APPLIED AND ENVIRONMENTAL MICROBIOLOGY, May 1991, p. 1540-1545 Vol. 57, No. 5 0099-2240/91/051540-06$02.00/0
Copyright ? 1991, American Society for Microbiology
Subgroups of the Cowpea Miscellany: Symbiotic Specificity within
Bradyrhizobium spp. for Vigna unguiculata, Phaseolus lunatus,
Arachis hypogaea, and Macroptilium atropurpureumt
JANICE E. THIES, B. BEN BOHLOOL,* AND PAUL W. SINGLETON
NifTAL Project, University of Hawaii, 1000 Holomua Avenue, Paia, Hawaii 96779-9744
Received 4 December 1990/Accepted 19 February 1991
Rhizobia classified as Bradyrhizobium spp. comprise a highly heterogeneous group of bacteria that exhibit differential symbiotic characteristics on hosts in the cowpea miscellany cross-inoculation group. To delineate the
degree of specificity exhibited by four legumes in the cowpea miscellany, we tested the symbiotic characteristics of indigenous cowpea bradyrhizobia on cowpea (Vigna unguiculata), siratro (Macroptilium atropurpureum), lima bean (Phaseolus lunatus), and peanut (Arachis hypogaea). The most-probable-number counts of indigenous bradyrhizobia at
three sites on Maui, Hawaii, were substantially different on the four hosts: highest on siratro, intermediate on cowpea, and significantly lower on both lima bean and peanut. Bradyrhizobia from single cowpea nodules from the most-probable-number assays were inoculated onto the four hosts. Effectiveness patterns of these rhizobia on cowpea followed a normal distribution but were strikingly different on the other legumes. The effectiveness profiles on siratro
and cowpea were similar but not identical. The indigenous cowpea-derived bradyrhizobia were of only moderate effectiveness on siratro and were in all cases lower than the inoculant-quality reference strain. Between 5 and 51 % of
the bradyrhizobia, depending on site, failed to nodulate peanut, whereas 0 to 32 % failed to nodulate lima bean. No significant correlation was observed between the relative effectiveness of the bradyrhizobia on cowpea and their corresponding effectiveness on either lima bean or peanut. At all sites, bradyrhizobia that were ineffective on cowpea but that effectively nodulated lima bean, peanut, or both were found. Eighteen percent or fewer of the bradyrhizobia were as effective on lima bean as the reference inoculant strain; 44% or fewer were as effective on peanut as the reference strain. Only 18% of all cowpea-derived bradyrhizobia tested were able to form N,-fixing nodules on both
lima bean and peanut. These results indicate the need to measure indigenous bradyrhizobial population characteristics
directly with the crop of interest to obtain an accurate assessment of the need to inoculate.
Rhizobial classification is based first and foremost on host specificity on the part of these hosts or whether overlap exists specificity. In the genus Bradyrhizobium, the bacteria that nodulate between the various groups. soybean are classified as Bradyrhizobium japonicum and all others Size and effectiveness of indigenous rhizobial populations are are assigned to the Bradyrhizobium spp., previously referred to as primary factors that determine the incidence and magnitude of cowpea rhizobia or more loosely as tropical rhizobia. This has been legume inoculation response (11, 14). Singleton and Tavares (11) the source of misconceptions about the promiscuity of tropical isolated indigenous bradyrhizobia from cowpea, lima bean, and legumes and the ubiquity of tropical bradyrhizobia (10) and has led peanut growing in four Hawaiian soil samples and tested the to false recommendations regarding the inoculation requirements of effectiveness of the bradyrhizobia on the legumes from which they tropical legumes. For example, cowpea (Vigna unguiculata), lima were isolated. They found that, within a soil sample, the range of bean (Phaseolus lunatus), peanut (Arachis hypogaea), and siratro effectiveness of indigenous rhizobial isolates for these hosts (Macroptilium atropurpureum) are all thought to nodulate with the differed. Their results indicated that considerable diversity in the cowpea rhizobia, now classified as Bradyrhizobium spp. But relative effectiveness of indigenous Bradyrhizobium spp. individual bradyrhizobial isolates may not be equally infective or populations on these legumes exists. Such population diversity is effective on these four host legumes. Observations of differential also reflected in measured differences in the sizes of indigenous effectiveness led Burton (2) to separate these hosts into bradyrhizobial populations capable of nodulating these legumes (11, effectiveness groupings within the cowpea miscellany. Cowpea and 14), differences in the ability of bradyrhizobia to compete with siratro were included in one group, while lima bean and peanut inoculant strains for nodulation of the different hosts (14), and composed separate groupings. Under this system, it is assumed that differences in the incidence and magnitude of inoculation responses legume hosts within a given effectiveness grouping have similar obtained on these hosts in the same soil samples (11, 14). bradyrhizobial requirements. While these divisions may be useful In their studies, Singleton and Tavares (11) did not characterize for making strain recommendations and formulating an inoculant, the effectiveness of indigenous bradyrhizobial isolates from any they fail to indicate the degree of one of the hosts on the others. Hence, the nature of observed differences in the range of effectiveness of these isolates could not
be determined. In this study, we examined the nature of these †differences by assessing the degrees of specificity, in terms of both * Corresponding author. Journal series no. 3521 of the Hawaii nodulation and effectiveness, exhibited by siratro, lima bean, and Institute of Tropical Agriculture and Human Resources. peanut
SYMBIOTIC SPECIFICITY OF BRADYRHIZOBIA FOR LEGUMES 1541 VOL. 57, 1991
TABLE 1. Location and characteristics of three sites on the island of Maui, Hawaii"
pH Legume genera Site no. Elevation Soil Median annual rainfall` (mm/year) and name (m) subgroup' present at site 6.8 Leucaena, Prosopis Torroxic 1, Hashimoto Farm 37 322 7.5 Leucaena, Indigofera, 375 2, Kula Agricultural Park 366 haplustoll Macroptilium, Torroxic Prosopis haplustoll 5.3 Desmodium, Trifolium, 3, Haleakala Station 660 Humoxic tropohumult 1,800 Acacia, Crotalaria ? For more detailed descriptions of these sites, see reference 18. b From reference 10. C From reference 3.
when inoculated with bradyrhizobia from cowpea nodules. Our each test host. Nodulation characteristics were recorded, and leaf
results further substantiate division of the cowpea miscellany into chlorophyll content (chlorophyll a plus chlorophyll b) on six leaf disks (diameter, 0.635 cm) per plant (7), taken from the most effectiveness subgroups, particularly as they relate to inoculation response, and illustrate that measurement of indigenous recently fully expanded trifoliate leaf 32 days after inoculation for Bradyrhizobium spp. population characteristics should be made cowpea and lima bean and 41 DAI for peanut and siratro, was determined. directly with the host of interest to obtain an accurate assessment of the symbiotic capability of the population. Data analysis. Effectiveness of individual nodule bradyrhizobia
in symbiosis with the test hosts was divided into four categories: highly effective, effective, moderately effective, and ineffective.
The bradyrhizobia were considered ineffective if the chlorophyll MATERIALS AND METHODS content of host plant leaf disks was within the 95% confidence interval for the chlorophyll content of uninoculated (nonnodulated) Soil sampling and enumeration of indigenous Bradyrhizo-control plants; moderately effective if host leaf disk chlorophyll bium spp. populations. Soil samples were collected from fallow content was higher than the upper confidence limit for areas at three field sites on the island of Maui, Hawaii (Table 1). uninoculated control plants but less than the lower confidence limit After the top 1 cm of soil was removed, 20 to 25 for chlorophyll content of plants nodulated by the reference strains 2.54-cm-diameter soil cores to a depth of 25 cm were taken in a TAL 658 for cowpea, peanut, and siratro and TAL 644 for lima grid pattern around each field area. Soil cores were pooled, mixed, bean; effective if leaf disk chlorophyll content was within the 95% subsampled for determination of moisture content, and stored at confidence interval; and highly effective if leaf chlorophyll content 4?C. The most-probable-number (MPN) of indigenous was higher than the upper confidence limit for chlorophyll content bradyrhizobia in each soil sample was determined (for four test of plants inoculated with the known reference strains listed above. hosts: V. unguiculata cv. Knuckle Purplehull, P. lunatus cv. Kendall tau b rank correlation analysis (8) of host leaf chlorophyll Henderson's Baby, A. hypogaea cv. Burpee Florunner, and M. content was used to assess the degree of relatedness between the atropurpureum cv. Siratro. Serial 1:2, 1:5, or 1:10 soil dilutions effectiveness of nodule occupants on cowpea and their corre-were prepared and inoculated onto four replicate test plants per sponding effectiveness on siratro, peanut, and lima bean. dilution growing in either plastic growth pouches (cowpea, lima
bean, and peanut) or test tubes (siratro) (13). Plants in pouches were kept supplied with an adequate volume of N-free nutrient
solution (9), except that micronutrients were supplied by adding a RESULTS commercial micronutrient mix (0.25 ml liter-1; Hawaiian Horticultural). Plants were scored for nodulation 21 to 28 days At all sites, MPN counts of indigenous Bradyrhizobium spp. after inoculation. The MPN counts were determined by using the were highest on siratro (Table 2). Counts of indigenous Most-Probable-Number Enumeration System (17). bradyrhizobia were 2.4- to 18.6-fold lower when cowpea was used Assay for the effectiveness of indigenous Bradyrhizobium spp. as the trap host. Population counts on peanut and lima bean were A representative sample of nodules was taken from each of the from 6- to more than 1,000-fold lower than those obtained on MPN assays performed on cowpea. Nodules were selected from siratro and from 2.3- to 62-fold lower than those obtained on all dilutions when present. Three or more replicate nodules per test cowpea. host formed by inoculant-quality reference strains TAL 644 (CIAT
257) and TAL 658 (CIAT 71) were used as positive controls. Nodules were surface sterilized by immersion in 70% ethanol for 1 min, which was followed by several rinses in sterile water. Individual nodules were crushed in 0.1 ml of yeast extract-mannitol broth (YMB) (16), nodule remnants were
removed, and 4 ml of YMB was added. After 2 days of incubation at room temperature, 1 ml of YMB containing bradyrhizobia from individual cowpea nodules was inoculated onto one plant each of cowpea, lima bean, peanut, and siratro growing in plastic growth
pouches. No fewer than seven uninoculated and four uninoculated control plants were maintained for
APPL. ENVIRON. MICROBIOL. 1542 THIES ET AL.
FIG. 1. Nodulation patterns and effectiveness profiles of indige-
nous cowpea rhizobia from site 1 (A), site 2 (B), and site 3 (C) on cowpea, siratro, lima bean, and peanut. Chlorophyll content of control plants inoculated with known strains was set at 100%. Chlorophyll content of plants inoculated with indigenous rhizobia is expressed as a percentage of the control. Areas between dotted lines represent the confidence intervals for ineffective (I), moderately effective (ME), effective (E), and highly effective (HE) symbioses. Lane I, inoculated control; lane U, uninoculated control.
All bradyrhizobia from cowpea nodules used in the effec-
tiveness test nodulated both cowpea and siratro. When tested on cowpea, the effectiveness of cowpea-derived nodule bradyrhizobia from the three sites was roughly normally distributed, with approximately two-thirds or more of the bradyrhizobia forming
moderately effective to effective symbioses, and the remainder of the bacteria were divided between forming highly effective or ineffective symbioses (Fig. lA to C). Two of the distributions were slightly skewed; a greater proportion of the bradyrhizobia from site 1 (Fig. lA) formed effective to highly effective symbioses on cowpea compared with those at the other sites, whereas site 2 (Fig.
113) yielded a larger proportion of ineffective bradyrhizobia. Across sites, 43% or more of the bradyrhizobia tested on cowpea
were as effective as, or more effective than, the reference strain, TAL 658, and at least 76% formed symbioses of moderate effectiveness or better. At all sites, bradyrhizobia that were more effective on cowpea than the reference strain were present. Effectiveness and infectiveness profiles of these bradyrhizobia were significantly different on the other legumes.
Whereas all of the cowpea-derived bradyrhizobia were able to
nodulate siratro, 45 to 77% of the rhizobia across sites formed symbioses of only moderate effectiveness. The
SYMBIOTIC SPECIFICITY OF BRADYRHIZOBIA FOR LEGUMES 1543 VOL. 57, 1991
largest percentage of bradyrhizobia forming ineffective nodules on when in symbiosis with cowpea, lima bean, peanut, and siratro in siratro was observed at site 2 (Fig. 1B), whereas the fewest order to interpret large differences in MPN counts obtained for ineffective and largest proportion of effective bradyrhizobia were these species from the same soil samples and to account for obtained from site 3 soil (Fig. 1C). A highly significant positive differential inoculation responses observed with these hosts at the correlation between the effectiveness of the indigenous same sites (11, 14). bradyrhizobial population on cowpea at site 2 and their Siratro is considered one of the more promiscuous hosts in the corresponding effectiveness on siratro (Table 3) was obtained. In cowpea miscellany (5, 16) and has recently been shown to general, the best agreement between effectiveness profiles on the nodulate with some fast-growing rhizobial isolates (6). The four legumes was obtained between cowpea and siratro. consistently higher population counts we obtained on this species On lima bean, 56 to 70% of the bradyrhizobia either failed to compared with the others (Table 2) support this conclusion. nodulate or formed ineffective nodules, 12 to 32% formed However, the frequency of effectiveness was much lower on this symbioses of moderate effectiveness, and only 7 to 18% were as species compared with cowpea. This indicates that the effective as the reference strain, TAL 644 (Fig. lA to C). With bradyrhizobial requirement for these two hosts may differ more peanut, 30 to 81% failed to nodulate or formed ineffective nodules, than previously suggested (2). 5 to 31% were moderately effective, and 14 to 44% were as Lima bean was the most specific of the hosts examined, with effective as the reference strain, TAL 658. Those bradyrhizobia greater than one-half of the bradyrhizobia at each site either failing forming effective symbioses on lima bean and peanut did not to nodulate or forming ineffective symbioses. Peanut displayed a necessarily coincide with those forming effective symbioses on similar, though somewhat lower, degree of specificity. These cowpea and coincided with each other only infrequently (Fig. IA results demonstrate that screening of strains for effectiveness to C). Overall effectiveness of indigenous bradyrhizobia on should be performed with the host of interest, since results are cowpea was not significantly correlated with their effectiveness on clearly not transferable from host to host within the cowpea either peanut or lima bean at any of the sites (Table 3). In all but miscellany. one instance inverse relationships were obtained. The inability of The large variation in numbers of indigenous rhizobia counted some bradyrhizobia to nodulate either lima bean or peanut was not on the four legumes could be due to the differential detection limits related to their effectiveness on cowpea (Fig. lA to C). of the plant infection assay with different host legumes. However, The widest divergence in effectiveness profiles was observed in a separate study (data not shown), we obtained satisfactory between lima bean and peanut. Of the bradyrhizobia exhibiting agreement between population counts of liquid pure cultures of moderate effectiveness or better on lima bean, 25, 81, and 60% rhizobia on petri plates and those obtained in plant infection assays from site 1, 2, and 3 soil samples, respectively, were ineffective or with these hosts. Therefore, the observed variability in population failed to nodulate peanut (Fig. lA to C). Of those that were counts on the four legumes is more likely due to the differential ineffective or failed to nodulate lima bean, 39, 19, and 88% from specificities of these hosts. The observation that many of the site 1, 2, and 3 soils, respectively, were effective on peanut. A cowpea-derived bradyrhizobia tested failed to nodulate peanut and highly significant inverse correlation between effectiveness of the lima bean supports this view. Size of the indigenous, homologous bradyrhizobia from site 3 on lima bean and their corresponding rhizobial population is one of the primary determinants of legume effectiveness on peanut (Table 3) was observed. No correlation for inoculation response (11, 14, 15). The large differences in MPN the other sites between effectiveness on these two hosts was population counts obtained for the four legumes indicate that use observed. In general, these legumes shared in common a much of the appropriate legume host in the MPN procedure may be larger proportion of bradyrhizobia with cowpea and siratro than critical for the accurate prediction of the impact of the use of with each other. legume inoculants.
Analysis of the extent to which effective indigenous bradyrhizobia were shared by cowpea, lima bean, and peanut (Fig. DISCUSSION 2) reveals that the highest frequency of effectiveness was always Division of the cowpea miscellany into effectiveness groupings observed on cowpea. Since all bradyrhizobia originated from has been proposed previously (2). Yet, little is known about the nodules formed on cowpea, this is not surprising. Bradyrhizobia comparative specificities of members in these different groupings effective on peanut and lima bean appear to occupy subgroups that and how the groupings might affect evaluation of the symbiotic are not completely distinct from each other but are mostly within capacity of indigenous bradyrhizobial populations. We the group effective on cowpea. At all sites, a small percentage of investigated the degree of compatibility of indigenous cowpea bradyrhizobia effectively nodulated peanut, lima bean, or both Bradyrhizobium spp. species but were ineffective on cowpea. Hence, while
APPL. ENVIRON. MICROBIOL. 1544 THIES ET AL.
from single nodules as inoculants in this study and found that lima
bean was considerably more specific and peanut was less specific
than he reported. Effectiveness patterns within each population differed markedly
among the three sites and were likely dependent on the types of
indigenous legumes present at each site (Table 1), which, in part, may dictate the types of rhizobia and bradyrhizobia present (18).
These patterns did not appear to be related to population size at a
particular site. In this study, effectiveness of occupants of single nodules
formed by indigenous bradyrhizobia was assessed by using a
chlorophyll assay (7). The results were used to develop an effectiveness index for each site by which indigenous
bradyrhizobial populations for a particular legume species could be
compared between sites. For example, the bradyrhizobial population estimates (by MPN) for cowpea (Table 2) can be
multiplied by the percentage of each effectiveness grouping at a
site to yield estimates of the size of the effective portion of the cowpea population at each site. This effectiveness index (the
adjusted population size estimate) could then be used to calculate
more precisely the potential yield response of legumes to inoculation by using the response models proposed by Thies et al.
(15). Similar percentages and estimates of effective indigenous
population size can be generated for other host legumes in the cowpea miscellany by testing in MPN assays the effectiveness of
nodules formed on the host of interest. FIG. 2. Degree of relatedness of cowpea rhizobia: extent to The fact that a few nonnodulated lima bean and peanut plants which indigenous cowpea rhizobia of moderate effectiveness or had chlorophyll contents that placed them in the moderately better were shared by cowpea, lima bean, and peanut. Numbers effective grouping points out a limitation in the use of this assay. are percentages of the total that effectively nodulate the legume This result indicates that division between ineffective and indicated. moderately (or weakly) effective rhizobia may not be clear-cut
when large-seeded legumes growing in growth pouches are used. the subgroups overlap considerably, in terms of effectiveness, In summary, we have presented evidence supporting Burton's (2) none was completely contained within any of the others. A close separation of members of the cowpea miscellany into different relationship was observed between the effectiveness of effectiveness groupings. In this regard, cowpea and siratro had the bradyrhizobia on lima bean and their ability to nodulate peanut, most similar profiles in terms of both invasiveness and since 79% of those effective on lima bean failed to nodulate effectiveness but did not show complete homology. Lima bean and peanut. This again indicates a very high degree of specificity on peanut showed greater specificity for both effectiveness and the part of these two hosts and strongly supports their separation nodulation than either cowpea or siratro but did share some into different effectiveness groupings. bradyrhizobia in common with these hosts. Peanut and lima bean Lima bean has been shown previously to exhibit a high degree shared in common only one nodule occupant that was as effective of specificity compared with other tropical legumes (1). In their on both species as their respective reference strains. This cross-inoculation studies, Allen and Allen (1) showed that demonstrates that a relatively clear division into separate bradyrhizobia isolated from nodules on lima bean grown in effectiveness subgroups can be made between these two species tropical soil either failed to nodulate or formed ineffective but that separations from other groupings are not as clear-cut as symbioses on all of 28 tropical host legumes examined, including Burton (2) suggests. However, from the standpoint of response to Vigna sinensis (now V. unguiculata), the legume host used for inoculation and determining the need to inoculate, all of these reference. In our study a large proportion of the bradyrhizobia hosts could be considered to occupy effectiveness groups separate tested were able to nodulate lima bean and all soil samples from each other. A significant proportion of the bradyrhizobia contained bradyrhizobia that were effective on both lima bean and failed to form nodules on peanut and lima bean. These results cowpea, indicating that lima bean is not as exclusive as Allen and explain large differences in indigenous bradyrhizobial population Allen (1) found it to be. Less agreement in effectiveness profiles counts obtained from the same soil samples and from legumes and more clearly separated subgroups might have been observed, which are currently classified in the same cross-inoculation group. however, if nodules from lima bean or peanut had been used as a
source of rhizobia. Doku (4) used bradyrhizobia from a number of crushed nodules
formed on lima bean, peanut, soybean, bambara groundnut, and ACKNOWLEDGMENTS cowpea to examine the cross-infection patterns between these This research was supported by National Science Foundation grant hosts. He found that crushed-nodule mixtures from either cowpea BSR-8516822 and the U.S. Agency for International Development Cooperative or lima bean that contained effective bradyrhizobia for these two Agreement DAN-4177-A-00-6035-00 (NifTAL Project). hosts failed to nodulate peanut. He also reported that lima bean We thank G. Haines and K. Keane for soil sampling and D. Turk for nodulated freely with effective bradyrhizobia from peanut, assistance with chlorophyll analyses. soybean, cowpea, and bambara groundnut nodules. We used
VOL. 57, 1991 SYMBIOTIC SPECIFICITY OF BRADYRHIZOBIA FOR LEGUMES 1545
REFERENCES response to rhizobial inoculation in the tropics: myths and realities. In R. 1. Allen, O. N., and E. K. Allen. 1939. Root nodule bacteria of some tropical Lal and P. Sanchez (ed.), Myths and science of soils in the tropics. ASA leguminous plants. II. Cross-inoculation tests within the cowpea group. Soil special publication, in press. ASA-SSSA-CSSA, Madison, Wis. Sci. 47:63-76. 2. Burton, J. C. 1979. Rhizobium species, p. 29-58. In H. J. Peppler and D. 11. Singleton, P. W., and J. W. Tavares. 1986. Inoculation response of legumes Penman (ed.), Microbial technology, 2nd ed., vol. 1. Microbial processes. in relation to the number and effectiveness of indigenous rhizobium Academic Press, Inc., New York. populations. Appl. Environ. Microbiol. 51:1013-1018. 3. Department of Land and Natural Resources. 1982. Median rainfall. Circular C88. Division of Water and Land Development, Department of Natural 12. Soil Conservation Service. 1972. Soil survey of the islands of Kauai, Oahu, Resources, Honolulu, Hawaii. Maui, Molokai, and Lanai, State of Hawaii. Soil Conservation Service, U.S. Doku, E. V. 1969. Host specificity among five species in the cowpea Department of Agriculture, Washington, D.C. cross-inoculation group. Plant Soil 30:126-128. 5. Gibson, A. H., B. L. Dreyfus, and Y. R. Dommergues.1982. Nitrogen 13. Somasegaran, P., and H. Hoben. 1985. Methods in legumeRhizobium fixation by legumes in the tropics, p. 37-73. In Y. R. Dommergues and H. G. technology. University of Hawaii NifTAL Project, Paia, Hawaii. Diem (ed.), Microbiology of tropical soils and plant productivity. Martinet Nijhoff/Dr. W. Junk Publishers, The Hague, The Netherlands. 14. Thies, J. E., P. W. Singleton, and B. B. Bohlool. 1991. Influence of the size of indigenous rhizobial populations on establishment and symbiotic Lewin, A., C. Rosenberg, H. Meyer, C. H. Wong, L. Nelson, J.-F. Manen, 6. performance of introduced rhizobia on fieldgrown legumes. Appl. Environ. J. Stanley, D. N. Dowling, J. Denarie, and W. J. Broughton. 1987. Multiple Microbiol. 57:19-28. host-specificity loci of the broad host-range Rhizobium sp. NGR234 15. Thies, J. E., P. W. Singleton, and B. B. Bohlool. 1991. Modeling symbiotic selected using the widely compatible legume Vigna unguiculata. Plant Mol. performance of introduced rhizobia in the field based on indices of Biol. 8:447459. indigenous population size and nitrogen status of the soil. Appl. Environ. Microbiol. 57:29-37. 7. Mirza, N. A., B. B. Bohlool, and P. Somasegaran. 1990. Nondestructive 16. Vincent, J. M. 1970. A manual for the practical study of root chlorophyll assay for screening of strains of Bradyrhizobium japonicum. Soil nodule bacteria. Blackwell Scientific Publications Ltd., Oxford. Biol. Biochem. 22:203-207. 8. SAS Institute. 1986. SAS user's guide: basics. SAS Institute, Woomer, P. W., J. Bennett, and R. S. Yost. 1990. Overcoming the Inc., Cary, N.C. inflexibility in most-probable-number procedures. Agron. J. 82:349-353. 9. Singleton, P. W. 1983. A split-root growth system for evaluating the effect of salinity on components of the soybean/Rhizobium 18. Woomer, P., P. W. Singleton, and B. B. Bohlool. 1988. Ecological indicators of native rhizobia in tropical soils. Appl. Environ. Microbiol. japonicum symbiosis. Crop Sci. 23:259-262. 54:1112-1116. 10. Singleton, P. W., B. B. Bohlool, and P. L. Nakao. 1991. Legume