Di-nitrogen fixation at the early and late growth stages of soybean

Soybean derives a significant portion of the required nitrogen (N) from the symbiosis with rhizobia bacteria. However, information on the available genetic variation for N2 fixation capacity in different growth stages of soybean is limited. The objective of this study was to investigate the N2 fixation capacity of 22 soybean lines compared with that of non-nodulating and supernodulating checks at the early and late growth stages and identify the most informative traits for selection. Two cycles of greenhouse experiments were carried out to estimate the percentage of N derived from the atmosphere (%Ndfa) as well as 10 different traits related to N2 fixation. The results showed that %Ndfa was significantly different among the lines at the early and late growth stages. SPAD readings showed the highest correlation with the early N2 fixation, whereas shoot dry weight with the late N2 fixation. Early and late %Ndfa could be used to select superior lines for N2 fixation and study the underlying physiological and molecular mechanism.


Introduction
The nitrogen (N) requirements of soybean are probably the highest among all major crops, since it demands 29 mg of N g -1 photosynthate compared with 11 mg of N g -1 photosynthate in corn and 26 mg of N g -1 photosynthate in cowpea (Sinclair & Wit, 1975).The crop derives 36-69% of its total N through its symbiotic relationship with rhizobia bacteria (Salvagiotti et al., 2008) and the rest from the available soil N.
Yield improvement in soybean is relatively high and averages 31.2 kg ha -1 year -1 .Thus, some concerns have been raised regarding the capacity of the new high-yielding cultivars to reach their yield potential without the input of nitrogenous fertilizers (Specht, Hume, & Kumudini, 1999;Salvagiotti et al., 2008).Moreover, previous studies have suggested a reduction in the N 2 fixation capacity of new cultivars compared with that of old ones (Nicolás, Arias, & Hungria, 2002;Van Kessel & Hartly, 2000).Therefore, the evaluation of lines and traits, aiming to improve the N 2 fixation capacity, is considered crucial (Nicolás et al., 2002).
Multiple evaluations of the N 2 fixation activity of the nodulating line Chippewa and its non-nodulating isoline were performed and unveiled significant variation in the proportion of N derived from fixation at different growth stages (Zapata, Danso, Hardarson, & Fried, 1987).Using the acetylene reduction activity assay for evaluating nitrogenase activity, Pazdernik, Graham, Vance, and Orf (1996) screened 20 soybean lines and identified considerable variation in early nodulation traits and N 2 fixation activity, whereas Fabre and Planchon (2000) concluded that the studied lines differed in N 2 fixation activity at the R5 and R6 growth stages, but not at the R2 growth stage.Herridge, Bergersen, and Peoples, (1990) used the ureide abundance and 15 N natural abundance methods, collecting data at the V8-R6.5 growth stages of seven soybean lines and demonstrated an overall increase in N 2 fixation with the advance of growth cycle; however, the pattern of fixation was genotype-and environment-dependent.
Variations in the N 2 fixation activity throughout the growth cycle of soybean have been attributed to differences in the expression of genes that control nodule development (Kaewsuralikhit, Yokoyama, Kouchi, & Arima, 2005).However, information on the underlying genetic mechanism of symbiosis in soybean is still limited mainly due to the difficulties (Santos et al., 2013) and relatively high cost of accurately measuring the N 2 fixation rate.In this context, a limited number of studies have focused on the study of traits related to N 2 fixation (Nicolás et al., 2002).In addition to nodulation, some other traits, such as photosynthesis parameters, seed weight, seed protein, plant height, and time to maturity, have been used to discriminate soybean genotypes regarding their N 2 fixation capacity (Vollmann, Walter, Sato, & Schweiger, 2011).
Therefore, information on measurements that can help to identify soybean lines with enhanced N 2 fixation capacity throughout the soybean growth cycle is needed.Furthermore, evaluations of different soybean lines at different growth stages are critical for identifying genetic variation for high N 2 fixation.The objectives of this study were to: (1) assess the value of early and late evaluations of soybean lines for high N 2 fixation capacity; (2) test the association of measurements with traits directly and indirectly related to N 2 fixation; and (3) select lines that can be used for studying the underlying mechanisms involved in the genetic control and inheritance of traits responsible for high N 2 fixation capacity.

Plant material
Twenty-two soybean lines of different maturity groups (II-IX) and genetic backgrounds were selected for this study based on their potential N 2 fixation, nodulation, and higher yield in order to investigate the association of growth stages and N 2 fixation (Table 1).Additionally, three checks were included; the nonnodulating soybean cultivars Nitrasoy (Burton, Israel, & Bishop, 2006) and D68-099 (Hartwig, 1994) for estimating the amount of N derived from fixation as well as the supernodulating mutant SS2-2 (Youn et al., 2008;Youn et al., 2009) for evaluating the levels of N 2 fixation.

Greenhouse Experiments
The study was conducted under greenhouse conditions at Southern Illinois University, Carbondale, IL, USA.All the lines were sown in plastic pots (15 cm in diameter; 14 cm in depth) filled with Fafard Growing Mix 2 (Conrad Fafard Inc., Agawam, MA, USA), consisting of 70% peat moss, 20% perlite, and 10% vermiculite.A solution, containing 14 mg of P and 18 mg of K, was applied to each pot prior to sowing.The first cycle of the study was sown on February 6, 2015 and the second on December 12, 2015; plants were then grown until they reached the R7 growth stage (Fehr, Caviness, Burmood, & Pennington, 1971).Three seeds were planted per pot, and two weeks later thinned to a single plant per pot.The lines were assigned to the experimental units following a randomized complete block design with four blocks and four replications within each block.Plants were kept throughout the cycle in a 16-h photoperiod at 26/22 o C day/night.All the pots were watered once a day with a volume of water to prevent leaching of the 15 N fertilizer.

Inoculation with Bradyrhizobium japonicum
The B. japonicum strain USDA 110 was selected to inoculate soybean seeds due to its vastly and long term use in N 2 fixation soybean studies, and no additional strains were applied to prevent any interaction effects.USDA 110 was kindly provided by the USDA Soybean Genomics and Improvement Lab in Beltsville, MD, USA.The supplied sample was grown in liquid modified arabinose gluconate medium, pH 6.6 (Van Berkum, 1990) for 7 days in a shaking incubator until reaching the stationary phase.The inoculation was performed on the seed prior to sowing at a rate of 2 × 10 5 cells seed -1 .

N labelling
A solution of 15 N labeled urea with 10% atom 15 N excess was carefully applied to the growing medium of each pot prior to sowing.The medium was removed individually from each pot and mixed with the labeled solution using a plastic bag to achieve uniform distribution, and then, the labeled medium was returned to each pot.The total amount of labeled N was 7.82 mg N kg -1 of potting medium.A plastic tray was placed underneath each pot to collect any leaching and prevent the 15 N labeled material to spread on the greenhouse benches.

Sampling
Sampling was performed at 35 d after emergence (DAE; early growth stage) and R7 (late growth stage).The upper most fully developed leaf was collected at the early growth stage and pod walls at the late growth stage.These samples were then oven dried at 70 o C for 72h in a convection oven (Isotemp 500; Fisher Scientific, Waltham, MA, USA) and ball milled in steel vials using a 5100 Mixer/Mill (SPEX SamplePrep LLC, Metuchen, NJ, USA) for 15 min.After weighing and packing in tin capsules, the samples were placed in 96well plates and sent to the Mass Spectrometry Facility of Southern Illinois University for the analysis of 15 N enrichment.Prior to the collection of leaf samples at the early growth stage, three measurements of chlorophyll concentration were obtained from the center of the leaflets using a Minolta SPAD-502 chlorophyll meter (Konica Minolta Sensing Inc., Osaka, Japan), and the mean value was recorded as SPAD meter readings (SPD).Shoot, nodule, and seed samples were collected at the late growth stage and oven dried at 70 o C for 72h prior to weighing and recording of shoot dry weight (SDW), nodule number (NN), nodule dry weight (NDW), total seed number (TSN), total seed weight (TSW), and mean seed weight (MSW).The number of days to maturity was recorded as DTM.

N Analysis and Total %N
For estimating 15 N enrichment and total %N, leaf and pod wall samples were analyzed using continuous flow elemental analysis isotope ratio mass spectrometry (CF-EA-IRMS) with a Thermo-Scientific Delta V Plus isotope-ratio mass spectrometer (Bremen, Germany) connected to a Costech 4010 Elemental Combustion System (Costech Analytical Technologies Inc., Valencia, CA, USA) via a Conflo IV unit.The measured 15 N abundance of each sample represented by the parameter atom% 15 N was subtracted from the natural abundance of 15 N in the atmosphere (0.3663 atom% 15 N) to obtain the atom% 15 N excess (Unkovich et al., 2008) in the leaf (LAT%) and pod wall samples (PAT%); the analysis also provided the percentage of N in the leaf (%LN) and pod wall (%PN) samples.The percentage of N derived from the atmosphere (%Ndfa) was estimated as follows (Unkovich et al., 2008): The non-nodulating lines Nitrasoy and D68-099 were used as references.

Statistical Analysis
All statistical analyses were performed using JMP 13 (SAS Institute, Cary, NC, USA).Analysis of variance (ANOVA) was performed in conjunction with Tukey's test to identify differences among lines or Dunnett's multiple comparison test for comparing the %Ndfa of each line at the early and late growth stages with that of the supernodulating check SS2-2.Evaluation cycle (EC) and the interaction between genotype and EC (G × EC) were considered as fixed effects, whereas block as a random effect.
Pearson's correlation was used to determine the relations among measurements and traits.A correlation coefficient was estimated for each pair-wise combination of traits related to N 2 fixation (Table 3) using JMP 13.To analyze the association of these traits with early and late fixation activity, the estimates of leaf (-LAT%) and pod (-PAT%) atom% 15 N excess were included.For these two parameters, the data were transformed using the negative value of each data point.This approach allowed to have an estimation of N 2 fixation activity for all genotypes, including the non-nodulating checks.Since the negative values of atom% 15 N excess were used, all the traits associated with the fixation activity yielded a positive correlation with -LAT% and -PAT%.The Ward's minimum variance method (Ward, 1963) was used for clustering analysis.ANOVA was performed using the minimum variance criterion in the sum of squares to separate the lines in different clusters.These analyses were performed separately for the early and late growth stage and consequently, two dendrograms were constructed.

Assessment of Di-nitrogen Fixation
Measurements of di-nitrogen fixation parameters at the early and late growth stages are presented in Table 1.The estimates of %Ndfa using D68-099 as a check were significantly different from those obtained using Nitrasoy as a check at the early (t = -9.5;p < 0.01) and late growth stages (t = 5.5; p < 0.01).However, differences in ranking were minor, and the estimates for %Ndfa obtained from the two checks were highly correlated at both the early (r = 0.97; p < 0.0001) and late growth stages (r = 0.91; p < 0.0001).The LSmean of atom% 15 N excess was used as an auxiliary criterion for distinguishing the capacity of N 2 fixation of each line (Table 1).
All the estimates of %Ndfa were significantly affected by G (p < 0.0001), EC (p < 0.0001), and G×EC (p < 0.0001), except for that at the late growth stage that was not affected by EC.These results showed that the host control of N 2 fixation in soybean has an important contribution to the overall symbiotic N 2 fixation, but also that this trait is strongly affected by environmental factors.
At the early growth stage, S.J.2 and Davis showed the lowest and highest %Ndfa, respectively, and the estimates varied from 4-31% and 2-29% when using D68-099 and Nitrasoy as a check, respectively.The proportion of N derived from fixation at the early growth stage was in accordance to that reported by George and Singleton (1992), in which the average was 28% for the total N fixed during all the vegetative growth stages (George and Singleton, 1992).The limitation of soybean to supply N through N 2 fixation during the early growth stages has been well documented and can be attributed to the inability of the symbiotic rhizobia to supply all the required N by the crop (Phillips & DeJong, 1984;Keyser & Li, 1992).
At the late growth stage, PI96169B and Bragg showed the lowest and highest %Ndfa, respectively, and the estimates varied from 0-90% and 0-91% when using D68-099 and Nitrasoy as a check, respectively.Except for PI96169B that did not form any nodules with USDA 110, %Ndfa values estimated in the present study were in accordance to those reported by Salvagiotti et al. (2008) that in a review paper of 61 studies on soybean N 2 fixation reported a range of 58-98%.
Our results showed that atom% 15 N excess in leaf (LAT%) and pod walls (PAT%) were significantly influenced by G (p < 0.0001), EC (p < 0.0001), and G × EC (p < 0.0001), except for PAT% that was not affected by EC.Of these parameters, the high values for atom% 15 N excess indicated a reliance of the line on the N derived from the soil instead of that obtained from N 2 fixation.This pattern was confirmed by the higher enrichment of leaf and pod wall samples in both non-nodulating lines.The only exception was LAT% and PAT% of PI96169B, probably due to an incompatibility between the line and USDA 110; however, further studies need to be conducted to confirm this assumption.Davis and S.J.2 showed the lowest and highest LAT%, whereas Bossier and PI96169B showed the lowest and highest PAT%, respectively.
The overall high proportions of N derived from fixation identified at the late growth stage were consistent with those found by Harper (1987) that studied N 2 fixation under low soil-N conditions and reported that N derived from fixation was 80-94%.Although the average contribution of N 2 fixation at the late growth stage (mean %Ndfa = 66.3; data not shown) was markedly higher than that at the early growth stage (mean %Ndfa = 17.3; data not shown), the level of genetic variation followed the opposite trend.These results were in agreement with those reported by Pazdernik et al. (1996) that analyzed 20 soybean lines for early N 2 fixation and nodulation efficiency and suggested that early nodule formation can markedly improve N 2 fixation capacity.
We used the Dunnett's test to compare the N 2 fixation capacity of each line with that of the supernodulating check SS2-2.At the early growth stage, the %Ndfa of nine lines (Davis, Enrei, R01-416F, Williams, PI96171, Osage, Hardee, Bossier, and Ozark) did not show any significant differences from that of SS2-2 (Figure 1), indicating their high N 2 fixation capacity at the early growth stage.The remaining 13 lines yielded a significantly lower %Ndfa compared with that of SS2-2, demonstrating a limited N 2 fixation capacity at the early growth stage.
At the late growth stage, the %Ndfa of 19 lines (Bossier, S.J.2, Davis, Bragg, Hardee, Centennial, JTN-4307, Clark, Ozark, R01-416F, Jake, Saluki 4910, PI471938, JTN-5203, J-200, Williams, Osage, R05-3239, and Jackson) did not show any significant differences from that of SS2-2 (Figure 2).PI96169B, PI96171, and Enrei showed %Ndfa significantly different from that of the supernodulating check, indicating their poor performance in N 2 fixation at the late growth stage.The contribution of N 2 fixation to the total N accumulated during the vegetative period is estimated to be 21%, markedly lower than that at the R3-R7 growth stages that is approximately 56% (Zapata et al., 1987).In the present study, given the substantial variability found among the lines in the measurement of early N 2 fixation, we compared the mean of the two groups of lines discriminated using the Dunnett's test, and a difference of 14% in %Ndfa was found (data not shown); this difference in early N 2 fixation capacity might help to increment the accumulation of N derived from fixation during the vegetative period.
Further studies are needed to investigate the identified restriction of N 2 fixation in the vegetative phase.This limitation is markedly higher in the beginning of growth cycle, and thus, various studies have reported the benefits of N application at the early growth stages of soybean, a practice that is commonly known as 'starter N'.Osborne and Riedell (2006) obtained a 6% yield increase when 16 kg N ha -1 was applied at planting.

Traits related to N 2 fixation
The mean values of the 10 N 2 fixation-related traits (SPD, SDW, TSW, TSN, MSW, L%N, P%N, NN, NDW, and DTM) assessed in the 22 studied lines as well as the non-nodulating and supernodulating checks are presented in Table 2.The results showed that all the 10 N 2 fixation-related traits assessed in this study were significantly (p < 0.001) affected by G, EC, and G×EC, except for MSW that was not affected by G×EC or L%N that was not affected by EC.Among the 10 N 2 fixation-related traits, NN and NDW were the most directly related to nodulation; and the supernodulating check was significantly different from all the studied lines (p < 0.0001) for these two nodulation traits.This indicates that the studied lines had an adequate nodulation, but none of them reached the nodulation pattern of the supernodulating check.Although a superior nodulation ability is desirable for breeding lines with higher N 2 fixation capacity, the level of nodulation displayed by supernodulating mutants can restrict root growth and lead to a 30-40% yield reduction (Day, Lambers, Bateman, Carroll, & Gresshoff, 1986;Gremaud & Harper, 1989;Wu & Harper, 1991;Herridge, 2003).The MSW of Enrei was significantly (p < 0.0001) higher than that of the supernodulating check and all the other lines.The non-nodulating checks Nitrasoy and D68-099 performed poorly or showed the lowest values for P%N, SDW, MSW, L%N, TSW, and TSN.Additionally, Bossier showed the highest SDW, Enrei the highest SPD, and R01-416F the highest P%N.
A previous study that assessed soybean plants under controlled conditions demonstrated that the mean seed weight, total biomass, net photosynthetic output, and total plant N are related to both N 2 fixation and seed yield and that these effects were more pronounced at the pod-fill stage (Imsande, 1989).Analogous traits studied herein also accompanied the N 2 fixation activity among the studied lines; the supernodulating check showed the highest values of TSW, TSN, and L%N, whereas the non-nodulating checks the lowest values, indicating an overall robust relationship of these traits with the N 2 fixation activity.SPD, SPAD readings (mg chlorophyll m -2 ); SDW, shoot dry weight (g plant -1 ); TSW, total seed weight (g plant -1 ); TSN, total seed number per plant; MSW, mean seed weight (mg seed -1 ); L%N, leaf % nitrogen; P%N, pod % nitrogen, NN, number of nodules per plant; NDW, nodule dry weight (mg plant -1 ); DTM, d to maturity.*, **.Significance at p < 0.05 and p < 0.01, respectively.NN, number of nodules per plant; NDW, nodule dry weight (mg plant -1 ); TSW, total seed weight (g plant -1 ); TSN, total seed number per plant; MSW, mean seed weight (mg seed -1 ); DTM, d to maturity; SPD, SPAD readings (mg chlorophyll m -2 ); SDW, shoot dry weight (g plant -1 ); L%N, leaf % nitrogen; P%N, pod % nitrogen; -LAT%, negative leaf atom% 15 N excess; -PAT%, negative pod atom% 15 N excess.

Correlation analysis
A weak, but significant, correlation between the early -LAT% and the late -PAT% (r = 0.15; p < 0.01) suggested significant differences in N 2 fixation profile among the lines, showing that the superior N 2 fixation during the early growth stages did not last until the late growth stages.The strongest correlation was identified between -LAT% and SPD as well between -PAT% and SDW (Table 3).Since N 2 fixation was related with different traits at the early and late growth stages this confirmed the independence of the two evaluations.The variability in the N 2 fixation pattern of each line and the traits related to each of the measurements might be a promising tool for selecting lines with improved N 2 fixation capacity in breeding programs.
The nodulation parameters NN and NDW were found to be highly correlated (r = 0.88; p < 0.01), indicating that only one of those needs to be used for screening lines for N 2 fixation capacity.Additionally, both NN and NDW were moderately correlated with SDW (r =0.39; p < 0.01).This result might reflect the influence of the N derived from the nodules in the biomass accumulation.Mastrodomenico and Purcell (2012) reported that the N derived from fixation is largely accumulated into the biomass; however, a large portion of this N is not remobilized to the seed.
At the late growth stage, -PAT% was moderately correlated with SDW (r = 0.47; p < 0.01), similarly as reported by Herridge et al. (1990), in which P that represented the proportion of N derived from fixation was significantly correlated with the crop dry matter (r = 0.41; p < 0.05); however, the coefficient of correlation between NDW and -PAT% (r = 0.26; p < 0.01) found in the present study was markedly lower than that between nodule weight and P (r = 0.79; p < 0.001).
The coefficient of correlation between NN and -PAT% (r = 0.27; p < 0.01) was similar to that reported by Houngnandan et al. (2008) between nodule number and δ 15 N (r = 0.325*) in a diverse group of soybean cultivars.Additionally, Pazdernik, Graham, and Orf (1997) studied a soybean population to investigate early nodulation traits and found that %Ndfa at R5 was significantly correlated with nodule fresh weight (r = 0.33; p < 0.01).
Correlation analysis also revealed associations with traits that might be used to reduce the cost of population screening for N 2 fixation capacity.SDW was moderately correlated with NN (r = 0.39; p < 0.01) and NDW (r = 0.39; p < 0.01), which are commonly used to improve N 2 fixation in soybean, but can impose practical and economical limitations in the screening of large populations.The correlation between SPD and L%N (r = 0.44; p < 0.01) might also be useful due to the lower cost to assess the former.Additionally, the correlation of TSN with NN (r = 0.33; p < 0.01) and NDW (r = 0.39; p < 0.01) could be advantageous, since the screening of the former is less laborious than that of the latter two.

Cluster analysis
Considerable variation was identified among the 22 soybean lines and the supernodulating check for N 2 fixation at the early and late growth stages, which was also demonstrated by the weak correlation between %Ndfa at the early and late growth stage (r = 0.14; p < 0.001) using both checks.These results confirmed the overall different performance of lines regarding the N 2 fixation capacity at different growth stages.
At the early growth stage, cluster analysis classified the 22 lines and SS2-2 into three different clusters (Figure 3): Cluster 1 included 10 lines with a high N 2 fixation capacity; Cluster 2 included 12 lines with a low N 2 fixation capacity; and Cluster 3 included only SS2-2 that had the highest N 2 fixation activity.
At the late growth stage, the 22 lines and the check SS2-2 were also classified into three different clusters: Cluster 1 included 20 lines with a high N 2 fixation capacity; Cluster 2 included Enrei and PI96171 with a low N 2 fixation capacity; and Cluster 3 included PI96169B that had the lowest N 2 fixation activity.
Soybean is estimated to spend 5.2-18.8g of C g -1 of fixed N 2 (Minchin & Witty, 2005).To meet both the requirements of photosynthate from the plant and the N 2 -fixing rhizobia, the elevated N 2 fixation activity is likely to be accompanied by a higher photosynthetic rate (Kaschuk, Hungria, Leffelaar, Giller, & Kuyper, 2010).Thus, the identification of this pattern may allow to develop new lines without compromising the photosynthate required for other physiological processes.In the present study, the identified variation in N 2 fixation activity during the plant cycle might be associated with the time and pattern of nodule formation on soybean roots.Nodules that are formed during the early growth stages are usually located at the top of the main root and are known to last on average 65 days.Later in the growth cycle, the N 2 fixation activity is maintained by a secondary set of nodules that is usually located on deep and lateral roots (Keyser & Li, 1992;Zapata et al., 1987).In the nodule aging process, the peak of N 2 fixation activity occurs at 4-5 weeks after the infection of plant cells by rhizobia; then, the nodules start to senesce and the rate of N 2 fixation is decreased (Dupont et al., 2012).Nonetheless, further research is needed on the formation and aging of nodules in soybean and its effect on N 2 fixation activity.

Conclusion
Overall, our results revealed significant variation in N 2 fixation capacity among the studied lines and also, between and within the early and late growth stages.The traits that were found closely related to N 2 fixation as well as the lines with contrasting N 2 fixation capacity at the early and late growth stages could be used in future studies to better understand the underlying physiological and molecular mechanisms of N 2 fixation in soybean.

Figure 1 .
Figure 1.Comparison of %Ndfa at the early growth stage (35 d after emergence) between each of the 22 lines and the supernodulating line, SS2-2, using the Dunnett's multiple comparison test.Different letters indicate significant differences between the line and the check.Error bars indicate one standard error of the mean.

Figure 2 .
Figure 2. Comparison of %Ndfa at the late growth stage (R7) between each of the 22 lines and the supernodulating line, SS2-2, using the Dunnett's multiple comparison test.Different letters indicate significant differences between the line and the check.Error bars indicate one standard error of the mean.

Figure 3 .
Figure 3. Clustering of 23 soybean lines and the two non-nodulating lines, Nitrasoy and D68-099, based on %Ndfa at the early (35 days after emergence; left) and late (R7; right) growth stages using the Ward's method.

Table 1 .
Leaf (LAT%)and pod (PAT%) enrichment of 15 N and least squares means for early and late percentage of nitrogen derived from atmosphere (%Ndfa) for 22 lines, one supernodulating line, and two non-nodulating lines grown under greenhouse conditions for two growing seasons in 15 N-labelled soil.

Table 2 .
Comparison of nitrogen fixation-related traits among 22 lines, one supernodulating line, and two non-nodulating lines grown under greenhouse conditions for two growing seasons in 15 N-labelled soil.

Table 3 .
Correlation matrix of 15 N enrichment and N fixation related traits for 22 lines, one supernodulating line, and two nonnodulating lines grown under greenhouse conditions for two growing seasons in 15 N-labelled soil.