Large genotypic variation but small variation in N2 fixation among rhizobia nodulating red clover in soils of northern Scandinavia


Mette M. Svenning, Department of Biology, Faculty of Science, University of Tromsø, N-9037 Tromsø, Norway.


Aims:  To analyse the symbiotic variations within indigenous populations of rhizobia nodulating red clover (Trifolium pratense L.) in soils of northern Norway and Sweden at different times of the growing season.

Methods and Results:  A total of 431 nodule isolates sampled under field conditions in summer and autumn, were characterized genetically by targeting both chromosomal and symbiotic genes. The Enterobacterial Repetitive Intergenic Consensus polymerase chain reaction (PCR) fingerprinting of chromosomal DNA revealed considerable variation within the isolated populations that was more influenced by geographical origin than sampling time. Analysis of PCR amplified nodEF gene on the symbiotic plasmid by restriction fragment length polymorphism revealed a high proportion of nod types common to the two studied sites. The symbiotic efficiency of the isolates, representing both dominating and rare nodEF genotypes, showed high N2 fixation rates in symbiosis with the host plant in a greenhouse experiment using the 15N isotope dilution method.

Conclusions:  Effective N2-fixing strains of Rhizobium leguminosarum bv. trifolii nodulating red clover are common and genetically diverse in these northern Scandinavia soils.

Significance and Impact of the Study:  This study provides information on the variability, stability and dynamics of resident populations of rhizobia nodulating red clover in Scandinavian soils which has practical implications for applying biological nitrogen fixation in subarctic plant production.


Rhizobia are well known for their contribution to the soil nitrogen pool through N2 fixation in symbioses with leguminous plants. In many areas where legumes are cultivated or make up a part of the native flora, indigenous or naturalized populations of rhizobia are abundant in the soil. Studies of these populations have been motivated by ecological and agricultural considerations. In terms of agriculture, high abundance of rhizobia in the soil facilitates cultivation of legumes as they can be grown without inoculation. On the other hand, soils containing indigenous rhizobia are problematic as they create a barrier to the establishment of introduced efficient inoculants in nodules of target host plants (Vlassak and Vanderleyden 1997). Also, there is evidence of widespread suboptimal efficiency of native strains with agricultural legumes (Brockwell et al. 1995). However, it is generally argued that indigenous populations are highly adapted to their local soil environments and may form comparable or more effective symbioses than commercial inoculants isolated from a distant and unrelated soil environment (Gandee et al. 1999). Thus, selection of indigenous strains with high nitrogen-fixing capacity, adapted to a range of environmental conditions at a specific site, remains an important strategy to maximize legume production.

Indigenous soil populations of rhizobia (Rhizobium leguminosarum bv. trifolii) nodulating clovers (Trifolium) are often highly genetically diverse, but typically consist of one or a few dominating genotypes and several genotypes present at lower frequency (Leung et al. 1994a; Zézéet al. 2001; Fagerli and Svenning 2005). There is evidence that different clover species select for specific subtypes of rhizobia from a natural population (Robinson 1969; Harrison et al. 1989). In northern Scandinavia, R. leguminosarum bv. trifolii populations associated with white clover (T. repens L.) have been particularly emphasized in studies of rhizobial diversity. For example, Svenning (1991) extensively analysed the phenotypic and genotypic diversity of indigenous isolates collected from nodules of white clover grown in different regions of Norway. On the contrary, information on diversity of rhizobia nodulating red clover (T. pratense L.) in northern areas remains limited in spite of red clover being the most important forage legume in Scandinavia (Frame et al. 1998). Recent breeding programmes have generated new cultivars of red clover with improved winter survival and high forage quality. To ensure effective symbiosis with these new cultivars, it is essential to obtain data on their symbiotic efficiency in association with indigenous rhizobia adapted to northern climatic conditions. Two recent studies have indicated large genetic diversity among nodule isolates sampled with the improved red clover cultivar Betty in cold temperate soils (Fagerli and Svenning 2005; Duodu et al. 2006). However, no studies have considered the symbiotic efficiency of the indigenous populations of red clover-nodulating rhizobia with this host, nor has the influence of seasonal changes in the composition of active soil bacteria been addressed. Monitoring such seasonal variation may be of agronomical importance, given the influence that an established or varying rhizobial population can have on the symbiotic N2-fixation efficiency of particular plant cultivars.

In this study, seasonal changes in the composition and effectiveness of rhizobial populations isolated from field-grown red clover trap plants were analysed at two geographical sites: northern Norway (Tromsø) and northern Sweden (Umeå). The aim was to reveal the genotypic and phenotypic variations among R. leguminosarum bv. trifolii genotypes at different parts of the growing season. The genetic variation among sampled isolates was analysed based on variations within their chromosomal as well as their symbiosis (sym) plasmid genomes. DNA fingerprinting with Enterobacterial Repetitive Intergenic Consensus (ERIC) primers in the polymerase chain reaction (PCR) method was used to characterize the chromosomal diversity (Versalovic et al. 1991; de Bruijn 1992), and the sym portion of the genome was analysed by restriction fragment length polymorphisms (RFLP) of a PCR-amplified nodEF gene sequence. As genes important for nodulation and nitrogen fixation are primarily located on sym plasmid, representative isolates of the nodEF types were tested for their N2-fixation efficiency using the 15N isotope dilution (ID) method in a controlled environment.

Materials and methods

Field sites

The two field sites used in this study were located on agricultural land, at the Plant Research Centre (Holt, Tromsø, Norway) (69°40′N 18°56′E) and at the Swedish University of Agricultural Sciences (SLU) in Umeå, Sweden (63°45′N 20°17′E). The Norwegian site was a managed grass field that had not been sown to clovers since 1993. At the time of sampling, June (summer) 2001 and September (autumn) 2001, the selected plot (measuring 1·5 m × 3·0 m) was covered with grasses, although a few plants of white clover were observed in the vicinity. The soil was sandy loam (USDA soil classification). The field has been routinely fertilized with 106 kg N per ha, 16 kg P per ha and 88 kg K per ha. Sampling of R. leguminosarum bv. trifolii population was carried out in the same plot in the summer and in autumn. The soil temperatures measured at 5-cm depth during trapping of Rhizobium were in the range of 10–15°C and 5–16°C for the summer and autumn samplings, respectively.

The soil at the Swedish site was fine silty sand with little clay. The cropping history before the sampling dates, July (summer) 2001 and September (autumn) 2001, was a crop rotation with barley, potato and grass/clover leys. Sampling of the summer and autumn rhizobial populations was performed in the same field site, but in two separate plots (7·0 m × 5·0 m) distanced about 150 m apart. In the sampling year, the plot for summer sampling carried barley under-sown with a grass/red clover ley and was fertilized in spring with 42 kg N per ha, 16·5 kg P per ha and 31·5 kg K per ha. The plot for autumn sampling carried a first-year grass/red clover ley and was fertilized with 14 kg P per ha, 50 kg K per ha in the sampling year. The soil temperatures measured at 5-cm depth during trapping of Rhizobium were in the range of 8–15°C and 11–15°C for the summer and autumn samplings respectively. At both sites, soil was sampled to 10-cm depth and analysed for pH, concentration of carbon, total N, available P, Mg, Ca and K (Table 1). The total infective rhizobial population size in soil was estimated using the standard most probable number test (Vincent 1970).

Table 1.   Soil chemistry and population size of infective Rhizobium leguminosarum bv. trifolii in the two studied sites
SiteSampling seasonmg 100 g−1pH in H2OMPN§ (infective cells per g dry soil)
  1. *Extracted with NH4-lactate in Tromsø and with NH4-acetate in Umeå.

  2. †Loss-on-ignition in Tromsø, C/N autoanalyser in Umeå.

  3. ‡Kjeldahl-N in Tromsø, C/N autoanalyser in Umeå.

  4. §MPN, most probable number (Vincent 1970).

TromsøSummer2033·111·817·53·50·315·51·2 × 104
Autumn1993·811·818·03·20·305·88·4 × 103
UmeåSummer10724·417·40·152·10·156·43·0 × 105
Autumn93·29·462·160·271·80·136·32·0 × 104

Plant material and isolation of R. leguminosarum bv. trifolii strains

Red clover (Trifolium pratense L.) cv. Betty (Svalöf-Weibull AB, Lännäs, Sweden) was used as a trap host to sample red clover-nodulating rhizobial populations from the field plots. Seeds were surface sterilized in 3% (v/v) H2O2 for 5 min (Somasegaran and Hoben 1994), rinsed four to five times in sterile water, and incubated on 0·9% water agar. The seeds were pre-conditioned in the dark at 4°C overnight before germinating at room temperature. Seedlings were transferred to sterilized Magenta jars (Sigma, St Louis, MO, USA) and grown in vermiculite or perlite moistened with sterile N-free modified Evans solution (Huss-Danell 1978) diluted to 1/10 of full strength. These were kept in a greenhouse for 7 days with 17 h of supplemental light from Philips HPI/T 400 W lamps (Philips, Turnhout, Belgium) at 25°C and 7 h without supplemental light at 15°C.

At least 30 clover seedlings were uniformly transplanted to each field plot, except the summer sampling in Umeå, where red clover seeds were sown directly in the field plot together with barley and grasses. In this particular case, the clover seeds were not surface sterilized, but did not carry rhizobia as non-sterilized seeds from the same seed lot grown in the greenhouse did not develop nodules. After 5–6 weeks (8 weeks for seeds sown directly in field plot), all plants were carefully dug up with the roots intact and 20 plants were arbitrarily selected and transported to the laboratory. At sampling, the plants were still in their leaf rosette stage with three to four leaves per plant. All nodules on the roots of the 20 trap plants were collected and rhizobia isolated from crushed nodules as described by Svenning et al. (2001). Isolates were stored at −80°C in tryptone yeast (TY, Somasegaran and Hoben 1994) broth containing 25% glycerol.

Isolation of DNA from reference strain

Genomic DNA of the reference strain, R. leguminosarum bv. trifolii USDA2063 (US Department of Agriculture, Agricultural Research Service, National Rhizobium Germplasm Collection, Beltsville, MD, USA) was extracted from liquid cultures harvested at late exponential phase according to the protocol of Wilson (1987).

Genetic characterization of isolates by genomic ERIC fingerprinting and PCR-RFLP

A total of 431 field isolates were screened for their chromosomal heterogeneity by rep-PCR using ERIC primers (Versalovic et al. 1991; de Bruijn 1992). These included: 115 isolates from the summer sampling and 82 isolates from the autumn sampling in Tromsø; 162 isolates from the summer sampling and 72 isolates from the autumn sampling in Umeå. Individual isolates from the frozen stocks were grown on TY agar medium and fingerprinted by PCR amplification. The PCR reaction was performed in a total volume of 25 μl containing 1 μl of cell suspension or 50 ng extracted DNA of the reference strain as template, 1 U DynaZymeTM DNA Polymerase (F500L) (Finnzymes Oy, Espoo, Finland), 1x reaction buffer with 10% DMSO, 0·2 mmol l−1 dNTP (Finnzymes Oy) and 50 pmol of ERIC1 and ERIC2 (Eurogentec, Seraing, Belgium). Amplification was performed in a PTC 200, Peltier Thermal Cycler (MJ Research, Waltham, MA, USA). The cycling conditions were as described by de Bruijn (1992). Twelve microlitre aliquots of amplified DNA were separated on 1·5% Seakem®LE agarose (FMC BioProducts, Rockland, ME, USA) gels in Tris-borate–EDTA at 80 V for 3 h. All gels stained by ethidium bromide were photographed using the Gel DocTM 2000 Gel Documentation System (Bio-Rad, Hercules, CA, USA). The generated ERIC patterns were reproducible and differences in fingerprints served as the basis for grouping the isolates into ERIC types (Fig. 1). Representative isolates (varying from 1 to 5) from each ERIC type were further characterized by PCR-RFLP analysis of the nodEF gene region and the 16S–23S rDNA sequences of the internally transcribed spacer (ITS) region. The primers and reaction conditions used for amplification of the nodEF gene fragment were as described before (Fagerli and Svenning 2005). A 1·5 kb 16S–23S ITS gene fragment was amplified using primers and cycling conditions as described by de Oliveira et al. (1999) in a final volume of 50 μl containing: 2·0 U Taq DNA polymerase (Qiagen Inc., Valencia, CA, USA), 1 × Taq buffer, 2·5 mmol l−1 MgCl2, 0·2 mmol l−1 dNTP mix and 0·05 μmol l−1 of each primer. For both gene markers, the amplified DNA fragments were checked on 0·9% (w/v) Seakem®LE agarose (FMC BioProducts) gels at 100 V for 2 h. The sizes of the PCR product were estimated based on a molecular weight standard (pGEM; Promega, Madisson, WI, USA). Four restriction endonucleases, AluI, TaqI, CfoI and HinfI (Promega Corporation) were selected for their highly discriminatory power based on in silico cutting ( Digestion of PCR products were carried out as recommended by the manufacturer in total volume of 18 μl, containing 7 μl of PCR product (250–300 ng), 5 U enzyme (Promega), and 2·5 μl of 10x buffers specific for the enzyme. The digests were separated on 2·5% (w/v) Metaphor® agarose (FMC BioProducts) gels.

Figure 1.

 A picture of an agarose gel showing different fingerprint patterns of Rhizobium leguminosarum bv. trifolii isolates generated by rep-polymerase chain reaction (PCR) using Enterobacterial Repetitive Intergenic Consensus (ERIC) primers. The isolates were recovered from nodules of red clover grown in the field site in Tromsø. The numbering of the lanes corresponds to specific ERIC-PCR genotypes. Lanes labelled with numbers >30 represent groups with less than five isolates which are included in ‘others’ in Table 3; pGEM is molecular size marker (Promega).

Estimation of N2-fixation efficiency

To relate genomic identity and abundance in the field soils to symbiotic efficiency, eight isolates representing dominating and rare genotypes were tested for their N2-fixation efficiency in symbiosis with the host plant. The isolates included both common and site-specific nodEF types. Two reference strains, R. leguminosarum bv. trifolii 20–15 (local collection, Department of Biology, University of Tromsø, Tromsø, Norway) and R. leguminosarum bv. trifolii HAMBI 461 (Elomestari Oy, Kukkola, Finland), isolated from native soil populations in Norway and Finland, respectively, were also included in the N2-fixation measurement. The commercial strain R. leguminosarum bv. trifolii HAMBI 461 is originally isolated from red clover, whereas strain 20–15 is originally isolated from a white clover nodule (Svenning et al. 2001). Red clover seeds were surface sterilized and germinated as described before. Seedlings were transferred singly to sterilized Magenta jars (Sigma, St Louis, MO, USA) containing sterile vermiculite. Seedlings were watered with sterile modified Evans nutrient solution (Huss-Danell 1978) diluted to 1/10 of full strength. To measure N2 fixation according to the 15N ID method (McAuliffe et al. 1958), the solution was supplemented with 0·14 mmol l−1 N added as 15NHinline imageNO3 (1 atom %15N excess). Each plant was inoculated with 200 μl of inoculum of the appropriate bacterial suspension, initiated by diluting 100 μl of culture (OD500 0·2–0·3) into 10 ml of yeast mannitol broth (Somasegaran and Hoben 1994). Immediately after inoculation, the vermiculite was covered with a layer (c. 10 mm) of sterile fine sand (150–300 μm grain size) to minimize the risk of contamination. There were five replicate plants for each tested Rhizobium isolate, along with uninoculated controls. All jars were placed in a greenhouse with the same conditions as described previously. The nutrient solution was renewed at least every 14 days. Roots were shielded from light with aluminium foil. At 18 days after inoculation, the lid of each jar was replaced by a plastic bag covering the shoot, so as to give the plants a less enclosed environment but still maintain a contamination barrier. Plants were harvested at 98 days after inoculation. Due to the stringent aseptic precautions that were taken throughout the experiment, the uninoculated controls remained nodule free. ERIC-PCR fingerprinting of rhizobial isolates from ten nodules in each treatment, arbitrarily chosen at harvest, confirmed that there had been no contamination of rhizobia during the experiment. All plants were separated into shoots and nodulated roots, dried for 24 h at 60°C, weighed (dry matter) and milled in a ball-mill (Retsch MM 2000; Haan, Germany). Shoots were analysed for N concentration and 15N content using an online CN analyser (Europa Scientific ANCA-NT; Europa Scientific, Crewe, UK) coupled to an isotope ratio mass spectrometer (Europa Scientific Europa 20-20) at the IRMS Laboratory, Department of Forest Ecology, Swedish University of Agricultural Sciences, Umeå (Ohlsson and Wallmark 1999). The proportion of N in the shoots coming from N2 fixation (Ndfa) was calculated according to the 15N ID method (McAuliffe et al. 1958): Ndfa = 1 − (15Nfix/15Nref), where 15Nfix is the atom %15N excess in the N2-fixing plant and 15Nref is the atom %15N excess in the non-N2-fixing reference plant (uninoculated control plant). The amount of N coming from N2 fixation (Nfix) was calculated as: Nfix = Ndfa × N concentration (g N/g shoot dry matter) × shoot dry matter (g).

Data analysis

Fingerprints of ERIC-PCR patterns were analysed visually, and the isolates were manually grouped into ERIC types. Diversity among isolates was estimated by using the strain richness index (SRI), calculated by dividing the number of ERIC types identified by the number of isolates sampled. The fingerprint patterns generated by restriction of the PCR-amplified nodEF and 16S–23S ITS genes were analysed both visually and by the Molecular Analyst® Fingerprinting program, version 1.6 (Applied Maths BVBA, Kortijk, Belgium). Smaller fragments (100 bp and less) were not included in the analysis, because they appeared diffuse in the gel. A dendrogram was constructed using Dice correlation coefficient for band-based similarity calculations and the average linkage (UPGMA) method. In the case of N2-fixation efficiency, analysis of variance was performed using Minitab statistical software, release 14 (Minitab Inc., State College Pennsylvania, PA, USA, 2003), and the two-sample Student's t-test (unequal variance) was used to test for possible differences between treatment mean values (P < 0·05).


ERIC fingerprinting of R. leguminosarum bv. trifolii isolates

ERIC PCR-DNA fingerprinting of field nodule isolates showed multiple ERIC types within the sampled populations (Table 2). Among the isolates from Tromsø, 43 and 31 ERIC types were detected in summer and autumn samplings, respectively (Table 2). These two samplings shared 12 ERIC types, constituting 61·4% (121 of 197) of all isolates sampled from the soil population. Most of these common genotypes were recovered at nearly similar frequencies at both sampling times. The same three ERIC types (types 14, 23 and 26; Table 3) dominated at each sampling time. However, there were many unique genotypes recovered that were encountered only at one sampling time with a majority represented by only one isolate.

Table 2.   Number of isolates, Enterobacterial Repetitive Intergenic Consensus (ERIC) types and strain richness index (SRI) of Rhizobium leguminosarum bv. trifolii populations isolated from nodules of red clover trap plants grown at the field sites in Tromsø and Umeå during the summer and autumn seasons
SiteSampling seasonNo. isolatesNo. ERIC typesSRI*No. shared ERIC types
Within TromsøWithin UmeåBetween Tromsø and Umeå
  1. *SRI has been calculated by dividing the number of ERIC types identified by the number of isolates recovered.

TromsøSummer115430·3712 1
UmeåSummer162440·27 10 
Table 3.   Distribution of Enterobacterial Repetitive Intergenic Consensus (ERIC) types among nodule isolates obtained from summer and autumn sampling of red clover Rhizobium leguminosarum bv. trifolii field populations
ERIC typePercentage of isolates from
  1. *The nodEF types associated with each ERIC types are given within parentheses.

  2. †The results from ERIC types represented by fewer than five isolates and restricted to only one site or season.

1 (a)*37  
2 (b, n)  153
3 (d)  5 
4 (j, n)  23
5 (b)31  
6 (a)24  
7 (a)  23
8 (e)  7 
9 (a)13  
10 (g, i)  7 
11 (i)13  
12 (a)  414
13 (d)17  
14 (e)2716  
15 (c, j)  11
16 (g)  11
17 (e)  13 
18 (c)13  
19 (e)4   
20 (a)  110
21 (d, n)   7
22 (d)  11
23 (d, j)810731
24 (j)36  
25 (e)23  
26 (d)86  
27 (j)  33
28 (e)  3 
29 (i)11  
30 (g)1   

Rhizobial isolates trapped from soils in Umeå also revealed large chromosomal variation (Table 2). There were 44 and 19 ERIC types identified in the summer and autumn collections, respectively. Ten of these ERIC types were common to both sampling times, and constituted 45% (106 of 234) of all isolates sampled. The dominating ERIC types changed with sampling time. In summer, ERIC type 2 (15% of isolates) dominated the sampled population; while in autumn isolates of ERIC type 23 (31% of isolates) were found at the highest frequency. Some ERIC types with unique profiles appeared at a relatively high frequency in the summer (ERIC types 3, 8, 10 and 17) and autumn (ERIC type 21) samplings respectively. There were also many unique genotypes recovered at low frequency at each sampling time.

One ERIC type (23) was common across both sites. This was the most dominant ERIC type (14% of isolates) in Umeå and the second largest (9% of isolates) ERIC type recovered from Tromsø. Arbitrarily selected isolates from this group, encountered at both sites, were further examined with RFLP analysis of PCR amplified 16S–23S rDNA ITS region. The results showed identical ITS PCR-RFLP patterns for all these isolates (data not shown), indicating that they belonged to the same strain.

SRI (the number of ERIC types identified divided by the number of isolates sampled) calculated for each population was in the range of 0·26–0·38 (Table 2), with greater richness in the Tromsø than in the Umeå populations. The strain richness was similar for isolates obtained in the summer and autumn samplings at each site even though we studied twice as many isolates from summer samplings as from autumn samplings.

PCR-RFLP analysis of the nodEF region

Variation in the nodEF gene region of arbitrarily selected isolates representing the different ERIC types was examined by RFLP analysis. PCR amplification resulted in detection of a single band with the expected size of 1130 bp. Restrictions of the nodEF product with AluI, TaqI, CfoI and HinfI, generated two to six bands and four distinct RFLP patterns each. By combining these patterns, 14 different nodEF types were found among the 431 isolates. Of these 14 genotypes, three were unique to the populations from Tromsø or Umeå, while eight were common to both geographical sites (Table 4). The majority of the isolates (>90%) shared nodEF types either within or between sites, and very few isolates (<2%) had a nodEF type exclusively detected at a single sampling time at each site. The nodEF type e was the most dominant genotype at both sites, but was not recovered in Umeå in the autumn population (Table 4). Isolates belonging to the three next most frequent nodEF types (j, a and d; in order of abundance) were detected from all sampled populations, but were more frequent in autumn than in summer samplings in Umeå. The predominant nodEF type e was found to be associated with the majority of the ERIC types (15 of 59) recovered from the populations in Umeå. For the populations in Tromsø, nodEF type e was associated with the second largest group of ERIC types (11 of 62), which included the most prevalent ERIC type 14. It is noteworthy that some of the isolates sharing chromosomal identity by their ERIC fingerprint patterns (ERIC types 2, 4, 10, 15, 21 and 23) had different nodEF types (Table 3). The ERIC type 23 isolates from Tromsø harboured only nodEF type d, whereas those found in Umeå carried nodEF types d and j.

Table 4.   Distribution and frequency of nodEF types among nodule isolates obtained from summer and autumn sampling of red clover Rhizobium leguminosarum bv. trifolii field populations
SiteSampling seasonPercentage of isolates in each nodEF type
  1. ERIC, Enterobacterial Repetitive Intergenic Consensus.

Autumn211161128   5161   
UmeåSummer9174635 8 811 1 1
Autumn281811  11  36  13
No. ERIC types in each nodEF type
 Tromsø 2028711121361   
 Umeå 846615 7 34 213

Clustering analysis using Dice coefficient revealed two main clusters for the nodEF types, which joined at level of similarity of 52·3% (Fig. 2). The distribution of the genotypes between these two clusters was not specific to any one site or sampling time. However, cluster I included two of the most dominant nodEF types found in Tromsø (genotype e and a), whereas cluster II included nodEF type j which was mostly encountered at the site in Umeå. The reference strain, R. leguminosarum bv. trifolii USDA2063, showed identical restriction patterns to nodEF type d, and was grouped together in cluster II at a 100% level of similarity (Fig. 2).

Figure 2.

 Dendrogram representing the genetic relatedness between nod genotypes based on restriction fragment length polymorphism analysis of the polymerase chain reaction amplified nodEF fragments (1130 bp). Included is also the reference strain USDA2063. The dendrogram was constructed using the unweighted pair group method with arithmetric mean implemented in the Molecular Analyst® Fingerprinting program.

Symbiotic efficiency of nodEF types

The eight isolates tested for N2 fixation efficiency in the greenhouse experiment represented one isolate each of the nodEF types a, b, e, h, i and m (Table 4). Two isolates were included for the nodEF type c, designated as c and c-1, collected in Tromsø and Umeå respectively. All inoculated plants formed nodules which were dispersed along the whole root system. Each plant had at least 100 nodules. Plants inoculated with the isolate characterized by the very rare nodEF type m formed small and pale nodules (about 1 mm in diameter and length), and had pale yellow leaves as the uninoculated control plants that were devoid of nodules at the end of the experiment. The shoot heights in these two treatments were about 10 cm, and the whole-plant dry weights were about 0·1 g per plant. In all other treatments the plants had dark pink nodules (about 1 mm in diameter and 1- to 5-mm long), dark-green leaves, shoot heights of about 20 cm, and whole-plant dry weights of about 0·8 g per plant. The proportion of plant N derived from N2 fixation, Ndfa, was around 0·7 in all treatments except for plants inoculated with the nodEF type m, which gave an extremely low Ndfa value of 0·1 (Table 5). The amount of N2 fixed (Nfix) varied depending on the isolate used as inoculant (Table 5). Plants inoculated with the isolate characterized by nodEF type m fixed almost no nitrogen, and produced only slightly more dry matter than the uninoculated control plants (data not shown). The highest level of N2 fixation was obtained with isolates characterized by the nodEF types b and c-1, but the only statistically significant (P < 0·05) difference in Nfix was that nodEF type c-1 fixed more N2 than nodEF type c (Table 5).

Table 5.   Proportion of plant N derived from N2 fixation (Ndfa) and amount of N2 fixed (Nfix) with different nodEF types and two reference strains
Inoculation treatmentNdfaNfix mg N/shoot
  1. Values presented are mean of five replicates and those appearing in the same column and with the same letter are not statistically different by the Student's t-test (P > 0·05). SD values are in parentheses.

nodEF a0·72 (0·01)a11·81 (1·38)ab
nodEF b0·71 (0·01)a12·56 (1·60)ab
nodEF c0·71 (0·01)a10·67 (1·24)a
nodEF c-10·71 (0·01)a12·77 (0·92)b
nodEF e0·71 (0·01)a11·023 (1·50)ab
nodEF h0·70 (0·02)a11·31 (3·58)ab
nodEF i0·71 (0·02)a10·66 (2·67)ab
nodEF m0·11 (0·02)b 0·06 (0·01)c
Rhizobium leguminosarum bv. trifolii 20–150·71 (0·01)a10·90 (2·32)ab
Rhizobium leguminosarum bv. trifolii HAMBI 4610·70 (0·02)a 8·27 (3·52)ab


The composition of natural populations of Rhizobium has been shown to change with the sampling time (Bromfield et al. 2001). The basic question addressed in this study was to analyse for seasonal changes in the population of active rhizobia, associated with red clover grown in northern Scandinavia soils and to determine whether this has any relevance to the N2-fixing potential of the cultivar Betty. The two field sites used in this study differed markedly in their cropping history. While the site at Umeå has been subjected to recent cultivation of arable crops, the site at Tromsø was a grass field that had received less agricultural activity in the past 8 years. We observed that the same pool of genotypes dominate the population in Tromsø over the two sampling dates, suggesting a stable structure for the symbiotic strains. Leung et al. (1994b) also observed minor seasonal shifts of rhizobial symbionts in the rhizosphere of orchard grass compared with a strong rhizospheric effect for subclover. This and the findings reported here, therefore suggests that grassland soils probably create a more favourable environment for uniform selection of rhizobial genotypes that dominate the population over time. On the contrary, the differences in genetic composition of the isolated populations between the two samplings in Umeå could be results of several possible mechanisms including: (i) the occurrence of greater disturbance from repeated cultivation; and (ii) distinct microsites due to spatial separation of the plots or unique environmental niches to which different rhizobial strains are adapted (Handley et al. 1998; Palmer and Young 2000). Soil-specific factors like differences in magnesium concentrations have been invoked to explain changes in rhizobial genotypes present in a population (Labes et al. 1996). This might also apply here as the two plots sampled differed greatly in their magnesium content (Table 1). Although differences in the temperature range for growth and infection of the trap plant could also generate differences in the isolates recovered (Lipsanen and Lindstrom 1986; Prévost et al. 2003), this might be of less significance as the soil temperatures were very similar during the summer and the autumn samplings. Other environmental factors such as differences in soil moisture availability and length of diurnal cycle, which might cause changes in nodulation by different rhizobial genotypes (Balatti and Giménez 1989; Postma et al. 1989; Svenning 1991), were not studied in this work. The importance of agricultural practice in influencing the genomic stability and diversity of the R. leguminosarum bv. trifolii populations was also demonstrated by different nodEF types found in the same chromosomal background and the lower SRI measured for the nodulating populations in Umeå compared with Tromsø (Tables 2 and 3). Mixed results have been reported on the effect of land management practice on rhizobial diversity in arable vs uncultivated sites or grasslands (Handley et al. 1998; Palmer and Young 2000; Zhang et al. 2001). Whereas Palmer and Young (2000) indicated higher diversity in arable fields with repeated cultivation, Handley et al. (1998) demonstrated that diversity is not affected by cultivation. From this study, we may suggest that continuous cultivation of arable crops may reduce the diversity of rhizobia in symbiotic nodules. The detection of multiple nodEF types among isolates with the same chromosomal identity suggests exchange or transfers of sym plasmid which gives a corroborative support to the theory that the rate of plasmid transfer might be higher in cultivated than in relatively undisturbed soils (Wernegreen et al. 1997).

Generally, the isolated ERIC types were geographically separable, agreeing with other studies (Harrison et al. 1989; Fagerli and Svenning 2005). However, in contrast to most previous studies on rhizobial populations (Martinez-Romero 2003; Vessey and Chemining'wa 2006), one ERIC type (type 23) was detected at high frequencies at both sites and was associated with some of the dominant nodEF types (Tables 3 and 4). There is evidence that the sym plasmid contributes significantly to the ecological fitness of Rhizobium strains in soil (Wernegreen et al. 1997; Brom et al. 2002). Also, competitive traits important for nodulation have been localized to both the chromosome and the sym plasmid (Brewin et al. 1983; Laguerre et al. 2003). Thus, the dominance of ERIC type 23 isolates suggests better local adaptation and/or greater competitiveness for host nodulation. It needs to be recognized that the ERIC 23 isolates also carried the same nodEF type (type d) as the reference strain, USDA2063 from the USA. Even though no records of previous inoculation have been reported for these fields, the wide distribution of this nodEF plasmid could result from agricultural activities, as rhizobia attached to seed coats provide an important mechanism for their broad dispersal (Pérez-Ramírez et al. 1998).

Despite the large genetic differences, only small variations in symbiotic efficiency among the nodulating genotypes were observed in the sampled populations. Nearly all tested nodEF types formed effective symbioses with the host plant as the estimated Ndfa value was at least 0·7. Such Ndfa values have frequently been found in field studies (Carlsson and Huss-Danell 2003), including fields with red clover cultivar Betty (Huss-Danell and Chaia 2005). These findings are intriguing as resident clover rhizobial populations often consist of strains that are ineffective or poorly effective in N2 fixation (Holding and King 1963; Bergersen 1970). However, there is evidence that different clover plants preferentially nodulate with effective symbionts in native populations (Robinson 1969), possibly explaining the dominance of effective strains with this cultivar. The failure of the commercial inoculant strain (HAMBI 461) to achieve the highest level of N2 fixation (Table 5) probably reflects a cultivar preference or influence of environmental variables or both.

It is worth saying that this study employed testing of single genotypes of rhizobia in a controlled environment, whose symbiotic efficiency may or may not be the same under field conditions. However, a high correlation is observed previously between the N2-fixing effectiveness of strains tested in the greenhouse and in the field, based on total plant weight and total N (Bremer et al. 1990). Thus, the strains that showed effective symbioses in our tested greenhouse conditions are also likely to be efficient in the field, in spite of the expected differences in nutrient availability and growing conditions. The fact that both rare and dominant isolates were equally efficient in N2 fixation, suggests that nodule dominance of sym genotypes is not necessarily correlated to greater symbiotic efficiency. The relatively efficient genotypes recovered following a long absence of clover hosts, suggest some relation between saprophytic growth and N2-fixing efficiency. Our results also suggest that there might be other chromosomally or plasmid-encoded genes determining the N2-fixation efficiency of these isolates, based on the observed significant differences in the amount of N2 fixed between the two isolates harbouring nod type c. However, before assuming such conclusions, we need to analyse a large sample of isolates with the same nodEF types.

In conclusion, the results of this study indicate that effective strains of R. leguminosarum bv. trifolii are abundant and genetically diverse in northern Scandinavia soils. Fewer seasonal effects on the genetic composition of the resident populations were observed. The ability of the cultivar Betty to form efficient symbiosis with nearly all isolates recovered at different time of the growing season makes Betty an agronomically valuable red clover cultivar for northern Scandinavia including subarctic regions. The evidence of plasmid transfers and a lower genetic diversity in the intensively cultivated Swedish site, demonstrates the impact of agricultural practices on the natural rhizobial populations in Scandinavian soils.


We are grateful to Petri Leinonen, Elomestari Oy, Kukkola, Finland, who kindly provided the reference strain HAMBI 461, and to Svalöf Weibull AB, Lännäs, Sweden, who provided the red clover seeds. We thank Coby Weber for her technical assistance. Financial support from the Research Council of Norway, the Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning and CL Behms Foundation for Legume Cultivation is gratefully acknowledged.