Selection of effective strains of Bradyrhizobium for Caatinga stylo ( Stylosanthes seabrana ) Selección de cepas efectivas de Bradyrhizobium para Stylosanthes seabrana

Caatinga stylo (Stylosanthes seabrana) is recommended as a forage legume for permanent and long-term ley pastures on clay soils of southern (cv. Primar CPI92838B) and central (cv. Unica CPI110361) Queensland. The release of the 2 cultivars was contingent on the availability of an effective and persistent strain of Bradyrhizobium, because suitable effective nitrogen-fixing strains do not occur naturally in the soils of the target regions. Effective strains of Bradyrhizobium (strains CB3480 and CB3481), suitable as inocula for Caatinga stylo, were selected from nodule material collected in Bahia, Brazil. This paper documents soil-pot and field experiments that led to the selection of these persistent and effective strains of Bradyrhizobium and the eventual release of CPI92838B and CPI110361, respectively, as cvv. Primar and Unica.


Introduction
The selection and release of adapted accessions of Caatinga stylo [Stylosanthes seabrana; by the taxonomic database GRIN (www.ars-grin.gov),now considered to be S. scabra] for planting on clay soils and suitable effective and persistent strains of Bradyrhizobium arose from a series of experiments over (R.A. Date unpublished data).Similar trials with strains of bradyrhizobia, isolated from other species of Stylosanthes but from the same geographical area (i.e.states of Bahia and Minas Gerais, Brazil) and already held in CSIRO's germplasm collection, similarly failed to form effective N-fixing associations with the exception of strains CB2126 and CB3053 (Date 2010).These failures led to the collection of new nodule material from Brazil in 1992 and 1994.The isolation and authentication of new strains from these nodules were completed in 1995 and preliminary effectiveness was demonstrated on a range of accessions of S. seabrana the following year (Date 2010).
In addition to the absence of suitable effective strains of bradyrhizobia, soil-pot experiments determined that plants, grown in soil from Edye's evaluation sites (Edye 1994;Edye et al. 1998), did not nodulate while soil nitrogen was available but subsequently nodulated effectively and grew satisfactorily, provided suitable Bradyrhizobium had been applied at the time of sowing.This observation indicated that, under field conditions, applied inoculum strains would need to persist for 1 or more growing seasons before plants would nodulate.Nitrogen inhibition of nodulation has been documented for a number of legumes (Streeter 1988;Carroll and Mathews 1990) and is demonstrated in the Caatinga stylo soil-pot experiments, where regrowth of uninoculated controls showed classical nitrogen-deficiency symptoms, whereas inoculated treatments did not (see plant dry weight differences between uninoculated and strain treatments in the ST series experiments below).
Strains CB3480 and CB3481 were isolated and selected from the nodule samples collected in Brazil in 1992.This success led to further collections of S. seabrana germplasm and nodule material in 1994.Morphological classification of the new germplasm (Date et al. 2010) and isolation and selection of additional strains of Bradyrhizobium have provided additional plant and bradyrhizobial lines for further selection work.
On these clay soils producers prefer to sow the smallseeded legumes on the surface (rolled or with shallow scarifying) prior to the onset of seasonal rainfall.Soilsurface temperatures under these conditions frequently exceed 50-60 °C for 4-6 h/d and are lethal to rhizobia on surface-sown inoculated seed where rhizobia may not survive more than 2-3 days (McInnes and Date 2005).An alternative method of introducing the inoculum to the system is required.
The work reported in this paper was completed over several years, as new germplasm of both S. seabrana and bradyrhizobia became available, with the aim of providing effectively nodulating cultivars of Caatinga stylo for the clay soil regions of southern and central Queensland.

Bradyrhizobium
New strains of bradyrhizobia obtained from the 1992 and 1994 nodule collections in Brazil were authenticated as Bradyrhizobium by reinoculation of seedling plants of S. seabrana growing in aseptic tube culture (Norris and Date 1976), i.e. completing Koch's postulate.The dry weights of the same plants also served as an initial assessment of the nitrogen-fixation effectiveness of the isolates.From a bacteriological aspect, the new isolates (strains) were not typical of bradyrhizobia isolated from other species of Stylosanthes.They grew only on acidified nutrient agar and were very slow-growing (10-12 vs. 6-8 days) and then only to pin-head size (± 20% that of other bradyrhizobia; R.A. Date unpublished data).

Soil-pot technique
Air-dried soil from some of Edye's field sites (Table 1) was passed through a 5 mm sieve, then potted to within 15 mm of the top of 15 cm pots lined with plastic bags.The amount of dry soil varied between 1,800 and 1,950 g/pot depending on soil type.Pots were then watered to 95% field capacity and maintained at this level by being mounted on circular tables (10 pots per table) on an automatic watering machine (Andrew and Cowper 1973), which circulated the tables with pots on an endless chain belt around a glasshouse twice each 24 hours.There were 2 replicates of each strain x soil treatment.No fertilizer/nutrients were added to the potted soils.
Surface-sterilized and pre-germinated seed of S. seabrana CPI110361, later released as cv.Unica, was sown at 10 seeds per pot and later thinned to 6 plants, which were inoculated with 1 mL suspension of a peat culture of the test strains of bradyrhizobia.Plant tops were harvested at 6-8 weeks, dried at 60 °C for 48 hours and weighed.In instances where there were few differences between uninoculated controls and strain treatments, plants were allowed to regrow for a further 6-8 weeks and again plants tops were harvested and weighed.Dry weights of whole tops and leaf were used as indices of the effectiveness of the strain treatments.After harvest plant roots were washed out of soil and observations made on level of nodulation.

Experimental series
Four soil-pot experiments were completed:  ST09 tested 8 strains of bradyrhizobia: 7 new strains from the nodule material collected in 1992 and strain CB3053 from the early screening work (Date 2010) plus an uninoculated control using soil from the Narayen granite, Narayen brigalow, Brian Pastures and Holyrood sites (Table 1). ST25 tested 23 new strains from nodule material collected in 1992, an uninoculated control and control with N added using soil from Narayen granite and Narayen brigalow sites (Table 1). ST50 tested 47 new strains from nodule material collected in 1994, 1 of the best strains (CB3480) from ST25, an uninoculated control and a control with N added using soil from the Narayen granite, Narayen brigalow, Hillgrove and Holyrood field sites (Table 1). ST130 tested a further 94 strains from nodule material collected in 1994, 12 existing strains, 1 'diagnostic strain' (Date and Norris 1979), 3 of the best strains from ST50 and 8 strains isolated from other species of Stylosanthes, plus uninoculated controls and controls with N added (1 of each on alternate tables of automatic watering machine) using soil from Narayen granite and Holyrood field sites (Table 1).

Data analysis and classification methods
For the soil-pot experiments data sets of plant top dry weights, leaf dry weights and nitrogen content were used as ratio attributes for each strain.These data were analyzed using simple ANOVA [strain x replication (x harvest time where appropriate)] and the strain means used as attributes in PATN (Belbin 1989) analysis.
The means were range-standardized (TRND module) to provide the input data for ASO (with the Gower Metric option) to obtain symmetric matrices, which were classified by the hierarchical routine FUSE (UPGMA option).The routine GDEF was used to determine group composition and DEND (Dendrogram) to display group structure.

Field experiments
Two series of field trials were established: 1 in January 1995 to assess the strains of bradyrhizobia currently available; and another in January 1996 to assess new strains developed from glasshouse soil-pot assessment of bradyrhizobia arising from new material collected in Brazil in 1994.In each year experiments were established at the CSIRO field stations at Lansdown (solodic soil) and Narayen (granite soil) and in a red earth soil at Holyrood, near Roma (see Table 1).Field plots were comprised of a series of 5 m rows, each row representing a single strain of Bradyrhizobium.There were 3 replications of each strain treatment, randomized within replications.Each strain was prepared as a peat-based inoculum with more than 100 million cells per gram.Seed of cv.Unica was inoculated at the commercial rate of 250 g peat per 25 kg seed using a 5% solution of methyl cellulose, allowed to air dry and then sown the same day.Supplementary irrigation was used to ensure establishment of the Caatinga stylo.Plots were maintained weed-free in the establishment year only.
Harvests each year were made by taking 5 randomly selected 10 cm diameter core samples (to a depth of 15 cm) per row, recording the dry weight of plant tops from this area and washing out root samples to collect nodules for strain identification.Fluorescently labelled specific antibody typing (Somasegaran and Hoben 1985) was used to identify strains forming nodules.

Inoculum delivery
Experiments assessing alternative methods of delivery of the inoculum strains were established at 3 sites in southern Queensland.Treatments (Table 2), comparing surface vs. deep placement (8-10 cm) of the inoculum either on the seed of a preceding wheat crop or on inert plastic prills, were established at Roma Research Station, Holyrood and Banoona (see Table 1).Bradyrhizobium strain CB3546 was used to inoculate wheat seed (Treatments 1 and 2), plastic prills and Caatinga stylo seed in Treatments 3 and 4, respectively, and CB3481 to inoculate Caatinga stylo in Treatment 2 (see Table 2).The use of these 2 strains was to assess which method of inoculum introduction contributed most to the nodule population.Strain CB3546 is an antibiotic (rifamycin and streptomycin) -resistant variant of CB3481 and was equally effective in nitrogen fixation in sand-jar assay (6.2, 4.3, 5.5, 5.9 and 0.2 g dry weight/jar, respectively, for CB3481 mother culture, CB3481 commercial peat, CB3546 mother culture, CB3546 peat culture used in experiment and uninoculated control).It became obvious that strain CB3546 was serologically distinct from CB3481 and this method of distinguishing between the 2 strains proved simpler than attempting antibiotic-resistance assay of the strains forming the nodules.
In an adjacent area at the Banoona site, 2 commercial 10-ha areas (1 of cv.Primar and 1 of cv.Unica Caatinga stylo) were established by Queensland Department of Agriculture, Forestry and Fisheries (QDAFF) using commercially prepared peat inoculant of CB3481, 2 months later in 1996 than the alternative delivery experiment.

Results
More than 700 isolates, obtained from the nodules collected in 1992 and 1994, nodulated Caatinga stylo in the aseptic tube-culture authentication tests.In a sand-jar glasshouse nitrogen-fixation effectiveness assessment, based on plant dry weight as an index of nitrogenfixation effectiveness (Date 2010), 154 of these were selected for further evaluation in the soil-pot experiments (Table 3).

ST09
Most strain treatments were better than uninoculated controls in the Narayen granite, Narayen brigalow and Holyrood soils and there were significant differences among strains within these 3 soils (Table 4).There were no significant differences between strain treatments and the uninoculated control plants growing in the Brian Pastures soil at the corresponding harvest time (Table 4), nor were there any differences after 4 successive regrowth periods.When soil was washed from plant roots after the fourth regrowth period, none of the plants was nodulated, suggesting that soil nitrogen remained at levels high enough for good plant growth and too high for plants to nodulate.

ST25
At the 5-group level most strains formed effective nitrogen-fixing associations (Figure 1) but for some strains this was not obvious until after the regrowth harvest.Dry weight and leaf nitrogen data for a representative of each of the similarity groups are recorded in Table 5. Strains CB3480 and CB3586 were overall the most effective strains.

ST50
This group of strains showed some interaction responses between soils.Some differences were significant at the 1% level.Strains in Groups 1, 2, 3 and 4 were the most effective and had smaller dissimilarity values Continued than the least effective strains in Group 5 (Figure 2 and Table 6).Strains CB3480, CB3489, CB3490, CB3491, CB3640, CB3673 and CB3684 were the best overall performers.
Group 5 strains had only 3-10 small (<0.5 mm) white nodules compared with many 0.5-1.5 mm nodules on plants in Groups 1, 2, 3 and 4. All nodules occurred at the root/lateral root/secondary root junctions.

ST130
As with Experiment ST50 there was a range of responses but most strains were effective (Figure 3 and Table 7).Dry weight data for the best-performing strains from the groups included those for strains CB3480, CB3481 and CB3489.There was a high level of dissimilarity between the effective strains in Groups 1, 2, 3 and 4 and those lesseffective strains in Group 5.

Dissimilarity Index
Dissimilarity Index Significantly there was no difference between samples for uninoculated and N-fertilized treatments.There were scattered small (<0.5 mm) nodules on the roots of plants in Group 5, whereas in Group 1 both tap and lateral roots were profusely populated with larger (1-1.5 mm) nodules.Similar but less numerous nodulation was recorded for plants in Groups 2, 3 and 4. Overall there were fewer nodules on plants grown in Narayen granite soil than in Holyrood soil.

Field experiments
In the trials sown in 1995 only strains CB3053, CB3480 and CB3481 had dry weight yields greater than the uninoculated controls.Strains CB2126, CB3599 and CB3600 failed to produce responses and were not harvested (Tables 8 and 9).Strains CB3053 and CB3481 formed the majority of nodules and only CB3481 maintained this high level in successive years.A significant level of 'contamination' of control plots by CB3481 was observed in later years (Tables 8 and 9).
As with the trials sown in 1995 only those plots in the trials sown in 1996 that demonstrated better growth responses than the uninoculated controls were harvested for plant dry weight yield and determination of the proportion of nodules formed by the inoculum strains (Tables 10 and  11).The proportional differences between inoculated and uninoculated treatments for plant dry weight yield increased in successive years and were greater at the Narayen granite and Lansdown sites than at Holyrood (Table 10).Strains CB3481, CB3494 and CB3495 formed a high proportion of the nodules at all sites and in all years (Table 11), although CB3494 failed to improve yield and form nodules in the year of sowing (1996).Strains CB3488 and CB3489 also formed a large proportion of the nodules at the Holyrood and Narayen granite sites but failed at Lansdown.There was significant 'contamination' of some control plots from adjacent inoculated rows in some replications (see ad hoc notes Table 12).Values in control columns are for positive identification of CB3481.Values in brackets refer to the strain indicated.

Inoculum delivery
For Caatinga stylo the introduction of the bradyrhizobia by inoculation of wheat seed sown in June (T1) was better than placing the bradyrhizobia 10 cm below the legume seed in the normal December-January sowing (T3) (Table 13).At the Holyrood site the combined total of nodules identifiable as CB3481 and CB3546 in T2 was similar to that for CB3546 alone in T1, but there was a trend for nodules from CB3481 to increase with time in T2 with a concomitant decline in those from CB3546.Nodule formation by strain CB3546 also declined in T1.There were similar but less marked declines at Roma Research Station and Banoona sites.Deep placement of the inoculum either on a preceding cereal crop (T1 and T2) or on inert plastic prills (beads) (T3) at the time of sowing the Caatinga stylo provided more nodules than when introduced as inoculum on the surface-sown legume seed (T4).The poor result in T4 confirms that inoculation of surfacesown seed for this legume is a risky practice (Table 13).
For farming systems for which these new cultivars have been selected, the preferred method of sowing the legume is as an associated crop with the cereal, often into dry soils and at high soil surface temperatures (>50 °C for 4-6 h/d; D.A. Eagles and R.A. Date unpublished data).Corresponding bare-soil temperatures at 2, 5 and 10 cm depths were, respectively, 40-50 °C for 6-8 h/d, 30-40 °C for 4-6 h/d, and frequently >30 °C for 8-10 h/d (D.A. Eagles and R.A. Date unpublished data).Temperature profiles at 0, 2, 5 and 10 cm depth for the Holyrood site are recorded in Figure 4 for the 1996-1997 season.Similar profiles were recorded for the 1997-1998 season and for the Narayen granite, Narayen brigalow and Roma Research Station sites in both seasons.

Discussion
This series of soil-pot assessments was aimed at selecting suitable strains of Bradyrhizobium for Caatinga stylo (Stylosanthes seabrana) in order to advance the release of the 2 new cultivars, Primar and Unica.Several strains, which showed particular promise, were selected and evaluated for their ability to survive in field situations and to form nodules in second and subsequent growing seasons, when soil nitrogen had been depleted.Twenty-four strains based on plant dry weight yield (used as an index of nitrogen-fixation effectiveness) were selected: 6 in trials sown in 1995 and 18 in 1996 at the Holyrood, Narayen granite and Lansdown sites.
These strains produced 2−7 fold increases in plant dry weight, especially in the 2nd, 3rd and 4th (for Holyrood) growing seasons, and accounted for the majority of nodules formed.Based on data from trials sown in 1995 (Tables 8 and 9), strain CB3481 was released to industry for the inoculation of the new cultivars, Primar and Unica, in 1997.As well as confirming the ability of CB3481 to satisfactorily nodulate S. seabrana over a 3-year period, the data from field trials sown in 1996 (Tables 10  and 11) have identified several additional strains that could serve as replacement strains for CB3481, if this became necessary.There was good evidence of the ability of these strains to spread to neighbouring areas as indicated by the proportions of strains CB3481, CB3488 and CB3495 found in uninoculated control plots (Tables 8, 10 and 12).
Only cv.Unica was used in these trials, due to limited availability of seed of cv.Primar seed at the time of experimentation; however, nitrogen-fixation effectiveness responses of the 2 cultivars in N-free glasshouse assessments (Date 2010) and in separate soil-pot experiments In random samples of nodules from adjacent QDAFF trials tested for strain CB3481 in May 1998, there were no nodules on Primar and 3% on Unica and in March 1999 values were 70 and 40%, respectively, for the 2 cultivars.
confirmed that both cultivars responded in the same way with the recommended strains of Bradyrhizobium.The good percentages of nodules identified as CB3481 on both cultivars in the commercial sowings (see Table 13) in the third growing season compared with the recoveries in the second growing season provide separate evidence of the inability of Caatinga stylo to nodulate, while soil nitrogen is available in the initial years, and that the recommended inoculum strain CB3481 persists into the second and third growing seasons.
The alternative delivery experiments suggest that deep placement of the inoculum, either by inoculation of a prior crop, e.g.wheat, or on an inert carrier, e.g.plastic prills, may confer an advantage over inoculation of surfacesown Caatinga stylo seed for producers in clay-soil cropping systems, where soil temperatures are detrimental to the inoculum Bradyrhizobium.
Commercially prepared inoculum (peat-based and freeze-dried forms) of strain CB3481 has been available since 1998.If new germplasm material is required for additional cultivar and strain selection and evaluation, the genetic resource collection of S. seabrana germplasm now resides with SARDI/Australian Pastures Genebank and the root-nodule bacteria (Bradyrhizobium) collection with the Centre for Rhizobium Studies (CRS), Murdoch University, Western Australia.

Table 3 .
List of strains of Bradyrhizobium, host species of isolation and country of origin.
1 Strains used in 1995 and 1996 field trials. 2 Strains used in 1996 field trials.3 Strains used in 1995 field trials.4 All strains isolated from nodule material collected in 1992 and 1994 from Bahia State, Brazil.5 Strains isolated from nodules formed on trap host Macroptilium atropurpurem inoculated with crushed nodules from Stylosanthes and authenticated by checking nodule-forming ability and effectiveness on S. seabrana.

Table 5 .
Top dry weight, leaf dry weight and leaf nitrogen content of Caatinga stylo grown in soil from Narayen granite (NG) and Narayen brigalow (NB) sites after inoculation with new strains of Bradyrhizobium collected in 1992 (ST25) with an unfertilized control and a N-fertilized control.

Table 6 .
Plant-top dry weights for Caatinga stylo grown in soils from the Narayen granite (NG), Narayen brigalow (NB), Hillgrove and Holyrood sites inoculated with 47 new strains of Bradyrhizobium collected in 1994 (ST50) plus an unfertilized control and a N-fertilized control.
2 na = not available -many samples accidently destroyed.Of those recoverable top dry weight values ranged from 8.5 to 17.5 g/pot.

Table 7 .
Plant top dry weights (g/pot) for best-performing strains of Bradyrhizobium on Caatinga stylo (Stylosanthes seabrana) grown in soils from Narayen granite and Holyrood sites inoculated with additional new strains of Bradyrhizobium collected in 1994 (ST130) plus an unfertilized control and a N-fertilized control.R.A. DateTropical Grasslands-Forrajes Tropicales (ISSN: 2346-3775)
Data for CB3481 Lansdown May-95 and CB3480 in one replication may be unreliable due to accidental damage by wind drift herbicide.Yields of controls for Narayen Mar-98 and Lansdown May-97 and Mar-98 high due to contamination by effective N-fixing strains.

Table 12 .
Ad hoc observational evidence for spread of strains CB3481 and CB3495 in strain trials sown in 1996.

Table 13 .
Percentage recovery of nodules containing inoculum strains CB3546 and CB3481 for alternative delivery trials at Holyrood, Roma Research Station and Banoona sites 1 .