Variability in plant microbe interaction between Lupinus lines and Bradyrhizobium strains

Transcription

1 Plant Science 159 (2000) Variability in plant microbe interaction between Lupinus lines and Bradyrhizobium strains Keith O. Robinson a, Desta A. Beyene b, Peter van Berkum c, Regina Knight-Mason a, Harbans L. Bhardwaj d, * a Department of Life Sciences, Virginia State Uni ersity, Petersburg, VA 23806, USA b Uni ersity of Maryland, College Park, MD 20742, USA c USDA-ARS, Belts ille Agricultural Research Center, Belts ille, MD 20705, USA d Agricultural Research Station, Virginia State Uni ersity, PO Box 9061, Petersburg, VA 23806, USA Received 13 March 2000; received in revised form 18 July 2000; accepted 19 July 2000 Abstract Even though lupin (Lupinus albus L.) is known to potentially fix kg/ha nitrogen for the use of a succeeding crop, precise information about lupin Bradyrhizobium strain interaction under the climatic conditions prevalent in the mid-atlantic region of the United States is unknown. We conducted two greenhouse experiments with the objective of characterizing this symbiotic relationship and to evaluate potential interaction between Bradyrhizobium strains and lupin lines. In the first experiment, performance of 60 bradyrhizobial strains was evaluated by inoculating three lupin cultivars and using combined score, which consisted of an arithmetic total of plant vigor, nodulation scores from crown root, nodulation scores from fibrous roots, shoot dry weight, and root dry weight. In the second experiment, performance of 80 lupin lines was evaluated by inoculating with three selected Bradyrhizobial strains and using the combined score, which consisted of an arithmetic total of plant vigor, acetylene reduction activity, nodule number per plant, nodule weight per plant, and dry shoot weight. Significant variation existed for all traits in both experiments except for nodule number in the second experiment. Significant Bradyrhizobial strain by lupin line interaction existed for nodulation score, shoot and root dry weights, and the combined scores. Comparison of relative ranks indicated that nodulation effectiveness was dependent on specific strain and lupin line combinations. It was concluded that specific Bradyrhizobial strain and lupin line combinations would need to be identified for successful utilization of lupin s capability to fix atmospheric nitrogen for use in low-input and sustainable agriculture Elsevier Science Ireland Ltd. All rights reserved. Keywords: Nitrogen fixation; Legumes; Sustainable crop production 1. Introduction Contribution of Virginia State University and US Department of Agriculture. Journal Article No The use of trade names or vendors does not imply approval to the exclusion of other products or vendors that may also be suitable. * Corresponding author. Tel.: ; fax: address: hbhardwj@vsu.edu (H.L. Bhardwaj). There is renewed interest in lupin (Lupinus albus L.) as a crop for use in sustainable and environment-friendly agriculture because of its high potential for biological nitrogen fixation, which may contribute kg nitrogen per ha for use by the succeeding crops [1 3] and its potential in providing nutritious food and feed (32 38% protein). The nitrogen-fixation potential of lupin makes it suited ideally to low-input sustainable and environment friendly agriculture because mineral nitrogen fertilizer may not be needed for crop production. Lupin, being a legume, benefits potentially from a nitrogen-fixing symbiosis with soil bacteria. These soil bacteria belong usually to the genus Bradyrhizobium [4,5] /00/$ - see front matter 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S (00)

2 258 K.O. Robinson et al. / Plant Science 159 (2000) Several species of lupin have potential in agriculture and within each species there is significant genetic diversity [6]. The different traits within lupin have not been explored to identify cultivars or lines, which are most suitable for crop production in the mid-atlantic region of the USA. The identification of suitable cultivars and lines is important because the potential benefit from nitrogen fixation may be influenced by the plant genotype and bradyrhizobial strain interaction [7]. Therefore, our objective was to evaluate the interaction between Bradyrhizobium strains and lupin lines for nodulation and plant performance Bradyrhizobial material In the first experiment, we used 60 strains of Bradyrhizobium (Table 3) whereas the second experiment was conducted with three selected bradyrhizobial strains (USDA 3044, 3051, and 4114). These strains were obtained from the National Rhizobium Culture Collection, maintained by US Department of Agriculture in Beltsville, MD. The bradyrhizobial strains were grown in 10 ml modified arabinose-gluconate (MAG) broth [8] at room temperature to a density of approximately 10 9 cells per ml before they were used as liquid inoculants at the time of sowing. 2. Material and methods 2.1. Plant material These studies were conducted with three cultivars (L. albus cv Lunoble, Lupinus angustifolius cv Gungurru, and Lupinus luteus cv Juno) in the first experiment and 80 lines of L. albus in the second experiment (Table 4) Experimental conditions The first experiment was arranged as a factorial design of the three lupin cultivars and the 60 bradyrhizobial strains and was conducted in a greenhouse with supplemental lighting at the Beltsville Agricultural Research Center, USDA, Beltsville, MD. Lupin seeds were sterilized by Table 1 Analysis of variance (mean squares) for lupin plant characteristics following inoculation of three lupin cultivars with 60 Bradyrhizobium lupini strains a Source Plant vigor Nodulation score Dry weight Combined score Crown Roots Shoot Root Strain (S) 0.91** 7.59** 3.31** 0.52** ** Cultivars (C) 6.71** 62.07** 22.20** 11.68** 0.77** ** S C ** 0.88** 0.23* 0.05** 3.13* Error Mean S.E a *, **, Significant at 5 and 1% levels, respectively. If the strain line interaction was significant, the significance of strains and cultivars was obtained by using strain line interaction mean square as the error term. Table 2 Analysis of variance for lupin plant characteristics following inoculation of 80 lupin lines with three Bradyrhizobium strains a Source of variation Acetylene reduction activity Plant vigor Nodule Plant weight Combined score Weight Lines 59.53* 0.60*** * 0.20*** ** Strains *** 3.09*** *** 0.003* 3.63*** *** Error Mean S.E a *, **, ***, Significant at 10, 5, and 1% level, respectively. The line strain interaction was used as an error term.

3 K.O. Robinson et al. / Plant Science 159 (2000) Table 3 Overall performance of 60 Bradyrhizobium strains, based on combined score, and relative rank when used to inoculate three selected lupin cultivars a Strain Combined score b Gungurru c Juno d Lunoble e Mean 1 USDA (19) 11.2 (13) 7.5 (58) USDA (20) 10.5 (26) 8.2 (49) USDA (12) 9.6 (37) 8.6 (44) USDA (34) 8.6 (48) 8.5 (46) USDA (1) 12.8 (1) 9.2 (29) USDA (15) 11.3 (10) 10.9 (6) USDA (8) 11.2 (12) 11.2 (2) USDA 3047a 10.2 (4) 9.7 (35) 8.4 (47) USDA (2) 11.8 (6) 9.3 (27) USDA (16) 11.3 (11) 10.4 (12) USDA (26) 11.6 (9) 11.1 (4) USDA (39) 8.0 (53) 7.3 (59) USDA (35) 11.7 (7) 9.5 (20) USDA (50) 5.9 (60) 7.9 (55) USDA 3054a 10.1 (5) 9.9 (30) 10.1 (15) USDA (17) 10.8 (20) 10.6 (10) USDA 3057a 7.1 (37) 8.6 (49) 8.7 (42) USDA (6) 11.6 (8) 10.5 (11) USDA (42) 12.6 (2) 9.7 (18) USDA (45) 12.2 (5) 10.8 (8) USDA (30) 11.0 (18) 9.5 (21) USDA (48) 8.1 (52) 8.1 (50) USDA (51) 9.9 (33) 7.6 (57) USDA 3063a 5.7 (49) 9.9 (32) 8.0 (52) USDA (11) 11.0 (16) 9.1 (33) USDA (7) 10.5 (25) 8.6 (43) USDA 3514 (.) f 9.9 (34) 9.3 (28) USDA 3644 (.) f 10.1 (29) 9.1 (35) USDA (22) 11.0 (17) 10.4 (13) USDA (47) 6.9 (59) 7.9 (54) USDA (24) 9.9 (31) 9.2 (31) USDA (52) 7.3 (57) 8.1 (51) USDA 3714 (.) f 9.4 (39) 9.1 (36) USDA (21) 9.0 (45) 7.9 (53) USDA (46) 7.7 (55) 8.6 (45) USDA ( 9) 12.3 (3) 10.3 (14) USDA 3930 (.) f 12.3 (4) 9.2 (30) USDA (31) 8.8 (47) 8.8 (41) USDA 3932 (.) f 9.1 (43) 8.9 (40) USDA (27) 6.9 (58) 6.1 (60) USDA (18) 10.8 (21) 9.9 (17) USDA (28) 10.7 (23) 9.1 (34) USDA (29) 10.2 (28) 9.5 (23) USDA (41) 9.2 (42) 8.4 (48) USDA (10) 9.2 (41) 10.8 (9) USDA (13) 9.4 (38) 9.5 (24) USDA (33) 10.8 (22) 11.2 (3) USDA (43) 8.9 (46) 9.7 (19) USDA (32) 11.1 (14) 8.9 (38) USDA (25) 8.5 (50) 9.2 (32) USDA (36) 9.7 (36) 11.4 (1) USDA 4114 (.) f 11.1 (15) 9.0 (37) USDA (44) 10.6 (24) 8.9 (39) 8.6

4 260 K.O. Robinson et al. / Plant Science 159 (2000) Table 3 (Continued) Strain Combined score b Gungurru c Juno d Lunoble e Mean 54 USDA (3) 9.2 (40) 10.0 (16) USDA (23) 7.7 (54) 10.9 (7) USDA 4118 (.) f 9.1 (44) 9.4 (25) USDA 4119 (.) f 7.6 (56) 11.0 (5) USDA (14) 8.4 (51) 9.4 (26) USDA (38) 10.3 (27) 9.5 (52) USDA (40) 10.9 (19) 7.8 (56) 8.5 Mean a LSD (5%) for comparing lupin cultivars, 0.40; LSD (5%) for comparing Bradyrhizobium strain averaged over three lupin cultivars, 1.81; LSD (5%) for comparing Bradyrhizobium strains within Gungurru, 3.13; LSD (5%) for comparing Bradyrhizobium strains within Juno, 2.87; LSD (5%) for comparing Bradyrhizobium strains within Lunoble, b An unweighted total of plant vigor score, score based on nodulation on the crown of the root, score based on nodulation on fibrous roots, root dry weight, and stem dry weight. c L. angustifolius. d L. luteus. e L. albus. f The plants did not survive. soaking for 5 min in 1% bleach and then were rinsed five times with sterile distilled water. These surface-sterilized seeds were held at room temperature between sterile paper towels moistened with sterile distilled water until they germinated. The germinated seeds were planted in furrows in autoclaved vermiculite, which had been placed in 20- cm sterilized plastic pots and had been moistened with full strength N-free plant nutrient solution [9]. Each germinated seed was inoculated with 0.5 ml of a bradyrhizobial suspension. Each lupin line bradyrhizobial strain combination was replicated three times. The pots were watered with sterile distilled water as needed. The second experiment, a factorial design of three bradyrhizobial strains (USDA 3044, 3051, 4114) and 80 lines of L. albus, was also conducted in a greenhouse with supplemental lighting at the Beltsville Agricultural Research Center. The protocols used in the second experiment were similar to those used in the first Data collection and analysis In the first experiment, the plants were grown for 52 days before they were turned out of the pots and the vermiculite was washed off the roots. Each plant was assigned a vigor score from 1 to 3 with 1, poor; 2, intermediate; and 3, vigorous plant growth. A nodulation score was also assigned separately to each plant based on nodules at the crown of each root and those located on the fibrous roots with 1, no nodulation; 2, a few small nodules; 3, many small nodules; 4, a few large nodules; and 5, many large nodules. Shoot and root dry weights were also recorded. In the second experiment, the plants were grown for 54 days before harvesting. After harvesting, each plant was given a vigor score and nodulation score similar to that for the first experiment. The numbers of nodules, nodule dry weight, and shoot dry weight were recorded. The acetylene reduction activity (ARA) as described by [10] was used to compare nitrogenase activities by the strain and lupin line combinations. The nodulated roots were separated from shoots at the cotyledonary nodes and were enclosed quickly in 250 ml mason jars and incubated with 10% acetylene for 15 min before collecting 0.5 ml gas samples for determinations of ethylene as described by [11]. A selection index, referred to as the Combined Score, was developed to determine the efficacy of each of the 60 bradyrhizobial strain with each of the three lupin cultivars in the first experiment and to compare the 80 lupin lines in the second experiment. In the first experiment, the combined score consisted of an arithmetic total of plant vigor score (value from 1 to 3), nodulation scores from crown root (value from 1 to 5), nodulation scores from fibrous roots (value from 1 to 5), shoot dry weight in grams, and root dry weight in grams. As

5 K.O. Robinson et al. / Plant Science 159 (2000) Table 4 Performance of 80 lupin lines, based on combined score, and relative rank when inoculated with three selected Bradyrhizobium strains a Line Combined score a USDA3044 USDA3051 USDA4114 Mean 1 PI (70) (2) (12) PI (71) (29) (36) PI (7) (25) (18) PI (8) (38) (33) PI (5) (10) (4) PI (30) (6) (5) PI (64) (8) (9) PI (2) (11) (1) PI (11) (9) (15) PI (67) (7) (16) PI (76) (19) (6) PI (63) (12) (11) PI (60) (37) (48) PI (50) (30) (20) PI (53) (35) (13) PI (41) (22) (23) PI (20) (18) (39) PI (39) (15) (17) PI (29) (21) (2) PI (33) (36) (26) PI (35) (34) (29) PI (25) (54) (66) PI (1) (47) (63) PI (40) (56) (64) PI (4) (1) (47) PI (9) 1.52 (77) (71) PI (3) 7.16 (45) (40) PI (77) (20) (73) PI (6) 7.57 (43) (38) PI (80) (13) (27) PI (13) (33) (30) PI (27) 1.47 (78) (49) PI (47) 6.25 (51) (14) PI (23) 8.79 (40) (7) PI (14) 4.09 (60) (54) PI (45) (17) (10) PI (19) 3.43 (64) (19) PI (10) 3.41 (66) (44) PI (37) 6.85 (48) (22) PI (28) 4.46 (59) (3) PI (18) (4) (46) PI (54) (24) (25) PI (16) 3.89 (62) (8) PI (34) 3.29 (68) (28) PI (15) 5.35 (57) (21) PI (32) (26) (42) PI (24) (23) (69) PI (46) (31) (43) PI (52) (28) (60) PI (48) 4.09 (61) (76) PI (44) 6.92 (46) (55) PI (12) 1.86 (72) (59) PI (59) 1.60 (75) (75) 03.75

6 262 K.O. Robinson et al. / Plant Science 159 (2000) Table 4 (Continued) Line Combined score a USDA3044 USDA3051 USDA4114 Mean 54 PI (17) 1.60 (74) (80) PI (75) 3.33 (67) (68) PI (62) 1.53 (76) (35) PI (21) 3.03 (69) (31) PI (65) 1.62 (73) (57) PI (56) (14) (56) PI (61) 1.42 (79) (79) PI (22) 1.31 (80) (51) PI (26) (3) (52) PI (51) 6.12 (53) (50) PI (49) (16) (78) PI (38) 6.61 (49) (37) PI (58) 6.15 (52) (32) PI (43) 2.86 (70) (34) PI (36) 4.65 (58) (74) PI (57) 2.31 (71) (58) PI (31) 8.87 (39) (41) PI (72) 5.49 (55) (77) PI (42) 8.29 (41) (65) PI (68) (32) (72) PI (79) (5) (62) PI (66) (27) (53) PI (55) 3.48 (63) 5.86 (67) PI (74) 7.61 (42) (61) PI (78) 6.50 (50) (70) Lycayne 2.52 (73) 3.42 (65) (24) LUNABLE 2.83 (69) 7.55 (44) (45) Mean a LSD (5%) for comparing lupin line means, 15.36; LSD (5%) for comparing bacterial strain means, an example, for the combination of Gungurru with strain 3040, the values for plant vigor score, nodulation scores from crown root, nodulation scores from fibrous roots, shoot dry weight in grams, and root dry weight in grams for replication 2 were, 2, 3, 2, 0.332, and g, respectively, thus, leading to a combined score of ( =8.182). In the second experiment, the combined score consisted also of an arithmetic total of plant vigor score (value from 1 to 3), ARA (The amount of ethylene expressed as mol per jar per h), nodule number per plant, nodule weight per plant in grams, and dry shoot weight in grams. All data on individual variables and combined scores were analyzed using generalized linear model procedure in SAS [12]. 3. Results and discussion 3.1. E aluation of bradyrhizobial strains Due to differential plant survival in the first experiment, data were available only for 52 bradyrhizobial strains with Gungurru (L. angustifolius). An evaluation of 60 bradyrhizobial strains was possible for the other two lupin cultivars. For the statistical analysis, the significance of bradyrhizobial strains and lupin cultivars was tested by using the interaction mean squares as the error term. Significant interaction was detected between bradyrhizobial strains and lupin cultivars for nodulation of the crown and fibrous roots, dry weights of shoots and roots, and the combined score (Table 1). The variation among the lupin

7 K.O. Robinson et al. / Plant Science 159 (2000) cultivars for root dry weight and bradyrhizobial strains was not significant (Table 1) except for root dry weight for bradyrhizobial strains (Table 2). Based on the combined score (Table 3), Juno was most efficient lupin cultivar for nodulation with a mean score of 9.9, followed by Lunoble with a mean score of 9.3. Within Juno, the score varied from 5.9 (USDA 3054) to 12.8 (USDA 3044). Within Lunoble, the score varied from 6.1 (USDA 3935) to 11.4 (USDA 4113). Within Gungurru, the score varied from 4.3 (USDA 3712) to 11.1 (USDA 3044). The combined score averaged over three lupin cultivars varied from 6.4 (USDA 3054) to 11.0 (USDA 3044). Based on the combined analysis, which included all strains and three lupin cultivars and used strain lupin cultivar interaction mean squares as the error term, the most efficient five strains of Bradyrhizobium were, USDA 3044, USDA 3930, USDA 3058, USDA 4114, and USDA The five least efficient strains of Bradyrhizobium were, USDA 3054, USDA 3712, USDA 3935, USDA 3710, and USDA E aluation of lupin lines Significant variation existed among lupin lines and bradyrhizobial strains for all the variables tested except for nodule number in an evaluation of 80 lupin lines inoculated with three strains of Bradyrhizobium (Table 4). It was not possible to estimate line strain interaction or to derive the error variance because of lack of replications. Therefore, the line strain mean squares were used as the error term. The combined scores for 80 lupin lines varied from 2.7 to 38.0 (Table 4). The highest ranking five lupin lines were, PI , PI , PI , PI , and PI The combined score of these five lines varied from 38.0 to 24.6 whereas the mean combined scores for Lunoble and Lycayne were 8.3 indicating that these five lines potentially are desirable over the current lupin cultivars under conditions of nitrogen fixation and might be useful as parents in a breeding program. However, the data were obtained in a greenhouse study and similar evaluations under field conditions would be needed Bradyrhizobial strain and lupin line specifity Results of both experiments indicated specificity between Bradyrhizobium strains and lupin lines. In the first experiment (Table 3), the strain USDA 3040 had a combined score of 8.1 (ranked 19), 11.2 (ranked 13), and 7.5 (ranked 58), for Gungurru, Juno, and Lunoble cultivars, respectively. Similarly, the combined score for USDA 3044 also varied depending upon the lupin line. In this case, the combined scores were 11.1 and 12.8 when evaluated with Gungurru (ranked 1) and with Juno (ranked 1), respectively, but only 9.2 (ranked 29) with Lunoble. In the second experiment, lupin line PI had the highest mean score of This lupin line had a combined score of (ranked 2), (ranked 11), and (ranked 1) when evaluated with USDA 3044, 3051, and 4114, respectively. The lupin line PI had a combined score of (ranked 1), 6.89 (ranked 47), and 6.56 (ranked 63) when evaluated with USDA 3044, 3051, and 4114, respectively. In general, it was not uncommon for the combined score to fluctuate depending upon the combination of bradyrhizobial strain and lupin line (Table 4). In the majority of cases, the lupin lines nodulated satisfactorily with most bradyrhizobial strains. The combined score of PI line was (ranked 5), (ranked 10), and (ranked 4) when evaluated with USDA 3044, 3051, and 4114, respectively. Similar results were also obtained with PI There is little information about the potential variation for nitrogen fixation among lupin cultivars and lines as affected by the bradyrhizobial strain. However, information about soybean [Glycine max (L.) Merr.] is available [13 16] and the principles with this crop may be of relevance to lupin. For instance, [17] observed a significant interaction between soybean cultivars and bradyrhizobial strains and concluded that breeding programs for soybean should consider the interaction between cultivars and strains as well as the soybean cultivars and bradyrhizobial strains separately. This consideration may be of relevance to lupin since variation for nitrogen fixation ability among bradyrhizobial strains has been reported [18] and since we have found significant variation in lupin when inoculated with the 60 bradyrhizobial strains. Furthermore, [7] reported significant lupin species bradyrhizobial strain interaction in a greenhouse experiment similar to our observations in this study. The data presented indicate that the most important aspect of lupin bradyrhizobial interaction is not to identify the superior lupin lines or bradyrhizobial strain but rather to emphasize the specificity between the host and the bacteria. Even though

8 264 K.O. Robinson et al. / Plant Science 159 (2000) some lupin cultivars might be chosen as a reasonable choice in a field situation where inoculant is not controlled, but a different choice might be made where both the host and bacteria could be specified. Therefore, we conclude that if the resurgence of lupin as a nitrogen-fixing component in sustainable crop production is to be successful, specific lupin cultivar and bradyrhizobial strain combinations need to be identified. These results also imply that breeding programs to develop high yielding lupin cultivars, adapted to the mid-atlantic region, should use the most suitable bradyrhizobial strain as inoculant to precisely determine lupin productivity for sustainable crop production systems. Acknowledgements Assistance of Charles Simon (USDA-ARS, Pullman, Washington, USA), and Blaine Schatz (Carrington Research Extension Center, North Dakota State University, Carrington, ND) for providing seeds and technical assistance is acknowledged gratefully. Major author acknowledges the financial support from M.S. to Ph.D. program of Virginia State University and resources of Agricultural Research Station, Virginia State University. References [1] E. van Santen, D.W. Reeves, G.L. Mullins, White lupin, a potential new crop for Alabama. Alabama Agriculture Expansion Station, Highlights Agric. Res. 40 (1994) 15. [2] D.W. Reeves, Experiences and prospects for lupin in the South and Southeast, in: Prospects for Lupin in North America, Proceedings of the Symposium Sponsored by the Center for Alternative Plant and Animal Products, University of Minnessota, St. Paul, MN, March 21 22, 1991, pp [3] D.W. Reeves, J.T. Touchton, R.C. Kingery, The use of lupin in sustainable agriculture systems in the Southern Coastal Plain, in: Abstracts of Technical Papers, No. 17, Southern Branch ASA, Little Rock, AR, February 3 7, 1990, p. 9. [4] P.J. Bottomley, H.H. Cheng, S.R. Strain, Genetic structure and symbiotic characteristics of a Bradyrhizobium population recovered from a pasture soil, Appl. Environ. Microbiol. 60 (1994) [5] L.L. Barrera, M.E. Trujillo, M. Goodfellow, F.J. Garcia, G. Davila, P. van Berkum, E. Martinez-Romero, Genetic diversity of bradyrhizobia nodulating Lupinus spp., Int. J. Syst. Bacteriol. 47 (1997) [6] L.A. Field, D.H. Putnam, Crop description, growth and development, in: R.A. Meronuck, M. Harvey, D.H. Putnam (Eds.), Lupin Production and Utilization Guide, Center for Alternative Plant and Animal Products, University of Minnesota, St. Paul, MN, 1993, pp [7] B. Lagacherie, M. Bours, J.J. Giraud, G. Somer, Interaction between Rhizobium lupini strains and species or cultivars of lupin (Lupinus albus, Lupinus luteus and Lupinus mutabilis), Agron. Sci. Prod. Veg. Environ. Paris 3 (1983) [8] P. van Berkum, Evidence for a third uptake hydrogenase phenotype among the soybean bradyrhizobia, Appl. Environ. Microbiol. 56 (1990) [9] D.O. Norris, Techniques used in the work with Rhizobium. Commonwealth Bureau of Pasteures and Field Crops, Hurley Berkshire Bull. 47 (1964) [10] R.W.F. Hardy, R.D. Holsten, E.K. Jackson, R.G. Burns, The acetylene ethylene assay for N 2 -fixation: laboratory and field evaluation, Plant Physiol. 43 (1968) [11] P. van Berkum, C. Sloger, Immediate acetylene reduction by excised grass roots not previously pre-incubated at low oxygen tensions, Plant Physiol. 64 (1979) [12] SAS, SAS System for Windows, SAS Institute Inc., Cary, NC, [13] J.E. Harper, Soil and symbiotic nitrogen requirements for optimum soybean production, Crop Sci. 14 (1974) [14] S.L. Neuhausen, P.H. Graham, J.H. Orf, Genetic variation for dinitrogen fixation in soybean of maturity group OO and O, Crop Sci. 28 (1988) [15] T.R. Sinclair, A.R. Soffes, K. Hinson, S.L. Albrecht, P.L. Pfahler, Genotypic variation in soybean nodule number and weight, Crop Sci. 31 (1991) [16] M.L. Ferrey, P.H. Graham, M.P. Russelle, Nodulation efficiency of Bradyrhizobium japonicum strains with genotypes of soybean varying in the ability to restrict nodulation, Can. J. Microbiol. 40 (1994) [17] D.L. Pazdernik, P.H. Graham, C.P. Vance, J.H. Orf, Host genetic variation in the early nodulation and dinitrogen fixation of soybean, Crop Sci. 36 (1996) [18] K.K. Ayisi, D.H. Putnam, C.P. Vance, P.H. Graham, Bradyrhizobium inoculation and nitrogen fertilizer effects on seed yield and protein of white lupin, Agron. J. 84 (1992)