Introduction

Rhizobia are soil bacteria that are capable of colonising plant rhizospheres and forming a symbiosis with legumes (Poole et al. 2018). Infection of legume roots by rhizobia results in the formation of root nodules, inside which atmospheric N2 is reduced or “fixed” by rhizobia into ammonia, which is subsequently secreted and assimilated into host plant tissues. This process is exploited in crop and pasture systems that include legumes, where fixed N2 is a biological source for the crop or pasture and also enters the soil following plant senescence and decay or via livestock deposition (Peoples et al. 2009).

In Australia and New Zealand, agricultural forage and grain legumes are often inoculated with highly effective rhizobia, frequently sourced from other parts of the world, to overcome a lack of pre-existing compatible soil rhizobia for these exotic plants (Corbin et al. 1977; Davidson and Davidson 1993; Howieson et al. 1988, 1995; Lange 1961; Nichols et al. 2012). In the case of Mesorhizobium spp. that form a symbiosis with Cicer arietinum (chickpea), Biserrula pelecinus and Lotus spp., horizontal gene transfer of symbiosis genes from inoculant strains to resident soil bacteria has resulted in the evolution of novel nodulating rhizobia (Hill et al. 2021; Nandasena et al. 2007b; Sullivan and Ronson 1998). The symbiosis genes in Mesorhizobium spp. are generally chromosomally encoded on mobile regions of DNA referred to as Integrative and Conjugative Elements (ICEs), which can be either monopartite or tripartite in structure. ICEs can excise from the host bacterial chromosome and transfer by bacterial conjugation, inserting site-specifically into a recipient cell’s chromosome, converting it into a legume nodulating microsymbiont (Haskett et al. 2016b; Ramsay et al. 2006; Sullivan and Ronson 1998). Symbiosis ICEs were first discovered in M. japonicum (formerly M. loti), following its inoculation on Lotus corniculatus in New Zealand, leading to the evolution of genetically distinct rhizobia capable of nodulating this introduced forage legume (Sullivan and Ronson 1998; Sullivan et al. 2002). A similar scenario occurred in Australia with the inoculation of the forage legume Biserrula pelecinus with Mesorhizobium ciceri WSM1271 and WSM1497 (Nandasena et al. 2007a, 2007b). In the case of both legume introductions, newly evolved rhizobia were able to compete with the inoculant strain to nodulate the target legume, either occupying the majority (Ramsay and Ronson 2015) or almost half of the nodules surveyed (Nandasena et al. 2007b). Crucially in the case of B. pelecinus, some of these competitive novel rhizobia were less effective at fixing N2 than the inoculant strain (Nandasena et al. 2007b), thereby potentially undermining the efficacy of inoculation and long-term benefits of symbiotic N2 fixation.

Australia is one of the world’s largest exporters of high quality of C. arietinum (chickpea), predominately grown in eastern Australia, with smaller growing areas extending into regions in western Australia (GRDC 2017; Muehlbauer and Sarker 2017; Wood and Scott 2021). Mesorhizobium ciceri symbiovar (sv.) ciceri CC1192, a highly effective strain from Israel, was introduced into Australia in the 1970’s and since then, has been the only commercial inoculant for C. arietinum (Corbin et al. 1977; Farquharson et al. 2022). In spite of this and an apparent lack of pre-existing compatible rhizobia for this grain legume in Australian soils, several studies have now reported the isolation of rhizobia genetically distinct to the CC1192 inoculant strain that are capable of nodulating C. arietinum (Oparah et al. 2024; Zaw et al. 2021). In particular, an extensive survey by Elias and Herridge (2015) across 26 farms in one of eastern Australia’s main C. arietinum growing areas reported that 53% of nodule occupants were genetically distinct to CC1192. Although these isolates were not genomically characterised, subsequent sequencing and analysis of a subset of 11 symbiotically effective novel strains originating from a single farm showed they had all acquired the mobile 419-kb M. ciceri CC1192 symbiosis ICE (ICEMcSym1192) by environmental transfer from the inoculant strain (Hill et al. 2021). However, in their study, Elias and Herridge (2015) also tested 29 novel strains from across eight fields and found that 41% of these strains were significantly less effective than CC1192 at fixing N2 with C. arietinum. As these strains were lost following their characterisation, their genomic diversity, symbiotic gene complement and competitiveness could not be further investigated.

The Ord River Irrigation Area (ORIA) is a semi-arid tropical area in north-western Australia, where irrigation agriculture was first undertaken in 1963 (DPIRD 2021). Since this time, a range of crops have been cultivated in the ORIA, including maize, cotton, mango, citrus, melons and pumpkins. C. arietinum is also an important legume in the region, and was first commercially grown and inoculated with CC1192 in 1985 (DPIRD 2021; Riley 1994). Geographically, the ORIA is isolated from Australia’s other centres of agricultural production in the south-west, east and south-east of the country by considerable distance and climatic variation. This isolation, along with the established history of C. arietinum in the ORIA, provide an ideal location to study the genotypic and symbiotic diversity of C. arietinum symbionts. Here, we investigate the genomic diversity, phylogeny and origin of strains nodulating cultivated C. arietinum in the ORIA as well as their effectiveness at N2 fixation compared, to the commercial inoculant, CC1192.

Material and methods

Isolation, strains and media

Root nodule bacteria were obtained from cultivated C. arietinum sampled from ten sites across eight farms within the Ord River Irrigated Agriculture (ORIA) area (Table S1/Fig. 1). Whole nodules were collected from C. arietinum roots and desiccated in 5 mL vials containing silica gel with a cotton wool plug (Howieson et al. 2016). The desiccated nodules were immersed in deionised (DI) water for 4 h, and then surface sterilised, crushed and the nodule contents cultured with half-strength lupin agar (½LA) or broth at 28 °C (Hungria et al. 2016). Log phase pure cultures of the isolates were stored at -80 °C in broth containing 10% (v/v) dimethyl sulfoxide (DMSO). Bacterial strains used in this work are shown in Table 1 and Table S1.

Fig. 1
figure 1

Locations of sites sampled for C. arietinum nodules within the Ord River Irrigation Area (ORIA) of northwest Australia (a), in relation to Australia (b) and the Lake Argyle catchment area (c). Map images prepared with Google My Maps [Google Imagery ©2023]

Table 1 M. ciceri CC1192 and selected strains representative of rhizobial genotypes isolated from C. arietinum from the Ord River Irrigation Area (ORIA). A full list of isolated strains can be found in Table S1

RAPD-PCR and ICEMcSym1192 screening

RAPD-PCR fingerprints were generated with the RPO1 primer (5’-AATTTCAAGCGTCGTGCCA-3’), designed to bind the nifHDK promoter consensus region (Richardson et al. 1995). Resultant PCR fingerprints were compared to CC1192, to identify strains as inoculant reisolates or novel strains. DNA template for PCR consisted of whole cell pellets (OD600 of 6), with PCR reaction and conditions as described by Gerding et al. (2013). PCR products were visualized on a UV transilluminator after electrophoresis on 2% (w/v) agarose gel and staining with ethidium bromide (0.5 µg mL−1). RAPD-PCR fingerprints of the isolates were compared to that produced by CC1192 and those that were different were assigned a genotype group based on similarity of the fingerprints. Novel isolates were further PCR screened for the presence of the CC1192 symbiosis (ICEMcSym1192). A primer pair was designed to target the intS region (5’-GAGGAGCAGCTTGATACGG-3’) of the ICEMcSym1192 and within the flanking Ser-tRNA gene (5’-ATCGGGGGTTCGAATCCC-3’) to yield a 499 bp product, using the cells prepared for the RPO1-PCR.

Authentication and effectiveness of nodule isolates

C. arietinum cv. CBA Captain was grown in free-draining sterile sand in a glasshouse maintained at 24 °C, as described by Yates et al. (2016), where growth of legumes is limited by N-deficiency, except when plants are nodulated by N2-fixing rhizobia. Briefly, seeds were surface sterilized in 70% (v/v) ethanol (1 min), followed by 4% (v/v) NaOCl (3 min), rinsed in six successive washes with DI water, and imbibed in the final wash for 5 min. Surface sterilized seeds were pregerminated on 0.9% (w/v) agar for 24 h at 25 °C, and following emergence of radicles, the seedlings were sown aseptically into 1-L pots (170 mm x 80 mm × 80 mm), containing a coarse sand mixture prepared and steam sterilized as described by Yates et al. (2016), with the sand mix at pH(CaCl2) 6.98. Each pot was sown with two seedlings and thinned to one plant per pot upon emergence of shoots.

To prepare the inoculum, strains of rhizobia were cultured in 10 mL ½LA broth on a gyratory shaker (at 250 rpm) to an OD600 of 1 (≈109 cells mL−1). C. arietinum treatment pots were either uninoculated (N-starved and N-fed controls) or inoculated separately with 1 mL of the cell culture, while uninoculated pots received 1 mL of sterile deionized water. All pots were protected from airborne contamination by sterile alkathene beads as described by Yates et al. (2016). Each treatment consisted of either three or eight replications, which were randomized and maintained with sterile water and a plant growth nutrient solution as required (Yates et al. 2016). Uninoculated N-fed control treatments received 2 mL of 0.1 M NH4NO3 twice weekly.

Two separate glasshouse experiments were conducted with C. arietinum. The first experiment was to perform Koch’s postulates and authenticate, with three replicates, all nodule isolates retrieved from ORIA sampling (Table S1). The second experiment measured the effectiveness of the representative RPO1-PCR group strains, WSM4875, WSM4884, WSM4887, WSM4898, WSM4904, WSM4906, WSM4935, WSM4962, WSM4976, WSM4983 and wild-type CC1192 with eight replications (Table 1).

To authenticate strains, plants were harvested at 42 days post-inoculation and presence or absence of nodules in the three replicates for each treatment noted. To assess N2 fixation effectiveness of the ten representative strains plus CC1192, plants were harvested at 62 days post inoculation, by carefully removing the roots from the soil and washing the root systems. Nodules were excised from the roots, and plant shoots were separated from the roots at the hypocotyl. Both shoots and nodules were then incubated at 60 °C to complete dryness prior to weighing. After weighing, shoot dry weight (SDW) samples for each treatment were pooled and shoot N content was determined using the Dumas high combustion method on a Leco F528 Nitrogen Analyzer (CSBP Plant and Soil Laboratory, Perth, Australia). The variance of the means of the SDW and nodule dry weight (NDW) was assessed by performing one-way analysis of variance (ANOVA), and the significant difference between the treatment means was analyzed by the Tukey honestly significant difference (HSD) post hoc test at an α value of 0.05 using IBM SPSS Statistics v29.4.

Sequencing, assemblies, and analysis

Genomic DNA for short-read sequencing was extracted according to the manufacturer’s instructions using a Qiagen blood and tissue DNAeasy extraction kit. DNA for long-read sequencing was isolated by phenol:chloroform extraction as previously described (Ramsay et al. 2021). Concentration and purity of the extracts were analyzed with a NanoDrop One spectrophotometer (Thermo Fisher Scientific) and gel electrophoresis. Short-read sequence data were produced from Illumina NextSeq 500 2 × 150 bp paired end reads (Illumina, Inc., USA). Long-read sequence data were produced from Oxford Nanopore Technologies MinION sequencing platform (R. 9.4.1 flowcells and MinION Mk1B). Assemblies were produced using Flye v. 2.9 (Kolmogorov et al. 2019) or SPAdes v. 3.14.1 (Bankevich et al. 2012) from long-read sequence data and short-read sequence data respectively. Flye assemblies were polished five times with filtered nanopore reads using Racon v. 1.4.17 (Vaser et al. 2017) and six times with Illumina reads using Pilon v. 1.23 (Walker et al. 2014a). Genomes were annotated with Prokka v. 1.13 (Seemann 2014). For confirmation of ICEMcSym1192 presence in incomplete assemblies, scaffolds were mapped to the ICEMcSym1192 sequence using Geneious mapper 2023.0.4. Average nucleotide identity (ANI) was determined using fastANI (Jain et al. 2018). Hierarchical clustering of ANI distance matrices (Table S2) were plotted using R v. 4.2.2 (R Core Team 2022) and the heatmaply package (Galili et al. 2018).

Phylogenetic analysis

Genomes generated from this work were submitted to the NCBI database and these, plus other strains included in the Mesorhizobium core gene phylogenetic analysis are listed in Table S3. Proteinortho v. 6.0.31 (Lechner et al. 2011) was run with default setting and identified 1873 single copy core genes shared between these strains. The nucleotide sequences were aligned with MAFFT v. 7.487 (Katoh et al. 2002), then the alignments were cleared of gaps and concatenated using Goalign v. 0.3.5 (Lemoine and Gascuel 2021). The nonsymbiotic ICE tree was constructed from a set of 15 single copy ICE backbone genes identified by Colombi et al. (2021). The sequences of these proteins from M. japonicum R7A were used to query each mobile region using TBLASTN (Camacho et al. 2009) with an E value cutoff of 1 × 10–5 (Table S4). Gene sequences were aligned using clustalo v. 1.2.4 (Sievers and Higgins 2018) and concatenated with SeqIO for Python v. 3.8.3. Phylogenies were constructed using RAxML v. 8.2.11 (parameters: -f a -p $RANDOM -x $RANDOM -N 100 m GTRCATX -T 12) (Stamatakis 2014). Trees were visualized with FigTree v. 1.4.4 (https://github.com/rambaut/figtree/releases).

Results

Distribution and diversity of Mesorhizobium strains that nodulate C. arietinum

Nodules were collected from commercially growing C. arietinum Desi and Kabuli types across 69.0 km2 at ten sites within the Ord River Irrigation Area (ORIA) in northwest Australia in June 2017 (Fig. 1). All crops had been inoculated at sowing with M. ciceri CC1192, but with a substantial variation in inoculum formulations, inoculation methods and site cropping histories (Table 2). Five sites had inoculated C. arietinum sown within the three-four years prior to sampling (Sites B, C, D, E and I), site F was sown 14 years prior and site J had no recorded history of C. arietinum cultivation (Table 2). Sites G and H had not been sown with C. arietinum in the four years prior to sampling, with no site history available prior to this time. Several inoculant formulations were applied across the ten sites, including peat, freeze dried and granuler inoculants, with multiple formulations used at three of the sites (Table 2). Isolates were obtained from nodules collected from five plants at sites B, C, H, and I, four plants at sites A, E and F, three plants at sites D and G and one at site J. A total of 68 strains were isolated and authenticated from sampled C. arietinum nodules. All strains were positive when screened by PCR with primers targeting the CC1192 symbiosis ICE (ICEMcSym1192) integrase (intS) and the 3’ conserved region of the flanking Ser-tRNA gene, consistent with insertion of an ICE within this gene. Genotyping by RPO1-PCR identified 12 strains with banding patterns consistent with being CC1192 reisolates, with the remaining 56 divided into 10 unique groups (Table S1). Inoculant formulation and delivery had no effect on rhizobia genotype across the different sites. However, for site G where the site history was unknown and site J that had not previously grown chickpea, only CC1192 was detected. For each of the 10 unique genotypic groups, representative strains were selected for further study, which were WSM4875 (KN1), WSM4884 (KN3), WSM4887 (KN4), WSM4898 (KN5), WSM4904 (KN6), WSM4906 (KN7), WSM4976 (KN8), WSM4935 (KN12), WSM4962 (KN14) and WSM4983 (KN18) (Figure S1).

Table 2 Ord River Irrigation Area (ORIA) sample site locations and Cicer arietinum sowing history, at time of sampling in 2017

The distribution of novel strains differed across the ten sites (Table 3). Sites B, H and I showed the greatest genomic richness, with five or six nodule genotypes identified within a single site (Table 3). Inoculant CC1192 and novel genotype KN1 were the most widely distributed strains, isolated from six of the ten sites sampled, while genotypes KN5, KN7 and KN14 were found at three sites, KN12 at two, and the remainder at single sites only (Table 3). Across the nodules surveyed, KN1 was the most abundant genotype, identified in 33.8% of nodules, with CC1192 and KN7 the next most frequent occupants, each at 17.6% (Table 3). The remaining eight novel genotypes were dispersed across the sites and made up 30.9% of the isolates screened. At five of six sites where CC1192 was detected along with other strains, nodules were dominated by novel strains, with between 80–90% of nodule occupants genotypically different to CC1192. Thus, the surveyed inoculated C. arietinum in the ORIA were nodulated by a diverse suite of novel rhizobia genotypes.

Table 3 Distribution of isolated ICEMcSym1192 recipients and CC1192 from C. arietinum nodules across 10 sampling sites within the ORIA. Nodule isolates were grouped by RPO1-PCR genotype, and their frequency of isolation at each site is indicated by percentage occurrence.

Symbiotic capacity of novel C. arietinum strains

To assess the symbiotic capacity of the these rhizobia, the nodulation and N2 fixation efficiency of selected strains from each of the ten genotypes was assessed in a controlled glasshouse experiment on C. arietinum cv. CBA Captain. Nine of the ten strains tested against CC1192 produced shoot dry weights (SDW) not significantly different to CC1192, with mean values ranging from 0.348 to 0.471 g plant−1 or 70.5% to 95.3% SDW of CC1192 (Fig. 2a). Moreover, the inoculant strain CC1192 and strains WSM4875 and WSM4962 were equivalent to the uninoculated N-fed control in terms of SDW production. The nine effective strains, along with CC1192, represent genotypic groups that constitute 98.5% of the isolates obtained from the ORIA sites. The %N of the inoculated and control treatments correlated strongly to the SDW, producing a linear regression R2 value of 0.8337 (Fig. 2b) indicating, there was a strong relationship between N2 fixed and biomass production in the test system, even though most treatments did not differ significantly in their SDW production. Nodule dry weights were not significantly different across any of the inoculated treatments (Fig. 2c) and apart from a strong correlation between nodule number (Figure S2) and nodule weight, there were no other significant correlated relationships between any of the other assessed harvest parameters (plant dry weight, nodule score, nodule number, total plant nodule mass, total nodule mass per nodule and % N; Table S5).

Fig. 2
figure 2

Symbiotic effectiveness of representative Mesorhizobium sp. harbouring ICEMcSym1192 and CC1192 on C. arietinum cv. CBA Captain. Plants were, inoculated separately with the indicated strains, and grown for 55 days under N-limited conditions. Uninoculated and N-fed (supplied as NH4NO3) plants were included as negative and positive controls, respectively. Data are mean shoot dry weight (a), correlation of total shoot % N versus mean shoot dry weights (b) and mean dry nodule weight per plant (c). Error bars denote standard errors of the means, and treatments that share a letter are not significantly different according to the Tukey’s test (P ≤ 0.05)

Only WSM4904 (the solitary member of KN6 genotype group) produced a mean SDW (0.228 g plant−1) which was significantly lower (P < 0.05), at 54% that of CC1192 inoculated plants, and equivalent to the N-starved uninoculated control (Fig. 2a). Total foliage %N-concentration of WSM4904 inoculated plants was also 2.23-fold lower than CC1192 (P < 0.001), and visually WSM4904 treatment plants were clearly smaller and paler than those inoculated with CC1192 (Fig. 3a), indicating that WSM4904 was poorly effective at fixing N2 with C. arietinum. Nodule number per plant with WSM4904 was significantly greater (P = 0.009) than CC1192-inoculated plants with one-way ANOVA, and the poorly effective strain elicited nodules that outwardly were very similar in morphology to those produced in the CC1192 treatment (Fig. 3b, d).Transverse sectioned nodules showed a less intense pigmentation in the pink apical region, corresponding to the likely infection and N2 fixation zone, as well as the distal green region, representing the probable nodule senescent zone, compared to nodules from the CC1192 treatment (Fig. 3c,e). Thus, although SDW was significantly different between wild-type and WSM4904 treatments, phenotypic differences between these strains at the nodule level were very small.

Fig. 3
figure 3

Nodules and whole plants of C. arietinum cv. Captain inoculated with CC1192 and WSM4904 (KN6). Entire plants shown (a) with CC1192 and WSM4904 from left to right. Nodules of CC1192 treatments shown whole (b) and transverse section (c), WSM4904 shown whole (d) and transverse section (e)

Genomic diversity of Mesorhizobium strains that nodulate C. arietinum

To explore the genomic diversity of the ten selected strains and identify their symbiosis ICEs, the strains were genome sequenced on an Illumina NextSeq platform. Analysis of the genome sequences confirmed all ten strains carried ICEMcSym1192, consistent with the environmental acquisition of this ICE from the inoculant strain CC1192 (Fig. 4). Further analysis of their genomic diversity from a maximum likelihood phylogeny showed the ten strains clustered into four distinct groups (Fig. 5), all of which were only distantly related to the M. ciceri group containing the C. arietinum commercial inoculant CC1192, as well as other known symbionts such as M. mediterraneum USDA3392 isolated from Spain and M. wenxiniae WYCCWR 10195 isolated China (Nour et al. 1995; Zhang et al. 2018). They also were only distantly related to other ICEMcSym1192 recipients isolated from Australian soils (Hill et al. 2021), with the exception of WSM4313 which grouped more closely with strains from this study (Fig. 5). The majority of the strains (WSM4898, WSM4962, WSM4887, WSM4983 and WSM4875) were most similar to each other and formed a clade with M1A and P13.3 (Fig. 5), C. arietinum strains from India and South Africa, respectively (Greenlon et al. 2019; Wanjofu et al. 2022). The isolates WSM4904 and WSM4906 are very similar to each other and collectively grouped with WSM4884. Together, these three strains distantly grouped with M1B, an isolate of C. arietinum from Ethiopia (Greenlon et al. 2019). The isolate WSM4935 also grouped distantly with M. plurifarium ORS1032T, isolated from Acacia senegai in Senegal (De Lajudie 1994). The remaining isolate WSM4976, distantly grouped to M2A and ORS3359, isolates from C. arietinum and Acacia seyal from Ethiopia and Senegal (Diouf et al. 2007; Greenlon et al. 2019).

Fig. 4
figure 4

Schematic representation of ICE regions from strains isolated in this study. Scaffolds that span the ICE region were assembled from short-read sequence data, accession numbers associated with these sequences are listed on the right with specific designations above each scaffold. Scaffold orientation is marked with an arrowhead, those that extend beyond the ICE region are represented with jagged edges, and those less than 2 kb in length were excluded. WSM4875, WSM4904 and WSM4906 were also sequenced by ONT MinION long-read sequencing, which allowed generation of closed genomes for these strains, consisting of a single contig for the chromosome. Therefore, ICEMCSym1192 was identified in its entirety in these strains, as indicated in above by the unbroken arrows

Fig. 5
figure 5

Core-genome phylogeny of Mesorhizobium isolated from nodules during this study. In total, 1, 873 single-copy core-genes were used to build a maximum likelihood tree using RAxML. The tree was rooted with the M. sp. HJP. Coloured boxes represent the grouping of strains as probable species based on > 96% ANI values within the group. The scale bar indicates substitutions per site, and only those bootstrap values < 100 are shown

To confirm that the four clusters represented distinct genospecies, the average nucleotide identity (ANI) of the ten strains was compared using hierarchical clustering analysis (Fig. 6). ANI values of less than 95–96% are generally accepted as representing distinct bacterial species (Chun et al. 2018). This analysis grouped the strains with > 96% ANI into four clusters, consistent with the core genome phylogeny (Fig. 5) and supporting the separation of the ten strains into four genospecies, with WSM4898, WSM4962, WSM4887, WSM4983 and WSM4875 in one group, WSM4884, WSM4904, and WSM4906 in a second group, and WSM4935 and WSM4976 making up two more individual groups (Fig. 6). The largest group also contains M1A and P13.3, however none of these strains are similar enough to a sequenced type strain for definitive species assignment. Therefore, the ten strains represent four novel Mesorhizobium genospecies that have acquired the CC1192 ICEMcSym1192 by environmental transfer.

Fig. 6
figure 6

Whole-genome heat map showing the pairwise genome conservation distances of the genome group representative strains. Hierarchical clustering of ANI distance matrices were plotted using R v. 4.2.2 and the distance matrix was plotted as an heat map using heatmaply. Dendrograms across the top and right of the heat map show the relationship of genomes based on genome conservation. The strain names are indicated to the left and bottom of the heat map. %ANI range from 87.5 to 100, which are depicted by the gradient of colours ranging from dark blue (indicating low similarity between genomes) to yellow (high similarity between genomes). The matrix of percentage ANI values used to construct this figure is available in Table S2

Genomic comparison between poorly effective WSM4904 and effective WSM4906

The poorly effective strain WSM4904 and effective strain WSM4906 clustered within the same genospecies (Fig. 5), sharing a very high sequence similarity of 98.7% (Fig. 6), yet they exhibited very different symbiotic phenotypes (Fig. 2), with WSM4904 producing 0.228 g plant−1 SDW, while WSM4906 produced 0.364 g plant−1 SDW. Importantly, both strains were originally isolated from separate nodules collected from the same plant growing at site C. While a Tukey test (P < 0.05) indicated the SDW were not significantly different (Fig. 2a), a single factor ANOVA comparison between the two isolates was significant (P = 0.01), with the SDW of WSM4904 and WSM4906 49.0% and 73.6% to that of CC1192. The sequence similarity between these two strains therefore presented an opportunity to explore the possible genetic causes of their phenotypic differences.

Both strains were sequenced using long read ONT MinIon sequencing, to generate closed genomes and facilitate more detailed genomic comparisons. Analysis of the genomes revealed no sequence divergence in any known symbiosis genes (i.e. nod, nif or fix) within ICEMcSym1192 between either strain and CC1192. A wider comparison of the ICEMcSym1192 region outside of the symbiosis genes from both strains to the CC1192 ICEMcSym1192 revealed five variations, four of which were shared between WSM4904 and WSM4906. These four mutations were; an 8 bp insertion within succinate semialdehyde dehydrogenase (locus tag in CC1192 genome A4R28_21220), a single bp insertion in an intergenic region upstream of hypothetical protein (A4R28_21300), a single bp deletion in an intergenic region downstream of the putative Major Facilitator Superfamily transporter gene (A4R28_21530) and a T to G transversion in a putative transposase (A4R28_20730). To ascertain whether these mutations were unique to the WSM4904 and WSM4906 strains that were isolated from the same plant at site C, strain WSM4875 isolated from site A and which is highly effective at fixing N2, was also sequenced. The same four mutations were also present in the WSM4875 ICEMcSym1192, indicating that these mutations alone cannot explain the poor effectiveness of WSM4904. The remaining mutation, a single base pair deletion within a putative integrase gene (A4R28_21015) in the WSM4904 ICESymMc1192, was the only unique mutation in the ICESym region between this strain and effective strains WSM4906, WSM4875, and CC1192. The deletion causes a frameshift in the integrase sequence which introduces a premature stop codon 636 bp into the gene.

Outside of ICEMcSym1192, a mauve alignment of the WSM4904 and WSM4906 chromosomes showed 92.4% sequence identity. The 7.6% difference between these two genomes translated to 75,866 sequence variations, consisting of 525,410 residue differences across WSM4904 and WSM4906. This sizable number of variations are distributed across coding sequences and intergenic regions alike. In the course of comparing these chromosomal replicons, a second 57 kb region was detected in the WSM4906 genome upstream of a Gln-tRNA (QAZ22_RS22720) that is absent from WSM4904. This locus encodes several phage assembly proteins, suggestive of insertion of a lysogenic phage within WSM4906. Additionally, an approximately 300 kb region, was identified inserted at one of the three Ile-tRNA genes in both strains (QAZ47_RS13780 in WSM4904 and QAZ22_RS13720 in WSM4906). This region is flanked by 16-bp inverted repeats, with an integrase gene (QAZ47_RS13785 and QAZ22_RS13725) directly upstream of the Ile-tRNA insertion site. The region also encodes a relaxase (QAZ47_RS14390 and QAZ22_RS14325) and an excisionase (QAZ47_RS14570 and QAZ22_RS14505), but with no symbiotic (i.e. nod, nif or fix) genes, which are features consistent with both regions being nonsymbiotic ICEs. Phylogenetic comparison of 15 single copy genes encoded across both ICEMsp.WSM4904, ICEMsp.WSM4906, 28 symbiotic and 24 non-symbiotic ICEs showed that the ICEMsp.WSM4904 and ICEMsp.WSM4906 genes surprisingly group strongly with the symbiotic ICEs known to nodulate Cicer, Lotus and Biserrula, but not with other known non-symbiotic ICEs (Fig. 7). Although ICEMsp.WSM4904 and ICEMsp.WSM4906 are phylogenetically closely related, they do differ in size significantly, with ICEMsp.WSM4904 being 308,640 bp, while ICEMsp.WSM4906 is 292,670 bp., with the majority of these excess nucleotides present at the 3’ end of the ICEMsp.WSM4904 adjacent to attR (Figure S3). This region is absent in ICEMsp.WSM4906 and contains 13 genes with putative roles in carbohydrate metabolism and transport (Table S6). Therefore, while the WSM4904 and WSM4906 genomes are phylogenetically very similar, with each strain encoding near-identical copies of the symbiotic ICEMcSym1192, the variation in sequence across both chromosomes and non-symbiotic ICEs are the most likely cause of their differing N2 fixation effectiveness in symbiosis with C. arietinum.

Fig. 7
figure 7

Phylogeny of non-symbiotic ICEMsp.WSM4904 and ICEMsp.WSM4906 in comparison to selected symbiotic and non-symbiotic ICEs from a range of Mesorhizobium spp.. The tree was constructed with RAxML and is based on the concatenation of alignments of 15 single copy backbone genes. It was rooted at the midpoint and the scale bar indicates substitutions per site. Only bootstrap values < 100 are shown. ICESyms that associate with Biserrula are highlighted in blue, Lotus in green, Cicer in pink, an undefined plant in grey, and those identified in this study are in brown

Discussion

This study has demonstrated that a phylogenetically diverse suite of Mesorhizobium strains nodulate C. arietinum in the ORIA. At the core genome level, these strains are genetically distinct to M. ciceri CC1192, yet they all harbor the 419-kb monopartite symbiosis ICE (ICEMcSym1192) from this inoculant strain. The symbiotic effectiveness of nine of the ten representative strains was not significantly different to CC1192, with only one strain (WSM4904) poorly effective at fixing N2 with C. arietinum. Interrogation of the WSM4904 genome and comparison to closely related but effective WSM4906 showed that the genetic cause of the poorly effective phenotype is likely to be chromosomal, and not ICEMcSym1192, encoded.

The 68 authenticated isolates grouped into ten genotypes distinct to CC1192, and sequencing of representatives from each group showed the strains to cluster into four distinct genospecies, based on their core-genome phylogeny and ANI values. The genomes of these strains are genetically distinct to M. ciceri CC1192, and to the inoculant strains M. ciceri WSM1271 and WSM1497 for the temperate pasture legume Biserrula pelecinus and M. jarvisii SU343 for Lotus spp. (Nandasena et al. 2007a; Perry et al. 2020). Therefore, the novel strains isolated in this study have not been derived from other inoculant Mesorhizobium spp. commercially used in Australia. The high proportion of novel strains (i.e. 82% of the 68 strains isolated) reported in this study, aligns with the work of Elias and Herridge (2015), where of 570 strains from field grown C. arietinum sampled from eastern Australia, 86% were not CC1192 (based on RAPD-PCR); and Zaw et al., (2021) where 74% of the 77 strains acquired from agricultural soils sampled across Australia, were novel (based on 16S-23S rRNA IGS). Nodulation of Cicer spp. by genetically diverse Mesorhizobium is well known outside of the Australian context in regions of the world such as China, Ethiopia, India, Morocco, Myanmar, Portugal and Syria, where the grain legume has been cultivated for many thousands of years (Alexandre et al. 2009, 2006; Greenlon et al. 2019; Maâtallah et al. 2002; Nour et al. 1995; Singh et al. 2019; Soe et al. 2020; Zhang et al. 2018). Thus, it is clear that genetically diverse rhizobia now nodulate cultivated C. arietinum in Australia, even though only one inoculant strain has been commercially available across the country since 1977 (Bullard et al. 2005).

ICEMcSym1192 has previously been shown to be mobile in vitro (Hill et al. 2021), so the presence of this element in the genomes of all strains analysed in this study is consistent with the environmental transfer for this ICE from the CC1192 inoculant strain. ICEMcSym1192 was previously found to have been acquired by 11 novel strains isolated from C. arietinum cultivated in Moree, in eastern Australia (Elias and Herridge 2015; Hill et al. 2021). Similarly, novel C. arietinum-nodulating strains isolated from soils sampled across Australia all harboured nodC (encoding the Nod factor synthesis enzyme N-acetyl glucosamine transferase) and nifH (encoding the Fe-protein in mature nitrogenase) genes identical to the corresponding genes in ICEMcSymCC1192 sequence, suggesting that these strains had also acquired this ICE from CC1192 (Zaw et al. 2021). Oparah et al. (2024) did report isolation of four novel strains from C. arietinum nodules collected from the Department of Primary Industries and Regional Development’s Frank Wise Institute of Tropical Agriculture in Kununurra in western Australia, however the symbiosis ICE of these strains and their genomic relatedness to CC1192 were not investigated. Although other Mesorhizobium strains introduced into Australia also harbor symbiosis ICEs (WSM1271, WSM1497 and SU343), these tripartite ICEs are distinct to the monopartite ICEMcSym1192 (Colombi et al. 2021; Haskett et al. 2016b; Perry et al. 2020) and do not confer an ability to nodulate C. arietinum (Nandasena et al. 2006). Therefore, apart from the CC1192 symbiosis genes encoded in ICEMcSymCC1192, no other C. arietinum-nodulating ICEs have been detected in Australian soils.

Although Australian native legumes are nodulated by diverse root nodule bacteria genera, such as Bradyrhizobium, Paraburkholderia, Rhizobium and Sinorhizobium, there have been comparatively few symbiotic Mesorhizobium identified (Hoque et al. 2011; Lafay and Burdon 1998, 2007; Walker et al. 2014b; Yates et al. 2004). Mesorhizobium have been isolated from areas of south-eastern Australian as symbionts of several Acacia spp., Goodia lotifolia and Oxylobium sp. (Hoque et al. 2011; Lafay and Burdon 1998) and of Indigofera spp. from Kakadu National Park in the Northern Territory (Lafay and Burdon 2007). However, these organisms appear to have been lost from their respective culture collections, and because their isolation pre-dated large-scale genome sequencing, they could not be genotypically compared to the Mesorhizobium strains reported in this study. In contrast, from soils collected at just seven sites across Australia’s 154,000 km2 south-west agricultural region, 126 non-symbiotic Mesorhizobium (i.e. lacking any nod, nif or fix genes) were isolated from soils, and these strains clustered into 22 genospecies based on their core genome phylogeny, with most genospecies likely representing new Mesorhizobium spp. (Colombi et al. 2023). In vitro ICE transfer experiments with a subset of these non-sym strains also showed that some, but not all, were able to acquire ICEMjSymR7A from symbiotic M. japonicum R7A, converting them into Lotus-nodulating microsymbionts. Therefore, although symbiotic Mesorhizobium spp. are rarely isolated from native legumes, Australian soils appear replete with a wide diversity of non-symbiotic Mesorhizobium spp, which can act as recipients for symbiosis ICE transfer from introduced inoculants. This is likely the origin of the novel C. arietinum-nodulating rhizobia reported in this work from the ORIA.

WSM4904 was the only strain tested that was ineffective at fixing N2 with C. arietinum, while the genetically similar WSM4906 was an effective microsymbiont. Both strains were isolated from separate nodules from the same plant and yielded similar nodule morphology on C. arietinum. The ICEMcSym1192 in both strains were almost identical in sequence with only one unique mutation of a single base pair deletion within a putative integrase gene. While it would render this gene inactive, it is unlikely to be the cause of the poor symbiotic performance of WSM4904. The genetic cause of this difference in effectiveness is most likely due to sequence divergence in the WSM4904 and WSM4906 chromosomes, or in the approximately 300-kb non-sym ICE detected in each chromosome. Indeed, ICEMsp.WSM4904 and ICEMsp.WSM4906 differ in size by almost 16-kb, and phylogenetically were more closely related to the symbiotic ICE clade than the ICE clade of non-symbiotic strains. In their analysis of 38 Mesorhizobium genomes, Colombi et al. (2021) detected 56 ICEs, comprising 34 symbiotic ICEs (i.e. they encoded sym genes) and 22 non-symbiotic ICEs (i.e. they lacked symbiosis genes). While most genomes harboured a single symbiotic or non-symbiotic ICE, some harboured multiple ICEs, including three strains (M. ciceri WSM1497, M. loti NZP2042 and M. loti TONO) with both a symbiotic and non-symbiotic ICE. ICEs are therefore reasonably abundant in the genomes of Mesorhizobium, and the co-existence of multiple ICEs in the same genome likely provides an opportunity for homologous recombination between these elements (Ramsay et al. 2023). However, whether cross-talk between expression of symbiotic and non-symbiotic ICE genes has any impact on N2 fixation, is a question that requires further investigation, and these two strains could prove useful in identifying these genetic determinants.

Since cultivation of C. arietinum began in the ORIA in 1985, M. ciceri CC1192 has been the sole inoculant strain. It is evident that the CC1192 symbiosis ICE (ICEMcSym1192) has transferred to a wide diversity of Mesorhizobium spp. that are presumably present in the soils of the ORIA. These evolved strains are dispersed across the cultivated areas and are competitive with CC1192 for nodulation of C. arietinum. The plant with nodules containing the only ineffective strain (WSM4904) found in this survey, was also carrying an effective strain (WSM4906) in a separate nodule. It is unknown whether N2 fixation for this plant was impacted by the presence of the ineffective strain. While individual strains may be ineffective, the likely impact on symbiotic N2 fixation would depend on their distribution and abundance in the rhizosphere, persistence in the soil between cropping seasons and competitive ability for nodulation. Based on the suite of strains isolated, C. arietinum cultivation is not constrained by N2 fixation when nodulated by these diverse and evolved strains. These strains are likely highly adapted to the local conditions in the ORIA, providing them with an advantage over the introduced inoculant strain CC1192.