivars introduced in the derived savanna of Benin from IITA, Ibadan, Nigeria (TG× lines), the Cerrados in Brazil (Conquista) and East and Central Africa (Yezumutima). Nitrogen fixation was determined by the 15N natural abundance technique, using as reference plants two non-fixing soybean species (Chippewa and Clay) from the Seibersdorf in Vienna.

Materials and methods

Site description

The study was carried out at the experimental farm of the Faculty of Agricultural Sciences of the University of Abomey-Calavi, Benin from July 30 to October 31, 2005. The site is located between 6°5′ and 6°30 N and 2°10′ and 2°40′ E and is characterized by a subequatorial humid, hot climate with a bimodal rainfall distribution. Wet seasons occur from March to July and from September to November. Data from the nearest weather station (ASECNA, Cotonou), averaged over 20 years, showed a total annual rainfall ranging between 1,000 and 1,200 mm. The average daily temperature was 27°C, with a minimum of 26°C during August and a maximum of 36°C between February and April.

The experimental farm has been established on a typical “terre de barre” soil, classified by Food and Agriculture Organization-United Nations Educational, Scientific and Cultural Organisation as Rhodic Ferralsol and by USDA as Eutrustox. The soil had a sandy texture, and its main chemical characteristics are presented in Table 1.

Table 1 Physico-chemical properties of soil
Full size table

Field experiment

Seedling of two non-nodulating reference plants [“Chippewa” and “Clay” (Peña-cabriales et al. 1993)] and 17 nodulating soybean cultivars were planted on July 30, 2005 at the University of Abomey-Calavi’s experimental farm. We used a randomised complete block design with three replicates per treatment. The size of plots was 1 × 1.5 m, and plots were maintained at 350 g kg−1 water-holding capacity. Seeds were pre-germinated in Petri dishes and transferred into field plots.

Soybean seeds were inoculated with IRAT FA3 strain of B. japonicum containing an approximate density of 107 viable rhizobia per seed before sowing. IRAT FA3 strain was a more specific strain, selected by the Nitrogen Fixation in Tropical Agricultural Legumes/Microbial Resource Centre Rhizobium project and used to inoculate strict nodulating soybean varieties adapted to tropical agricultural zones in Africa and Latino America. Each plot contained four planting lines, and the space between rows was 30 cm. On the planting row, soybean was seeded at a 5-cm within-row space. Two seeds were sown per hill and later thinned to one 14 days after.

Muriate of potash (KCl) was applied to each plot at the rate of 15 kg K2O ha−1 at sowing.

Pot experiment

Nodulating and non-nodulating cultivars of soybean [G. max (L.)] were sown on August 4, 2005 and grown for 45 days in 20-cm diameter pots containing a sandy loam soil which was N-deficient (Shearer and Kohl 1986; Okito et al., 2004) for the whole experiment (Table 2). A modified bottle-Masson jar technique described by Vincent (1970) was used. The pots were equipped with a 4-mm thick drainage mat and filled with 3 kg of autoclaved soil sampled from the beach near the Atlantic Ocean in Cotonou.

Table 2 Characteristics of the N-free substrate collected from the Atlantic Ocean beach to determine isotopic fractionation (β) values
Full size table

Each pot received solution with the following concentrations (Norris and Date 1976): 0.436 g of P, 0.830 g of K, 0.25 g of Ca, 0.06 g of Mg, 0.03 g of S and 5 mg of Fe, Cu, Zn, Mn, Mo and Bo using, respectively, NaH2PO4·H2O, KH2PO4, CaCO3, MgCl2, Na2SO4·10H2O, and [FeSO4·7H2O] = (CuSO4·5H2O), (ZnSO4·7H2O), (MnCl2·4H2O), [(NH4)6Mo7O24·4H2O], H3BO3.

Pots were arranged following a complete randomised block design with three replicates per treatment. In each block, the pots containing the 17 nodulating and two non-nodulating soybean cultivars were randomly distributed before sowing the seeds. Seeds were first inoculated with Bradyrhizobia strain IRAT FA3 at a density of 107 viable rhizobia per seed before sowing (Vincent 1970). One week after germination, seedlings were thinned to two plants per pot. Some physico-chemical properties of the substrate were shown in Table 2.

Plant sampling

In the two trials (field and pot), soybean shoots were sampled at the flowering time both in the field and plot trials. In the field, six plants were randomly sampled in each plot at the first sampling, while all shoots were harvested at 45 days after sowing (DAS) in the pots. The plants in the N-free pots grew poorly.

After harvesting the shoot, roots and nodules were separated. After washing all these samples, shoots were put in paper. Roots were stored at 4°C. Shoots were dried at 65°C for 72 h and ground-milled to pass 0.4-mm sieve with an ultra centrifugal mill (ZM 200, Retsch Germany) as described by Okito et al. (2004).

Plant shoots were analysed for total N and δ15N using an elemental analyzer (ANCA-SL, PDZ Europa, UK) coupled to isotope ratio mass spectrometer (20-20, SerCon, UK).

Methodology and calculations

  • Calculation of % Nitrogen derived from Atmosphere (Ndfa)

The expression for calculating the fraction of N derived from atmospheric N2 in N2-fixing plants (%Ndfa) was given by an isotope dilution expression, which was also the contribution of atmospheric N2 to the total N in the N2-fixing soybean cultivars calculated as follows (Shearer and Kohl 1986; Nguluu et al. 2001):

$$\% {\text{Ndfa = }}\frac{{\delta ^{15} {\text{Nref - }}\delta ^{{\text{15}}} {\text{Nfix}}}}{{\delta ^{15} {\text{Nref - }}\delta ^{{\text{15}}} {\text{N}}}} \times 100{\kern 1pt} \,$$
(1)

Where:

  • δ15Nref was the δ15N value of reference plants (value of N from sources other than atmospheric N2); Clay and Chippewa were the non-fixing soybean cultivars in the study.

  • δ15Nfix was the δ15N value of the total N in the N2-fixing soybean grown under conditions in which atmospheric N2 and N from other sources are available (N-fixing soybean cultivars grown in field).

  • δ15N was the isotopic discrimination (B value) of fixed N in the N2-fixing soybean (as measured in N-fixing soybean forced to depend solely on fixed N by growing them hydroponically with N-free nutrient pots; Shearer and Kohl 1986).

The Isotopic discrimination (B value) of soybean grown in N-free pot was read directly from the mass spectrometer. However, B value can also be calculated as follows (Shearer and Kohl 1986):

(2)

Where sample = soybean

  • 15N/14N (standard) was a known 15N natural abundance of atmospheric N2, 0.3663 atom% 15N, which is usually the ultimate reference (Shearer and Kohl 1986).

  • 15N/14N (sample) was obtained from soybean grown in the N-free pots by subtraction from Eq. 2.

The isotopic fractionation factor

The isotopic fractionation factor (β) was calculated as follows (Bergersen et al. 1988):

$${\text{ $ \beta $ }}\,{\text{N}}_{\text{2}} \,{\text{fixation}} = \frac{{^{15} {{\text{N}} \mathord{\left/{\vphantom {{\text{N}} {^{14} {\text{N}}}}} \right.\kern-\nulldelimiterspace} {^{14} {\text{N}}}}\left( {{\text{standard}}} \right)}}{{^{15} {{\text{N}} \mathord{\left/{\vphantom {{\text{N}} {^{14} {\text{N}}}}} \right.\kern-\nulldelimiterspace} {^{14} {\text{N}}}}\,{\kern 1pt} \left( {{\text{legume}}} \right)}}\,$$
(3)

Where legume = soybean

  • 15N/14N (standard) was a known 15N natural abundance of atmospheric N2, 0.3663 atom% 15N, which is usually the ultimate reference (Shearer and Kohl 1986).

  • 15N/14N (legume) = 15N/14N (sample) and was obtained by subtraction from Eq. 2.

Statistical analysis

Statistical analysis was done using “Proc ANOVA” (SAS 2001) to determine the statistical differences among the soybean cultivars and their interactions. Specific pair-wise comparisons of treatment levels were done using the least significance differences (LSD) test at P < 0.05. Correlation study was done among traits.

Results

There were significant differences (P < 0.001) in plant δ15N among nodulating soybean cultivars grown in inorganic N-free medium, with values ranging from −2.7756‰ (Jupiter, a local variety) to 0.1951‰ (Conquista; Table 3). The non-fixing soybean cultivars (Chippewa and Clay) had higher δ15N values (2.67–9.30‰) than the other cultivars evaluated (Table 3). In this study, δ15N values were used to calculate the isotopic fractionation factor, β.

Table 3 Aboveground dry matter yield harvested 45 days after flowering, total N content, isotopic discrimination (B value) and isotopic fractionation factors (β) for 17 soybean cultivars grown under N-free conditions
Full size table

The δ15N of the total N accumulated by nodulating soybean cultivars in N-free medium reflects the isotopic fractionation during the N2-fixing process. The isotopic fractionation values (β) calculated during N2 fixation by the different cultivars of soybean from the δ15N in the aboveground portions showed significant differences among the soybean cultivars (P < 0.01) since it varied of about 1.000 for most soybean cultivars and around 1.002 for TG× 1448 2E, TG× 1894 3F, TG× 1908 3F, TG× 1910 14F and Jupiter (Table 3). The cultivars with β values >1 showed 15N discrimination during N2 fixation. The isotopic discrimination (B values) was inversely related to the isotopic fractionation factor β (r = −0.42; P < 0.05; n = 51) but more related to the percentage of N derived from atmosphere when Clay was used as reference (r = 0.60; P < 0.05; n = 51), than when Chippewa was the reference plant (r = 0.29; P < 0.05; n = 51; Table 6).

Dry matter yield was significantly (P < 0.001) greater by 72% to 110% in Conquista, Yezumutima and two of the promiscuous soybeans (cvs. TG×1908 3F and TG×1908 8F) compared with the popular local cv. Jupiter (Table 3). The two non-nodulating cultivars (Chippewa and Clay) and all other promiscuous soybean cultivars had similar DM as cv. Jupiter, ranging from 195 to 320 mg per plant (Table 3).

Among the N2-fixing soybean cultivars, N content in the whole plant ranged from 2.6 mg N per plant (TG×1448 2E and TG×1895 3F) to 8.1 mg N per plant (Yezumutima) compared with only 1.8–1.9 mg N per plant for the non-fixing cultivars, confirming that these latter cultivars relied only on soil N for growth and N acquisition. Among the N-fixing cultivars, only three cultivars, Yezumutima, TG× 1908 3F and TG× 1888 29F had a higher N content.

There were highly significant differences (P < 0.001) among the N2-fixing soybean cultivars for biomass production, N content and N accumulation (Table 4). Among the N2-fixing cultivars, cv. TG× 1910 2F showed the greatest dry matter yield, and it was the only cultivar with dry matter production higher than Jupiter (local cultivar; Table 4). Two promiscuous soybean cultivars (TG× 1910 2F and TG× 1908 3F) and cv. Conquista from Brazil ranked among the best three cultivars in DM accumulation (Table 4). Among the N-fixing cultivars, TG× 1440 1F, TG× 1888 29F and TG× 1908 3F, all promiscuous, had the highest δ15N fixation (Table 4). The yield of non-nodulating plants cvs. Chippewa and Clay was 1.7–2.2 g DM per plant in the field experiment (Table 4), and the plant showed the lowest N content but the highest δ15N value compared with the nodulating cultivars under field conditions (Table 4).

Table 4 Aboveground dry matter yield harvested 45 days after flowering and total N yield of 17 soybean cultivars grown under field conditions
Full size table

Nodulation was observed on all soybean cultivars, and differences among cultivars in nodule number were significant (Table 4). The number of nodules per plant varied from 6 (TG× 1910 2F and Jupiter) to 15 (Conquista) and were, on average, related to the nodule fresh weight (r = 0.29; P < 0.05; n = 51) and inversely related to δ15N‰ calculated in the field (r = −0.33, P < 0.05; n = 51; Table 6).

Percentage and amount of N derived from air were significantly different (P < 0.01) among the soybean cultivars (Table 5) and depended on the selected reference plant. Significant correlation was observed in percentages of N derived from N using either clay or Chippewa as the reference plant (r = 0.67; P < 0.05; n = 51; Table 6). When Clay was used, the average percentage of Ndfa was 1.4 times higher than when Chippewa was the reference plant. However, both reference plants consistently ranked promiscuous soybean cvs. TG × 1894 3F and TG × 1908 8F as the best two cultivars and cv. TG × 1888 29F as the least cultivar in percentage Ndfa. Similar trends were found for the amount of N derived from air. But the reference plants ranked the cultivars slightly differently. Whereas Chippewa ranked TG× 1910 11F (94), TG× 1910 2F (78), and Conquista (65 mg N per plant) as the best three cultivars, Clay ranked TG× 1910 2F, Conquista and TG× 1908 3F as the best three performing cultivars with, respectively, 117, 88 and 81 mg N per plant derived from the atmosphere (Table 5).

Table 5 Percentage (%) and amount of N derived from the atmosphere in aboveground part of 17 soybean cultivars using two reference plants (Clay and Chippewa)
Full size table
Table 6 Coefficient of correlation between nodulation (nodule number and nodule fresh weight), percentage of N derived from atmosphere, isotopic discrimination (B value) and isotopic fractionation (β) of soybean cultivars grown in southern Benin in 2006
Full size table

Discussion

Our results indicated that 15N natural abundance method can be used to evaluate the percentage and amount of N derived from atmosphere by the 17 soybean cultivars. Although different values were obtained using either Clay or Chippewa as the reference plant, the ranking of the cultivars were still similar. Moreover, the data obtained using the two reference crops were highly correlated (r = 0.67; P < 0.05; n = 51). The choice of the reference plant is somewhat difficult thus limiting the use of this technique for the assessment of percentage and amount of N derived from the air. However, as reported by Gathumbi et al. (2002) and Somado and Kuehne (2006), if the objective is not to obtain very precise quantification of N2 fixation, the use of the δ15N natural abundance method may be adequate for estimating the amount of N derived from atmospheric N2 fixation by field-grown herbaceous and woody legumes and for screening and ranking of legumes for biological nitrogen fixation. Because the method does not require the use of costly 15N fertilizers, it will be more affordable compared with other methods particularly in low-income countries such as Benin.

Our results confirmed the inverse relationship generally observed between the isotopic discrimination (B value) and the isotopic fractionation factor (β). The observed negative values of plant δ15N for most of the soybean cultivars grown in N-free medium except TG× 1895 49F, Conquista and the two non-nodulating cultivars (Chippewa and Clay) using only shoot biomass harvested at 45 DAS might have been due to differences in the genomes of the symbiotic partners (Steele et al. 1983; Yoneyama et al. 1984) and the growth stages of these cultivars (Nguluu et al. 2001). The δ15N values obtained for shoots in this study were higher than those earlier reported for soybean (Bergersen et al. 1989; Nguluu et al. 2001) and for Phaseolus vulgaris (Yoneyama et al. 1984) but similar to those reported by Yoneyama et al. (1984) for cowpea and for Stylosanthes hamata by Peoples et al. (1989). The observed differences in δ15N among N2-fixing plants and the consistently lower values of δ15N from N2-fixing plants compared with the non-fixing plants is in agreement with results earlier reported (Amarger et al. 1979; Kohl et al. 1980; Shearer et al. 1980; Gathumbi et al. 2002).

In most of the available studies, differences among plant parts and within species in δ15N were also strongly considered (Nguluu et al. 2001). The screening of the cultivars revealed broad genotypic variability for total N and δ15N. Our results showed that the soybean genotypes could be classified broadly into two main groups: (1) those with positive δ15N and (2) those with negative δ15N. The positive group utilized soil N efficiently, whereas the negative group absorbed limited N from the soil.

The calculated isotopic fractionation factor (β) showed that there was a positive isotope effect on legume shoots discrimination against 15N. However, the effects were small in all legumes indicating that only small fractionation effects were associated with the accumulation of N derived from biological N2 fixation. Most of the values found in our study were less than (1.0014 ± 0.0001) the values reported for soybean (Mariotti et al. 1983; Shearer and Kohl 1986), for shoot tissue of Stylosanthes humilis (Yoneyama et al. 1986) and for Stylosanthes spp. and cowpea (Nguluu et al. 2001).

Amarger et al. (1979) measured the isotopic effect of four species of soybeans on N2 fixation to be about −1‰, while Kohl et al. (1980) reported isotopic effect on N2 fixation of +1‰ for some of the varieties of soybean used in their experiments. The δ15N values for seed N were between −0.2‰ and −1.6‰ (Kohl et al. 1980). With the exception of nodules, none of the plant parts deviated in 15N from the entire plant by more than two δ15N units (Kohl et al. 1980). In most cases, the deviation was one unit or less. Kohl et al. (1980) emphasized these specific aspects and reported that as much as differences in 15N abundance between plant parts and whole plants vary with experimental conditions, it would not be feasible to attempt to correct the BNF estimates for these different plant fractions.

The differences we observed in the values of BNF and N fixation based on the reference plant used for assessment supported earlier report by Högberg (1997) that estimates of N2 fixation using 15N natural abundance method largely depended on the selected non-N2-fixing reference plant. Kohl et al. (1980) reported, on average, 51.4% Ndfa for soybeans on the basis of the δ15N value of the whole plant while Maskey et al. (1997) estimated an average of 62% for the proportion of %N derived from air for herbaceous legumes. However, the range of values for N2 fixation observed in our study (Table 5) was similar to those reported for studies on soybean elsewhere (Chapman and Myers 1987; Bergersen et al. 1989; Peoples and Herridge 1990; Herridge and Holland 1992).

Some limitations of the δ15N method for measuring N2-fixation have been highlighted by Boddey et al. (2000), Handley and Scrimgeour (1997), and Unkovich et al. (1994), who suggested that N2-fixation estimates were more dependent on the δ15N values of the legume than that of the reference crop. However, this was not the case in our study because the δ15N values differed among the reference plants, and these differences significantly influenced the estimates of N2 fixation.

In conclusion, the study showed that any of the non-nodulating reference soybean cultivars could be used in δ15N natural abundance method for assessing N2-fixation to rank soybean cultivars for their relative contribution of fixed N to the production systems. Because this method is relatively inexpensive compared with the use of 15N fertilizer and also more reliable compared with the use of a cereal as the reference plant, the method will be most adaptable to low-income developing countries. Promiscuous soybean cultivars, TG× 1894 3F and TG× 1908 8F, were identified to have the highest capacity to fix N and could be included in a soybean extension program for West African farming systems.