Research
Global biogeography of Alnus-associated Frankia actinobacteria Sergei P~ olme1,2, Mohammad Bahram1, Urmas K~ oljalg1,2 and Leho Tedersoo1,2 1
Department of Botany, Institute of Ecology and Earth Sciences, University of Tartu, 40 Lai St., 51005 Tartu, Estonia; 2Natural History Museum of Tartu University, 46 Vanemuise Street,
51014 Tartu, Estonia
Summary Author for correspondence: ~ lme Sergei Po Tel: +372 7376222 Email:
[email protected] Received: 16 May 2014 Accepted: 3 July 2014
New Phytologist (2014) 204: 979–988 doi: 10.1111/nph.12962
Key words: alder (Alnus), coevolution, DNA barcoding, Frankia alni complex, macroecology, nitrogenase reductase (nifH) gene, phylogeny effect, sequence similarity threshold.
Macroecological patterns of microbes have received relatively little attention until recently. This study aimed to disentangle the determinants of the global biogeographic community of Alnus-associated actinobacteria belonging to the Frankia alni complex. By determining a global sequence similarity threshold for the nitrogenase reductase (nifH) gene, we separated Frankia into operational taxonomic units (OTUs) and tested the relative effects of Alnus phylogeny, geographic relatedness, and climatic and edaphic variables on community composition at the global scale. Based on the optimal nifH gene sequence similarity threshold of 99.3%, we distinguished 43 Frankia OTUs from root systems of 22 Alnus species on four continents. Host phylogeny was the main determinant of Frankia OTU-based community composition, but there was no effect on the phylogenetic structure of Frankia. Biogeographic analyses revealed the strongest cross-continental links over the Beringian land bridge. Despite the facultative symbiotic nature of Frankia, phylogenetic relations among Alnus species play a prominent role in structuring root-associated Frankia communities and their biogeographic patterns. Our results suggest that Alnus species exert strong phylogenetically determined selection pressure on compatible Actinobacteria.
Introduction Nitrogen (N) is an essential nutrient for growth of all organisms. By fixation of atmospheric dinitrogen, prokaryotes compensate for N losses during biochemical processes in soil. In many terrestrial ecosystems, N is the main growth-limiting nutrient for plants and therefore association with N-fixing mutualists provides plants with a competitive advantage in early successional environments and in extremely N-poor soils (Stacey et al., 1992). To overcome N limitation, numerous plant families have developed mutualistic relationships with various prokaryote groups, mostly Actinobacteria and Cyanobacteria (Postgate, 1998). The Frankiaceae and Rhizobiaceae families are the most common mutualistic prokaryote root symbionts in angiosperms. While Rhizobiaceae associate with Fabaceae, Frankiaceae inhabit root nodules of eight unrelated angiosperm families (Normand et al., 1996). The mutualistic Frankia species fall into three genetic lineages that exhibit distinct genome characteristics, physiological properties and capabilities to infect specific plant families (Normand et al., 2007). Frankia are considered to be ubiquitous free-living soil organisms and they are distributed outside the geographic range of their compatible hosts (Benson & Dawson, 2007; Chaia et al., 2010). Alnus is the most widely distributed actinorrhizal plant genus that associates with the Frankia alni species complex. Frankia alni isolates form a monophyletic group, but they exhibit substantial genetic differences (Normand et al., 2007). In addition to Alnus, other host families Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust
within Fagales such as Casuarinaceae and Myricaceae form associations with this distinct clade of Actinobacteria that are regarded as ‘alder strains’ (Normand et al., 1996, 2007; Benson et al., 2004). The association of plant species with Frankia is a function of edaphic parameters and genotypes of both organisms. Depending on soil conditions, host species/genotype and successional stage, Alnus spp. always associate with a narrow range of isolates within the potentially suitable pool of soil Frankia strains (Anderson et al., 2009, 2013; Welsh et al., 2009a; Kennedy et al., 2010b; Lipus & Kennedy, 2011; Higgins & Kennedy, 2012). Besides Actinobacteria, Alnus species associate with ectomycorrhizal (EcM) and arbuscular mycorrhizal fungi, and thus this multipartite symbiosis is a good model in which to study host– symbiont interactions on a wide geographic scale (Tedersoo et al., 2009; P~olme et al., 2013). These root symbionts may improve the establishment and growth of Alnus spp. in pioneer habitats (Chatarpaul et al., 1989; Yamanaka et al., 2003). DNA sequencing-based molecular methods have tremendously increased knowledge of the ecology of microbes including Frankia. Sequences of specific marker genes are typically grouped into operational taxonomic units (OTUs) by using arbitrarily selected similarity thresholds. In eukaryotes, such similarity thresholds are usually determined based on DNA barcoding, in which sequences from multiple well-identified specimens from closely related species are compared (Hollingsworth et al., 2009; Schoch et al., 2012). Ideally, interspecific variation distinctly exceeds intraspecific variation, generating a ‘barcoding gap’, but New Phytologist (2014) 204: 979–988 979 www.newphytologist.com
New Phytologist
980 Research
in many situations such a gap is an artefact of limited taxonomic and geographic sampling (Bergsten et al., 2012). Because many prokaryotes are unculturable, DNA barcoding in the strict sense cannot be applied. The optimal similarity thresholds can, however, be determined statistically using the pairwise similarity of obtained sequences as a reference (e.g. Puillandre et al., 2012; Links et al., 2012; Tedersoo et al., 2014). In most prokaryote studies, a sequence similarity threshold of 97% for the highly conserved rDNA small subunit (SSU) gene is used for discrimination of OTUs, but this is clearly too conservative and lumps many species together (Vetrovsky & Baldrian, 2013). In Frankia, the nitrogenase reductase (nifH) gene provides greater resolution among OTUs and it has been widely used for detection and identification of Frankia in soil (R€osch et al., 2002; B€ urgmann et al., 2004; Izquierdo & N€ usslein, 2006; Mirza et al., 2009b). For Frankia, 93–99.8% nifH gene similarity has been inconsistently used to delimit OTUs in various studies (Mirza et al., 2009a,b; Welsh et al., 2009a,b; Kennedy et al., 2010a,b; ; Lipus & Kennedy, 2011). Mirza et al. (2009a) proposed that a 97% similarity cut-off suits best for assigning Frankia strains to their proper genomic groups, but simultaneously Welsh et al. (2009a) argued that the 97% threshold is too conservative and underestimates Frankia diversity. Biologically meaningful sequence similarity thresholds that best approximate species or ecotype level potentially improve the recovery of ecological and biogeographic patterns in any organism. Compared with eukaryotic macroorganisms, the global-scale biogeographic patterns of prokaryotes have remained poorly understood. Similarly to animals and plants, biogeographic phenomena such as the taxon richness–area relationship, dispersal limitation and biogeographic provincialism are evident in prokaryotes (Horner-Devine et al., 2004; Martiny et al., 2006). However, several microbial taxa exhibit exceptions to the biogeographic concepts established for macroorganisms (Martiny et al., 2006). Comprehensive large-scale studies demonstrate the prominent role of edaphic variables – particularly pH and the carbon (C) : N ratio – in structuring the richness and diversity of soil Bacteria and Archaea, respectively (Lauber et al., 2009; Bates et al., 2010). Conversely, environmental energy (a function of
climatic variables and resources) constitutes the strongest predictor of plant and animal diversity (Hawkins et al., 2003; Griffiths et al., 2011). Clearly, intimate biotrophic associations between macro- and microorganisms affect the distributions of both partners (P~olme et al., 2013). In this study, we determined statistically the best suitable sequence similarity thresholds for the nifH gene in the F. alni complex on a global scale. This study aimed to disentangle the relative effects of ecological and environmental parameters on the global distribution patterns of Alnus-associated actinorrhizal bacterial communities. We hypothesized that host phylogenetic identity has a prominent role in structuring the Frankia community on a global scale, as seen in EcM fungi (P~olme et al., 2013). In light of the facultative nature of symbiosis in the F. alni complex, we predicted that there would be no coevolution between Alnus and Frankia. We also hypothesized that the distribution of Frankia communities occurs independently from that of EcM fungi, because there is no direct interaction between these microbial symbionts.
Materials and Methods Sampling design The study was performed in 90 alder stands in Europe, North and Central America, and East and West Asia and covered subalpine, subarctic, boreal, temperate and subtropical ecosystems. Most of these sites were studied simultaneously for EcM fungal diversity in P~olme et al. (2013) (Fig. 1; for detailed information see Supporting Information Table S1). We sampled 22 alder species, out of 29–44 valid species, representing all three subgenera – Alnobetula, Clethropsis and Alnus (Chen & Li, 2004; Catalogue of Life, www.catalogueoflife.org (accessed 15.04.2013)). Alnus glutinosa (L.) Gaertn served as a host tree at 15 sites, Alnus viridis (Chaix) DC. at 10 sites, Alnus hirsuta (Spach) Rupr. at nine sites, Alnus subcordata C. A. Mey. at five sites, Alnus incana (L.) Moench at six sites, Alnus nepalensis (D. Don) at six sites, Alnus rubra (Bong.), Alnus serrulata (Aiton) Willd, Alnus japonica (Thunb.) Steud. and Alnus mandshurica (Callier ex C. K.
Fig. 1 Map of study sites indicating the number of host species/number of sites studied in the region. The following host species were sampled: Alnus glutinosa, Alnus viridis, Alnus hirsuta, Alnus subcordata, Alnus incana, Alnus nepalensis, Alnus rubra, Alnus serrulata, Alnus japonica, Alnus mandshurica, Alnus maritima, Alnus maximowiczii, Alnus acuminata, Alnus fauriei, Alnus firma, Alnus formosana, Alnus orientalis, Alnus sieboldiana, Alnus pendula, Alnus matusmurae, Alnus rhombifolia, and Alnus trabeculosa. New Phytologist (2014) 204: 979–988 www.newphytologist.com
Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust
New Phytologist
Research 981
Schneider) at four sites, Alnus maritima (Marshall) Muhl. ex Nutt. and Alnus maximowiczii (Callier) at three sites, Alnus acuminata (Kunth), Alnus fauriei (H. Lev. & Vaniot), Alnus firma (Siebold & Zucc.), Alnus formosana (Burkill) Makino, Alnus orientalis Deckne, Alnus sieboldiana (Matsum.) and Alnus pendula (Matsum.) at two sites, and Alnus matusmurae (Callier), Alnus rhombifolia (Nutt.) and Alnus trabeculosa (Hand.-Mazz.) at a single site (Table 1). At each study site, sampling was performed in a c. 2500-m2 area. Root nodules were sampled from six host trees, situated at least 10 m apart. Two healthy-looking nodule lobes of actinorrhiza were separated from each host tree and kept in CTAB buffer (1% cetyltrimethylammonium bromide, 100 mM Tris–HCl (pH 8.0), 1.4 M NaCl, and 20 mM ethylenediaminetetraacetic acid) for further molecular analyses. Molecular analyses Nodules were carefully peeled and washed in tap water. Only a small fraction of a single nodule per host tree was subjected to DNA extraction using Qiagen DNeasy 96 Plant (Qiagen GmbH, Hilden, Germany) or Thermo Scientific Phire Plant Direct PCR Kit (Thermo Scentific, Waltham, MA, USA) according to the manufacturer’s instructions. In the course of the study, PCR was performed by use of three alternative products: puReTaq Ready-To-Go PCR Beads (GE Healthcare, Little Chalfont, UK), 59 HOT FIREPol Blend Master Mix Ready to Load (Solis BioDyne, Tartu, Estonia) or Fermentas PCR mixture (Fermentas, Vilnius, Lithuania). Use of different chemistry did not affect the results, because we did not use a cloning step
Table 1 List of host species by sampling sites and number of associated operational taxonomic units (OTUs)
Host
No. of sampling sites
No. of associated Frankia OTUs
Alnus acuminata Alnus fauriei Alnus firma Alnus formosana Alnus glutinosa Alnus hirsuta Alnus incana Alnus japonica Alnus mandshurica Alnus maritima Alnus matsumurae Alnus maximowiczii Alnus nepalensis Alnus orientalis Alnus pendula Alnus rhombifolia Alnus rubra Alnus serrulata Alnus sieboldiana Alnus subcordata Alnus trabeculosa Alnus viridis
2 2 1 2 15 10 6 4 4 3 1 3 6 2 2 1 5 4 2 5 1 9
1 4 2 4 7 14 2 6 1 6 2 2 7 4 3 2 11 4 3 5 4 5
Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust
and we re-extracted and re-amplified DNA from samples that failed. To identify actinorrhizal bacteria from nodules, a 606-bp fragment of the nifH gene was amplified with the Frankia-specific primer pair nifHf1 (50 -ggcaagtccaccacccagc-30 ) and nifHr (50 -ctcgatgaccgtcatccggc-30 ) (Welsh et al., 2009a,b). DNA extracts that yielded no PCR products were re-amplified with nifHf1 in combination with one of the newly designed internal reverse primers nifHr260 (50 -gatggaggtgatgacgccac-30 ) or nifHr361 (50 -ggcggatcggcatcgcgaa-30 ) that, respectively, yielded 260- or 361-bp fragments from the least conservative 50 end. For amplification of the nifH gene, thermal cycling conditions were set as follows: 96°C for 5 min; 35 cycles at 96°C for 30 s, 64°C for 30 s, and 72°C for 45 s; and a final 7-min 72°C extension. PCR products were separated by electrophoresis through an 1.5% agarose gel in 0.59 TBE buffer (45 mM Tris base, 45 mM boric acid, and 1 mM EDTA (pH 8.0)), visualized under UV light and purified using Exo-Sap enzymes (Sigma, St Louis, MO, USA). Sequencing was performed with the primer nifHf1. Sequences were assembled, checked, trimmed and manually corrected using the SEQUENCHER 4.10.1 software (GeneCodes Corp., Ann Arbor, MI, USA). Sequences were confirmed to belong to Actinobacteria by use of blastn searches against the International Nucleotide Sequence Databases Consortium (INSDC). Sequences were aligned by use of MAFFT 6 (Katoh & Toh, 2008). The alignments were checked and corrected manually in SEAVIEW (Gouy et al., 2010). Sequences of different length were included in phylogenetic analyses without trimming to the same length. Maximum likelihood (ML) and fast bootstrap analyses were performed by use of default settings in RAXML (Stamatakis et al., 2008). In order to delimit OTUs of Alnus-associated Actinobacteria, we downloaded all uncloned sequences belonging to Frankia alni from INSDC. Sequences < 520 bp in length and not associated with Alnus were excluded. With these criteria, we retrieved 215 sequences from INSDC in addition to 184 original sequences of sufficient length. Sequences were trimmed to 520 bp and their pairwise raw distances (i.e. Levenshtein metric) were calculated using the dist.dna command in the Ape package of R (Paradis et al., 2004). To confirm the suitability of a similar threshold for shorter sequences, we included all our 406 sequences along with these INSDC sequences. All sequences were aligned and trimmed to 210 bp. The sequence similarity threshold of 99.3% derived from analysis of full-length sequences was used to assign all sequences into OTUs as implemented in BLASTclust (http:// toolkit.tuebingen.mpg.de/blastclust). The nucleotide sequence data reported in this paper have been submitted to the INSDC database and assigned accession numbers JN088542–JN088686 and KM017147–KM017399. Environmental data Estimates of the mean annual precipitation and temperature were retrieved from the database of Earth surface climate (Hijmans et al., 2005) using ARCGIS 9.3 (ESRI, Redlands, CA, USA). This database represents a global model of the average monthly surface New Phytologist (2014) 204: 979–988 www.newphytologist.com
New Phytologist
982 Research
climate characteristics with raster size of 30 s latitude and longitude (c. 0.81 km2 on the equator). From each sampling site, c. 50 g of rhizosphere soil was pooled from the six samples for analysis of soil parameters. Concentrations of soil N, phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg) and organic matter and pH were measured as described in Tedersoo et al. (2009). All soil nutrient concentrations were log-transformed before analyses. To address the effect of host species on Frankia community composition, we obtained phylogenetic eigenvectors from an ITS phylogram of Alnus species (P~olme et al., 2013). The effect of geographic distance was taken into account by use of principal coordinates of neighbour matrices (PCNM) eigenvectors that account for spatial autocorrelation at different scales (Borcard & Legendre, 2002). Significant PCNM vectors (a = 0.05) were forward-selected in the Packfor package of R (Dray et al., 2007) and used in multivariate analyses. Host phylogeny To address the effect of phylogenetic relationships among Alnus host species on Frankia community composition, we created a phylogeny of Alnus species based on the ITS region, using Betula pendula as an outgroup. ITS sequences of each species were downloaded from the INSDC. To construct a phylogram of Alnus spp., ML and fast bootstrap analyses with 1000 replications were performed using the GTR + CAT evolutionary model in RAXML. To account for the node ages in the host phylogeny, we used the chronopl function (k = 0) in the Ape package of R (Paradis et al., 2004). This function uses a trade-off between a parametric formulation where each branch has its own rate and a nonparametric term where changes in rates are minimized between contiguous branches (Sanderson, 2002). We used MESQUITE (Maddison & Maddison, 2008) to generate a patristic distance matrix from the derived host phylogeny. Phylogenetic eigenvectors of the PCNM were derived from the patristic distance matrix, forward-selected (a = 0.05) in the Packfor package of R (Dray et al., 2007), and used in further statistical analyses. Community structure To estimate sampling completedness of global Frankia diversity, we calculated rarefied OTU interpolation (Mau Tau algorithm sensu Colwell et al., 2004) and the minimum richness estimators Chao1 and Chao2 (Chao, 1984) at different sampling depth. These statistics and their 95% confidence intervals were calculated in ESTIMATES 8.2 (Colwell, 2009). Based on the six sampled trees per site, we determined the frequency of Frankia OTUs. To address the relative importance of the effects of climatic, edaphic and biological factors on the community structure of Frankia as based on the frequency of OTUs, we applied a multivariate analysis of variance as implemented in the adonis function of the Vegan package of R (Oksanen et al., 2012). A Bray–Curtis dissimilarity metric was used to calculate distance matrices. Adonis partitions distance matrices among sources of variation and fits linear models to distance matrices by New Phytologist (2014) 204: 979–988 www.newphytologist.com
use of permutations. To complement adonis, we used the varpart function in the Vegan package to partition the variation of community dissimilarity by grouping host phylogenetic, edaphic, climatic (including altitude) and spatial variables. Variation partitioning is based on redundancy analysis (RDA) which uses Euclidean distance. Singletons were excluded from communitylevel analyses to reduce the effect of rare species. As edaphic data from the two Costa Rica sites were unavailable, these sites were omitted from community-level analyses and included only in the analyses of coevolutinary interactions. To investigate the concordance of Alnus-associated Frankia and EcM fungal communities (derived from P~olme et al., 2013), we used the protest function in the Vegan package. This algorithm permutes the configuration of one matrix to the maximum similarity with the other matrix (minimizing the sum of squared difference) and estimates the significance of the correlation derived from symmetric sum of squares (i.e. Procrustes statistic) (Oksanen et al., 2012). We conducted a partial Mantel test to assess the relative importance of environmental variables (a matrix combining the soil and climatic variables for each site) in structuring both microbial communities. The Mantel test recovers the correlation between two distance matrices, whereas the partial Mantel test determines the Mantel statistic that accounts for the third distance matrix (Oksanen et al., 2012). To illustrate the comparison of the regional distributions of bacterial and fungal OTUs, we constructed a two-way cladogram based on the Euclidean distance metric and Ward clustering method as implemented in PC-ORD 5 (McCune & Mefford, 2006). Coevolution To test the evolutionary independence between Frankia and Alnus phylogenies, we used the ParaFit function (Legendre et al., 2002) of the Ape package of R (Paradis et al., 2004). ParaFit tests whether the evolution of two groups of interacting organisms is independent based on phylogenetic trees and the set of association links between these organisms. The function converts phylogenetic distance matrices of both interacting organisms into matrices of principal coordinates and tests the congruence between the two phylogenies by accounting for the association links. We converted Alnus and Frankia phylogenies into ultrametric trees (cophenetic function) and derived patristic distances by use of the chronopl function (k = 0) and 999 permutations. We calculated unifrac distances from the ultrametric phylogeny of Frankia accounting for host associations using the unifrac function. To complement ParFit, the correlations of phylogenetic distances of the two symbionts (i.e. Frankia unifrac versus Alnus patristic distances derived from ultrametric phylogeny) were compared using a Mantel test. In order to assess phylogenetic distance decay among communities of Alnus-associated Frankia, we conducted a series of Mantel tests between the Bray–Curtis dissimilarity matrices of symbiont communities and (1) a dissimilarity matrix based on the distances of all Alnus ITS sequences included in the study; (2) a dissimilarity matrix based on the distances of all ITS sequences belonging to the Alnus subgenus included in the study; (3) a dissimilarity matrix based on the Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust
New Phytologist distances of all ITS sequences of Alnus sister species included in the study. To calculate the distance between aligned Alnus ITS sequences, we used the dist.dna function implemented in the Ape package (Paradis et al., 2004) with the K80 evolutionary model. Pairs of sister species were selected according to the phylogenetic reconstruction of Chen & Li (2004). The following pairs of sister species were tested: A. japonica and A. serrulata; A. orientalis and A. subcordata; A. maritima and A. formosana; A. firma and A. sieboldiana. The subgenus Alnus was chosen to represent the subgeneric level, because it was best represented in our data set. Biogeography To assess biogeographic patterns of Frankia, we separated the sites into eight regions: Northwest America, Northeast America, South Asia, North Asia, Iran-Turkey, South Europe, North Europe and Japan (P~olme et al., 2013). We compared similarities among biogeographic regions based on Frankia OTUs. In order to take into account the ratio of shared species to unique species between pairs of regions, we used the Jaccard distance metric and Ward clustering. Statistical support in area cladograms was tested using the Pvclust package of R (Suzuki & Shimodaira, 2006), which calculates P-values based on multiscale bootstrap resampling (1000 replications).
Results From 540 nodule samples, we obtained 184 long (520–612 bp) sequences with the original primers and 222 short (229–389 bp) sequences with the newly generated primers (Table S1). Mixed chromatograms were not observed, indicating that a single genotype was present in the nodules. Low amplification success with long amplicons suggests that many nodules may have been moribund in spite of their healthy visual appearance. Comparison of nifH similarities revealed the best similarity threshold to lie between 99.24% and 99.43% in the analysis of long sequences (Fig. 2a). This threshold corresponds to ≥ 4 nucleotide differences (over 520 bp) among OTUs. In short fragments, the optimal similarity interval was between 98.61% and 99.07% (Fig. 2b), which corresponds to ≥ 3 mismatches (over 229 bp) among OTUs. Because shorter sequences clustered relatively well with longer sequences, we selected the sequence similarity threshold of 99.3% for OTU construction. In only five cases, short sequences clustered together and formed distinct OTUs (#5, #13, #14, #15 and 16#). This criterion revealed 43 OTUs of Frankia (including 13 singletons) from root nodules of Alnus at the global scale. Usually these OTUs formed distinct well-supported clusters in the maximum likelihood phylogram (Supporting Information Fig. S1). The most frequently encountered OTU (#1) showed low specificity, associating with nine host species in all three continents. However, some relatively frequent OTUs were associated with a single host species or were found from a single biogeographic region. For example, OTU #43 was common exclusively on A. acuminata in Costa Rica and OTU #3 associated solely with A. glutinosa throughout Europe (Fisher’s exact test for host in Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust
Research 983
Europe: n = 27; df‘ = 2; v2 = 10.8; P < 0.005). Several rare OTUs observed in a few sites were usually restricted to a particular host, but this may stem from our limited sampling. Rarefied taxon accumulation and extrapolation curves tended to level off in the global sampling, suggesting the presence of at least 70 Alnus-associated Frankia OTUs (Fig. S2). According to the permutational multivariate analysis of variance, the addressed variables explained 58.0% of Frankia community structure (Table 2). Host phylogenetic eigenvectors had the strongest impact, explaining 37.5% of variation (Adonis: F9,68 = 7.91; P = 0.001). PCNM vectors representing spatial distances explained 4.5% of community variation (F8,68 = 1.94; P = 0.001), whereas climatic variables such as precipitation (F1,68 = 4.95; P = 0.001) and temperature (F1,68 = 2.55; P = 0.041) had a marginal effect, accounting for 1.9% and 0.3%, respectively. Edaphic parameters had no significant effect on Frankia community composition at the global scale. The variation partitioning analysis confirmed that host phylogenetic eigenvectors had the strongest effect on Frankia community (16.3%), followed by spatial PCNM vectors (4.8%), climatic variables (2.4%) and soil parameters (1.2%). Notably, the shared effect of host phylogeny and climatic variables explained an additional 4.0% of Frankia community composition (Fig. 3). According to the ParaFit method, our null hypothesis of independent evolution of Alnus and Frankia was rejected (SS = 2261.818; P = 0.001), corroborating the substantial effect of host phylogeny on the Frankia community. However, comparison of unifrac and patristic distances of Frankia and Alnus, respectively, did not support coevolution in the strict sense (Mantel r = 0.042; P = 0.596). Based on significantly positive Mantel r values, Frankia communities of alder sister species were more similar to each other (Mantel r = 0.307; P = 0.001) than Frankia communities within species of the subgenus Alnus (Mantel r = 0.248; P = 0.001) or the whole host genus (Mantel r = 0.116; P = 0.026; Fig. 4). We sought to determine whether the communities of EcM fungi and Frankia turn over in concordance. The protest function revealed significant correlation between communities of fungal and bacterial mutualists of Alnus (RProcrustes = 0.500; P = 0.001). The results of a partial Mantel test suggested that bacterial and fungal communities lack significant correlation when accounting for environmental variables (Mantel r = 0.057; P = 0.101), host phylogeny (Mantel r = 0.075; P = 0.056) or spatial PCNM eigenvectors (r = 0.079; P = 0.033). The regional clustering of both microbial groups is illustrated in Fig. S3. Cluster analysis of the biogeographic region-based distance matrix of Frankia communities revealed links between North Asia, Middle East and Europe (P > 0.001) (Fig. 5). The clustering of South Asia together with Japan and North America was also strongly supported (P = 0.001). Among smaller subdivisions, North Europe and South Europe shared significantly similar Frankia communities (P = 0.017). However, the remaining biogeographic links were poorly supported as a result of the host phylogeny effect. New Phytologist (2014) 204: 979–988 www.newphytologist.com
New Phytologist
984 Research 6000
(a)
Number of observations
5000
4000
3000
2000
9.06%
9.44%
8.31%
8.69%
7.93%
7.18%
7.55%
6.80%
6.42%
6.04%
5.29%
5.66%
4.53%
4.91%
3.78%
4.15%
3.02%
3.40%
2.64%
2.27%
1.89%
1.13%
1.51%
0.76%
0.00% 35 000
0.38%
1000
(b)
Number of observations
30 000
25 000
20 000
15 000
10 000
0
Discussion OTU separation Comparison of genetic differences in the nifH region revealed an interval of relatively low frequency in pairwise sequence comparisons at 99.2–99.4% sequence similarity in the F. alni complex at the global scale (Fig. 2). Our results therefore suggest that the nifH gene can be best used to discriminate among OTUs in the F. alni complex at a higher sequence similarity threshold than used in previous studies. Welsh et al. (2009a) found a tremendous decline in Frankia diversity when sequence similarity thresholds were reduced from 99.8% (a single mismatch) to New Phytologist (2014) 204: 979–988 www.newphytologist.com
13.88%
12.95%
12.03%
11.10%
9.25%
10.18%
Dissimilarity
8.33%
7.40%
6.48%
5.55%
4.63%
3.70%
2.78%
1.85%
0.93%
0.00%
5000
Fig. 2 Histogram of pairwise comparisons of Frankia alni nitrogenase reductase (nifH) sequences based on dissimilarity (a) among 520-bp long sequences and (b) among 210bp sequences.
99.0% to 97.0% to 95.0% in Arizona, but their recommendation of the 97% threshold was not backed by statistical tests or ecological niche differentiation. Our global data are in agreement with this finding and it is therefore important to determine biologically relevant criteria for OTU delimitation. With the proposed cut-off level, we detected 43 OTUs in 90 sampling sites from three continents associated with 22 Alnus species in the Northern Hemisphere. Many of the Frankia OTUs formed distinct and well-supported, sometimes host-specific clusters in the maximum likelihood phylogram, which indicates that our OTU delimitation standard is biologically relevant (Fig. S1). Compared with the SSU that is usually analyzed with general prokaryote primers, the nifH gene allows sequencing of Frankia without a costly Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust
New Phytologist
Research 985
Table 2 Relative importance of the effects of biological, spatial and climatic variables on the actinorrhizal community associated with 22 Alnus species at the global scale based on multivariate permutational ANOVA Degrees of freedom
Adjusted R2
F-value
P-value –
Host phylogeny Spatial vectors Precipitation Temperature Residuals Total
9 8 1 1 68 87
7.91 1.94 4.95 2.25
0.375 0.045 0.019 0.003 0.420 1.000
0.001 0.001 0.001 0.041
Spatial vectors 0.02 0.01
Host phylogeny 0.16
0.01
0.05 0.01
0.01
0.04
0.01 0.01
Soil 0.01
Climate 0.02
Fig. 3 Venn diagram demonstrating the independent and shared effects of host phylogeny, spatial, climatic and edaphic variables on the Alnusassociated Frankia community. Numbers indicate the proportion of explained variation.
Fig. 4 Phylogenetic distance decay among communities of Frankia with increasing host phylogenetic distance, computed on the basis of internal transcribed spacer (ITS) sequence distances among sister species, and among species at the subgenus and genus level.
cloning step and avoiding time-consuming next-generation sequencing analyses. The SSU is also regarded as being too conservative to address genetic variation within species complexes of Frankia (Ghodhbane-Gtari et al., 2010). However, the recommended sequence similarity threshold for the nifH gene in Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust
Fig. 5 Hierarchical clustering of biogeographic regions based on similarities between corresponding Alnus-associated Frankia communities. Numbers above branches indicate support from P-values, which are computed using multiscale bootstrap resampling.
Frankia alni (99.3%) is probably too high for error-prone nextgeneration sequencing, in which a few random errors may blur differences among OTUs. Community structure We employed multivariate dispersion and variation partitioning analyses to understand how host phylogeny, geographic distance, soil properties and climatic variables affect Frankia communities on roots of Alnus. Host phylogeny had the strongest effect on root-associated assemblages of Actinobacteria, explaining about one-third of community variation. These results indicate that much of the selectivity or compatibility of these associations is phylogenetically determined by the host and does not depend on local edaphic conditions. Earlier studies have demonstrated that host identity has an important role in structuring F. alni assemblages between two co-occurring Alnus species (Lipus & Kennedy, 2011; Anderson et al., 2013). Our study demonstrates that Frankia OTUs differ in their geographic range and breadth of association with host species. The A. glutinosa-specific OTU #3 was distributed from the Mediterranean to the Baltic countries. This finding indicates that, contrary to the general positive host phylogenetic similarity effect, sympatric host species (A. glutinosa and A. incana) may strongly differ in the association of Frankia OTUs independently of edaphic and climatic conditions. A substantial proportion of community variation was also shared between host phylogeny and climate, which indicates that much of the climatic effect is structured by Alnus phylogeny. Of the climatic variables, precipitation and temperature had a significant but weak impact on the Frankia community, altogether accounting for 2.2% of the variation. Although the regional-scale distribution patterns of Alnus species are often driven by climate, New Phytologist (2014) 204: 979–988 www.newphytologist.com
New Phytologist
986 Research
topography and edaphic properties, there is no information about the evolution of this genus in relation to these variables. Considering the niche differentiation of host species, negligible effects of climatic and edaphic variables on their actinobacterial and EcM symbionts (P~ olme et al., 2013) are surprising. In free-living soil prokaryotes, edaphic parameters such as pH largely determine the community composition, but climate has a negligible role over the large geographic scales (Lauber et al., 2009). The prominent roles of host phylogeny and spatial factors on Frankia community structure corroborate the results of a global study addressing Alnus-associated EcM fungal communities (P~olme et al., 2013). The Procrustes statistic revealed significant coupling of the two microbial communities, which probably results from the mutually shared host phylogeny effect as revealed by the partial Mantel test. Alternatively, the direct effects of symbionts on each other may be produced through their complementary effects on plant nutrition (Chatarpaul et al., 1989; Horton et al., 2013; Walker et al., 2013). As a result of N fixation, Alnusdominated ecosystems are phosphorus-limited (Uliassi & Ruess, 2002). Therefore, the efficiency of mineral nutrition may play a role in indirect or direct selection of symbiotic EcM fungi and Frankia by the host tree (Walker et al., 2013). Coevolution Comparisons among phylogenies of Frankia and Alnus revealed significant concordance that suggests a certain degree of correlated evolution among these organisms. However, this phenomenon could be largely ascribed to the strong host phylogeny effect on Frankia communities rather than coevolution in the strict sense. Mantel tests using unifrac phylogenetic distance revealed no tendency of closely related Alnus spp. to associate with phylogenetically more similar Frankia OTUs than expected by chance. Taken together, these results indicate that Alnus spp. are phylogenetically determined to associate with specific Frankia but not vice versa. These patterns may arise from the nature of the symbiosis, which is facultative in Frankia but obligate in Alnus. Therefore, Alnus species may have a strong selection pressure for Frankia OTUs to maintain the most beneficial strains (Mirza et al., 2009a; Welsh et al., 2009a,b; Kennedy et al., 2010b). This phenomenon, termed sanctioning, has been observed in plant–rhizobial relationships (Kiers et al., 2007), but it has not yet been addressed in Frankia. Selection by the host may have further resulted in the evolution of physiological compatibility between the symbionts. Given the origin of actinorrhizal symbiosis in Alnus at least 55 million yr ago (Benson et al., 2004), it is possible that certain beneficial and compatible Frankia taxa have a long symbiotic history with certain groups of Alnus across the radiation events of the host.
Frankia. In free-living prokaryotes, distance decay may be the strongest predictor of microbial community structure (Martiny et al., 2011). The Frankia community-based area cladogram was generally poorly resolved with several links receiving weak statistical support. This is ascribed to the overwhelming effect of host phylogeny and extinctions of certain Alnus clades in specific regions that have probably blurred regional biogeographic patterns of the F. alni complex. Nonetheless, based on Frankia communities, South Asia and Japan clustered together with Northeast and Northwest America, which supports the hypothesis of strong past connections over the Beringian land bridge. This connection is consistent with biogeographic relationships among Alnus-associated EcM fungal communities (P~olme et al., 2013) and many groups of terrestrial plants in general (Milne, 2006). Similarly to EcM fungi, North and South Europe clustered together with high support, possibly as a result of the common recent post-glacial history. In contrast to EcM fungal communities, South Asia and North Asia had substantially different Frankia assemblages and there was no evidence for past connections over the North Atlantic land bridge. In Frankia, North Asia was related to Iran-Turkey and these regions were both significantly clustered with Europe; in EcM fungi, Iran-Turkey differed from other regions by hosting a highly deviant community (P~olme et al., 2013). Conclusions The nitrogenase reductase gene provides the best delimitation of Frankia OTUs at 99.2–99.4% sequence similarity, which is ecologically relevant and is therefore recommended for future studies in the F. alni complex. Alnus-associated Frankia communities are characterized by a low diversity of host and habitat generalist OTUs with a few exceptional taxa that are strongly host specific. Phylogenetic relationships among Alnus spp. determine a large proportion of the associated Frankia community structure. The correlated evolution between Alnus and Actinobacteria is attributable to the host phylogeny effect on Frankia community composition. These evolutionary patterns probably result from phylogenetically influenced selection pressure from Alnus species rather than specific climatic and edaphic conditions which accounted for a negligible proportion of variation in Frankia communities. This indicates that Frankia OTUs are able to tolerate a wide variety of abiotic environmental conditions and their distribution is generally not dispersal limited. The cross-continental biogeographic relationships among Frankia OTUs provide evidence for historical connections over the Beringian land bridge but not over the North Atlantic, which is consistent with the large-scale biogeographic history of Alnus spp. and their EcM fungi.
Acknowledgements Biogeography Geographic structure of sites explained c. 5% of Frankia community composition, which suggests that dispersal limitation plays a relatively minor role in the distribution of Alnus-associated New Phytologist (2014) 204: 979–988 www.newphytologist.com
Financial support was provided by Estonian Science Foundation grants 7434, 9286, and PUT0171, Frontiers in Biodiversity Research (FIBIR), Doctoral Studies and Internationalisation Programme ‘DoRa’. We are grateful to numerous colleagues and Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust
New Phytologist kind people all over the world whose contributions made this study possible: Takashi Yamanaka, Hisayasu Kobayashi, Naoko Miyamoto, Keizo Hirai, Kiyoshi Ishida, Daisuke Sakaue, Maki Saito, Chih-Jen Tien, Ming-Hsiu Kao, Yu Cheng Dai, Wang Wei, Dr Li, Amin Fatahi, Kazuhide Nara, Tine Grebenc, Hojka Graigher, Mika Toivonen, Pi-Han Wang, Yosuke Matsuda, Corey Chantry, Catherine Morris, John Bishop, James Schrader, Peter G. Kennedy, Margit N~oukas, Katren Mikkel, Peeter Laas, Edgar Sepp and Ants Kaasik. Special thanks to The Office of Forest and Rangelands Organization and the Kheirudkenar Educational Forest of the University of Tehran in Iran. We also thank four referees for their useful comments.
References Anderson M, Ruess R, Myrold D, Taylor D. 2009. Host species and habitat affect nodulation by specific Frankia genotypes in two species of Alnus in interior Alaska. Oecologia 160: 619–630. Anderson M, Taylor D, Ruess R. 2013. Phylogeny and assemblage composition of Frankia in Alnus tenuifolia nodules across a primary successional sere in interior Alaska. Molecular Ecology 14: 3864–3867. Bates ST, Berg-Lyons D, Caporaso JG, Walters WA, Knight R, Fierer N. 2010. Examining the global distribution of dominant archaeal populations in soil. The ISME Journal 5: 908–917. Benson DR, Dawson JO. 2007. Recent advances in the biogeography and genecology of symbiotic Frankia and its host plants. Physiologia Plantarum 130: 318–330. Benson DR, Vanden Heuvel BD, Potter D. 2004. Actinorhizal symbioses: diversity and biogeography. In: Gillings M, ed. Plant microbiology. Oxford, UK: BIOS Scientific Publishers Ltd, 99–130. Bergsten J, Bilton DT, Fujisawa T, Elliott M, Monaghan MT, Balke M, Hendrich L, Geijer J, Herrmann J, Foster GN et al. 2012. The effect of geographical scale of sampling on DNA barcoding. Systematic Biology 61: 851– 869. Borcard D, Legendre P. 2002. All-scale spatial analysis of ecological data by means of principal coordinates of neighbour matrices. Ecological Modeling 153: 51–68. B€ urgmann H, Widmer F, Von Sigler W, Zeyer J. 2004. New molecular screening tools for analysis of free-living diazotrophs in soil. Applied Environmental Microbiology 70: 240–247. Chaia EE, Wall LG, Huss-Danell K. 2010. Life in soil by the actinorhizal root nodule endophyte Frankia. A review. Symbiosis 51: 201–226. Chao A. 1984. Non-parametric estimation of the number of classes in a population. Scandinavian Journal of Statistics 11: 265–270. Chatarpaul L, Chakravarty P, Subramaniam P. 1989. Studies in tetrapartite symbioses. Plant and Soil 118: 145–150. Chen Z, Li J. 2004. Phylogenetics and biogeography of Alnus (Betulaceae) inferred from sequences of nuclear ribosomal DNA ITS region. International Journal of Plant Sciences 165: 325–335. Colwell RK. 2009. EstimateS: statistical estimation of species richness and shared species from samples. Version 8.2. [WWW document] URL http://viceroy.eeb. uconn.edu/estimates [accessed 20 December 2012]. Colwell RK, Mao CX, Chang J. 2004. Interpolating, extrapolating, and comparing incidence-based species accumulation curves. Ecology 85: 2717– 2727. Dray S, Legendre P, Blanchet G. 2007. Packfor: forward selection with permutation. R package version 0.0-7. [WWW document] URL http://r-forge. r-project.org/R/?group_id=195. [accessed 20 July 2012] Ghodhbane-Gtari F, Nouioui I, Boudabous A, Gtari M. 2010. 16S–23S rRNA intergenic spacer region variability in the genus Frankia. Microbial Ecology 60: 487–495. Gouy M, Guindon S, Gascuel O. 2010. SeaView version 4: a multiplatform graphical user interface for sequence alignment and phylogenetic tree building. Molecular Biology and Evolution 27: 221–224. Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust
Research 987 Griffiths RI, Thomson BC, James P, Bell T, Bailey M, Whiteley AS. 2011. The bacterial biogeography of British soils. Environmental Microbioogy 13: 1642– 1654. Hawkins BA, Field R, Cornell HV, Currie DJ, Guegan JF, Kaufman DM, Kerr JT, Mittelbach GG, Oberdorff T, O’Brien EM. 2003. Energy, water, and broad-scale geographic patterns of species richness. Ecology 84: 3105–3117. Higgins LM, Kennedy PG. 2012. Symbiotic Frankia bacteria in Alnus forests in Mexico and the United States of America: is geographic location a good predictor of assemblage structure? Botany 90: 423–431. Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A. 2005. Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25: 1965–1978. Hollingsworth PM, Forrest LL, Spouge JL, Hajibabaei M, Ratnasingham S, van der Bank M, Chase MW, Cowan RS, Erickson DL, Fazekas AJ. 2009. A DNA barcode for land plants. Proceedings of the National Academy of Sciences, USA 106: 12794–12797. Horner-Devine MC, Lage M, Hughes JB, Bohannan BJ. 2004. A taxa–area relationship for bacteria. Nature 432: 750–753. Horton TR, Hayward J, Tourtellot SG, Taylor DL. 2013. Uncommon ectomycorrhizal networks: richness and distribution of Alnus-associating ectomycorrhizal fungal communities. New Phytologist 198: 978–980. Izquierdo JA, N€ usslein K. 2006. Distribution of extensive nifH gene diversity across physical soil microenvironments. Microbial Ecology 51: 441–452. Katoh K, Toh H. 2008. Recent developments in the MAFFT multiple sequence alignment program. Briefings in Bioinformatics 9: 286–298. Kennedy PG, Schouboe JL, Rogers RH, Weber MG, Nadkarni NM. 2010b. Frankia and Alnus rubra canopy roots: an assessment of genetic diversity, propagule availability, and effects on soil nitrogen. Microbial Ecology 59: 214– 220. Kennedy PG, Weber MG, Bluhm AA. 2010a. Frankia bacteria in Alnus rubra forests: genetic diversity and determinants of assemblage structure. Plant and Soil 335: 479–492. Kiers ET, West SA, Denison RF. 2007. Maintaining cooperation in the legume– rhizobia symbiosis: identifying selection pressures and mechanisms. In: Dilworth MJ, James EK, Sprent JI, Newton WE, eds. Nitrogen-fixing leguminous symbioses. Dordrecht, the Netherlands: Springer, 59–76. Lauber CL, Hamady M, Knight R, Fierer N. 2009. Pyrosequencing-based assessment of soil pH as a predictor of soil bacterial community structure at the continental scale. Applied Environmental Microbiology 75: 5111. Legendre P, Desdevises Y, Bazin E. 2002. A statistical test for host-parasite coevolution. Systematic Biology 51: 217–234. Links MG, Dumonceaux TJ, Hemmingsen SM, Hill JE. 2012. The chaperonin-60 universal target is a barcode for bacteria that enables de novo assembly of metagenomic sequence data. PLoS ONE 7: e49755. Lipus A, Kennedy PG. 2011. Frankia assemblages associated with Alnus rubra and Alnus viridis are strongly influenced by host species identity. International Journal of Plant Science 172: 403–410. Maddison WP, Maddison DR. 2008. Mesquite: a modular system for evolutionary analysis. Evolution 62: 1103–1118. Martiny JB, Eisen JA, Penn K, Allison SD, Horner-Devine MC. 2011. Drivers of bacterial b-diversity depend on spatial scale. Proceedings of the National Academy of Sciences, USA 108: 7850–7854. Martiny JBH, Bohannan BJ, Brown JH, Colwell RK, Fuhrman JA, Green JL, Horner-Devine MC, Kane M, Krumins JA, Kuske CR. 2006. Microbial biogeography: putting microorganisms on the map. Nature Reviews Microbiology 4: 102–112. McCune B, Mefford MJ. 2006. PC-Ord for Windows v. 5.15. Multivariate analysis of ecological data. Gleneden Beach, OR, USA: MjM Software. Milne RI. 2006. Northern Hemisphere plant disjunctions: a window on tertiary land bridges and climate change? Annals of Botany 98: 465–472. Mirza BS, Welsh AK, Rasul G, Rieder JP, Paschke MW, Hahn D. 2009a. Variation in Frankia populations of the Elaeagnus host infection group in nodules of six host plant species after inoculation with soil. Microbial Ecology 58: 384–393. Mirza BS, Welsh A, Rieder JP, Paschke MW, Hahn D. 2009b. Diversity of frankiae in soils from five continents. Sysematic and Applied Microbiolgy 32: 558–570. New Phytologist (2014) 204: 979–988 www.newphytologist.com
New Phytologist
988 Research Normand P, Lapierre P, Tisa LS, Gogarten JP, Alloisio N, Bagnarol E, Bassi CA, Berry AM, Bickhart DM, Choisne N et al. 2007. Genome characteristics of facultatively symbiotic Frankia sp. strains reflect host range and host plant biogeography. Genome Research 17: 7–15. Normand P, Orso S, Cournoyer B, Jeannin P, Chapelon C, Dawson J, Evtushenko L, Misra AK. 1996. Molecular phylogeny of the genus Frankia and related genera and emendation of the family Frankiaceae. International Journal of Systematic Bacteriology 46: 1–9. Oksanen J, Kindt R, Legendre P, O’Hara B, Stevens MHH, Oksanen MJ. 2012. vegan: community ecology package. [WWW document] URL http://vegan. r-forge.r-project.org/ [accessed 20 October 2012]. Paradis E, Claude J, Strimmer K. 2004. APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20: 289–290. P~olme S, Bahram M, Yamanaka T, Nara K, Dai YC, Grebenc T, Kraigher H, Toivonen M, Wang P, Matsuda Y. 2013. Biogeography of ectomycorrhizal fungi associated with alders (Alnus spp.) in relation to biotic and abiotic variables at the global scale. New Phytologist 198: 1239–1249. Postgate J. 1998. Nitrogen fixation, 3rd edn. Cambridge, UK: Cambridge University Press. Puillandre N, Lambert A, Brouillet S, Achaz G. 2012. ABGD, Automatic Barcode Gap Discovery for primary species delimitation. Molecular Ecology 21: 1864–1877. R€osch C, Mergel A, Bothe H. 2002. Biodiversity of denitrifying and dinitrogen-fixing bacteria in an acid forest soil. Applied Environmental Microbiology 68: 3818–3829. Sanderson MJ. 2002. Estimating absolute rates of molecular evolution and divergence times: a penalized likelihood approach. Molecular Biology and Evolution 19: 101–109. Schoch CL, Seifert KA, Huhndorf S, Robert V, Spouge JL, Levesque CA, Chen W, Fungal Barcoding Consortium. 2012. Nuclear ribosomal internal transcribed spacer (ITS) region as a universal DNA barcode marker for Fungi. Proceedings of the National Academy of Sciences, USA 109: 6241–6246. Stacey G, Burris RH, Evans HJ. 1992. Biological nitrogen fixation. New York, NY, USA: Springer, Chapman and Hall. Stamatakis A, Hoover P, Rougemont J. 2008. A rapid bootstrap algorithm for the RAxML web servers. Systematic Biology 57: 758–771. Suzuki R, Shimodaira H. 2006. Pvclust: an R package for assessing the uncertainty in hierarchical clustering. Bioinformatics 22: 1540–1542. Tedersoo L, Bahram M, Ryberg M, Otsing E, K~oljalg U, Abarenkov K. 2014. Global biogeography of the ectomycorrhizal/sebacina lineage (Fungi, Sebacinales) as revealed from comparative phylogenetics analyses. Molecular Ecology 23: 4168–4183. Tedersoo L, Suvi T, Jairus T, Ostonen I, P~olme S. 2009. Revisiting ectomycorrhizal fungi of the genus Alnus: differential host specificity, diversity and determinants of the fungal community. New Phytologist 182: 727–735.
New Phytologist (2014) 204: 979–988 www.newphytologist.com
Uliassi DD, Ruess RW. 2002. Limitations to symbiotic nitrogen fixation in primary succession on the Tanana River floodplain. Ecology 83: 88–103. Vetrovsky T, Baldrian P. 2013. The variability of the 16S rRNA gene in bacterial genomes and its consequences for bacterial community analyses. PLoS ONE 8: e57923. Walker JK, Cohen H, Higgins LM, Kennedy PG. 2013. Testing the link between community structure and function for ectomycorrhizal fungi involved in a global tripartite symbiosis. New Phytologist 202: 287–296. Welsh A, Mirza BS, Rieder JP, Paschke MW, Hahn D. 2009b. Diversity of frankiae in root nodules of Morella pensylvanica grown in soils from five continents. Systematic Applied Microbiology 32: 201–210. Welsh AK, Dawson JO, Gottfried GJ, Hahn D. 2009a. Diversity of Frankia populations in root nodules of geographically isolated Arizona alder trees in central Arizona (United States). Applied and Envrionmental Microbiology 75: 6913–6918. Yamanaka T, Li CY, Bormann BT, Okabe H. 2003. Tripartite associations in an alder: effects of Frankia and Alpova diplophloeus on the growth, nitrogen fixation and mineral acquisition of Alnus tenuifolia. Plant and Soil 254: 179– 186.
Supporting Information Additional supporting information may be found in the online version of this article. Fig. S1 Maximum likelihood phylogram of Frankia sequences. Fig. S2 Confidence intervals of the Frankia OTU accumulation curves. Fig. S3 Comparison of the regional distribution of Alnus-associated EcM fungal and Frankia OTUs based on cluster analysis. Table S1 Detailed sampling data Please note: Wiley Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.
Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust
