Diversity and Geographic Distribution of Microsymbionts Associated With Invasive Mimosa Species in Southern China

In order to investigated diversity and geographic distribitution of rhizobia associated with invasive Mimosa species, Mimosa nodules and soils around the plants were sampled from five provinces in southern China. In total, 361 isolates were obtained from Mimosa pudica and Mimosa diplotricha in 25 locations. A multi-locus sequence analysis (MLSA) including 16S rRNA, atpD, dnaK, glnA, gyrB, and recA identified the isolates into eight genospecies corresponding to Paraburkhleria mimosarum, Paraburkholderia phymatum, Paraburkholeria carbensis, Cupriavidus taiwanensis, Cupriavidus sp., Rhizobium altiplani, Rhizobium mesoamericanum, and Rhizobium etli. The majority of the isolates were Cupriavidus (62.6%), followed by Paraburkholderia (33.5%) and Rhizobium (2.9%). Cupriavidus strains were more predominant in nodules of M. diplotricha (76.2) than in M. pudica (59.9%), and the distribution of P. phymatum in those two plant species was reverse (3.4:18.2%). Four symbiotypes were defined among the isolates based upon the phylogeny of nodA-nifH genes, represented by P. mimosarum, P. phymatum–P. caribensis, Cupriavidus spp., and Rhizobium spp. The species affiliation and the symbiotype division among the isolates demonstrated the multiple origins of Mimosa rhizobia in China: most were similar to those found in the original centers of Mimosa plants, but Cupriavidus sp. might have a local origin. The unbalanced distribution of symbionts between the two Mimosa species might be related to the soil pH, organic matter and available nitrogen; Cupriavidus spp. generally dominated most of the soils colonized by Mimosa in this study, but it had a particular preference for neutral-alkaline soils with low fertility whereas. While Paraburkholderia spp. preferred more acidic and fertile soils. The Rhizobium spp. tended to prefer neutral–acidic soils with high fertility soils.


INTRODUCTION
Leguminous plants are important for their ability to fixnitrogen in symbiosis with rhizobia, which makes them critical ecologically and economically. Ecologically, the symbiotic N-fixation not only supply the N nutrition to the host legume, but also enhance the soil N content by its root and shoot remnants . In addition, the wide distribution and specificity between the legume species and their microsymbionts make the same legume species form symbiosis with distinct rhizobial populations/species in the different geographic regions. Therefore, the growth of a legume plant in a certain region could enrich its corresponding rhizobia adapted to the local environment, e.g., the rhizobia are selected by both the host legumes and the soil conditions, mainly soil pH, nutrient (N, P, K and organic material) contents, and salinity . With the mentioned concern, characterization rhizobia associated with the same legume species grown in different regions will help to understand the evolution or diversification of rhizobia under the double selection from both the host plant and the soil condition, as well as help for screening the high effective rhizobial strains in agricultural sustainable development.
Mimosa species are able to form nitrogen fixing symbiotic associations with soil bacteria collectively termed "rhizobia" (Sprent, 2009;Sprent et al., 2017). Currently, rhizobia are found in two classes: Alpha-rhizobia including species in the well-known genus Rhizobium and other genera in the class Alphaproteobacteria, and Beta-rhizobia covering the symbiotic species in genera Paraburkholderia (splited from Burkholderia), Cupriavidus and Trinickia symbiotica in the class Betaproteobacteria (Gyaneshwar et al., 2011;Peix et al., 2015;Beukes et al., 2017;Sprent et al., 2017;Los Santos et al., 2018). Species in Mimosa genus and another large mimosoid genus Calliandra mainly nodulate with Beta-rhizobia, particularly Paraburkholderia and Trinickia, in its native range in South America, suggesting that the two partners co-evolved (Chen et al., 2005a;Bontemps et al., 2010;dos Reis et al., 2010;Los Santos et al., 2018;Silva et al., 2018). In addition, the three main invasive Mimosa species (M. diplotricha, C. Wright, M. pigra L., and M. pudica L., originated from the netotropics) in Asia, Australia and the Pacific region also preferred Beta-rhizobia for nodulation (Chen et al., 2001(Chen et al., , 2003a(Chen et al., ,b, 2005bLiu et al., 2007Liu et al., , 2011Liu et al., , 2012Parker et al., 2007;Elliott et al., 2009;Andrus et al., 2012;Klonowska et al., 2012;Gehlot et al., 2013;Melkonian et al., 2014), and they are closely related to the microsymbionts of Mimosa and related genera in their original regions (Bournaud et al., 2013;da Silva et al., 2012;Mishra et al., 2012;Taulé et al., 2012;Platero et al., 2016;de Castro Pires et al., 2018). For exotic nodulating legumes, access to compatible rhizobial strains in new environments is a critical factor for their successful establishment, and hence, their ability to survive and spread will depend on the presence of compatible symbionts in the soil . Several studies have indicated that invasive legumes, such as Mimosa species and Dipogon lignosus, have been introduced into their invasive environments together with their symbionts (Liu et al., 2014). Although Alpha-rhizobia have occasionally been isolated from Mimosa species in South American, they either failed to nodulate their hosts of origin or did so ineffectively (Barrett and Parker, 2006;Elliott et al., 2009;Klonowska et al., 2012;Mishra et al., 2012). In addition, Alpha-rhizobia (Rhizobium or Ensifer species) appear to be the dominant symbionts of native Mimosa spp. in central Mexico, central Brazil and India, where the soils presented neutralalkaline pH values (Wang et al., 1999;Gehlot et al., 2013;Baraúna et al., 2016;de Castro Pires et al., 2018). These discrepancies in symbiont preference between Mimosa species in different regions might be attributed to the soil characteristics, particularly pH, as the Beta-rhizobia, are highly tolerant to the low fertility acidic soils de Castro Pires et al., 2018).
The herbaceous perennial legume Mimosa pudica was first introduced into Taiwan Province of China in 1645 as an ornamental plant (Wu et al., 2003) and it has been dispersed throughout the tropical and subtropical China. It is a plant serving as valuable bio-resource for various uses, such as green manure, fodder crops, honey source, as well as a medicine used in zoster therapy (Wang, 2014) and treating kidney disease. However, this naturalized plant is highly invasive causing considerable ecological damage e.g., by affecting the growth of grass lawns and as a common exotic weed in rice paddy fields (Guan et al., 2006). Therefore, M. pudica and M. diplotricha another invasive plant without history record, are considered to be serious pests widely dispersed in wastegrounds and city suburbs of China. It has been estimatied that the access to compatible rhizobial strains in new environments is a critical factor for the successful establishment of exotic nodulating legumes . For getting the compatible symbionts in the introduced region. The invasive legumes, such as Mimosa species (see above) and the invasive papilionoid legume Dipogon lignosus (L.) Verdc., may have been introduced into their invasive environments together with the symbionts from their original region (Liu et al., 2014). Or they may form the symbiosis with rhizobia adapted to the local environment and adopted the corresponding symbiotic genes through lateral gene transfer, like the cases of chickpea (Cicer arietinum L.) rhizobia in China (Zhang et al., 2017(Zhang et al., , 2018. Previously, Burkholderia spp. and Cupariavidus taiwanensis have been isolated from Mimosa species grown in China (Chen et al., 2001(Chen et al., , 2005aLiu et al., 2007Liu et al., , 2011Liu et al., , 2012. In which C. taiwanensis was recognized as native to Taiwan and the Burkholderia spp. were estimated as rhizobia introduced together with the host plants (Chen et al., 2005a). Furthermore, preference for Cupriavidus by M. pudica and Burkkholderia by M. pigra in Taiwan (Chen et al., 2005b), while Cupriavidus by M. diplotricha and Burkkholeria by M. pudica in Yunnan (Liu et al., 2012) demonstrated that both the host plants and the geographic regions affected the symbiosis combination between the legume and the rhizobia in the introduced regions. However, no soil conditions were considered in these previous studies about the Mimosa rhizobia in China.
In order to explore how the environmental factors and the host species influenced the composition and competitiveness of Mimosa symbionts, we performed this study to investigate the Mimosa symbionts in in different geographic regions for evaluating the competitiveness of different rhizobial species associated with invasive Mimosa spp. under varied soil traits (organic matter, N, P, K, and pH).

Isolates and Strains
Root nodules were sampled from M. pudica and two varieties of M. diplotricha var. inermis (Adelbert) Verdcourt. and var. diplotricha, growing in 25 locations of five Chinese provinces in the subtropical and tropic regions (Figure 1), as described previously (Liu et al., 2007). Root nodules were collected from three plant individuals at each site and were stored over silica gel in closed vials until their isolation in the laboratory (Vincent, 1970). Root nodule bacteria were isolated and purified from the nodules on yeast mannitol agar (YMA) using the standard procedure (Vincent, 1970). The nodule isolates obtained in this study were maintained in yeast mannitol broth (YMB) supplied with 20% (v/v) glycerol at −80 • C.

Molecular Typing of Rhizobia
For grouping the isolates by genomic analysis, total DNA of each isolate and the reference strains Paraburkholderia mimosarum LMG23256 T , Paraburkholderia phymatum LMG21445 T , Paraburkholderia caribensis LMG18531 T , Cupriavidus taiwanensis LMG19424 T , Cupriavidus sp. SWF66294 (Liu et al., 2011) was extracted from 5 mL of culture in YMB (Vincent, 1970). The extracted genomic DNA was used as template DNA for BOX-AIR and the PCR-based RFLP (restriction fragment length polymorphism) of 16S rRNA gene (rDNA). The rDNA primers were fD1 (5 -AGAGTTTGATCCTGGCTCAGA-3 ) and rD1 (5 -AAGGAGGTGATCCAGCC-3 ) (Weisburg et al., 1991). The BOXAIR primer 5 -CTA CGG CAA GGC GAC GCT GAC G-3 (Versalovic et al., 1991) was used for BOX-PCR. Both PCRs were carried out in a total volume of 25 µL of the reaction mixture with the PCR procedure of Nick et al. (1999) and the products were checked by electrophoresis in 1% (w/v) of a garose gel. For analysis of RFLP, aliquot (5-10 µL, depending on the concentration) of PCR products was digested separately with the restriction endonucleases Hae III (GG|CC), Rsa I (GT|AC), Hif I (G|ANTC), and Msp I (C|CGG) (Laguerre et al., 1994) as specified by the manufacturer with an excess of enzyme (5 U per reaction). The restriction fragments were separated by horizontal electrophoresis in agarose (2%, w/v) gels (14 cm in length) at 80 V for 3 h and were visualized by staining with ethidium bromide. Strains or isolates with different RFLP patterns were designated into distinct rDNA types. The BOX-AIR products were separated by electrophoresis in 1.5% (w/v) agarose gels containing ethidium bromide and were photographed under UV light. The BOX profiles were distinguished by their different band patterns, e.g., the isolates sharing the same pattern were designed as the same BOX pattern.

Characterization of Whole Cell Protein by SDS-PAGE
Bacterial strains were grown until the end of the exponential phase at 28 • C for 2 days on YMA. The cells were collected and washed twice in 10 mM Tris-HC1, pH 7.6, the pellet obtained by centrifugation (5000 × g, 10 min at 4 • C) was weighed and the cells were resuspended in 10 mM Tris-HC1 to a concentration of 10 mg ml −1 using. Then, the same volume of 2 × treatment buffer (0.5 g of SDS, 3 ml of glycerol, 1 ml of 2-mercaptoethanol, 4 mg of bromophenol blue, 2 ml of 1 M Tris-hydrochloride, and distilled water to make a final volume of 10 ml at pH 6.8) was added. The samples were incubated at 100 • C for 20 min and immediately stored at −20 • C after cooled on ice. The SDSpolyacrylamide gel (200 mm × 200 mm × 1 mm) were used for electrophoresis according to Laemmli (1970). The samples were incubated at 100 • C for 10 min before the sample loading. Twenty-five samples per gel were subjected to the discontinuous slab gel electrophoresis at 250 V in an SDS-Tris-glycine buffer system, as described by Laemmli (1970). The protein patterns were visualized by silver staining (Tan et al., 1997). The protein bands were scanned with a Densitometer Extra-Scanner and strains sharing the identical band patterns were designed into the same SDS-PAGE pattern.

Phylogenetic Analyses of Housekeeping Genes and Symbiotic Genes
The 16S rRNA amplified by PCR as described above was purified and sequenced directly (Weisburg et al., 1991) commercially in the Beijing Genomics Institute (BGI). The sequences acquired in this study were aligned with related sequences extracted from GenBank using Clustal W (Thompson et al., 1997). Maximum likelihood phylogenetic trees were constructed and were bootstrapped with 1000 pesudo-replicates using Mega 6.1 (Tamura et al., 2013).
Multilocus sequence analysis (MLSA) based on the five housekeeping genes atpD (encoding for the ATP synthase betachain), recA (recombinase A), dnaK (DnaK chaperone), gyrB (DNA gyrase, beta-subunit), and glnA (glutamine synthetase I) widely used to differentiate rhizobial species (Vinuesa et al., 2005;Martens et al., 2007Martens et al., , 2008 was also employed in the present study. The five genes were independently amplified using corresponding primer pairs reported in previous studies (Payne et al., 2005;Vinuesa et al., 2005;Martens et al., 2008), or designed in this study (Supplementary Table S1). The PCR products were checked by electrophoresis in 1% (w/v) agarose gel. After purified with the Solarbio DNA purification kit (Beijing Solarbio Science and Technology Co., Ltd.), the amplicons were sequenced directly using the same primers in BGI mentioned above. The separated sequences of recA and the combined sequences of atpD, glnA, gyrB, and dnaK, and their combined sequences were aligned using Clustal W with those from type strains of the defined bacterial species (obtained from the NCBI database). Distance calculation and construction of the gene phylograms were performed using the Maximum likelihood method and the bootstrapping algorithms with 1000 pseudo-replicates were carried out in MEGA 6.0 (Tamura et al., 2013). Phylogenies were also constructed using the concatenated sequences of 16S rRNA and the five housekeeping genes by Maximum likelihood method.
Fragments of the symbiosis genes nifH and nodA genes were amplified and sequenced using primers reported previously (Haukka et al., 1998;Laguerre et al., 2001;Liu et al., 2012) as well as with the new primers designed in this study (Supplementary Table S1). The visualization purification and sequencing of the nifH and nodA amplicons were performed same as that mentioned for the housekeeping genes. The sequences were deposited in the NCBI database and were used for alignment and construction of the phylogenies using the same methods described above for the 16S rRNA gene.
The obtained nucleic acid sequences were submitted in GENEBANK, and the accession numbers in this paper was MT337483 as list in Supplementary Table S2.

Nodulation Tests
A total of 98 representative strains were used in the nodulation tests that were selected according to their affiliations of genotypes based on the results of 16S rRNA sequencing, protein patterns in SDS-PAGE, and genomic fingerprinting by BOX-PCR. M. pudica seeds were scarified using concentrated sulfuric acid for 10 min, rinsed several times with sterile water, and then surface-sterilized in 3.2% (w/v) sodium hypochlorite followed by several rinses with sterile water. They were then placed on 0.8% wateragar at 4 • C for 3 days, and after germinated at 28 • C until the seedlings developed roots of 0.5-1 cm in length. Two seedlings then were transplanted into a sterile glass tubes (30 cm × 200 cm) with nitrogen-free plant nutrient solution (Vincent, 1970) in 0.8% agar. The seedlings were then inoculated separately with 0.1 mL liquid cultures of each test strain (about 10 8 cells mL −1 ). Five replicates were used and controls without inoculation were included. The plants were placed in a growth cabinet under conditions described previously (Zhang et al., 2012). The representative strains Paraburkholderia spp. SWF66044, SWF66029, and C. taiwanensis SWF66166, SWF66194, and SWF66322 from Liu et al. (2012) were also used for cross-inoculation tests with ten other leguminous species: Glycine max (Linn.) Merr., Pisum sativum L., Galega officinalis L., Phaseolus vulgaris Linn., Vigna unguiculata L. Walp, Lotus corniculatus L., Medicago sativa L., Trifolium repens L., Macroptilium atropurpureum (Moc. and Sessé ex DC.) Urb. and Leucaena leucocephala (Lam.) de Wit. Seed treatments and inoculation details were the same as described above. Plants were checked for nodule formation at 35 d after inoculation.

Correlation Between Soil Types and Distribution of Rhizobial Groups
In order to evaluate the influence of soil characters on the symbiosis between Mimosa spp. and different rhizobial types, soil, and root nodules were sampled intensively from 13 locations Frontiers in Microbiology | www.frontiersin.org (59 sites) including Hepu, Beihai, and Nanning city in Guangxi (GXh,GXb, and GXn), Zhangjiang, Leizhou, Mazhang, and Foshan town in Guangdong (Gzj, Gl, Gm, and Gf), Jinhong and Mangshi town in Yunan (Yj and Ym) and Ledong, Wuzhishan, Wanning and Danzhou in Hainan (Hl, Hw, Hwn, and Hd), which were main districts for Mimosa speies and habitats for diverse rhizobia. For most locations, four or more sites with minimum distance of 5 km between them were samples, except the location Mangshi town in Yunnan where rhizobial strains were isolated from only one sampling site. Soils were sampled compositely from the root zone of nodule sampled plants (5-20 cm in depth). The soil samples were dried and milled until they could pass through an 80-mesh sieve. Soil alkali-hydrolysable N, available P (using Bray's hydrochloric acid fluoride ammonium by extraction method), and available K (by ammonium acetate extraction plus flame photometry) were determined with the standard procedures (Du and Gao, 2006). Soil pH was measured using a pH meter (Mettler Toledo) by suspending 5 g soil in 5 mL of distilled water, and organic matter was measured using the potassium dichromate volumetric method (Du and Gao, 2006). Rhizobial isolation, and genus/rRNA type identification by PCR-based RFLP of 16S rRNA gene were performed same as mentioned above.
Based on the soil characters, the soil samples in the 59 sites were sorted into soil types by SPSS 13.0 (SPSS Inc., Chicago, IL, United States), in terms of their pH values and the nutritional characteristics, including organic matter (OM), alkali-hydrolysable N, available P, and available K.
The data was standardized, then using construct UPGMA dendrogram (Sneath and Sokal, 1973) for soil clustering. Principal component analysis (PCA) on a correlation matrix was used to evaluate the distribution of the different rhizobial rRNA types in the 59 sites to see if they correlated with the soil characteristics. Data analysis and graphs were performed using Past 3.0.

Isolation and Genotyping of the Rhizobia
In total, 361 strains were isolated from the nodules of M. pudica and M. diplotricha sampled in the 25 locations in southern China (Figure 1 and Table 1). The majority of the isolates were obtained from M. pudica (83.7%), and minor from M. diplotricha (16.3%), which most likely reflects the relative abundance of these two plant species in the sampling sites. By 16S rRNA PCR-RFLP analysis, six rRNA types were revealed (Table 1), which were recognized as members of Paraburkholderia (three rRNA types with 55, 57, and 9 strains; 33.5%), Cupriavidus (two rRNA of types with 98 and 128 strains; 62.6%) and Rhizobium (a single rRNA type with 9 strains; 3.9%). Paraburkholderia genotypes I and II, Cupriavidus genotypes I and II, as well as Rhizobium genotype were isolated from both M. pudica and M. diplotricha; while. Paraburkholderia genotype III was only isolated from M. pudica.
The multivariate statistical analysis (Supplementary Table S4) for conducting population distribution of the six genotypes associated with M. pudica and M. diplotricha in the four provinces (Guangdong, Guangxi, Hainan, Yunnan, and Yunnan data also from previous study in Liu et al., 2012) of China showed no significant difference (p = 0.094, >0.05), but it was significantly different (p = 0.039, <0.05) for distribution of the three genera Paraburkholderia, Cupriavidus, and Rhizobium in the four provinces, for example, the Rhizobium isolates were mainly from Hainan.

Fingerprinting of the Isolates for Estimation of Genetic Diversity
In analyses of genetic diversity, a total of 97 protein profiles and 51 BOX PCR profiles were distinguished among the 361 strains (Table 1), revealing a high level of diversity among them. Paraburkholderia rRNA type I contains 18 protein profiles and 10 BOX PCR profiles; Paraburkholderia rRNA type II contains 11 protein profiles and 6 BOX PCR profiles; and Paraburkholderia rRNA type III contains 3 protein profiles and 3 BOX PCR profiles; Cupriavidus rRNA type I contains 27 protein profiles and 16 BOX PCR profiles; and Cupriavidus rRNA type II contains 31 protein profiles and 9 BOX PCR profiles. Rhizobium rRNA type contains 7 protein profiles and 7 BOX PCR profiles. The greater number of patterns in the Cupriavidus populations suggested that they were more diverse than the Paraburkholderia populations.

Phylogenies by MLSA and Affiliation of the Isolates
Out of the total collection, 34 strains (Supplementary Table S2) were selected according to their different rRNA types, protein patterns, BOX profiles, host species and collected sites for full 16S rRNA and housekeeping gene sequencing. The relationships of the Mimosa isolates were high similar in 16S rRNA gene phylogeny (Supplementary Figure S2) and in the MLSA-based phylogeny deduced from the concatenated sequences of 16S rRNA and the five housekeeping genes ( Figure 2B). Five groups were defined among the betarhizobial isolates at the species level (similarities ≥96.4%), which corresponded to (1) P. mimosarum (Paraburkholderia rRNA type I), (2) P. phymatum (Paraburkholderia rRNA type II), (3) P. caribensis (Paraburkholderia rRNA type III), (4) C. taiwanensis (Cupriavidus rRNA types I), and (5) Cupriavidus sp. SWF66294 (Cupriavidus rRNA types II). The Rhizobium isolates obtained in this study were grouped into three species corresponding to R. etli, R. mesoamericanum, and R. altiplani. The phylogenies of the individual housekeeping genes (atpD, recA, dnaK, gyrB, and glnA) (Supplementary Figures S1-S5) were generally consistent with that of 16S rRNA gene and the MLSA (Figure 2), except the isolates of R. etli that was a unique linage separated from all the defined species in atpD phylogenetic tree (Supplementary Figure S3).

Sequencing and Phylogenetic Analysis of Symbiosis Genes
This analysis were performed for the 34 representative strains mentioned above. Four nodA and four nifH lineages were defined among them (Figure 3 and Supplementary Figure S6). The phylogenies of the nodA and nifH genes of the isolates were the   same and they were incongruent in several cases with that of the housekeeping genes, e.g., both the Alpha-and Beta-rhizobia formed two clades and they were intermingled. P. caribensis strains and P. phymatum strains shared similar symbiosis genes, while P. mimosarum presented another lineage (Figure 3 and Supplementary Figure S6). The strains in both Cupriavidus genotypes I and II, including C. taiwanensis LMG19424 T , formed the third lineage in the symbiosis gene phylogeny, which was inserted between the two lineages of the Paraburkholderia species.
All the three Rhizobium species identified in the present study shared the same symbiosis gene and formed the forth lineage represented by that of R. mesoamericanum STM3625 isolated from M. pudica in Mexico.

Nodulation Test
All 98 representative strains selected from different groups according to their patterns in 16S rRNA PCR-RFLP, SDS-PAGE protein and BOX-PCR profiles (  Table S3).
The locations and the soil parameters within each sampling site were analyzed via cluster analysis by SPSS and PCA, which revealed main four groups with distinct soil patterns, an unusual soil site as another type for Cupriavidus spp. surviving (because it is distant from other soil point) was not considered (Figure 4). We obtained 200 rhizobial strains from the 59 sites and they were identified into four groups as Cupriavidus spp., P. mimosarum, P. phymatum, and Rhizobium spp., and each group grown soil nutrition and pH range displayed as Table 2.
In PCA, only the existence (not abundance) of the different group in each site was considered, and they were in relation to their spatial distribution (Figure 4 and Table 2). The soil PCA resulted in three components with eigenvalues greater than one which explained 75.3% of the total variance (first component: 47.7%, second component: 27.6%, third component: 14.2%). However, the third component did not provide any further information above the first two components, and hence was excluded from the interpretation. The first component revealed the positive correlation of OM and available N, and a contrasting correlation of these with pH (loading factors = 0.58, 0.60, and 0.37, respectively) with soil types I, II, III, and IV; the second component is characterized by a positive correlation of available P and available K (loading factors = 0.55, 0.73). On the PCA scatter plot, the four soil patterns are almost completely separated.
In total, the 59 sampling sites were plotted onto the soilsite PCA (Figure 4), and the rhizobial species that were isolated from each site and isolates numbers are indicated in Supplementary Table S3. Four soil types corresponded to the localization site of the different rhizobial types associated with Mimosa in southern China. Soil category I (32 sites, 115 isolates) was the major soil type characterized by low fertility, relatively high available P, and neutral-alkaline pH and occupied by the majority of the Mimosa symbionts obtained in this study. In this type of soil, the rhizobial community contained 37.4% Cupriavidus, 23.5% P. mimosarum, 23.5% P. phymatum and 15.5% Rhizobium spp. strains which covered 75% of the Cupriavidus isolates, 50% of the P. mimosarum, 43% of the P. phymatum and 62% of the Rhizobium spp. strains. Soil category II (10 sites, 28 isoates) harbored 35.7% Cupriavidus, 17.9% P. mimosarum, 35.7% P. phymatum, and 10.7% Rhizobium spp. strains; it characterized by intermediate fertility and neutral pH. Soil category III (11strains, 37 isoates) tended to acidneutral pH, with lower fertility and lower available P and K; it contained 2.7% Cupriavidus (1 strain), 16.2% P. mimosarum, 48.6% P. phymatum, and 32.5% Rhizobium spp. strains. Soil category IV (5 sites, 17 isoates) was quite acidic with high fertility, but with low available P; it harbored the rhizobial community with 17.6% Cupriavidus, 52.9% P. mimosarum and 29.5% P. phymatum strains.
Within all the 59 sites, Cupriavidus strains habitat in 29 soil sites (49.2% of total sites), but occupied 75%, 17.8%, 3.6, and 3.6% of the sites in soil categories I through IV, respectively. P. mimosarum strains habitat in 22 soil sites (37.3% of total sites), and scattered on 50, 18, 18, and 14% of the sites in categories I through IV, respectively, P. phymatum strains survive in 23 soil sites (about 39% of total sites), dispersed 43, 22, 26, and 14% of the sites in the soil categories I through IV, respectively, Rhizobium spp. strains habitat in 13 soil sites (22% of total sites) belonging to the soil categories I through III, appearing in 61, 3, and 12% of the sites, respectively.

Mimosa-Nodulating Rhizobial Community in Southern China
Based upon the MLSA resules (Figure 2), the six rRNA types of Mimosa rhizobia defined by PCR-based RFLP of 16S rRNA gene (Table 1), could be indentified as 8 species: Paraburkholderia rRNA type I as P. mimosarum, Paraburkholderia rRNA type II as P. phymatum, Paraburkholderia rRNA type III as P. caribensis, Cupriavidus rRNA types I as C. taiwanensis, Cupriavidus rRNA types II as Cupriavidus sp., Rhizobium sp. genotype as R. etli, R. mesoamericanum, and R. altiplani. These identifications might imply that the PCR-RFLP of 16S rRNA is an efficient method to identify the current beta-rhizobial species, but it is unable to differentiate the alpha-rhizobial species, as evidenced in many previous studies (for example, Huo et al., 2019;Li et al., 2019).
Although four of the six beta-rhizobial genotypes/species defined among the symbionts of Mimosa diplotricha and M. pudica in southern China (Table 1) were commonly associated with Mimosa species in both their native and invasive regions. The Cupriavidus sp. represented by strains SWF66294 covering 128 isolates was unique in China. The presence of identical nodA and nifH in Cupriavidus sp. and in C. taiwanensis might be evidence that lateral transfer of symbiosis gene between these two species has happened, as reported in the Lotusnodulation Mesorhizobium species (Bamba et al., 2019). The situation in the three Rhizobium species was similar. In addition, the intermingling of the Alpha-and Beta-rhizobial clades in the phylogenies of symbiosis genes (Figure 3 and Supplementary Figure S6) also demonstrating the lateral transfer of symbiosis gene between these two rhizobial categories, as estimated previously (see review of Andrews et al., 2018). Therefore, new symbionts of Mimosa has been evolved in China under the double selection from host plant (for symbiotic gene background) and the soil conditions (for survival in the local sites). Previously FIGURE 4 | Comparison of the distribution of four communities of rhizobial types associated with invasive Mimosa species in southern China by principal component analysis of total N, available K, available P, precipitation and pH. The principal component one revealed correlations of available K with available P and pH. The principal component two is characterized by precipitation and total N and organic matter. Ellipses represent 90% confidence limits. The word in text box means each occupied sites percent and species constitute in different soil types. The data use F-test by SPSS17.0, the result showed only soil pH was significant among different groups (p = 0.001, ≤0.01), different letters means subset by Turkey test. Gehlot et al. (2013) estimated that invasive Mimosa spp. do not interact with the symbionts of native legumes; however, the Mimosa species could get novel symbiosis adapted to their invaded region by lateral transfer of the symbiosis genes from its known rhizobia to the native relatives. Cupriavidus sp. may constitute a new species intermediate between C. taiwanensis and C. nantongensis (Sun et al., 2016), but more analyses are required to clarify its species affiliation, such as DNA-DNA hybridization (DDH) and/or average nucleotide identity (ANI) with the closest type strains. Cupriavidus taiwanensis is a very common symbiont of invasive Mimosa species in South East Asia (Chen et al., 2001(Chen et al., , 2003b(Chen et al., , 2005aElliott et al., 2009;Andrus et al., 2012;Liu et al., 2012Liu et al., , 2011Gehlot et al., 2013), and it could be the dominant symbiont in some locations for M. diplotricha and M. pudica, but not for M. pigra (Chen et al., 2001(Chen et al., , 2003b(Chen et al., , 2005bKlonowska et al., 2012). So, different Mimosa species may have distinct preferences for their symbionts (Chen et al., 2005a,b;Elliott et al., 2009). Although Cupriavidus is less commonly isolated in the native regions of Mimosa, C. taiwanensis has been isolated from M. pudica (Barrett and Parker, 2006;Mishra et al., 2012) and M. asperata  in Americas. Moreover, Cupriavidus strains closely related to C. necator and C. pinatubonensis were the dominant symbionts of native Mimosa in Uruguay (Platero et al., 2016). All these results imply that C. taiwanensis might originated in the Americans and was introduced to China together with the Mimosa plants.
In the present study, P. mimosarum, P. phymatum, and P. caribensis comprised 15.2, 15.8, amd 2.5% of the total isolates from 14, 11, and 3 sample sites, respecitively, P. mimosarum and P. phymatum have wide distribution in both the invaded and the original regions of the Mimosa species and P. caribensis (Bontemps et al., 2010;dos Reis et al., 2010;Chen et al., 2005aChen et al., ,b, 2006Elliott et al., 2007Elliott et al., , 2009Liu et al., 2011Liu et al., , 2012Mishra et al., 2012;Gehlot et al., 2013;Lardi et al., 2017). In addtion, its greater abundance in Yunnan Province (Liu et al., 2011(Liu et al., , 2012 than in neighbor provinces examined in the present study demonstrated that P. mimosarum strains were more adapted to the soil in Yunnan. P. caribensis was originally described for non-symbiotic strains isolated from soil (Achouak et al., 1999) and symbionts of Mimosa belonging to this species were isolated lately in China (Chen et al., 2003b;Liu et al., 2012). So, again, its possible that novel symbiont of Mimosa may have evolved in China after this plant was introduced.
Rhizobium was isolated as minor group from the Mimosa species in this study ( Table 1), and most of them (12 of the 14 strains) were from Hainan Province, with few isolates from Yunnan (Liu et al., 2012) and Guangxi ( Table 1). Based upon the MLSA results ( Figure 2) these strains were idertified as R. mesoamericanum, R. etli, and R. altiplani, while all of them were identified as sv. mimosa according to the nodA and nifG phylogenies (Figure 3 and Supplementary Figure S6). Previously, R. etli sv. mimosae has isolated from invasive Mimosa species in low frequency (Chen et al., 2001(Chen et al., , 2003b(Chen et al., , 2005aElliott et al., 2009;Klonowska et al., 2012;Mishra et al., 2012;Melkonian et al., 2014), while both R. etli and R. mesoamericanum were found to be predominate in Mimosa symbionts in Mexico, the second largest center of Mimosa diversity (Wang et al., 1999;Bontemps et al., 2016). The exception is R. altiplani, which is relatively common in central Brazil (Baraúna et al., 2016;de Castro Pires et al., 2018), but has never previously been isolated from Mimosa in its invasive range regions. So, the Mimosanodulation Rhizobium species in China might be also introduced together with their hosts, but they are not so adapted to the conditions in the invasive regions.
In addtition to the species definition, great genetic diversity represented by the 94 protein patterns and 63 BOX-PCR patterns ( Table 1) was revealed in this study among the Alpha-and Beta-rhizobial symbionts isolated from only two Mimosa species. This might be related to the vast area of the sampling locations, which covered diverse soil types and forced the diversification of rhizobia for their survival.
In summary, five beta-rhizobial species, especially Cupriavidus genotype II, with great genetic diversity as the dominant microsymbionts and three Rhizobium species as minor microsymbionts for M. pudica and M. diplotricha plants were detected in a vast sampling area in China. Most of them have been found in both the invasive regions and the centers of origin of these plants, but Cupriavidus genotype II might be a novel symbiont for Mimosa species evolved in China.

Soil Variables and the Distribution of Rhizobial Genotypes Associated With Mimosa Species
The worldwide distribution of various symbionts isolated from invasive Mimosa species may be the result of selection by soil characteristics and other ecological factors (Elliott et al., 2009;Melkonian et al., 2014;de Castro Pires et al., 2018). In the present study, the community composition in the five provinces were unbalanced: P. mimosarum was more abundant in Yunnan; P. phymatum was more in Guangdong and Guangxi; Rhizobium stains were mostly from Hainan; and the minor group P. caribensis was only isolated from M. pudica plants grown in Guangdong and Guangxi. These geographic distributions implied an interaction among the plants, the rhizobial species, and the environment factors, as described for rhizobia associated with soybean .
In the SPSS analysis and the principal component analysi (Figure 4), the correlation of Cupriavidus spp. and Rhizobium spp. with the infertile alkaline soil type (category I) and neutral soil (category II), and the accommodating of Paraburkholderia spp. in more acidic soils (category III and IV) were consistent with the previous reports that have shown that C. taiwanensis is abundant as a Mimosa symbiont in neutral to basic pH soils ( Table 2), often with relatively high fertility (Elliott et al., 2009;Klonowska et al., 2012;Mishra et al., 2012;Gehlot et al., 2013). In contrast to Cupriavidus, Paraburkholderia strains can tolerate acidic soils (Stopnisek et al., 2014), and indeed become dominant symbionts with the compatible legumes; it was similar for rhizobia nodulating with mimosoid and papilionoid legumes (Garau et al., 2009;dos Reis et al., 2010;Howieson et al., 2013;Liu et al., 2014;Lemaire et al., 2015Lemaire et al., , 2016a. The present study has clearly illustrated that pH appears to be the most important environmental factor in helping to explain the distribution of Mimosa symbionts in southern China (Table 2, p = 0.001, ≤0.01). For example, although higher soil fertility and N concentration improved the competitive nodulation of Paraburkholderia over Cupriavidus and Rhizobium on Mimosa spp. in reduced N concentrations/low fertility growth media (Elliott et al., 2009), this was not the case in the present study wherein the dominant symbionts in the alkaline-neutral soils (soil types I and II) were overwhelmingly Cupriavidus regardless of their low fertility. The opposite was the case for the acidic soils wherein the Paraburkholderia strains dominated even though the soils were relatively fertile.
Taken together, Cupriavidus spp. are the most common and competitive rhizobial type in southern China due to its ability to grow in the widest range of pH, soil nutrition/fertility, and soil moisture levels (from drought to flooding), and maybe also to its tolerance to heavy metals . C. taiwanensis was recognized as strong stress resistant bacteria able to survive and grow at phenol concentrations up to 900 mg/L (Chen et al., 2004). It is also particularly dominant in islands like Taiwan (Chen et al., 2003b), New Caledonia , and Hainan (Table 1). Certainly it is possible that the similar climate and soils in these islands have created habitats ideal for the two invasive Mimosa species (M. pudica, M. diplotricha), where the soils are also rich in P and K, and would favor C. taiwanensis rather than (for instance) Rhizobium.
On the other hand, the dominance of Cupriavidus as a Mimosa symbionts appears to be a phenomenon that occurs mainly in invasive ecosystems. For example, in some natural ecosystems, such as central Mexico (Bontemps et al., 2016), the Indian Thar Desert (Gehlot et al., 2013), and even in parts of central Brazil (Baraúna et al., 2016;de Castro Pires et al., 2018), where soils are neutral to alkaline, Alpha-rhizobia can be the dominant symbionts of the native/endemic Mimosa species. Indeed, Cupriavidus species are normally minor group or absent in their centers of origin (Bontemps et al., 2010;dos Reis et al., 2010), de Castro Pires et al., 2018. The exceptions come from Texas where a widespread species M. asperata was nodulated only with C. taiwanensis-like bacteria , and from Uruguay in which native/endemic Mimosa species grown in slightly acidic soils of a heavy metal mining area were exclusively nodulated with C. necator-and C. pinatubonensis-like bacteria with nod genes divergent from C. taiwanensis (Platero et al., 2016). Interestingly, the Uruguayan native/endemic Mimosa appeared to be incapable of nodulating effectively with Paraburkholderia strains suggesting that they had co-evolved with their Cupriavidus symbionts in a manner similar to that described for species in central Brazil and central Mexico with Paraburkholderia and Alpha-rhizobia symbionts, respectively (Bontemps et al., 2010(Bontemps et al., , 2016. In summary, Cupriavidus strains are highly adaptable and competitive symbionts of the two Mimosa species in an invasive context, particularly when soils are neutral-alkaline, but they even retain a high degree of competitiveness in soils less optimal for it (i.e., slightly acidic and with low levels of OM and N). The disparity between the dominance of Cupariavidus in an invasive environment, and its sporadic occurrence in their native regions is still not clearly explained, but it could be related to the fact that most Mimosa species are not capable of nodulating effectively with C. taiwanensis (Elliott et al., 2007;dos Reis et al., 2010). Interestingly, its preferred hosts, M. pudica and M. diplotricha, although common in lowland areas of the neotropics Parker, 2005, 2006;Mishra et al., 2012), are mainly restricted to disturbed sites (Bontemps et al., 2010;Baraúna et al., 2016). Therefore, it is possible that the same soil factors which encouraging the invasiveness of its Mimosa plants (fertile ground subject to anthropogenic disturbance) also created a niche which favoring C. taiwanensis over its usual competitors for nodulation in its native region (i.e., Paraburkholderia species).

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the Supplementary Material, Supplementary Table S2.

AUTHOR CONTRIBUTIONS
XYL conceived and designed the study. SHY, BJY, HYW, and FW collected the nodules from Mimosa and isolated the rhizobia strains. WDC and ZKL performed the soil nutrients detection, HJL conduct the PCA analysis, XYL and EKJ wrote and edited the manuscript. All authors read and approved the manuscript.

FUNDING ACKNOWLEDGMENTS
We thank Clare Gough (Laboratoire des Interactions Plantes-Microorganismes (LIPM), Institut National de la Recherche Agronomique, Castanet-Tolosan, France) for reviewing the manuscript. At end, we also thank En Tao Wang (ENCB-IPN, Mexico) a lot for helping to revise this manuscript.