• rhizobia;
  • endophytic bacteria;
  • nodulation;
  • plant growth promoting activity;
  • amplified fragment length polymorphism;
  • biogeography


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information

A total of 159 endophytic bacteria were isolated from surface-sterilized root nodules of wild perennial Glycyrrhiza legumes growing on 40 sites in central and northwestern China. Amplified fragment length polymorphism (AFLP) genomic fingerprinting and sequencing of partial 16S rRNA genes revealed that the collection mainly consisted of Mesorhizobium, Rhizobium, Sinorhizobium, Agrobacterium and Paenibacillus species. Based on symbiotic properties with the legume hosts Glycyrrhiza uralensis and Glycyrrhiza glabra, we divided the nodulating species into true and sporadic symbionts. Five distinct Mesorhizobium groups represented true symbionts of the host plants, the majority of strains inducing N2-fixing nodules. Sporadic symbionts consisted of either species with infrequent occurrence (Rhizobium galegae, Rhizobium leguminosarum) or species with weak (Sinorhizobium meliloti, Rhizobium gallicum) or no N2 fixation ability (Rhizobium giardinii, Rhizobium cellulosilyticum, Phyllobacterium sp.). Multivariate analyses revealed that the host plant species and geographic location explained only a small part (14.4%) of the total variation in bacterial AFLP patterns, with the host plant explaining slightly more (9.9%) than geography (6.9%). However, strains isolated from G. glabra were clearly separated from those from G. uralensis, and strains obtained from central China were well separated from those originating from Xinjiang in the northwest, indicating both host preference and regional endemism.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information

Glycyrrhiza, ‘sweet root’, originating from Greek ‘glykos rhiza’, is a leguminous plant genus that contains about 20 species native mainly to Eurasia (Lewis et al., 2005). In China, roots and rhizomes of three Glycyrrhiza species, Glycyrrhiza uralensis Fish., Glycyrrhiza glabra L. and Glycyrrhiza inflata BAT, are used for medical purposes and in food. In general, G. glabra (European liquorice) is a Mediterranean species with an eastern boundary of Iran, Iraq, Central Asia and the northwestern part of China. Glycyrrhiza uralensis (Chinese liquorice) is found further east, growing in Central Asia, Mongolia and China. Glycyrrhiza inflata is found mainly in the Chinese province Xinjiang (Kondo et al., 2007; Hayashi & Sudo, 2009; Zimnitskaya, 2009). Glycyrrhiza uralensis and G. glabra are halophytic perennial shrubs that grow to a height of 0.5–1.5 m. They have deep fusiform root systems, which in arid areas might be several meters long (Kushiev et al., 2005). In China, Glycyrrhiza species grow on dry grassy plains and sunny mountainsides in the northwestern provinces, where G. uralensis and G. glabra favour slightly alkaline and sandy soils.

Both in China and Europe, liquorice is a top-selling herb and is applied in sweets and food industry for its sweet taste. In China, G. uralensis, ‘the king of herbs’, is one of the most used herb species in traditional medicine. Glycyrrhiza uralensis produces glycyrrhizin, a triterpenoid compound linked to several medical and other favourable properties, e.g. expectorant and antipyretic activities as well as antibacterial and antiviral effects (Fiore et al., 2008; Hayashi & Sudo, 2009). While there exist a vast number of pharmaceutical studies on different compounds produced by Glycyrrhiza spp. (e.g. 1500 reports on glycyrrhizin), much less is known about the biology and ecology of the plant itself.

Roots and rhizomes of Glycyrrhiza have been collected from the wild. In China, collection of wild Glycyrrhiza plants was restricted in 1984 because it promoted desertification, and cultivation of G. uralensis was initiated in China in the early 1990s (Yamamoto & Tani, 2005). Glycyrrhiza can also be utilized in the rehabilitation of saline soils. In Uzbekistan, cultivation of G. glabra for 4 years in water-logged saline soils kept the water table below the critical level, decreased the soil and water salinity and successively increased the productivity of wheat and cotton crops compared to that of the local regional average (Kushiev et al., 2005).

Glycyrrhiza spp. are able to fix N2 from the atmosphere in symbiosis with rhizobia. In this paper, rhizobia are defined as bacteria capable of inducing symbiotic N2-fixing nodules in legume roots or stems. Rhizobia spend part of their life cycle as saprophytes in the soil. Although Glycyrrhiza plants are widely cultivated, e.g. in China and southern Europe, their N2-fixing capacity has not been utilized, presumably due to lack of knowledge about their symbionts. The type strain of the species Mesorhizobium tianshanense, A-1BS, was isolated from Glycyrrhiza pallidiflora in Xinjiang, China (Chen et al., 1995), but otherwise there is no information about which rhizobial species induce effective nodules on the most common liquorice species, G. uralensis and G. glabra.

Endophytic bacteria are defined as bacteria detected inside surface-sterilized plant parts or extracted from the inside of plant parts and having no visible harmful effects on plants. This definition includes internal colonizers with apparently neutral behaviour as well as symbionts (Hallman et al., 1997). In this paper, rhizobia are considered endophytic bacteria. Endophytic bacteria which are not able to form nodules on legumes are termed non-rhizobia. The host plant provides endophytes with a supply of nutrients and shelter from most abiotic stresses. In return for providing ‘bread and home’, plants may receive benefits from microbial associations by the enhancement of plant growth or reduction of plant stress, e.g. through the ability of bacterial ACC deaminase to modulate the level of ethylene produced by plants under stress (Lodewyckx et al., 2002; Hardoim et al., 2008).

The purpose of this study was to explore the diversity of rhizobia that nodulate wild Glycyrrhiza plants in China. In previous studies all strains not producing typical rhizobial colonies were considered contaminants and were discarded from further investigations. This also resulted in concomitant loss of non-rhizobial endophytic bacteria in the root nodules. However, they might also possess certain advantages, for instance assisting the symbiotic interaction between rhizobia and the host plant. Our strategy was to isolate all the rhizobia and non-rhizobial endophytes that inhabit root nodules. We also tested our non-rhizobial strains isolated from Glycyrrhiza for plant growth promoting (PGP) activities in cultures. Our third aim was to discover and identify rhizobial species that can be used as effective rhizobial inoculants in Glycyrrhiza cultivation in nitrogen-deficient soils or for rehabilitation of degraded soils.

Biogeography is the study of the distribution of biodiversity over space and time (which species, how many, where and why; Hughes Martiny et al., 2006; Ramette & Tiedje, 2007a). Very little is known about the biogeography of rhizobia isolated from nodules of particular host plants. By applying multivariate analyses, we wanted to find out whether the variation of amplified fragment length polymorphism (AFLP) genomic fingerprint patterns of the bacterial strains obtained from Glycyrrhiza nodules exhibit specific biogeographical patterns. For the distance-based discriminant analysis (db-DA), our hypothesis was that AFLP patterns would be more similar in strains originating either from one geographical region or from the same Glycyrrhiza species. To understand mechanisms that drive the colonization of Glycyrrhiza nodules by particular bacterial species, we used the canonical correspondence analysis (CCA) to explore possible relationships between bacteria and environmental variables. We assumed that the more similar the AFLP patterns of strains were, the more similar the environmental requirements of the strains. Finally, we used variance partitioning to determine how much of the total variation in AFLP patterns is explained by the original Glycyrrhiza hosts and/or the geographic locations of the sampling sites.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information


A total of 159 distinct bacterial strains, listed in Tables 1 and 2, were isolated from individual root nodules collected in June and August between 2001 and 2007 from G. uralensis, G. glabra, Glycyrrhiza squamulosa, Glycyrrhiza eurycarpa, G. inflata and a few taxonomically unidentified Glycyrrhiza plants growing on 40 sites in China. Twenty-seven sampling sites were located in the Xinjiang province in northwestern China. Because the Tianshan Mountains separate northern Xinjiang and southern Xinjiang, the sampling sites were further distinguished between them: 12 sites in the north and 15 sites in the south (Supporting Information, Fig. S1, Table S1). The rest of the sampling sites were situated relatively closely to each other in northern central China, in the Shaanxi (six sites), Gansu (four sites) and Ningxia provinces (three sites) (Fig. S1, Table S1). The distance between the closest sites in Xinjiang and central China was about 2000 km. The sampling sites were located between latitudes 34° and 48° (N) and longitudes 75° and 109° (E), and at elevations between 290 and 2510 m above sea level. These data were obtained using the global positioning system. The climate of these regions is arid or semi-arid, characterized by low rainfall, a long dry season and high evaporation, leading to high alkalization and salinization of soils. More detailed geographic and meteorological characteristics of each sampling sites are described in Table S1. Annual rainfall and annual mean temperature for each site were estimated from historical average data provided by the official regional websites. Soils of sampled sites were generally poor in organic matter and not fertilized. One plant species (several specimens) was sampled from each site. In Xinjiang, G. uralensis nodules were collected from 20 of 27 sites and G. glabra nodules were obtained from four sites. Nodules of G. squamulosa, G. eurycarpa and G. inflata were sampled only from a single site each. In central China, nodules from G. uralensis were obtained from six sites and those from unidentified Glycyrrhiza plants from seven sites (Table S1).

Table 1. List of rhizobial strains isolated from root nodules of Glycyrrhiza spp. growing in Gansu, Ningxia, Shaanxi and Xinjiang in central and northwestern China
StrainaGeographic originHost plantAFLP subgroupClosest species and strain (accession no.), max identity
  1. a

    Partial 16S rDNA was sequenced for strains written in bold.

NWXJ98Xinjiang BaichengG. uralensis1a 
NWXJ100Xinjiang BaichengG. uralensis1aRhizobium daejeonense L61T (AY341343) 97%
NWXJ114Xinjiang WuqiaG. uralensis1a 
NWXJ90Xinjiang BaichengG. uralensis1a 
NWXJ102Xinjiang BaichengG. uralensis1a 
NWXJ111Xinjiang TuofengxiaG. uralensis1a 
NWXJ79Xinjiang MaigaitiG. squamulosa1a 
NWXJ97Xinjiang BaichengG. uralensis1aRhizobium daejeonense L61T (AY341343) 97%
NWXJ58Xinjiang HabaheG. uralensis1aRhizobium daejeonense L61T (AY341343) 99%
NWXJ06Xinjiang BachuG. uralensis1b 
NWXJ09Xinjiang BachuG. uralensis1b 
NWXJ35Xinjiang ChahanG. glabra1bRhizobium cellulosilyticum ALA10B2T (DQ855276) 98%
NWXJ44Xinjiang HaermodunG. uralensis1cRhizobium cellulosilyticum ALA10B2T (DQ855276) 97%
NWXJ80Xinjiang MaigaitiG. squamulosa2Phyllobacterium sp. STM370 (AJ968698) 96%
NWXJ91Xinjiang BaichengG. uralensis2 
NWSX24Shaanxi YanglingG. uralensis3aMesorhizobium mediterraneum LMG17148T (AM181745) 97%
NWSX25Shaanxi YanglingG. uralensis3a 
NWSX19Shaanxi YananGlycyrrhiza sp.3a 
NWSX20Shaanxi YananGlycyrrhiza sp.3a 
NWXJ52Xinjiang BeitunG. uralensis3a 
NWXJ22-2Xinjiang ZiniquanG. glabra3b 
NWXJ36-2Xinjiang ChahanG. glabra3bMesorhizobium tianshanense A-1BST (AF041447) 99%
NWXJ25Xinjiang ZiniquanG. glabra3b 
NWXJ30Xinjiang ZiniquanG. glabra3b 
NWXJ27-2Xinjiang ZiniquanG. glabra3b 
NWXJ43-2Xinjiang HejingG. uralensis3b 
NWXJ28Xinjiang ZiniquanG. glabra3b 
NWXJ32Xinjiang ZiniquanG. glabra3bMesorhizobium tianshanense A-1BST (AF041447) 97%
NWGS07Gansu ChongxingGlycyrrhiza sp.3b 
NWNX05Ningxia ZhongweiGlycyrrhiza sp.3b 
NWGS10Gansu ChongxingGlycyrrhiza sp.3bMesorhizobium tianshanense A-1BST (AF041447) 97%
NWGS11Gansu ChongxingGlycyrrhiza sp.3b 
NWSX23Shaanxi YanglingG. uralensis3cMesorhizobium mediterraneum LMG17148T (AM181745) 97%
NWSX26Shaanxi YanglingG. uralensis3c 
NWSX08Shaanxi YulinG. uralensis3c 
NWSX11Shaanxi YulinG. uralensis3c 
NWGS03Gansu HuanxianG. uralensis3cMesorhizobium mediterraneum LMG17148T (AM181745) 99%
NWSX06Shaanxi ZiwulingG. uralensis3c 
NWGS01Gansu HuanxianG. uralensis3c 
NWGS09Gansu ChongxingGlycyrrhiza sp.3c 
NWSX09Shaanxi YulinG. uralensis3d 
NWSX10Shaanxi YulinG. uralensis3dRhizobium gallicum R602spT (U86343) 99%
NWGS02Gansu HuanxianG. uralensis3d 
NWSX01Shaanxi HuanglongG. uralensis3d 
NWSX02Shaanxi HuanglongG. uralensis3d 
NWSX05Shaanxi ZiwulingG. uralensis3d 
NWSX15Shaanxi YananGlycyrrhiza sp.3dRhizobium gallicum R602spT (U86343) 96%
NWXJ07Xinjiang BachuG. uralensis4a 
NWXJ12Xinjiang BachuG. uralensis4a 
NWXJ02Xinjiang BachuG. uralensis4aMesorhizobium sp. Mad-1 (EF364378) 96%
NWXJ05Xinjiang BachuG. uralensis4a 
NWXJ08Xinjiang BachuG. uralensis4a 
NWXJ14Xinjiang BachuG. uralensis4aMesorhizobium sp. Mad-1 (EF364378) 98%
NWXJ04Xinjiang BachuG. uralensis4a 
NWXJ15Xinjiang BachuG. uralensis4a 
NWXJ16Xinjiang BachuG. uralensis4a 
NWXJ17Xinjiang BachuG. uralensis4a 
NWXJ18Xinjiang BachuG. uralensis4a 
NWXJ19Xinjiang BachuG. uralensis4aMesorhizobium sp. Mad-1 (EF364378) 98%
NWXJ01Xinjiang BachuG. uralensis4a 
NWXJ42Xinjiang HejingG. uralensis4a 
NWXJ87Xinjiang MaigaitiG. squamulosa4b 
NWGS06Gansu MinqinGlycyrrhiza sp.4cSinorhizobium meliloti LMG6133T (X67222) 97%
NWSX04Shaanxi HuanglongG. uralensis4c 
NWSX13Shaanxi FuxianGlycyrrhiza sp.4c 
NWSX14Shaanxi FuxianGlycyrrhiza sp.4c 
NWXJ77Xinjiang MaigaitiG. squamulosa4c 
NWXJ78Xinjiang MaigaitiG. squamulosa4c 
NWXJ86Xinjiang MaigaitiG. squamulosa4c 
NWXJ115Xinjiang YuepuhuG. inflata4c 
NWXJ116Xinjiang YuepuhuG. inflata4cSinorhizobium meliloti LMG6133T (X67222) 99%
NWXJ81Xinjiang XiamaleG. eurycarpa4cSinorhizobium meliloti LMG6133T (X67222) 97%
NWXJ85Xinjiang MaigaitiG. squamulosa4c 
NWXJ82Xinjiang XiamaleG. eurycarpa4e 
NWXJ84Xinjiang XiamaleG. eurycarpa4e 
NWXJ83Xinjiang XiamaleG. eurycarpa4ePhyllobacterium ifriqiyense STM370T (AY785325) 97%
NWXJ71Xinjiang12tuanG. glabra4f 
NWXJ73Xinjiang12tuanG. glabra4fRhizobium giardinii H152T (U86344) 99%
NWXJ106Xinjiang TadaG. glabra4f 
NWXJ107Xinjiang TadaG. glabra4gRhizobium leguminosarum Rtl E12 (U73209) 97%
NWGS05Gansu LingtaiGlycyrrhiza sp.5aMesorhizobium amorphae ACCC19665T (AF041442) 94%
NWSX22Shaanxi YanglingG. uralensis5bUnidentified
NWXJ54Xinjiang GongliuG. uralensis5dRhizobium galegae LMG6214T (X67226) 97%
NWXJ03Xinjiang BachuG. uralensis5e 
NWXJ20Xinjiang BachuG. uralensis5e 
NWXJ21Xinjiang ZiniquanG. glabra5eMesorhizobium tianshanense ST-2 (AY225401) 99%
NWXJ22Xinjiang ZiniquanG. glabra5e 
NWXJ27Xinjiang ZiniquanG. glabra5e 
NWXJ29Xinjiang ZiniquanG. glabra5e 
NWXJ37Xinjiang ChahanG. glabra5e 
NWXJ36Xinjiang ChahanG. glabra5e 
NWXJ34Xinjiang ChahanG. glabra5e 
NWXJ38Xinjiang ChahanG. glabra5e 
NWXJ40Xinjiang ChahanG. glabra5e 
NWXJ33Xinjiang ChahanG. glabra5e 
NWXJ41Xinjiang ChahanG. glabra5e 
NWXJ31Xinjiang ZiniquanG. glabra5eMesorhizobium tianshanense ST-2 (AY225401) 99%
NWXJ26Xinjiang ZiniquanG. glabra5e 
NWXJ43Xinjiang HejingG. uralensis5eMesorhizobium tianshanense ST-2 (AY225401) 97%
NWNX01Ningxia YinchuanG. uralensis6aAgrobacterium tumefaciens LMG140T (AM181758) 96%
NWNX02Ningxia YinchuanG. uralensis6a 
NWXJ10Xinjiang BachuG. uralensis6a 
NWXJ75Xinjiang XiamaleG. eurycarpa6a 
NWXJ76Xinjiang XiamaleG. eurycarpa6a 
NWSX03Shaanxi HuanglongG. uralensis6a 
NWXJ23Xinjiang ZiniquanG. glabra6a 
NWGS04Gansu HuanxianG. uralensis6a 
NWXJ92Xinjiang BaichengG. uralensis6a 
NWXJ112Xinjiang TuofengxiaG. uralensis6a 
NWXJ88Xinjiang MaigaitiG. squamulosa6aAgrobacterium tumefaciens LMG140T (AM181758) 97%
NWXJ104Xinjiang BaichengG. uralensis6aAgrobacterium tumefaciens LMG140T (AM181758) 96%
NWXJ110Xinjiang TuofengxiaG. uralensis6a 
NWXJ109Xinjiang TuofengshangG. uralensis6b 
NWXJ53Xinjiang BeitunG. uralensis6cAgrobacterium tumefaciens LMG140T (AM181758) 96%
NWXJ108Xinjiang TuofengshangG. uralensis6c 
NWNX03Ningxia YinchuanG. uralensis6dUnidentified
NWXJ55Xinjiang GongliuG. uralensis6ePhyllobacterium bourgognense STM201T (AY785320) 97%
Table 2. List of non-nodulating endophytic bacteria isolated from root nodules of Glycyrrhiza spp. growing in Xinjiang, Gansu, Ningxia and Shaanxi, with data about their nodulation and plant growth promoting activities
Straina,b,cGeographic originHost plantAFLP subgroupClosest species and strain (accession no.) max identityIAAdP solubilizingeSiderophore productionfExo-enzymesg
  1. ng, Not growing in a medium used; nd, not done.

  2. a

    Partial 16S rDNA was sequenced for strains written in bold.

  3. b

    Strain was not able to nodulate the plant tested, G. uralensis.

  4. c

    In plant tests only strain NWXJ84 (subgroup 4e) induced ineffective, small and white nodules on G. glabra but no nodules on G. uralensis.

  5. d

    IAA = production of the plant hormone IAA. Pink color indicated indole production. Values in parentheses refer to the absorbance at 530 nm: − no production (< 0.1); + weak (0.1–0.2); ++ moderate (0.2–0.3); +++ good (> 0.3).

  6. e

    Solubilisation of phosphate: − no solubilisation; + production of acid (change in colour from blue to yellow); ++ solubilisation (a clear halo around the colony).

  7. f

    Siderophore production: − no production; + weak; ++ moderate or strong.

  8. g

    Plate assay.

  9. h

    Protease production: − no production; + weak halo around the colony; ++ clear halo around the colony.

NWXJ80bXinjiang MaigaitiG. squamulosa2Phyllobacterium sp. STM370 (AJ968698) 96%+++++
NWXJ91Xinjiang BaichengG. uralensis2 +++++
NWXJ11bXinjiang BachuG. uralensis4d +ng
NWNX04Ningxia TongxinGlycyrrhiza sp.4d ++++++
NWGS12Gansu ChongxingGlycyrrhiza sp.4d +++++++
NWXJ31-2Xinjiang ZiniquanG. glabra4d ++++++
NWSX17Shaanxi YananGlycyrrhiza sp.4dPaenibacillus amylolyticus NRS-290T (D85396) 95%ngngng
NWXJ93Xinjiang BaichengG. uralensis4d ++++
NWXJ99bXinjiang BaichengG. uralensis4d ++++
NWXJ49Xinjiang BeitunG. uralensis4d +
NWXJ51Xinjiang BeitunG. uralensis4d ++
NWXJ95Xinjiang BaichengG. uralensis4d +++
NWXJ48Xinjiang BeitunG. uralensis4d ++
NWXJ47Xinjiang BaishidunG. uralensis4d +++++
NWXJ56Xinjiang HabaheG. uralensis4d +
NWXJ57Xinjiang HabaheG. uralensis4d ++
NWXJ65bXinjiang 128tuanG. uralensis4dPaenibacillus graminis RSA19T (AJ223987) 93%ng+ng
NWXJ66Xinjiang 128tuanG. uralensis4d +
NWXJ113bXinjiang TuofengxiaG. uralensis4dPaenibacillus sp.+
NWXJ62Xinjiang HeshuoG. uralensis4d ngng
NWXJ63Xinjiang KelamayiG. uralensis4d +
NWXJ61Xinjiang HeshuoG. uralensis4d ++
NWXJ69Xinjiang TekesiG. uralensis4d +
NWXJ70Xinjiang ZeketaiG. uralensis4d ++
NWXJ67Xinjiang YishaoG. uralensis4d +
NWXJ68Xinjiang YishaoG. uralensis4d +
NWXJ74Xinjiang AlaerG. uralensis4d ndndndndndnd
NWXJ13Xinjiang BachuG. uralensis4dPaenibacillus sp.ndndndndndnd
NWSX18Shaanxi YananGlycyrrhiza sp.4d ++++
NWSX21Shaanxi YanglingG. uralensis4d ++++ng+
NWXJ64Xinjiang 128tuanG. uralensis4dPaenibacillus sp. +
NWGS08Gansu ChongxingGlycyrrhiza sp.4dPaenibacillus pasadenensis SAFN-007T (AY167820) 99%++++++
NWXJ82Xinjiang XiamaleG. eurycarpa4e +++
NWXJ84cXinjiang XiamaleG. eurycarpa4e ++++++
NWXJ83Xinjiang XiamaleG. eurycarpa4ePhyllobacterium ifriqiyense STM370T (AY785325) 97%++++
NWXJ45bXinjiang AletaiG. uralensis5cPantoea agglomerans SP1 (AF199029) 96%+++++
NWXJ96Xinjiang BaichengG. uralensis5c ++++++
NWXJ46bXinjiang BuerjinG. uralensis5fPaenibacillus polymyxa EJS-3 (DQ120522) 99%++++++
NWNX01Ningxia YinchuanG. uralensis6aAgrobacterium tumefaciens LMG140T (AM181758) 96%++++++
NWNX02Ningxia YinchuanG. uralensis6a ++++++
NWXJ10Xinjiang BachuG. uralensis6a ++++++
NWXJ75Xinjiang XiamaleG. eurycarpa6a +++++
NWXJ76Xinjiang XiamaleG. eurycarpa6a ++++++
NWSX03Shaanxi HuanglongG. uralensis6a ++++++
NWXJ23Xinjiang ZiniquanG. glabra6a +++++++
NWGS04Gansu HuanxianG. uralensis6a +++++
NWXJ92bXinjiang BaichengG. uralensis6a ++++++
NWXJ112Xinjiang TuofengxiaG. uralensis6a +++++
NWXJ88Xinjiang MaigaitiG. squamulosa6aAgrobacterium tumefaciens LMG140T (AM181758) 97%++++++
NWXJ104Xinjiang BaichengG. uralensis6aAgrobacterium tumefaciens LMG140T (AM181758) 96%+++++
NWXJ110Xinjiang TuofengxiaG. uralensis6a +++++
NWXJ109Xinjiang TuofengshangG. uralensis6b +++++
NWXJ53Xinjiang BeitunG. uralensis6cAgrobacterium tumefaciens LMG140T (AM181758) 96%+++++
NWXJ108Xinjiang TuofengshangG. uralensis6c +++++
NWNX03Ningxia YinchuanG. uralensis6dUnidentified+++++
NWXJ55bXinjiang GongliuG. uralensis6eP. bourgognense STM201T (AY785320) 97%++++
NWXJ89Xinjiang BaichengG. uralensis6f ++++++
NWXJ103Xinjiang BaichengG. uralensis6f ++++++
NWXJ101Xinjiang BaichengG. uralensis6fErwinia cypripedii ATCC29267T (U80201) 95%++++++
NWXJ59Xinjiang HabaheG. uralensis7aEnterobacter cloacae 766 (AM778415) 98%++++++
NWSX12Shaanxi FuxianGlycyrrhiza sp.7bUnidentified+++++
NWGS13Gansu ChongxingGlycyrrhiza sp. Rhodobacter changlensis JA139T (AM399030) 99%+++++
NWXJ24Xinjiang ZiniquanG. glabra Unidentified+++++++++
NWSX16Shaanxi YananGlycyrrhiza sp. Microbacterium oxydans DSM20578T (Y17227) 97%++++
NWSX07Shaanxi YulinG. uralensis Unidentified+++

Isolation of bacteria

Healthy, intact nodules were excised from the roots and placed in preservation vials containing a desiccant and brought to the laboratory. Nodules were surface-sterilized first with 95% ethanol for 5 min, then with 0.1% (w/v) mercuric chloride for 2–4 min and rinsed five times with sterile water according to Vincent (1970). Sterilized nodules were crushed, suspended in a small volume of sterile water, streaked out on yeast mannitol (YEM) agar (Vincent, 1970) and grown at 28 °C for 3–5 days. Single colonies were picked and checked for purity by repeated streaking and microscopic examination. When necessary, mixed bacterial cultures were purified by diluting cultures grown in YEM or tryptone-yeast extract (TY; Beringer, 1974) broth with Tween buffer (0.05 M phosphate, pH 6.6, 0.1% Tween 80) to separate different bacteria. The first dilution was shaken for at least 30 min, and dilutions of 10−5 and 10−6 were plated onto YEM agar containing 0.025 g kg−1 Congo red (Lindström et al., 1985) and grown for 3–5 days at 30 °C. All strains were maintained and stored on YEM agar and in 20% (v/w) glycerol at −70 °C.

DNA extraction

DNA was isolated from rhizobial cells grown in TY broth for 2–3 days at 30 °C using mostly a hexadecyltrimethyl ammonium bromide (CTAB) procedure suitable for polysaccharide producing bacteria. This procedure, described in more detail in Appendix S1, was largely based on the CTAB method of Wilson (1994). Salt-extraction (Aljanabi & Martinez, 1997) with some modifications (Appendix S1) was applied when the CTAB method was unsuccessful. In both methods, DNA was extracted from a cell pellet of approximately 50 μL. The extracted DNA was dissolved in 30–50 μL of sterile ultra-pure water.

AFLP fingerprinting

The AFLP procedure was largely based on the procedures described by Vos et al. (1995) and Dresler-Nurmi et al. (2000). The 25-μL restriction-ligation mixture contained 100–300 ng template DNA, 5 U EcoRI (Promega Corporation, Madison, WI) and Tru1l (MseI; MBI Fermentas), 10 pmol EcoRI, and 20 pmol Tru1l adapters and 1 U T4 DNA ligase in 1× ligase buffer (MBI Fermentas). This mixture was incubated at 37 °C and stored at −20 °C until use. Two primers with two selective bases (bold) were used in the PCR described in Appendix S1: EcoRI-ac (5′-gac tgc gta cca att cac-3′), and Tru1l-gc (5′-gat gag tcc tga agc-3′). The EcoRI-ac primer was fluorescently labelled at the 5′-end with blue 6-carboxyfluorescein (6-FAM; TAG, Copenhagen A/S, Denmark).

PCR products were checked, adjusted to a larger volume (50 μL) and purified with MicroSpin S-400 HR columns (Amersham Biosciences) according to the manufacturer's instructions, and stored at −20 °C until use.

Purified AFLP-PCR samples were analyzed by ABI Prism® 310 DNA Genetic Analyzer (PE Biosystems), the capillary of which was filled with denaturating polymer POP-6. A 3-μL volume of purified PCR product and 0.5 μL of internal red standard, GENESCAN™-500 TAMRA™ (Applied Biosystems, Carlsbad, CA) were added to 12 μL of deionized formamide. Samples were denatured for 2 min at 98 °C, chilled on ice and transferred to a sample tray. Samples were injected at 15 kV for 10 s into a 49-cm capillary and electrophoresed for 60 min at 50 °C.

AFLP data were examined and processed using abigenescan analysis software (PE Biosystems) and transformed and analyzed using the bionumerics software, version 5.0 (Applied Maths, Kortrijk, Belgium). Fragments between 35 and 500 bp were used for comparison. Fingerprints were normalized and checked manually. Fingerprint similarity values were calculated using the Dice coefficient with a position tolerance of 1.0% and optimization tolerance 0.5%. A dendrogram was constructed using the unweighted pair group method with arithmetic mean.

Identification of bacterial strains

Based on the AFLP results, representative strains of each AFLP subgroup were subjected to partial sequencing of the 16S rRNA gene (1000 bp). Sequencing was done directly from PCR products that were amplified (Appendix S1) using primers fD1 (5′-aga gtt tga tcc tgg ctc ag-3′) and rD1 (5′-aag gag gtg atc cag cc-3′) (Weisburg et al., 1991) by the Institute of Biotechnology, University of Helsinki. The most similar sequences were searched for and nucleotide alignments were constructed by silva comprehensive ribosomal RNA databases ( and manually corrected in arb software (Ludwig et al., 2004). Results were confirmed by comparing sequences against a nucleotide database (blast).

Nodulation tests

Thirty strains representing prevalent AFLP subgroups were tested for their nodulation ability on G. uralensis and G. glabra. Seeds were surface-sterilized by immersing them in concentrated H2SO4 solution for 1 h, rinsed with sterile water and germinated on 1% water agar for 2–3 days as described by Räsänen et al. (2001). Plants were grown in duplicate glass jars filled with a sterilized mixture of washed sand, vermiculite and ceramic Leca-gravel (Räsänen et al., 2001). Five seedlings were transferred to each glass jar and inoculated on either the same or the following day with 10 mL of bacterial cultures grown in YEM broth to the early stationary phase and diluted with sterile water at 106 CFU mL−1. Non-inoculated plants were used as controls. The seedlings were grown in a growth cabinet with a 18 h light period (at 20 °C for 1 h, 24 °C for 16 h and 20 °C for 1 h) and with a 6-h dark period at 16 °C, and watered alternatively with sterilized quarter strength Jensen's medium (Vincent, 1970) or water. At harvest, after 45 and 50 days, the colour and appearance of shoots were recorded. Roots were rinsed with tap water and the nodulation status of roots was checked. Nodulation results were recorded as positive (nodules were found) or negative (nodules were not found). Nodules were considered to be N2-fixing or effective if plants were green and healthy, and ineffective if plants were yellow with a similar appearance to uninoculated plants.

PGP activity tests

Indoleacetic acid (IAA) production was assayed from bacterial cultures grown in King's B broth (King et al., 1954) for 3–7 days at 30 °C using a colorimetric method (Egamberdieva & Kucharova, 2009). After centrifugation, bacterial supernatant was mixed with the Salkowski reagent (1 : 1 v/v). A pink colour indicating indole production was recorded after 30 min. Results were confirmed by measuring the absorbance at 530 nm.

Phosphate-solubilizing bacteria were screened on the solid Pikovskya's medium supplemented with Ca3(PO4)2 (5 g L−1) and Bromophenol Blue (0.025 g L−1) (Srividya et al., 2009), and incubated for 5–7 days at 30 °C. Production of acid changed the indicator colour from blue to yellow. The formation of a clear halo around the bacterial colonies indicated a positive reaction.

Siderophore production was assayed with chrome azurol S (CAS) agar (Alexander & Zuberer, 1991). After incubation for 5–7 days, siderophore-producing strains had an orange, purple or purplish-red halo around the colonies.

Protease activity was assayed with YEM agar containing 5% skimmed milk. After incubation for 5–7 days, a clear halo around the bacterial colonies due to hydrolysis of milk casein indicated a positive reaction.

Lipase activity was assayed with modified Sierra lipolysis agar (Sierra, 1957) supplemented with beef extract (3 g L−1) and ferrous citrate (0.2 g L−1). After autoclaving, 10 mL of Tween 80 and 50 mL of Victoria Blue B solution (0.1 g per 150 mL) was added to the medium. After incubation for 5 days, white calcium precipitates around the bacterial colonies indicated a positive reaction.

Cellulase activity was assayed with GYM Streptomyces medium (DSMZ;; medium 65) but without CaCO3 and supplemented with carboxymethyl cellulose (5 g L−1; Sigma) in place of glucose. After incubation for 5 days, plates were stained with Congo red solution and destained with NaCl according to Teather & Wood (1982). A clear or lightly coloured halo around the colonies indicated a positive reaction.

Statistical analyses

Nonparametric multivariate methods were used to analyze data, as the environmental variables and AFLP fingerprint data obtained from bacterial strains were not normally distributed. Permutations were used to calculate the significance. P-values were considered significant when  0.05.

The nonparametric discriminant analysis based on distances (db-DA) (Anderson & Robinson, 2003) was performed to find out whether the AFLP fingerprint patterns discriminate between geographic sampling regions and/or host plant species. Strains were divided in advance (a priori) into groups within each category (sampling region and host plant). The db-DA was performed in the R environment (R Development Core Team, 2007) using the package biodiversityr (Kindt & Coe, 2005) and its function ‘CAPdiscrim’. The function ‘permutations’, with 9999 permutations, was used to test the null hypothesis that strains were distributed randomly over a priori groups. The homogeneity of multivariate group dispersions (variances) of a priori groups (Anderson, 2006) was tested using the package vegan (Oksanen et al., 2008) and its function ‘betadisper’. The function ‘permutest’, with 9999 permutations, was used to compare pairwise multivariate dispersions of the groups, the null hypothesis being that there were no differences in dispersion between groups. For db-DA and the test for homogeneity of multivariate dispersions, the matrix of similarities (transformed to dissimilarities) of bacterial AFLP fingerprint data was obtained using the Dice coefficient.

A constrained ordination method, CCA, was used to determine whether environmental variables of sampling sites could explain the distribution of particular bacterial groups, species (AFLP subgroups) or strains. CCA can model the unimodal relationships, such as those between AFLP patterns representing strains and environmental variables. CCA is also well suited for absence–presence data, such as ours. The response of the distribution of strains to environmental variables was first explored by fitting environmental variables on the AFLP ordination based on canonical analysis of principal coordinates (CAP; Anderson & Gribble, 1998) in the R environment (R Development Core Team, 2007) using the vegan package (Oksanen et al., 2008) and its functions ‘capscale’ and ‘ordisurf’. CCA was performed using the R package vegan (Oksanen et al., 2008) and its function ‘cca’. The function ‘permutest’ with 9999 random permutations with a reduced model was used to test the null hypothesis that the distribution of AFLP subgroups (species) or strains was unrelated to environmental variables.

Variance partitioning (Borcard et al., 1992; Legendre & Legendre, 1998) was used to determine how much of the variation of AFLP patterns (strains) was explained by either the host plant species or the geographic location of the sampling site. First, a series of distance-based redundancy analysis (db-RDA; Anderson, 2001; McArdle & Anderson, 2001) was ran according to Legendre & Legendre (1998) using the program distml (Anderson, 2004) with 9999 permutations of raw data. RDA runs were used to partition the variation to pure proportions of host plant and geographic location and to the shared proportion of both host plant and geographic location. For db-RDA, the matrix of similarities (transformed to dissimilarities) of AFLP fingerprint data was obtained using the Dice coefficient. Geographic coordinates (latitude and longitude, WGS 84) of the sampling sites were projected in the geographical information system and converted to metric coordinate units (projection China Gauss-Kruger Beijing 1954 zone 17) (mapinfoprofessional version 10.0; Pitney Bowes Software Inc. 2009).


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information

Distribution of Glycyrrhiza isolates

Originally, 167 bacterial strains (Fig. S2) were isolated from surface-sterilized root nodules from five wild Glycyrrhiza species growing on 40 sites in four provinces in northwestern and central China. When identical strains were excluded, 159 distinct strains were identified from Glycyrrhiza nodules (Tables 1 and 2). Strains were considered to be identical when in the dendrogram of AFLP fingerprints, the similarity value of two or more strains was 100% and they originated from the same sampling site.

Most of the sampling sites (27) were located in Xinjiang in northwestern China (Table S1) and most of the bacterial strains (115) were also derived from this province (Table 3). Furthermore, 37 strains were obtained from 12 sites in northern Xinjiang and 78 strains from 15 sites in southern Xinjiang. The rest of the strains came from 13 sites in the provinces Shaanxi (26 strains), Gansu (13 strains) and Ningxia (five strains) in central China (Table S1, Table 3).

Table 3. Numbers of strains isolated from Glycyrrhiza nodules and their distribution among bacterial species and host plants growing in northern Xinjiang, southern Xinjiang and central China. Ninety-five strains were isolated from Glycyrrhiza uralensis (Gu), 28 strains from Glycyrrhiza glabra (Gg), 16 strains from Glycyrrhiza eurycarpa (Ge), Glycyrrhiza inflata (Gi) and Glycyrrhiza squamulosa (Gs), and 20 strains from taxonomically unidentified Glycyrrhiza species (Gsp)
Bacterial classAFLP subgroupNorthern XinjiangSouthern XinjiangCentral ChinaaStrain no. per species
GuGgGuGgGe Gi GsGuGsp
  1. a

    Provinces Shaanxi, Gansu and Ningxia.

  2. b

    Mesorhizobium amorphae (5a), Rhizobium leguminosarum (4g), Rhizobium galegae (5d), and Rhodobacter changlensis.

  3. c

    AFLP subgroups 5b, 6d and 7b with single strains, and strains NWXJ24 and NWSX07.

M.mediterraneum3a, 3c100009313
Mesorhizobium group A4a, 4b0014010015
Mesorhizobium group B5e063700016
Rhizobium daejeonense1a10701009
Rhizobium cellulosilyticum1b, 1c00310004
Rhizobium gallicum3d00000617
Sinorhizobium meliloti4c000071311
Agrobacterium tumefaciens6a, b, c117034016
Phyllobacterium spp.2, 4e, 6e10104006
Rhizobium giardinii4f00030003
Other speciesb5a, 4g, 5d10010024
Total 5133613162013116
Pantoea agglomerans5c10100002
Erwinia cypripedii6f00300003
Enterobacter cloacae7a10000001
Total 20400006
Gram-positive bacteria
Paenibacillus spp.4d, 5f1519001531
Microbacterium oxydans 00000011
Total 1519001632
Unidentified strainsc5b, 6d, 7b01000315
Total 22154913162420159
All strains per region  37  78 44 

Almost 60% of the bacteria were isolated from G. uralensis. Of these, 22 strains originated from northern Xinjiang (11 sites), 49 strains from southern Xinjiang (9 sites) and 24 strains from Central China (6 sites). The distribution of strains among other host plants, which were sampled only in Xinjiang, was: 28 strains from G. glabra, eight strains from G. squamulosa, six strains from G. eurycarpa and two strains from G. inflata. Twenty strains were obtained from nodules of unidentified Glycyrrhiza plants growing on seven sites in central China (Table 3, Table S1).

Diversity and identity of Glycyrrhiza isolates

In the AFLP analysis the strains were divided into seven genomic groups at the similarity level of 50%. Four single strains remained unclustered (Fig. S2). The seven groups were further divided into subgroups at 60% similarity level (Fig. S2). Sequencing of partial 16S rRNA gene (1000 bp) from representative strains revealed that each AFLP subgroup consisted of strains belonging to the same species or to a distinct group of closely related strains (Tables 1 and 2).

AFLP group 1 consisted of three subgroups (a, b, c), from which the largest subgroup 1a (nine strains) was identified as Rhizobium daejeonense. Subgroups 1b and 1c, altogether four strains, were identified as Rhizobium cellulosilyticum. AFLP group 2 contained two Phyllobacterium sp. strains. The majority of strains in AFLP group 3 belonged to the genus Mesorhizobium. Subgroups 3a and 3c with 13 strains represented Mesorhizobium mediterraneum and subgroup 3b with 12 strains M. tianshanense. Subgroup 3d with seven strains was related to Rhizobium gallicum (Table 1, Fig. S2).

AFLP group 4 contained several different rhizobial species. Subgroups 4a and 4b (altogether 15 strains) belonged to the genus Mesorhizobium, termed Mesorhizobium group A. This group was especially interesting because the sequenced strains of subgroup 4a were not closely related to any recognized species, showing only 96–98% identity to Mesorhizobium sp. strain Mad-1 (Table 1) isolated from Ononis tridenta in Spain (Rincón et al., 2008). Subgroup 4c with 11 strains was related to Sinorhizobium meliloti. The small subgroups 4e and 4f with three strains each were related to Phyllobacterium ifriqiyense and Rhizobium giardinii, respectively. The only representative of subgroup 4g was identified as Rhizobium leguminosarum (Table 1, Fig. S2).

AFLP group 5 contained the second largest Mesorhizobium subgroup, 5e (16 strains). The closest strain in the nucleotide database was M. tianshanense strain ST-2 (99% identity, Table 1) obtained from a chickpea nodule in Portugal (Laranjo et al., 2004). However, the taxonomy of this subgroup remains unresolved because other close strains in the database with equal identity belonged to the species M. mediterraneum, M. gobiense and M. tianshanense. Obviously, these closely related species cannot be distinguished reliably using the 16S rRNA gene alone for identification. Subgroup 5e was named Mesorhizobium group B. The single representatives of subgroups 5a and 5d were distantly related to Mesorhizobium amorphae and Rhizobium galegae, respectively.

The 15 strains in subgroups 6a and 6c belonged to Agrobacterium tumefaciens. The only representative of subgroup 6b was also included in A. tumefaciens because strain NWXJ109 was situated between subgroups 6a and 6c in the dendrogram of AFLP fingerprints (Fig. S2). The one strain in subgroup 6e was Phyllobacterium bourgognense (Table 1, Fig. S2).

Except for the single strain in subgroup 5f identified as Paenibacillus polymyxa, 30 Paenibacillus strains were found mainly in subgroup 4d. Compared to other subgroups this subgroup was genetically more homogeneous, although its strains could represent several species (e.g. Paenibacillus pasadenensis, Paenibacillus amylolyticus, Paenibacillus graminis-like sp.). The rest of the identified endophytic bacteria represented a mixed collection of species having one to three strains: Pantoea agglomerans (subgroub 5c), Erwinia cypripedii (subgroup 6f), Enterobacter cloacae (subgroup 7a) and unclustered Rhodobacter changlensis and Microbacterium oxydans (Table 2, Fig. S2). Five strains remained unidentified because PCR of the 16S rRNA gene was not successful [single representatives of subgroups 5b (NWSX22), 6d (NWNX03) and 7b (NWSX12) and two unclustered strains (NWXJ24 and NWSX07)].

In summary, many rhizobial and non-rhizobial species were identified among Glycyrrhiza isolates. In total, 73% of the strains were Alphaproteobacteria, with 57 strains representing Mesorhizobium, 25 Rhizobium, 11 Sinorhizobium (syn. Ensifer) and six Phyllobacterium. The rest of the 16 Alphaproteobacteria strains belonged mainly to the genus Agrobacterium (Table 3). Twenty percent of all the strains were Gram-positive bacteria, most of which belonged to the genus Paenibacillus within the Firmicutes (low G+C group). Six strains were members of the Gammaproteobacteria, representing the genera Erwinia, Pantoea and Enterobacter (Table 3).

Symbiotic properties of Glycyrrhiza isolates

Our work revealed that mesorhizobia are the most important rhizobial symbionts for Glycyrrhiza plants growing in China. Fifty-eight percent of all rhizobial strains identified were mesorhizobia, comprising M. mediterraneum (AFLP subgroups 3a, 3c), M. tianshanense (3b), Mesorhizobium group A (4a) and Mesorhizobium group B (5e) (Table 3). Nearly all mesorhizobial strains tested (14/15) induced N2-fixing nodules on G. uralensis and/or G. glabra (Table 4). Only the M. mediterraneum strain NWSX23 (3c) originally isolated from G. uralensis induced ineffective nodules on both test plants. The origin of a Mesorhizobium strain, i.e. host plant and isolation site, did not influence nodulation phenotype, and strains isolated from G. uralensis were capable of forming effective nodules on G. glabra and vice versa. Only one M. mediterraneum strain, strain NWSX20 (3a) isolated from an unidentified Glycyrrhiza plant species in central China, produced effective nodules on G. uralensis but ineffective ones on G. glabra (Table 4).

Table 4. Nodulation and nitrogen fixation of Glycyrrhiza uralensis and Glycyrrhiza glabra inoculated with rhizobial strains isolated from Glycyrrhiza. Most strains originated in Xinjiang in northwestern China. Eight strains equipped with an asterisk (*) were obtained from the three provinces located in central China. Plants were grown in growth cabinet for 7 weeks in two replicate jars containing sterilized mixture of sand, vermiculate and Leca-gravel
StrainHost plantClosest speciesAFLP groupG. uralensisG. glabra
  1. Nod, nodulation; Fix, nitrogen fixation.

  2. a

    This capability was deduced from the colour of plants. Green colour indicated effective nodules and yellow indicated ineffective ones.

NWXJ58G. uralensisR. daejeonense1a
NWXJ114G. uralensisR. daejeonense1a
NWXJ44G. uralensisR. cellulosilyticum1c+
NWSX20*Glycyrrhiza sp.M. mediterraneum3a+++
NWSX24*G. uralensisM. mediterraneum3a++++
NWXJ32G. glabraM. tianshanense3b++++
NWXJ36-2G. glabraM. tianshanense3b++++
NWXJ43-2G. uralensisM. tianshanense3b++++
NWGS10*Glycyrrhiza sp.M. tianshanense3b++++
NWGS03*G. uralensisM. mediterraneum3c++++
NWSX23*G. uralensisM. mediterraneum3c++
NWSX10*G. uralensisR. gallicum3d++++
NWSX15*Glycyrrhiza sp.R. gallicum3d
NWXJ02G. uralensisMesorhizobium sp. A4a++++
NWXJ14G. uralensisMesorhizobium sp. A4a++++
NWXJ19G. uralensisMesorhizobium sp. A4a++++
NWXJ81G. eurycarpaS. meliloti4c++++
NWXJ116G. inflataS. meliloti4c+++
NWGS06*Glycyrrhiza sp.S. meliloti4c++
NWXJ73G. glabraR. giardinii4f++
NWXJ107G. glabraR. leguminosarum4g+++
NWXJ54G. uralensisR. galegae5d++++
NWXJ21G. glabraMesorhizobium sp. B5e++++
NWXJ31G. glabraMesorhizobium sp. B5e++++
NWXJ36G. glabraMesorhizobium sp. B5e++++
NWXJ43G. uralensisMesorhizobium sp. B5e++++

The nodulation and N2-fixing capacity of the other rhizobial species, altogether 42 strains, was irregular. Four of 10 species, namely S. meliloti, R. gallicum, R. galegae and R. leguminosarum, had strains which were able to produce effective, N2-fixing nodules on G. uralensis and/or G. glabra (Table 4). Effective R. galegae and R. leguminosarum were not typical symbionts of Glycyrrhiza because they possessed only a single representative each among the Glycyrrhiza isolates (Table 3). Sinorhizobium meliloti and R. gallicum also included ineffective (S. meliloti NWGS06) and non-nodulating strains (R. gallicum NWSX15), both of which were originally isolated from unidentified Glycyrrhiza species (Table 4). A few strains – M. mediterraneum NWSX20, S. meliloti NWXJ116 and R. leguminosarum NWXJ107 – showed stricter host specificity, producing effective nodules on G. uralensis but ineffective ones on G. glabra (Table 4).

Strains belonging to R. giardinii and R. cellulosilyticum as well as Phyllobacterium strain NWXJ84 from subgroup 4e induced only ineffective nodules on both test plants (R. giardinii NWXJ73) or on G. glabra (R. cellulosilyticum NWXJ44, strain NWXJ84). The sequenced representative of subgroup 4e, strain NWXJ83, showed 97% identity to the P. ifriqiyense strain STM 370T (Table 2) obtained from Lathyrus numidicus nodule (Mantelin et al., 2006). Strains belonging to R. daejeonense, Phyllobacterium sp. and P. bourgognense did not nodulate at all. This was also the case for the endophytic species A. tumefaciens, Pantoea sp. and Paenibacillus spp. (Tables 2 and 4).

Altogether, of 115 strains belonging the Rhizobiales order, 57 rhizobial strains were effective Mesorhizobium species, 20 strains belonged to less effective species (S. meliloti, R. gallicum, R. galegae and R. leguminosarum) and nine strains to ineffective species.

When the nodulation data were compared with data about bacterial species isolated from Glycyrrhiza nodules on each site, it appeared that G. uralensis plants growing in Xinjiang were seldom nodulated by rhizobial species (4/20 sites) that had formed N2-fixing nodules in our plants tests (Table 4, Table S1). In contrast to strains originating in Xinjiang, strains obtained from central China generally belonged to effective rhizobial species (Table S1). Paenibacillus sp. was frequently isolated from G. uralensis sampled in Xinjiang, especially from those sampled in northern Xinjiang (Table 3, Table S1). The above points raised the question: can environmental conditions explain the differences in bacterial populations between central China and Xinjiang?

Discrimination of strains within the sampling regions and host plants

We used db-DA to test the hypothesis that bacterial AFLP fingerprint patterns representing strains were more similar to each other when the strains originated from one geographical region or from one host plant species. Besides the ability to discriminate strains between different sampling regions or host plants, DA also has a classification function which reveals how big a percentage of the strains is classified according to the a priori hypothesis.

DA was capable of discriminating bacteria between different Glycyrrhiza species and geographical regions but could not explain the mechanisms behind the divergence of bacteria. When bacterial strains were a priori divided into three groups according to the major sampling regions, i.e. northern Xinjiang, southern Xinjiang and central China, db-DA showed that 70% of all strains belonged to their a priori classes. Although there was some regional heterogeneity and overlap, strains obtained from central China were well separated from the Xinjiang strains along axis 1 (Fig. 1a).


Figure 1. Distance-based nonparametric discriminant analysis (DA) of a priori classified 159 bacterial strains obtained from root nodules of different Glycyrrhiza plant species growing in China. Coloured symbols indicate individual bacterial strains. Small numbers with letters refer to the major AFLP subgroups. Discrimination of strains based on (a) three geographical sampling regions and (b) four host plant species. In both cases, the P-value 0.0001 was based on 9999 permutations. The AFLP subgroup codes indicate bacterial species as follows: 1a, Rhizobium daejeonense; 1b and 1c, Rhizobium cellulosilyticum; 2, 4e and 6e, Phyllobacterium spp.; 3a and 3c, Mesorhizobium mediterraneum; 3b, Mesorhizobium tianshanense; 3d, Rhizobium gallicum; 4a, Mesorhizobium group A; 4c, Sinorhizobium meliloti; 4d, Paenibacillus spp.; 4f, Rhizobium giardinii; 4g, Rhizobium leguminosarum; 5e, Mesorhizobium group B; 6a, Agrobacterium tumefaciens; 6f, Erwinia cypripedii.

Download figure to PowerPoint

The DA method assumes that the dispersions (variances) of observations are approximately similar between the a priori groups to be analyzed. Strains of Paenibacillus sp., which were genetically rather homogeneous, constitute a big portion (40%) of the northern Xinjiang isolates (Table 3, Fig. S2). Therefore, the composition of different AFLP patterns (strains) from northern Xinjiang was less diversified in comparison with those of the other geographical groups. Thus, the dispersion (variance) in the northern Xinjiang group was greater than that of the other regional groups (Fig. S3) and the group dispersions between northern and southern Xinjiang and between northern Xinjiang and central China were not similar (P-values for pairwise comparisons were 0.0001 and 0.0056; Table S2). This argues slightly against the assumption of DA of similar group dispersions.

Along axis 2, most strains originating in southern Xinjiang also differed from those obtained from the north. Paenibacillus strains were an exception, as their southern variants were not distinguishable from their northern variants (Fig. 1a). The abundance of Paenibacillus strains in northern Xinjiang possible strengthened the discrimination of AFLP patterns between northern and southern Xinjiang.

Rhizobial AFLP subgroups (species) obtained from central China were clearly separated from those originating in Xinjiang. Only subgroups 3b (M. tianshanense) and 4c (S. meliloti) consisted of two geographically distinct populations (Fig. 1a). Strains representing A. tumefaciens (6a) were scattered near the origin of the ordination (Fig. 1a), indicating that the occurrence of A. tumefaciens was unrelated to the geography.

When strains were a priori classified in four groups according to their host species, i.e. G. uralensis, G. glabra, G. eurycarpa + G. inflata + G. squamulosa and the unidentified Glycyrrhiza sp., db-DA revealed that 82% of all bacterial strains belonged to their a priori classes. Strains isolated from G. glabra were clearly separated from those obtained from G. uralensis and unidentified Glycyrrhiza plants (Fig. 1b). Furthermore, most strains isolated from G. uralensis differed from those obtained from the pooled three Glycyrrhiza species and unidentified Glycyrrhiza plants along axis 2 (Fig. 1b).

The multivariate group dispersion of bacterial AFLP patterns was approximately the same in each plant group analyzed a priori. The dispersion was greater than in other plant groups only in the G. glabra group (Fig. S3) and the dispersions were different between G. glabra and G. uralensis (P-value 0.005; Table S2). Seventy percent of G. glabra isolates were represented by only two Mesorhizobium species (Table 3) and the strains, originating from three sampling sites (Table 1), were also genetically rather homogeneous (Fig. S2). Therefore, AFLP patterns obtained from the small number of G. glabra isolates showed less heterogeneity than those derived, for example, from numerous diversified G. uralensis isolates. This might partly explain why the G. glabra isolates separated so clearly from other bacteria in db-DA (Fig. 1b).

Rhizobial AFLP subgroups (species) were typical for the particular host plant species (Fig. 1a). Strains belonging to subgroups 4a (Mesorhizobium group A) and 1a (R. daejeonense) were characteristic for G. uralensis, whereas subgroups 3b (M. tianshanense), 5e (Mesorhizobium group B) and 4f (R. giardinii) were characteristic for G. glabra (Fig. 1b). In both DA plots, strains representing A. tumefaciens were scattered near the origin of the ordination (Fig. 1a and b), indicating that the variation in their AFLP patterns was not related to either the geography or the Glycyrrhiza hosts.

Relationships between environmental factors and bacteria

CCA was applied to explore possible relationships between changes in environmental variables and variations in bacterial strains colonizing Glycyrrhiza nodules. CCA used the variation in the environmental matrix to explain variation in the bacterial AFLP fingerprint matrix. Our assumption was that the greater the similarity of the AFLP patterns in the strains, the more similar the environmental requirements of the strains.

According to CCA all environmental variables used accounted for 8.4% of the total variation in the bacterial AFLP patterns. Although the variables used explained only a small part of the variation in AFLP patterns, CCA gave some clues to the factors that might be associated to bacteria isolated from Glycyrrhiza nodules.

The CCA biplot indicated that latitude showed the strongest correlation to the variation in AFLP patterns, followed by altitude, longitude, annual temperature and annual rainfall. Among geographical variables, a scattered group of rhizobial strains obtained from central China was related to the increasing longitude (Fig. 2a and b). The sampling sites of central China are located east of Xinjiang. Paenibacillus strains (cluster P; Fig. 2a), isolated almost exclusively from G. uralensis, correlated to the increasing latitude, corresponding to sampling sites in northern Xinjiang (Fig. 2). Strains that belonged to AFLP subgroups 4a (Mesorhizobium group A) and 1a (R. daejeonense) and isolated from G. uralensis in southern Xinjiang were associated with the increasing altitude (cluster MRGU; Fig. 2a). The geographic location of sampling sites did not explain the distribution of the other strains obtained from Xinjiang.


Figure 2. CCA showing associations between environmental variables and AFLP patterns of 159 bacterial strains isolated from Glycyrrhiza nodules in China. Strains are displayed as points and variables as vectors. The arrows point to the direction of the most rapid change in the environmental variable. The lengths of the arrows are proportional to the correlation between ordination and environmental variable. Coloured symbols indicating individual bacterial strains were printed according to (a) bacterial genera, (b) geographical sampling regions and (c) Glycyrrhiza host plants. P-value 0.0001 was based on 9999 permutations. Clusters indicate: P (Paenibacillus), M (Mesorhizobium) and MRGU (Mesorhizobium and Rhizobium from Glycyrrhiza uralensis). AFLP subgroup codes indicate bacterial species as follows: 1a, Rhizobium daejeonense; 3b, Mesorhizobium tianshanense; 4a, Mesorhizobium group A; 4d, Paenibacillus spp.; 5e, Mesorhizobium group B.

Download figure to PowerPoint

Paenibacillus strains were related to the decreasing rainfall. Thus, the drought could be a reason for the phenomenon that G. uralensis nodules in Xinjiang were commonly colonized by Paenibacillus spp. Strains belonging to subgroup 6a (A. tumefaciens) as well as most strains of subgroup 4c (S. meliloti) were associated with the increasing annual temperature and rainfall (Fig. 2a). In other words, these bacteria may favour warm and moist conditions.

Strains belonging to two mesorhizobial subgroups 3b (M. tianshanense) and 5e (Mesorhizobium group B) were not clearly associated with any environmental variables considered (cluster M, Fig. 2a). Although the strains originated from different host plants (G. glabra, G. uralensis, unidentified Glycyrrhiza spp.) as well as from different geographical regions (Xinjiang, central China) (Fig. 2b and c), CCA suggested that some unknown factor not included in this study united the strains in the Mesorhizobium cluster.

Influence of host plant species and geographical location on bacterial variation

We used partitioning of the variance to determine how big a proportion of the variation in bacterial AFLP patterns was explained by the geographic location and host plant species. This test showed that the host Glycyrrhiza species of the strains accounted for 9.9% (pure proportion of host species) of the total variation in the AFLP patterns. The geographic location of the sampling sites, when latitude and longitude coordinates were converted into metric coordinate units, accounted for 6.9% (pure spatial proportion) of the total variation. The shared proportion, indicating the variation in AFLP patterns explained by both the host plant species and geographic location, was 2.4%. This leaves 85.7% of the total variation in bacterial AFLP patterns unexplained. It is possible that the dependence between AFLP patterns and geographic distances was not totally linear, which could have affected the results of partitioning the variance. However, these results were parallel with those obtained from CCA, indicating that the geographical location of sampling sites has a minor effect on which bacteria colonize Glycyrrhiza nodules.

PGP activities among Glycyrrhiza isolates

Six different PGP activities were tested in vitro for 65 strains isolated from Glycyrrhiza nodules. Nearly all strains representing Alphaproteobacteria (Agrobacterium sp., Phyllobacterium spp.) and Gammaproteobacteria (Pantoea sp., Erwinia sp., E. cloacae) were able to produce cellulose and siderophores, i.e. molecules with a high affinity for iron, and at least a small amount of the auxin plant hormone indole-3-acetic acid (IAA) (Table 2). Only seven of 16 Agrobacterium strains and two strains, E. cloacae NWXJ59 and Microbacterium sp. NWSX16, showed a clear IAA production (Table 2). Only six strains representing Agrobacterium sp., Paenibacillus sp., Pantoea sp. and Erwinia sp. were able to solubilize phosphate, having a clear halo around the colony. However, except for the three P. ifriqiyense strains and R. changlensis strain NWGS13, all other Alpha- and Gammaproteobacteria tested produced organic acids (Table 2), indicating that they might be potential P-solubilizers (Vassilev et al., 2006). Besides the R. changlensis strain NWGS13 and unidentified strain NWXJ24, neither Alpha- or Gammaproteobacteria showed protease and lipase activity.

Protease activity was typical for the Gram-positive Paenibacillus spp. (61%), followed by the production of cellulase (57%) and siderophores (38%), and small amount of IAA (44%). Similarly to other bacteria, phosphate solubilization, the clear production of IAA and lipase production were also rare among Paenibacillus strains (Table 2).


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information

To avoid discarding putative novel symbiotic species and to obtain additional information on non-rhizobial endophytes colonizing Glycyrrhiza nodules, our strategy was to isolate all bacterial strains that inhabited the root nodules of the Glycyrrhiza plants sampled. Our work resulted in 159 distinct bacterial isolates, which represented mainly Alpha- and Gammaproteobacteria genera and Gram-positive Firmicutes. After testing for symbiotic properties (nodulation, N2 fixation) we divided the bacterial species into three categories (Fig. 3): (1) true symbionts, (2) sporadic symbionts and (3) non-nodulating species.


Figure 3. Grouping of nodule-associated endophytic bacteria isolated from Glycyrrhiza spp. by their ability to form a N2-fixing symbiosis with Glycyrrhiza uralensis and Glycyrrhiza glabra when isolates were reinoculated on the host plants.

Download figure to PowerPoint

The true symbionts, i.e. species that were common among Glycyrrhiza isolates and that, with few exceptions, were able to form N2-fixing nodules on the test plants G. uralensis and G. glabra, were all taxonomically Mesorhizobium sp. (Fig. 3). Four of five mesorhizobial AFLP subgroups represented the closely related lineage of M. mediterraneum, M. gobiense and M. tianshanense. The strains of the subgroup 4a were unrelated to any previously described Mesorhizobium species.

Sporadic symbionts, i.e. species within which there was a large variation between effective, ineffective and non-nodulating strains, usually had only a single or a few representatives among Glycyrrhiza isolates. Based on the nodulation phenotype, sporadic symbionts were further divided into two groups, those species with some effective strains and those with only ineffective strains. The first group included species for which N2 fixation ability was an exceptional characteristic (S. meliloti, R. gallicum) or which contained only a few but effective strains (R. galegae, R. leguminosarum). The species in the latter group with R. giardinii, R. cellulosilyticum and P. ifriqiyense-like strain NWXJ84 induced only ineffective nodules on G. uralensis and/or G. glabra. The non-nodulating species included the endophytic species Paenibacillus spp., A. tumefaciens, R. daejeonense, P. bourgognense, Phyllobacterium sp. and Pantoea sp. (Fig. 3).

The type strains of R. cellulosilyticum and R. daejeonense isolated from environmental samples were capable of inducing ineffective nodules on alfalfa and carried some symbiotic genes (nodD or nifH) (Quan et al., 2005; García-Fraile et al., 2007). Phyllobacterium spp., identified first from leaf nodules, have also been detected in root nodules (e.g. Mantelin et al., 2006). Besides Phyllobacterium trifolii, which nodulated Trifolium and Lupinus species (Valverde et al., 2005), they did not renodulate their original hosts (Zakhia et al., 2006; Mahdhi et al., 2007; Rincón et al., 2008; Bromfield et al., 2010), even though nodulation genes (nodD, nodC) were detected in a few strains (Valverde et al., 2005; Rincón et al., 2008). Although the above-mentioned Rhizobium and Phyllobacterium species belong taxonomically to the Rhizobiaceae and Phyllobacteriaceae families in the Rhizobiales order and may carry some nodulation genes, their nodulation seems to be coincidental. Besides, none of them has been shown to induce N2-fixing nodules on any legumes.

Among bacteria that induced ineffective nodules in our work, R. giardinii formed effective nodules, for example on Phaseolus vulgaris (Amarger et al., 1997), and therefore belong to the rhizobia. The ineffective R. cellulosilyticum and strain NWXJ84 representing P. ifriqiyense-like species as well as R. daejeonense, P. bourgognense and other Phyllobacterium spp., which did not nodulate Glycyrrhiza in our plant tests, could be considered transitional forms between nodulating rhizobia and non-nodulating endophytes.

Glycyrrhiza glabra and G. inflata grow mainly in Xinjiang in the northwest, whereas G. uralensis grows in all parts of northern China (Kondo et al., 2007). In our study the geographic distribution of the Glycyrrhiza species sampled was similar. Based on genetic markers, Kondo et al. (2007) concluded that G. glabra is more closely related to G. inflata than to G. uralensis, G. glabra and G. inflata being equidistant to G. uralensis. However, the genetic difference between them is not large because G. uralensis can hybridize with G. glabra and G. inflata (Hayashi et al., 2003; Kondo et al., 2007). This close genetic relationship might explain why in our work most nodulating strains were capable of forming nodules on both G. uralensis and G. glabra.

Endophytic bacteria identified from Glycyrrhiza nodules, namely Agrobacterium sp., Paenibacillus spp., Erwinia cypripedii, Pantoea agglomerans, E. cloacae, R. changlensis and Microbacterium oxydans, have been also previously detected in internal parts of roots, stems or nodules of various plants (Lodewyckx et al., 2002; Zakhia et al., 2006; Muresu et al., 2008). Even though inoculation of legumes with Agrobacterium alone hardly ever produces nodules (Mhamdi et al., 2005; Wang et al., 2006), our work as well those of others (e.g. de Lajudie et al., 1999; Bala et al., 2003; Mhamdi et al., 2005; Wang et al., 2006) indicated that Agrobacterium species frequently inhabit leguminous nodules. Only the phylogenetically different Agrobacterium strain IRBG74 was shown to induce effective nodules on tropical Sesbania plants (Cummings et al., 2009). In our work Paenibacillus spp. represented another big group of non-nodulating bacteria detected among Glycyrrhiza nodule isolates. Paenibacillus strains as well those of the closely related Bacillus genus have also been isolated from the nodules on other legumes (Bai et al., 2002; Zakhia et al., 2006; Muresu et al., 2008; Valverde et al., 2010). Thus, Gram-positive bacteria can also represent nodule dwellers even though they are usually considered laboratory contaminants.

Of the PGP activities tested, the cellulase production activity was common among all Glycyrrhiza isolates tested. Hydrolytic bacterial cellulase might be needed when endophytic bacteria specifically penetrate plant roots or stems (Lodewyckx et al., 2002). It is very likely that endophytic bacteria obtained from the nodules are capable of colonizing the internals of roots and stems (Wang et al., 2006). In legumes with indeterminate nodules, such as Glycyrrhiza (Sprent, 2001), rhizobia enter the host plant through root hair invasion. It is still unknown whether non-nodulating endophytes can enter the nodules as free passengers during the infection by symbiotic rhizobia. Bacteria that produce siderophores can inhibit plant pathogens by competing with them for iron in the rhizosphere. Some plants can also bind, transport and exploit bacterial iron-siderophore complexes (Lodewyckx et al., 2002).

Besides cellulase, protease produced by many Bacillus and Paenibacillus strains was proposed to degrade fungal cell walls, thereby inhibiting plant pathogens (Lodewyckx et al., 2002). Remarkable IAA and phosphate solubilizing activities were not common among Glycyrrhiza isolates. In fact, many PGP activities are considered to be strain-specific rather than species-specific. The biosynthesis of IAA is widespread among plant-associated bacteria, including plant pathogens. Presumably, PGP bacteria use IAA as a part of their colonization strategy, involving phytostimulation and circumvention of plant defence mechanisms. Moreover, IAA might act as a signal molecule in bacteria–bacteria communication (Spaepen et al., 2007). Perhaps some of these functions explain why the production of IAA was common among our Agrobacterium strains.

Recently, Muresu et al. (2008) demonstrated that the same root nodules of some Mediterranean legumes harboured prevailingly non-culturable rhizobia as well as a variety of culturable non-rhizobial endophytes. In our study, when only non-nodulating endophytes, such as Paenibacillus spp., were recovered from Glycyrrhiza nodules, the identity of the nodule-inducing organism remained unknown. As only intact, healthy nodules were collected from Glycyrrhiza roots, the reason for not getting rhizobia could either be that the original rhizobium or rhizobia died during the isolation or were not culturable under laboratory conditions.

In many cases, the distinguishing of bacterial strains on the basis of 16S rRNA gene techniques is too coarse to detect biogeographic patterns of bacterial divergence (Cho & Tiedje, 2000; Papke & Ward, 2004; Hughes Martiny et al., 2006; Ramette & Tiedje, 2007a). AFLP, a whole-genome fingerprinting method, has been subjected to multivariate analyses in plant ecology (e.g. Muchugi et al., 2008) but has been used hardly at all for bacterial ecology. Only Kölliker et al. (2005) included AFLP markers in multivariate analysis of grass pathogenic Xanthomonas isolates. Our work, among others (e.g. Portier et al., 2006), indicated that AFLP is a good method to identify strains belonging to the same genomic species or groups. In addition, AFLP has enough genetic resolution power for multivariate analyses to display biogeographic patterns of Glycyrrhiza isolates. A prerequisite of successful analyses and reasonable results would still be that a heterogeneous bacterial collection in particular consists mainly of approximately similar-sized AFLP groups and/or subgroups, which contain many strains.

Plants can dramatically modify their soil environment through the rhizosphere effect, which is largely induced by rhizodeposition (Jones et al., 2009). Above-ground plants shape soil microbial communities (Garbeva et al., 2006; Bremer et al., 2009) through root exudates, this effect even being plant species-specific (Haichar et al., 2008; Micallef et al., 2009). In the case of legumes, there is in addition an effect of plant species due to the exchange of specific signal molecules between the host plant and rhizobium preceding nodule formation (Dresler-Nurmi et al., 2009). Based on these assumptions, we postulated that the host plant species determine which endophytic bacteria are able to colonize Glycyrrhiza nodules.

Among the multivariate analyses used, DA classified most bacterial strains into four a priori Glycyrrhiza classes, with 82% of all strains showing host specificity. This was a bigger proportion than the proportion of rhizobial strains from all strains analyzed (54%), suggesting that in addition to effective and ineffective rhizobia, other endophytic bacteria might also show a preference for a particular Glycyrrhiza species. The variance partitioning test, however, showed that the host Glycyrrhiza species explained only a small portion (9.9%) of the total variation in bacterial AFLP patterns. These apparently contradictory results are due to the fact that DA is only capable of discriminating strains between different Glycyrrhiza species. In other words, DA distinguishes strains that are more similar within a plant group than between plant groups but is not capable of explaining the mechanisms behind the divergence of bacteria.

Recent studies dispute the first part of the paradigm ‘Everything is everywhere, but the environment selects’ presented by Baas-Becking in 1934 to explain microbial distribution (Hughes Martiny et al., 2006). Although many bacterial species, as defined by their 16S rRNA gene phylogeny, appear to be globally distributed, bacteria have shown site-specificity and endemism at the genotype or strain level (Cho & Tiedje, 2000; Kölliker et al., 2005; Ramette & Tiedje, 2007ab). Site-specificity or endemism was also detected among rhizobia isolated from leguminous nodules when ITS or repetitive element PCR (rep-PCR) fingerprinting were used for population analysis (Bala et al., 2003; Vinuesa et al., 2005).

In this work, rhizobial strains showed regional endemism. Several rhizobial AFLP subgroups and subpopulations of M. tianshanense and S. meliloti sampled from different sites were unique either to central China or Xinjiang. However, the variance partitioning test displayed that the geographic distance explained a small proportion (6.9%) of the variation in bacterial AFLP patterns. The CCA results paralleled this. Only a few strains were clearly associated with the geographic location of their sampling sites. In addition, geographic variables (latitude, longitude and altitude) and climatic variables (annual site temperature and rainfall) of the sampling sites explained only 8.9% of the total variation in AFLP patterns.

It is still controversial whether the geographic distance as such, affects bacterial community structure (Bissett et al., 2010; Chu et al., 2010). In our work the long distance between Xinjiang and central China (2000 km) and the isolated location of Xinjiang, which is surrounded by high mountains, explained only a small portion of the genetic variation of the bacteria. A reason for this could be that different soil bacteria have different dwelling or dispersion strategies. The study of Bissett et al. (2010) suggested that geographic distance has no influence on spore-forming taxa (e.g. Bacillaceae and Clostridiaceae), but does influence dispersal- or colonization-limited taxa such as Rhizobiaceae, Bradyrhizobiaceae and Xanthomonadaceae. Nemergut et al. (2011) demonstrated that relatively few taxa are cosmopolitan and the vast majority are restricted to individual assemblages. Moreover, the most abundant taxa appear also to be the most widely distributed.

The above-mentioned phenomena were also observed in our study. Most rhizobial groups originated from one region, whereas a portion of the spore-forming Paenibacillus strains did not show clear region-specificity. The Paenibacillus population was also genetically more homogeneous than other bacterial groups, suggesting a close relationship of easily dispersing bacteria. The distribution of the A. tumefaciens strains differed from that of the rhizobial ones, being unrelated to the geographic origin or Glycyrrhiza host. Agrobacterium tumefaciens is very commonly identified among nodule isolates regardless of geographic sampling sites and host plant (e.g. de Lajudie et al., 1999; Bala et al., 2003; Mhamdi et al., 2005; Wang et al., 2006). Perhaps A. tumefaciens is a representative of cosmopolitan bacteria.

Our results parallel the studies of Fierer & Jackson (2006) and Chu et al. (2010), which showed that the diversity of soil microbial populations is unrelated to variables predicting plant and animal diversity, such as annual site temperature and potential evapotranspiration, latitude and plant diversity. Only in the case of Paenibacillus strains, could the drought explain the phenomenon that G. uralensis nodules in Xinjiang were commonly colonized by Paenibacillus spp. Xinjiang has a typical continental climate with a low annual rainfall and high evaporation and a wide temperature range (Chen et al., 1995). Due to very resistant spores, Paenibacillus spp. are capable of surviving in dry environments. Edaphic factors such as soil pH, soil moisture, carbon and nutrient availability, and vegetation type (Fierer & Jackson, 2006; Chu et al., 2010) might be important factors in explaining variation of nodule isolates.

The variance partitioning together with CCA indicated that the host plant species and geographic location as well as climatic variables measured explained only a small part of the total variation in bacterial AFLP patterns, with the host plant explaining slightly more than the geography. However, DA showed that the AFLP fingerprint patterns, representing endophytic bacteria isolated from Glycyrrhiza nodules, showed host specificity and regional endemism. Our work involved strains isolated from closely related species of the Glycyrrhiza genus, most rhizobial species being able to cross-nodulate different Glycyrrhiza species. This may explain why only a small portion of the total genetic variation of bacteria was accounted for host plant species. A study involving specifically rhizobial strains obtained from different leguminous genera combined with several strains per each sampling site and with several environmental variables, including soil related ones, would provide more information about the role of host plant in driving the colonization of leguminous nodules by particular bacterial species.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information

Our work revealed that a wide variety of rhizobial and non-rhizobial endophytic bacteria can colonize Glycyrrhiza nodules. Glycyrrhiza plants are promiscuous in terms of nodulation because, besides mesorhizobia, the nodules harboured several other rhizobial genera. The classification of rhizobial species into true and sporadic symbionts is of practical importance. When looking for effective and stable inoculant strains for G. uralensis and G. glabra, our results suggest that these are to be found among the true symbiotic mesorhizobia.

To our knowledge this study is the first in which multivariate analyses were applied to resolve biogeographical patterns of endophytic bacteria, including those of rhizobia obtained from leguminous nodules. DA, CCA and partitioning of the variance with essential explorative and additional tests, for example the one examining the homogeneity of multivariate group dispersions, were appropriate methods to show when variation in bacterial AFLP fingerprint patterns exhibited specific biogeographical patterns, thereby providing additional useful knowledge related to the large collection of bacteria.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information

We thank Aneta Dresler-Nurmi and Petri Penttinen (University of Helsinki) for fruitful discussions and reading the manuscript and Jukka Rintala (Finnish Game and Fisheries Research Institute, RKTL) for projecting geographical coordinates in geographical information system and converting them to metric units. This work was funded by the National Science Foundation of China (31070444, 30970003, 31125007), the 973 project of China (2010CB126502), the Academy of Finland and the University of Helsinki. This work was shared with Northwest A & F University in China and University of Helsinki in Finland.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information
  • Alexander DB & Zuberer DA (1991) Use of chrome azurol S reagents to evaluate siderophore production by rhizosphere bacteria. Biol Fertil Soils 12: 3945.
  • Aljanabi SM & Martinez I (1997) Universal and rapid salt-extraction of high quality genomic DNA for PCR-based techniques. Nucleic Acids Res 25: 46924693.
  • Amarger N, Macheret V & Laguerre G (1997) Rhizobium gallicum sp. nov. and Rhizobium giardinii sp. nov., from Phaseolus vulgaris nodules. Int J Syst Evol Microbiol 47: 9961006.
  • Anderson MJ (2001) A new method for a non-parametric multivariate analysis of variance. Aust Ecol 26: 3246.
  • Anderson MJ (2004) DISTML v.5. Distance-Based Multivariate Analysis for a Linear Model. A Computer Program. Department of Statistics, University of Auckland, Auckland.
  • Anderson MJ (2006) Distance-based tests for homogeneity of multivariate dispersions. Biometrics 62: 245253.
  • Anderson MJ & Gribble NA (1998) Partitioning the variation among spatial, temporal and environmental components in a multivariate data set. Aust J Ecol 23: 158167.
  • Anderson MJ & Robinson J (2003) Generalized discriminant analysis based on distances. Aust N Z J Stat 45: 301318.
  • Bai Y, D'Aoust F, Smith DL & Driscoll BT (2002) Isolation of plant-promoting Bacillus strains from soybean root nodules. Can J Microbiol 48: 230238.
  • Bala A, Murphy P & Giller KE (2003) Distribution and diversity of rhizobia nodulating agroforestry legumes in soils from three continents in the tropics. Mol Ecol 12: 917930.
  • Beringer J (1974) R factor transfer in Rhizobium leguminosarum. J Gen Microbiol 84: 188189.
  • Bissett A, Richardson AE, Baker G, Wakelin S & Thrall PH (2010) Life history determines biogeographical patterns of soil bacterial communities over multiple spatial scales. Mol Ecol 19: 43154327.
  • Borcard D, Legendre P & Drapeau P (1992) Partialling out the spatial component of ecological variation. Ecology 73: 10451055.
  • Bremer C, Braker G, Matthies D, Beierkuhnlein C & Conrad R (2009) Plant presence and species combination, but not diversity, influence denitrifier activity and the composition of nirK-type denitrifier communities in grassland soil. FEMS Microbiol Ecol 70: 377387.
  • Bromfield ESP, Tambong JT, Cloutier S, Prévost D, Laguerre G, van Berkum P, Tran Thi TV, Assabgui R & Barran LR (2010) Ensifer, Phyllobacterium and Rhizobium species occupy nodules of Medicago sativa (alfalfa) and Melilotus alba (sweet clover) grown at a Canadian site without a history of cultivation. Microbiology 156: 505520.
  • Chen W, Wang E, Wang S, Li Y, Chen X & Li Y (1995) Characteristics of Rhizobium tianshanense sp. nov., a moderately and slowly growing root nodule bacterium isolated from an arid saline environment in Xinjiang, People's Republic of China. Int J Syst Bacteriol 45: 153159.
  • Cho J-C & Tiedje JM (2000) Biogeography and degree of endemicity of fluorescent Pseudomonas strains in soil. Appl Environ Microbiol 66: 54485456.
  • Chu H, Fierer N, Lauber CL, Caporaso JG, Knight R & Grogan P (2010) Soil bacterial diversity in the Arctic is not fundamentally different from that found in other biomes. Environ Microbiol 12: 29983006.
  • Cummings SP, Gyaneshwar P, Vinuesa P et al. (2009) Nodulation of Sesbania species by Rhizobium (Agrobacterium) strain IRBG74 and other rhizobia. Environ Microbiol 11: 25102525.
  • de Lajudie P, Willems A, Nick G et al. (1999) Agrobacterium bv. 1 strains isolated from nodules of tropical legumes. Syst Appl Microbiol 22: 119132.
  • Dresler-Nurmi A, Terefework Z, Kaijalainen S, Lindström K & Hatakka A (2000) Silver stained polyacrylamide gels and fluorescence-based automated capillary electrophoresis for detection of amplified fragment length polymorphism patterns obtained from white-rot fungi in the genus Trametes. J Microbiol Methods 41: 161172.
  • Dresler-Nurmi A, Fewer D, Räsänen LA & Lindström K (2009) The diversity and evolution of rhizobia. Microbiology Monographs, Volume 8: Prokaryotic Symbionts in Plants (Pawlowski K, ed), pp. 341. Springer-Verlag, Berlin.
  • Egamberdieva D & Kucharova Z (2009) Selection for root colonising bacteria stimulating wheat growth in saline soils. Biol Fertil Soils 45: 563571.
  • Fierer N & Jackson RB (2006) The diversity and biogeography of soil bacterial communities. P Natl Acad Sci USA 103: 626631.
  • Fiore C, Eisenhut M, Krausse R, Ragazzi E, Pellati D, Armanini D & Bielenberg J (2008) Antiviral effects of Glycyrrhiza species. Phytother Res 22: 141148.
  • Garbeva P, Postma J, van Veen JA & van Elsas JD (2006) Effect of above-ground plant species on soil microbial community structure and its impact on suppression of Rhizoctonia solani AG3. Environ Microbiol 8: 233246.
  • García-Fraile P, Rivas R, Willems A, Peix A, Martens M, Martínez-Molina E, Mateos PF & Velázquez E (2007) Rhizobium cellulosilyticum sp. nov., isolated from sawdust of Populus alba. Int J Syst Evol Microbiol 57: 844848.
  • Haichar FZ, Marol C, Berge O, Rangel-Castro JI, Prosser JI, Balesdent J, Heulin T & Achouak W (2008) Plant host habitat and root exudates shape soil bacterial community structure. ISME J 2: 12211230.
  • Hallman J, Quadt-Hallmann A, Mahaffee WF & Kloepper JW (1997) Bacterial endophytes in agricultural crops. Can J Microbiol 43: 895914.
  • Hardoim PR, van Overbeek LS & van Elsas JD (2008) Properties of bacterial endophytes and their proposed role in plant growth. Trends Microbiol 16: 463471.
  • Hayashi H & Sudo H (2009) Economic importance of licorice. Plant Biotechnol 26: 101104.
  • Hayashi H, Hattori S, Inoue K, Sarsenbaev K, Ito M & Honda G (2003) Field survey of Glycyrrhiza plants in central Asia (1). Characterization of G. uralensis, G. glabra and the putative intermediate collected in Kazakhstan. Biol Pharm Bull 26: 867871.
  • Hughes Martiny JB, Bohannan BJM, Brown JH et al. (2006) Microbial biogeography: putting microorganisms on the map. Nat Rev Microbiol 4: 102112.
  • Jones DL, Nguyen C & Finlay RD (2009) Carbon flow in the rhizosphere: carbon trading at the soil-root interface. Plant Soil 321: 533.
  • Kindt R & Coe R (2005) Tree Diversity Analysis: A Manual and Software for Common Statistical Methods for Ecological and Biodiversity Studies. World Agroforestry Centre, Nairobi.
  • King EO, Ward MK & Raney DC (1954) Two simple media for the demonstration of pyocyanin and fluorescein. J Lab Clin Med 44: 301307.
  • Kölliker R, Kraehenbuehl R, Boller B& Widmer F (2005) Genetic diversity and pathogenicity of the grass pathogen Xanthomonas translucens pv. graminis. Syst Appl Microbiol 29: 110119.
  • Kondo K, Shiba M, Yamaji H, Morota T, Zhengmin C, Huixia P & Shoyama Y (2007) Species identification of licorice using nrDNA and cpDNA genetic markers. Biol Pharm Bull 30: 14971502.
  • Kushiev H, Noble AD, Abdullaev I & Toshbekov U (2005) Remediation of abandoned saline soils using Glycyrrhiza glabra: a study from the Hungry Steppes of Central Asia. Int J Agric Sustainability 3: 102112.
  • Laranjo M, Machado J, Young JPW & Oliveira S (2004) High diversity of chickpea Mesorhizobium species isolated in a Portuguese agricultural region. FEMS Microbiol Ecol 48: 101107.
  • Legendre P & Legendre L (1998) Numerical ecology, 2nd edn. Developments in Environmental Modelling 20. Elsevier Science B.V., Amsterdam.
  • Lewis G, Schrire B, Mackinder B & Lock M (2005) (eds) Legumes of the World. Royal Botanic Gardens, Kew, London.
  • Lindström K, Sarsa M-L, Polkunen J & Kansanen P (1985) Symbiotic nitrogen fixation of Rhizobium (Galega) in acid soils, and its survival in soil under acid and cold stress. Plant Soil 87: 293302.
  • Lodewyckx C, Vangronsveld J, Porteous F, Moore ERB, Taghavi S, Mezgeay M & van der Lelie D (2002) Endophytic bacteria and their potential applications. Crit Rev Plant Sci 21: 583606.
  • Ludwig W, Strunk O, Westram R et al. (2004) ARB: a software environment for sequence data. Nucleic Acids Res 32: 13631371.
  • Mahdhi M, Nzoué A, Gueye F, Merabet C, de Lajudie P & Mars M (2007) Phenotypic and genotypic diversity of Genista saharae microsymbionts from the infra-arid region of Tunisia. Lett Appl Microbiol 45: 604609.
  • Mantelin S, Fischer-Le Saux M, Zakhia F, Béna G, Bonneau S, Jeder H, de Lajudie P & Cleyet-Marel J-C (2006) Emended description of the genus Phyllobacterium and description of four novel species associated with plant roots: Phyllobacterium bourgognense sp. nov., Phyllobacterium ifriqiyense sp. nov., Phyllobacterium leguminum sp. nov. and Phyllobacterium brassicacearum sp. nov. Int J Syst Evol Microbiol 56: 827839.
  • McArdle BH & Anderson MJ (2001) Fitting multivariate models to community data: a comment on distance-based redundancy analysis. Ecology 82: 290297.
  • Mhamdi R, Mrabet M, Laguerre G, Tiwari R & Aouani ME (2005) Colonization of Phaseolus vulgaris nodules by Agrobacterium-like strains. Can J Microbiol 51: 105111.
  • Micallef SA, Shiaris MP & Colón-Carmena A (2009) Influence of Arabidopsis thaliana accessions on rhizobacterial communities and natural variation in root exudates. J Exp Bot 60: 17291742.
  • Muchugi A, Muluvi GM, Kindt R, Kadu CAC, Simons AJ & Jamnadass RH (2008) Genetic structuring of important medicinal species of genus Warburgia as revealed by AFLP analysis. Tree Genet Genomes 4: 787795.
  • Muresu R, Polone E, Sulas L et al. (2008) Coexistence of predominantly nonculturable rhizobia with diverse, endophytic bacterial taxa within nodules of wild legumes. FEMS Microbiol Ecol 63: 383400.
  • Nemergut DR, Costello EK, Hamady M et al. (2011) Global patterns in the biogeography of bacterial taxa. Environ Microbiol 13: 135144.
  • Oksanen J, Kindt R, Legendre P, O'Hara B, Simpson GL, Stevens MHH & Wagner H (2008) Vegan: Community Ecology Package. R package version 1.13-1.
  • Papke RT & Ward DM (2004) The importance of physical isolation to microbial diversification. FEMS Microbiol Ecol 48: 293303.
  • Portier P, Fischer-Le Saux M, Mougel C, Lerondelle C, Chapulliot D, Thioulouse J & Nesme X (2006) Identification of genomic species in Agrobacterium biovar 1 by AFLP genomic markers. Appl Environ Microbiol 72: 71237131.
  • Quan ZX, Bae HS, Baek JH, Chen WF, Im WT & Lee ST (2005) Rhizobium daejeonense sp. nov. isolated from a cyanide treatment bioreactor. Int J Syst Evol Microbiol 55: 25432549.
  • R Development Core Team (2007) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna.
  • Ramette A & Tiedje JM (2007a) Biogeography: an emerging cornerstone for understanding prokaryotic diversity, ecology and evolution. Microb Ecol 53: 197207.
  • Ramette A & Tiedje JM (2007b) Multiscale responses of microbial life to spatial distance and environmental heterogeneity in a patchy system. P Natl Acad Sci USA 104: 27612766.
  • Räsänen LA, Sprent JI & Lindström K (2001) Symbiotic properties of sinorhizobia isolated from Acacia and Prosopis nodules in Sudan and Senegal. Plant Soil 235: 193210.
  • Rincón A, Arenal F, González I, Manrique E, Lucas MM & Pueyo JJ (2008) Diversity of rhizobial bacteria isolated from nodules of the gypsophyte Ononis tridentata L. growing in Spanish soils. Microb Ecol 56: 223233.
  • Sierra G (1957) A simple method for the detection of lipolytic activity of micro-organisms and some observations on the influence of the contact between cells and fatty substrates. Antonie Van Leeuwenhoek 23: 1522.
  • Spaepen S, Vanderleyden J & Remans R (2007) Indole-3-acetic acid in microbial and microorganism–plant signaling. FEMS Microbiol Rev 31: 425448.
  • Sprent JI (2001) Nodulation in Legumes. Royal Botanic Gardens, Kew, London.
  • Srividya S, Soumya S & Pooja K (2009) Influence of environmental factors and salinity on phosphate solubilization by a newly isolated Aspergillus niger F7 from agricultural soil. Afr J Biotechnol 8: 18641870.
  • Teather RM & Wood PJ (1982) Use of Congo red-polysaccharide interactions in enumeration and characterization of cellulolytic bacteria from the bovine rumen. Appl Environ Microbiol 43: 777780.
  • Valverde A, Velázquez E, Fernández-Santos F, Vizcaíno N, Rivas R, Mateos PF, Martínez-Molina E, Igual JM & Willems A (2005) Phyllobacterium trifolii sp. nov., nodulating Trifolium and Lupinus in Spanish soils. Int J Syst Evol Microbiol 55: 19851989.
  • Valverde A, Fterich A, Mahdhi M, Ramírez-Bahema M-H, Caviedes MA, Mars M, Velázquez E & Rodriguez-Llorente I (2010) Paenibacillus prosopidis sp. nov., isolated from nodules of Prosopis farcta. Int J Syst Evol Microbiol 60: 21822186.
  • Vassilev N, Vassileva M & Nikolaeva I (2006) Simultaneous P-solubilising and biocontrol activity of microorganisms: potentials and future trends. Appl Microbiol Biotechnol 71: 137144.
  • Vincent JM (1970) A Manual for the Practical Study of Root-Nodule Bacteria. IBP Handbook no. 15. International Biological Programme, London.
  • Vinuesa P, Silva C, Werner D & Martínez-Romero E (2005) Population genetics and phylogenetic inference in bacterial molecular systematics: the roles of migration and recombination in Bradyrhizobium species cohesion and delineation. Mol Phylogenet Evol 34: 2954.
  • Vos P, Hogers R, Bleeker M, Reijans M, van de Lee T, Hornes M, Frijters A, Pot J, Peleman J & Kuiper M (1995) AFLP: a new technique for DNA fingerprinting. Nucleic Acids Res 23: 44074414.
  • Wang LL, Wang ET, Liu J, Li Y & Chen WX (2006) Endophytic occupation of root nodules and roots of Melilotus dentatus by Agrobacterium tumefaciens. Microb Ecol 52: 436443.
  • Weisburg WG, Barns SM, Pelletier DA & Lane DJ (1991) 16S ribosomal DNA amplification for phylogenetic study. J Bacteriol 173: 697703.
  • Wilson K (1994) Preparation of genomic DNA from bacteria. Current Protocols in Molecular Biology (Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG, Smith JA & Struhl K, eds), pp. John Wiley & Sons Inc., New York.
  • Yamamoto Y & Tani T (2005) Field study and pharmaceutical evaluation of Glycyrrhiza uralensis roots cultivated in China. J Trad Med 22: 8697.
  • Zakhia F, Jeder H, Willems A, Gillis M, Gillis M, Dreyfus B & Lajudie P (2006) Diverse bacteria associated with root nodules of spontaneous legumes in Tunisia and first report for nifH-like gene within the genera Microbacterium and Starkeya. Microb Ecol 51: 375393.
  • Zimnitskaya SA (2009) State of the reproductive system of populations of species of the genus Glycyrrhiza L. (Fabaceae). Contemp Problems Ecol 2: 392395.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Conclusions
  8. Acknowledgements
  9. References
  10. Supporting Information
fem1198-sup-0001-Supplement1Procedures.pdfapplication/PDF69KAppendix S1. Procedures used for the extraction of DNA, selective AFLP-PCR and 16S rDNA-PCR.
fem1198-sup-0002-FigureS1.pdfapplication/PDF54KFig. S1. Geographic locations of the three sampling regions, northern Xinjiang, southern Xinjiang and central China, where Glycyrrhiza nodules were collected.
fem1198-sup-0003-FigureS2.pdfapplication/PDF19KFig. S2. Dendrogram based on the AFLP fingerprinting analysis showing the genetic diversity among endophytic bacteria isolated from the root nodules of Glycyrrhiza spp. growing in China.
fem1198-sup-0004-FigureS3.pdfapplication/PDF14KFig. S3. Box plots visualising the homogeneity of multivariate group dispersions obtained from bacterial AFLP fingerprint data when the matrix of similarities (transformed to dissimilarities) was obtained by using the Dice coefficient.
fem1198-sup-0005-TableS1.pdfapplication/PDF37KTable S1. Environmental parameters, host plant and bacterial species identified from each sampling site.
fem1198-sup-0006-TableS2.pdfapplication/PDF6KTable S2. Pairwise comparison of multivariate dispersions of a priori groups.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.