Bradyrhizobium STRAIN AND THE 15 N NATURAL ABUNDANCE QUANTIFICATION OF BIOLOGICAL N 2 FIXATION IN SOYBEAN

In commercial plantations of soybean in both the Southern and the Cerrado regions, contributions from biological nitrogen fixation (BNF) are generally proportionately high. When using the N natural abundance technique to quantify BNF inputs, it is essential to determine, with accuracy, the N abundance of the N derived from BNF (the ‘B’ value). This study aimed to determine the effect of four recommended strains of Bradyrhizobium spp. (two B. japonicum and two B. elkanii) on the ‘B’ value of soybean grown in pots in an open field using an equation based on the determination of δN natural abundance in a non-labelled soil, and estimate of the contribution of BNF derived from the use of N-isotope dilution in soils enriched with N. To evaluate N 2 fixation by soybean, three non-N 2 -fixing reference crops were grown under the same conditions. Regardless of Bradyrhizobium strain, no differences were observed in dry matter, nodule weight and total N between labelled and non-labelled soil. The N 2 fixation of the soybeans grown in the two soil conditions were similar. The mean ‘B’ values of the soybeans inoculated with the B. japonicum strains were -1.84 ‰ and -0.50 ‰, while those inoculated with B. elkanii were -3.67 ‰ and -1.0 ‰, for the shoot tissue and the whole plant, respectively. Finally, the ‘B’ value for the soybean crop varied considerably in function of the inoculated Bradyrhizobium strain, being most important when only the shoot tissue was utilised to estimate the proportion of N in the plant derived from N 2 fixation.


INTRODUCTION
Using the 15 N natural abundance technique to quantify the contribution of biological nitrogen fixation (BNF) to a legume has the almost unique attribute that it can be used without interference to the plant environment (Peoples et al., 1989).The technique is based on the commonly-observed phenomenon that N in most (but not all) soils is slightly naturally enriched with 15 N as compared to the natural abundance of atmospheric N 2 (Högberg, 1997;Boddey et al., 2000).In order to determine the proportion of N derived from the air (%Ndfa) it is necessary to determine: the 15 N abundance of the legume under study, and the 15 N abundance of both the N derived from the soil, and, that derived from the air via BNF (Shearer & Kohl, 1986).
The 15 N abundance of the N in the legume derived from BNF is known as the 'B' value.When contributions of BNF are high and/or the 15 N abundance of the soil-derived N is low, it is extremely important to accurately determine this value.
Using plants grown in N-free medium (Steele et al., 1983;Bergersen et al., 1986;Yoneyama et al., 1986) have reported that the 'B' value of a particular legume may vary with rhizobium strain.Doughton et al. (1992) developed a method to determine 'B' with plants grown in the field based on the comparison of the estimate of the % Ndfa from the use of the 15 N isotope dilution technique on soil enriched with 15 N with those derived from the use of the 15 N natural abundance technique.With this method Okito et al. (2004) found that there were large differences in 'B' value of either the shoot tissue or the whole soybean plants between the different strains.The 'B' values for the Bradyrhizobium strains 29 W (SEMIA 5019 -B.elkanii) and CPAC 7 (SEMIA 5080 -B.japonicum), respectively, were -4.54 and -2.39 ‰ for the shoot tissue.These two strains are both recommended for inoculant manufacture in Brazil and proportional BNF inputs to soybean are generally high (Alves et al., 2003(Alves et al., , 2006)).Therefore, the objective of this study was to determine the 'B' values in soybean plants grown in an open field soil inoculated with of all four recommended strains of B. elkanii and B. japonicum.

MATERIAL AND METHODS
An open field pot experiment was installed on December 13 th 2003 in Seropédica, Rio de Janeiro State (22º47' S, 43º40' W, 33 m above sea level).The soybean plants were grown in 140 plastic pots filled with 4 kg of dry topsoil (0-20 cm) taken from an area of the field station which had never been planted before with soybean and was known from previous studies not to possess significant numbers of rhizobium strains capable of nodulating this crop.The soil type was classified as an Orthic Acrisol based on the FAO classification, or as an Argissolo Vermelho -Amarelo distrófico using the Brazilian classification.Before fertiliser addition the main chemical analyses of the soil were: pH in H 2 O, 5.1 exchangeable; Al, Ca and Mg were 10, 10 and 8 mmol c dm -3 ; respectively, and P and K, 1 e 61 mg dm -3 , respectively (Embrapa, 1997).To each pot, 1.2 g dolomitic lime per kg of dry soil, and subsequently 230 mg P 2 O 5 kg and 120 mg K 2 O kg soil -1 were added.Half of the pots were amended with a solution containing 40 mg of 15 N-labelled ammonium sulphate enriched at 20 atom % 15 N excess.
Soybean seeds (cv Celeste) were surface sterilised in 70% aqueous ethanol for 3 min followed by 2 min in sodium hypochlorite solution (50 g L) and then washed 10 times in sterile distilled water.Pots were seeded with five seeds of each plant and 8 days after emergence these were thinned to two plants per pot.
The experiment was laid out in a randomised complete block design with split plots and five replicates.Main plots consisted of the two 15 N treatments (enriched with 15 N or not) and subplots of soybean plants inoculated with the four strains of Bradyrhizobium recommended for the manufacture of commercial inoculants in Brazil along with non-inoculated plants and three more pots for reference plants.The Brazilian recommended strains are 29W (SEMIA 5019) and SEMIA 587 which belong to the species Bradyrhizobium elkanii and CPAC 15 (SEMIA 5079) and CPAC 7 (SEMIA 5080) which are B. japonicum (Rumjanek et al, 1993;Sato et al., 1999).The inocula were peat based containing approximately 10 9 colonyforming units g dry weight -1 .To facilitate the quantification of BNF by the soybean and to evaluate the 'B' value, three non-N 2 -fixing reference plants, non-nodulating soybean (cv.T 201), rice (Oryza sativa cv.IAC 4440) and grain sorghum (Sorghum bicolor, cv.BR 305) were employed.
During the first 15 days the plants were irrigated with sterile distilled water to avoid possible contamination of the seedlings with rhizobia.Subsequently, ordinary tap water was used for irrigation.At 87 days after planting (DAP), coinciding with early grain filling, all plants were harvested and separated into shoot tissue, roots and nodules.Great care was taken to recover all roots and nodules by sieving the soil through a 1 mm sieve.All harvested materials were dried at 65ºC for >72 h and ground in a Wiley mill (<0.85 mm).Sub-samples were subsequently ground to a fine powder using a roller mill similar to that described by Smith & Myung (1990).Samples were analysed for total N content using the semi-micro Kjeldahl procedure as described by Urquiaga et al. (1992).The 15 N enrichment/abundance of aliquots of sub-samples containing approximately 35 mg N was determined using an automated continuousflow isotope-ratio mass spectrometer consisting of a Finnigan DeltaPlus mass spectrometer coupled to the output of a Carlo Erba EA 1108 total C and N analyser (Finnigan MAT, Bremen, Germany).Ground samples of the seeds were also analysed for total N and 15 N natural enrichment/ abundance using the same procedures.
The weighted mean of the 15 N natural abundance of the whole plants (discounting original seed N) can be calculated using a mass balance as described by Högberg et al. (1994): where Sh, Sd, Nod and Ro represent shoots, seeds, nodules and roots, respectively, and d 15 N is the 15 N natural abundance (‰) and N is the total N accumulated by each plant part.
For the soybeans planted in the soil enriched with 15 N, the proportion of N derived from the air (%Ndfa) via BNF was calculated according to the equation of Chalk (1985): where % 15 N f and % 15 N r are the values of atom % 15 N excess of the soybean and the reference plant, respectively.As three non-N 2 -fixing crops were used as reference plants, three independent estimates of the %Ndfa for each strains could be calculated.
The addition of a small amount of labelled ammonium fertiliser (40 mg per pot or 2.1 mg N kg soil -1 ) was insufficient to have any significant effect on N 2 fixation by the soybean.Hence the %Ndfa of non-fertilised plants was assumed to be equal to that of legumes grown in 15 N-labelled soil.In this case the 'B' value can be calculated for the non-labelled soybean plants using the equation (Okito et al., 2004): where d 15 N r and d 15 N f are the natural 15 N abundance of the reference and soybean plants, respectively.The data were subjected to standard analysis of variance to detect significant differences between treatment means.The Student LSD test was applied to verify similarities and differences of means for the various parameters for the reference plants and the soybean inoculated with the different Bradyrhizobium strains.
It was necessary to compare sub-plot treatment means for different main plots in order to evaluate whether there were effects of the Bradyrhizobium strain on accumulation of dry matter regime and N and 15 N enrichment in the labelled and unlabelled soil individually.This was achieved using the procedure described by Little & Jackson-Hills (1978): The calculation of the least significant difference between means (LSD -Student) becomes: where b is the number of subplot treatments, r the number of replicates, Ea and Eb are the mean squares of the sub-plot and main plot errors, respectively, and t ab is the weighted t value for main plots and subplots calculated as described by Little & Jackson-Hills (1978).

RESULTS AND DISCUSSION
The non-N 2 -fixing reference plants and the inoculated soybeans showed no difference in their dry matter accumulation when grown in 15 N-labelled or unlabelled soil (Tables 1 and 2).However, the reference sorghum and rice plants accumulated significantly more nitrogen in the soil amended with 15 N-labelled ammonium sulphate (Table 1), but the total N accumulated was far less (<6%) than that accumulated by the nodulated soybean plants (Table 3).For the nodulated soybeans, the small labelled-N addition had no significant effect on the dry weight or N content of (or N accumulation by) the nodules or the whole plants.These results indicate that the quantity of labelled N added to the soil in the 15 N-enriched treatment had no significant effect on the contribution of BNF to the soybean.This is a necessary prerequisite for the application of the technique used here, originally proposed by Doughton et al. (1992) for the determination of the 'B' value.
Amongst the nodulated soybean plants, inoculation with the CPAC 15 strain showed the greatest dry matter accumulation with 56.9 g DM per pot in the shoot tissue and 77.5 DM per pot for the whole plant (Table 2).Moreover, the soybean plants inoculated with this Bradyrhizobium strain also accumulated significantly more N than all other inoculation treatments (1729 mg N per pot).However, there were no differences in N accumulation between the other three strains (means ranged from 1380 to 1436 mg N per pot -Table 3).The uninoculated soybean plants grew more slowly than the inoculated plants.At 30 DAP, the plants were small with pale green leaves suggesting that there was very limited, if any, initial nodulation.Mean final dry matter accumulation was 8.5 g DM per pot (data not shown) approximately 8% of that accumulated by the nodulated plants.Mean N accumulation was 107 mg N per pot approximately 7.3% of that of the inoculated plants.Mean nodule weight of these plants was 0.50 g per pot were again far lower than that of the inoculated plants (mean 3.9 g per pot.These results point out that the native population of Bradyrhizobium capable of nodulating soybean in this Values are means of five replicates.Means within the same column followed by the same upper case letter, or in the same row followed by the same lower case letter are not different at p < 0.05 (Student 't' test).-----------------------------------------------------g per pot ----------------------------------------------------  A marked difference in nodule weight between inoculation treatments was observed.
The two B. elkanii strains (29 W and SEMIA 587), presented similar nodule dry weights which were on average 90% greater than the weight of the nodules formed by the two B. japonicum strains (CPAC 7 and CPAC 15) (Table 2).A similar division of different "Rhizobium japonicum" strains into two groups of low ("group I strains") and high ("group II strains") nodule weights was earlier reported by Döbereiner et al. (1970) and Neves et al. (1985).These authors found Sci.Agric.(Piracicaba, Braz.), v.65, n.5, p.516-524, September/October 2008 only minor differences between strains in the quantities of plant N derived from BNF.Those strains (Group I) inducing low nodule weights were described as highly efficient (higher ratio of N 2 fixed to nodule weight) and included the strain CB 1809 which has subsequently been found to be genetically almost identical to CPAC 7 (SEMIA 5080 - Nishi et al., 1996).Those of lower efficiency (Group II, high nodule weight) included the two strains used in this present study and are now classified as B. elkanii (SEMIA 587 and 29 W = SEMIA 5019 - Sato et al., 1999).Both strains that induced low nodule weight (CPAC 7 and CPAC 15) are now classified as B. japonicum (Rumjanek et al., 1993;Sato et al., 1999).This difference in nodule weight induced by the strains 29 W and CPAC 7 was also confirmed by Okito et al. (2004).
The three reference plants accumulated very small quantities of N, ranging from 4 to 16 mg per pot.In contrast the inoculated soybean plants accumulated over 1700 mg of N per pot indicating extremely large contributions of biologically-fixed N to them.
All the inoculated soybean plants grown in pots of soil enriched with 15 N showed 15 N enrichments of approximately 0.01 atom % 15 N excess (varying from 0.0091 to 0.0149 atom % 15 N excess), but the 15 N enrichments of the different reference plants, non-nod soybean, sorghum and rice were 0.618, 0.616 and 0.395 atom % 15 N excess, respectively (Table 4).The fact that the inoculated soybean plants had much lower 15 N enrichment than the three reference species shows conclusively that the soybean plants obtained very high proportions of their N from BNF.The rice plants showed considerably lower 15 N enrichment than the other two reference plants probably due to different spatial and temporal patterns of soil N uptake in soil which was not uniformly labelled with 15     al., 1995), although some input of unlabelled atmospheric N from N 2 -fixing bacteria associated with rice plants cannot be ruled out.The lower 15 N enrichment of the rice reference crop resulted in lower estimates of %Ndfa to the inoculated soybean than those resulting from the use of the other two reference crops.Although, this difference was very small (<1.4%) as the proportion of N derived from BNF was extremely high.
The %Ndfa was very high not only because soybean is an efficient N 2 fixer, but also because the amount of soil explored by the plant roots (2 kg per plant) was intentionally much less than would normally be exploited in the field.When the %Ndfa is high the 15 N enrichment of the reference plants can vary considerably with little impact on the resulting estimate of %Ndfa (Hardarson et al., 1988;Boddey & Urquiaga, 1992).Furthermore, the sensitivity of the %Ndfa estimate to small difference in 'B' value is maximised, so that this technique using labelled and unlabelled soil (Doughton et al, 1992;Okito et al., 2004) will give the most accurate values under these conditions.
While the differences in the estimates %Ndfa of the shoot tissue between inocula were very slight (Table 4), because of the much higher total N accumulation by the soybean plants nodulated with the B. japonicum strains, they obtained on average 26% more N from BNF than those nodulated by B. elkanii (Table 5).The B. japonicum strain CPAC 15 exceeded by 39% the mean of the N fixed by the two B. elkanii strains.This indicates that if these B. japonicum strains can compete with strains already established in the soil to dominate the nodule occupancy soybean growers could find that the use of single-strain inoculants of CPAC 7 or CPAC 15 will promote agronomically significant yield increases.The shoot tissue of the reference plants grown in the non-15 N-labelled soil showed positive values of 15 N abundance with values ranging between +6.1 and +8.8 ‰ (Table 6).However, there were smaller differences in the 15 N abundance for the whole plants between the three different reference plants indicating that these species tapped similar N pools and/or there was only a small temporal or spatial variation in soil 15 N abundance.In contrast, the 15 N abundance of shoot tissue of the inoculated soybean plants was always negative and plants nodulated by the B. elkanii strains (29 W and SEMIA 587) had more negative 15 N abun-dance values than those nodulated by the B. japonicum strains (CPAC 7 and CPAC 15) (Table 6).The roots of the soybean plants nodulated by the B. elkanii strains were slightly depleted in 15 N compared to that of the air, whereas those formed by the B. japonicum were slightly enriched.All nodules were strongly enriched with 15 N as has been observed by many authors for soybean as well as for several other legumes (Turner & Bergersen, 1983;Shearer & Kohl, 1986;Boddey et al., 2000).However, the weighted mean value (calculate as described in Equation 1 -Material and Methods) of the whole plants of all inoculated soybeans remained negative.
The 15 N natural abundance in the shoot tissue of the soybeans inoculated with the B. elkanii strains (29W and SEMIA 587) was more negative than those inoculated with the B. japonicum strains (CPAC 7 and CPAC 15) but there were no differences in d 15 N of the whole plants (Table 6).The 15 N abundance of the shoot tissue of the reference plants was +7.4 ‰, while the inoculated soybeans ranged from -3.54 and e -1.27 ‰ which confirms that the %Ndfa of all soybeans was very high and that the 'B' values of the shoot tissue (termed 'B s ' value by Okito et al., 2004) must be considerably negative.
The 'B' value of the shoot tissue ('B s ') and the whole plants ('B wp ') were calculated utilising the %Ndfa of the soybeans determined in the 15 N-enriched soil, along with the 15 N abundance of the soybean and the mean 15 N abundance of all three reference crops grown in the non-labelled soil.The 'B s ' values of the soybeans inoculated with the B. elkanii strains 29W    , respectively).The values for the strains 29W and CPAC 7 were similar to, but slightly less negative than, those determined using the same technique by Okito et al. (2004) for same strains (-4.54 and -2.39 ‰, respectively, -difference between strains significant at p < 0.05).These authors also determined the 'B s ' values for the shoot tissue of soybean inoculated with these two strains grown in N-free media in the greenhouse and again found that the value for B. elkanii strain 29W (-2.58 ‰) was lower than that of the B. japonicum strain CPAC 7 (-1.31‰).The less negative value of the 'B s ' values determined for plants grown in N-free medium may have been due to the lower total N accumulation of the plants under greenhouse conditions.Therefore Okito et al. (2004) suggested that determination of 'B s ' for plants gown in the open field was more appropriate for the application of the 15 N natural abundance technique.When the 'B wp ' values of the whole plants was determined these values were less negative than the 'B s ' values, principally because of the highly positive values of the d 15 N of the nodules.However, a single-tailed 't' test on each value showed that for the B. elkanii strain 29W and the B. japonicum strain CPAC 15 the 'B wp ' values were lower than zero.These results along with those of Okito et al. (2004) suggest that the isotope fractionation associated with N 2 fixation in the intact Bradyrhizobium/soybean symbiosis is significant and may contradict the conclusion of Unkovich & Pate (2000) from their study on chickpea (Cicer arietinum), lupin (Lupinus luteus), lentil (Lens culinaris) and Burr medic (Medicago polymorpha).
The advantage of the technique developed by Doughton et al. (1992), tested by Okito et al. (2004) and used in this study to determine 'B' values, is that plants can be grown in soil in full sunlight either in pots or even in the field.This means that the plants are essentially in the same environment as they would be in a farmer's field.
In most soybean production areas in Brazil farmers establish approximately 320,000 plants per ha and the mean grain yield of this crop in Brazil is close to 2,500 kg ha -1 .Assuming an N content of the grain of 6.5% N, this means that this crop would accumulate 163 kg N ha -1 in the grain, or approximately 500 mg N per plant.As both the proportion of N in the grain compared to the whole shoot (the N harvest index) and the %Ndfa are both usually close to 80% (Alves et al., 2002(Alves et al., , 2003(Alves et al., , 2006)), ideally soybeans used to determine 'B' value should accumulate approximately this amount of N from BNF.In the present study N derived from BNF by the soybeans inoculated with the 4 different Bradyrhizobium strains accumulated between 460 and 620 mg N per plant (910 to 1240 mg N per pot) from BNF (mean %Ndfa = 97%) very close to this value.In the study of Okito et al. (2004) the inoculated soybean plants accumulated between 340 and 675 mg N per plant in the shoot tissue when grown in pots in the field, but in monoaxenic N-free medium in the greenhouse these values were lower, ranging from 290 to 305 mg N per plant.
Estimates to date of 'B' values for soybean reported in the literature have almost all been for shoot Means for the same parameter in the same column followed by the same upper case letter are not different at p < 0.05 (Student 't' test).*Values corrected to discount the 15 N abundance of the original seeds.tissue only ('B s ' values) and all were determined in the greenhouse in monoaxenic N free culture (see review of Boddey et al., 2000).With the exception of the early study of Steele et al. (1983), all reported values were less negative than -1.7 ‰.
The results here presented show that the 'B' values of the shoot tissue ('B s ' values), when determined using plants of realistic size grown in conditions closely approximating those experienced in the field, are considerably more negative than results obtained from determinations in N-free greenhouse culture.Most significantly it was shown that the two B. elkanii strains, recommended for inoculant manufacture in Brazil, induced significantly lower 'B s ' values than the two recommended B. japonicum strains.In many important soybean-growing regions of Brazil, the 15 N natural abundance of plant-available soil N is between +2 and +4 ‰ (Macedo, 2003).
If we assume that the 15 N natural abundance in field-grown soybean and reference crop were -1.00 and -3.00 ‰, respectively, %Ndfa will be 60% using a 'B s ' value of -3.68 ‰ (mean for the two B. elkanii strains) and 82 % if a 'B s ' value of -1.85 ‰ (mean for the two B. japonicum strains).This difference for a total N shoot accumulation of 200 kg N ha -1 , would be 44 kg N ha -1 .This is a very considerable difference and equivalent to the quantity of N fertiliser usually added to wheat following soybean in Brazil.
This study, illustrates the importance of determining which recommended Bradyrhizobium strain is principally responsible for nodule formation if an accurate estimate of the BNF contribution is to be determined.In the case where there is a significant proportion of strains of both Bradyrhizobium species, it may become necessary to determine the proportion of nodule occupancy of each using either immunological or molecular techniques to establish the appropriate intermediate value for 'B s '.Further studies will be required to determine how 'B s ' value varies when nodules are occupied by strains of both B. elkanii and B. japonicum.
Values are means of five replicates.Means within the same column followed by the same upper case letter, or in the same row followed by the same lower case letter are not different at p < 0.05 (Student 't' test).

Table 2 -
Dry matter accumulation by soybean plants (cv.Celeste) inoculated with four different strains of Bradyrhizobium and harvested 87 days after planting.soil was extremely low and would not have any significant contribution to nodule occupancy.
Values are means of five replicates.Means within the same column followed by the same upper case letter, or in the same row followed by the same lower case letter are not different at p < 0.05 (Student 't' test).ns Differences between means not significant at p < 0.05.

Table 3 -
Nitrogen accumulation by soybean plants (cv.Celeste) inoculated with four different strains of Bradyrhizobium and harvested 87 days after planting.
Values are means of five replicates.Means within the same column followed by the same upper case letter are not different at p < 0.05 (Student 't' test).ns Values of %Ndfa calculated from the three different reference plants are not different at p < 0.05.*All values corrected for 15 N content of the seeds.

Table 4 -
15N enrichment* of the shoot tissue of the three reference plants (non-nod soybean, sorghum and rice) and of the soybean (cv.Celeste) inoculated with four different Bradyrhizobium strains grown in soil enriched with 15 N, and the three estimates of the proportion of N derived from BNF (%Ndfa) by the soybean calculated using the 15 N enrichment of the soybean and the three reference plants.

Table 5 -
Total nitrogen derived from BNF in shoot tissue by soybean plants (cv Celeste) inoculated with four different strains of Bradyrhizobium and harvested 87 days after planting.Values are means of five replicates.

Table 6 -
15N natural abundance (‰) of the reference plants and of the soybean (cv.Celeste) inoculated with 4 different strains of Bradyrhizobium, and the 'B' values calculated from these data and the %Ndfa derived from the 15 Nlabelled soil treatment.Plants harvested at 87 days after planting.Values are means of five replicates.