Molecular Phylogenetics and Evolution 77 (2014) 110–115
Contents lists available at ScienceDirect
Molecular Phylogenetics and Evolution journal homepage: www.elsevier.com/locate/ympev
Mosaic origins of Bradyrhizobium legume symbionts on the Caribbean island of Guadeloupe Matthew A. Parker a,⇑, Alain Rousteau b a b
Department of Biological Sciences, State University of New York, Binghamton, NY 13902, USA Laboratoire de Biologie et Physiologie Vegetales, Universite des Antilles et de la Guyane BP 592, 97159, Pointe-a-Pitre Cedex, Guadeloupe, France
a r t i c l e
i n f o
Article history: Received 14 February 2014 Revised 4 April 2014 Accepted 8 April 2014 Available online 18 April 2014 Keywords: Lateral gene transfer Legume symbiosis Multilocus sequence analysis Symbiont colonization Tropical community assembly
a b s t r a c t To analyze geographic affinities of Bradyrhizobium sp. symbionts associated with the diverse legume flora on the Caribbean island of Guadeloupe, 39 isolates from 18 legume genera were compared to a reference set of 269 Bradyrhizobium strains from North America, Central America, Puerto Rico and the Philippines. A multilocus sequence analysis (4192 bp) showed that nucleotide diversity in Guadeloupe equaled or exceeded that found in all other regional Bradyrhizobium populations examined. Bayesian phylogenetic tree analysis grouped the Guadeloupe Bradyrhizobium strains into clades with at least 20 distinct sets of non-Guadeloupe relatives, implying that the island was colonized numerous times from multiple source regions. However, for 18% of the Guadeloupe isolates, inferred geographic affinities for the nifD locus, in the symbiosis island region of the Bradyrhizobium chromosome, conflicted with the source region deduced from a tree based on six concatenated housekeeping genes. Geographic mosaic ancestry was therefore evident among Guadeloupe bradyrhizobia. Horizontal gene transfer subsequent to island colonization appears to have generated strains that carry combinations of genes from disparate source regions. Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction The ecological success of many legumes depends on symbiosis with nitrogen-fixing bacteria, and for such plants, island colonization poses some interesting challenges (Parker, 2001). No mechanism exists to ensure coordinated dispersal of legumes and their root-nodule bacteria. Thus, the absence of coadapted bacterial symbionts may act as an initial barrier to colonizing legumes when they arrive in a novel habitat (Parker et al., 2006). However, such barriers are expected to be transient, because establishment of a foothold plant population enhances the opportunity for subsequent colonization by symbionts (Parker, 2001). Bacterial horizontal gene transfer is another important factor affecting colonization dynamics. Two studies have shown that when exotic legumes and rhizobia were jointly introduced into a new geographic region, resident bacteria rapidly evolved the capacity to nodulate the novel host by acquiring symbiotic genes from the inoculant rhizobia (Sullivan et al., 1995; Nandasena et al., 2006). Similar outcomes have been reported for symbiont acquisition by invasive legume weeds that have spread into new
⇑ Corresponding author. Fax: +1 607 777 6521. E-mail address: [email protected] (M.A. Parker). http://dx.doi.org/10.1016/j.ympev.2014.04.011 1055-7903/Ó 2014 Elsevier Inc. All rights reserved.
regions without any deliberate human-mediated bacterial transport. In two case studies, exotic legumes acquired rhizobia that appeared to be indigenous to the invaded habitat based on housekeeping gene loci. However, these rhizobia possessed symbiotic genes from other bacterial lineages that were associated with the exotic legume in its ancestral range (Wei et al., 2009; Horn et al., 2014). Although legume nodule symbionts have been studied in many continental regions (e.g., Nzoue et al., 2009; Gyaneshwar et al., 2011; Stepkowski et al., 2012; Aserse et al., 2012; Koppell and Parker, 2012; Guerrouj et al., 2013), there has been little research specifically focused on the diversity and origins of legume symbionts on oceanic islands (Leary et al., 2004; Vinuesa et al., 2005). In this study, we analyzed Bradyrhizobium sp. symbionts associated with the rich legume flora of Guadeloupe, the largest oceanic island in the Lesser Antillean island arc (1440 km2). Guadeloupe consists of two islands separated by a narrow (50 m) sea channel. One is covered by marine limestone deposited in the past 400,000 years (Feuillet et al., 2004) and the other is largely volcanic with the oldest surface igneous rocks dated at 2.8 million years (Samper et al., 2007). The antiquity of terrestrial communities on Guadeloupe remains an open issue, because some organisms may have arrived from older islands in the region that are now below sea level (Iturralde-Vinent and MacPhee, 1999). Nevertheless, it appears
M.A. Parker, A. Rousteau / Molecular Phylogenetics and Evolution 77 (2014) 110–115
that Guadeloupe itself has been continuously habitable by terrestrial organisms for no more than 3 million years. To characterize Guadeloupe Bradyrhizobium, we sequenced seven genetic markers in strains sampled from a broad suite of legume hosts. To identify possible source regions that may have contributed to establishment of this population, we constructed a parallel data set for Bradyrhizobium strains from ten other regional populations (4 in the U.S., 2 in Mexico, and one each from Costa Rica, Panama, Puerto Rico, and the Philippines). We then used phylogenetic methods to search for the closest relatives of the Guadeloupe strains. To analyze the possible role of horizontal gene exchange in the evolution of the Guadeloupe bacteria, we compared phylogenetic tree structure for one symbiosis-related gene and for six non-symbiotic (housekeeping) loci. In Bradyrhizobium, symbiosis-related genes are clustered in a portion of the chromosome termed the symbiosis island (hereafter SI) region, which has hallmarks of being a mobile genetic element (Kaneko et al., 2002). Markers in the SI region commonly have a different genealogical history from other genes (Moulin et al., 2004; Stepkowski et al., 2007). Comparison of SI and non-SI loci can identify likely cases of lateral transfer of symbiosis-related traits, and may help to understand how bacteria have adapted to novel hosts (Parker, 2012).
2. Materials and methods 2.1. Sampling of nodule isolates The flora of Guadeloupe contains about 75 nodule-forming legume species in 42 genera (Howard, 1988; Saur et al., 2000; Fournet, 2002). Seven genera that associate primarily or exclusively with rhizobial lineages other than Bradyrhizobium were excluded from sampling, and nodules were sought from all other Guadeloupe legume species encountered. Samples were obtained from a total of 19 species in 18 legume genera at eight sites (Supplementary Data Table S1): Plage de Clugny (nodules from Canavalia, Chamaecrista, Dalbergia, Indigofera, Machaerium, Macroptilium, Stylosanthes), Chauffours, Pointe-a-Pitre (Lonchocarpus, Pterocarpus), Canal de la Belle Plaine (Inga, Pterocarpus), Bras David Riviere (Abarema, Desmodium, Inga, Ormosia, Swartzia), Vieux Fort near the southern tip of Basse-Tierre (Lonchocarpus), Soufrière Volcano (Rhynchosia), west coast near Bouillante (Piscidia), and Quai de la marina, Pointe-a-Pitre (Centrosema, Galactia). All of the sampled legume species appear to be indigenous to Guadeloupe except for Indigofera spicata, which is native to the Old World tropics (Morton, 1989). Only one of the legume species is endemic to the Lesser Antilles (Swartzia caribaea; Supplementary Data Table S1). Three others are restricted to northern South America plus the Lesser Antilles (Abarema jupunba, Inga ingoides, Lonchocarpus violaceus), but most are distributed throughout the Caribbean, Mexico, Central America and South America (Table S1). One bacterial isolate was purified from each nodule as described (Parker et al., 2002). Genomic DNA was obtained by heating cells at 95 °C for 5 min in a lysis buffer containing 1% Triton X-100, followed by chloroform extraction (Parker et al., 2002). Seven DNA regions were sequenced in 39 isolates chosen at random from among the available stains for each host legume. For 36 isolates, inoculation tests with the promiscuous legume Macroptilium atropurpureum verified that all were capable of inducing root nodule development. The three remaining isolates had multilocus haplotypes identical to other Guadeloupe strains that formed nodules on M. atropurpureum. For a marker in the SI region, we analyzed an 822 bp portion of nifD gene (which codes for dinitrogenase a subunit) as described (Parker et al., 2002). Six markers outside of the SI region were also
111
sequenced as described (Stepkowski et al., 2007; Parker, 2012; Koppell and Parker, 2012): dnaK (603 bp), gyrB (781 bp), a 50 portion of the 23S rRNA gene (497 bp), recA (513 bp), rplC (432 bp), and rpoB (537 bp). 2.2. Phylogenetic analyses Guadeloupe bacteria were compared to 269 Bradyrhizobium strains from other locations. Ten strains represented well-known taxa (B. elkanii USDA 76, B. diazoefficiens USDA 110 (formerly B. japonicum USDA 110; Delamuta et al., 2013), B. liaoningense USDA 3622, B. yuanmingense LMG 21827, B. canariense BTA1, B. jicamae PAC68T, B. pachyrhizi PAC48T, B. cytisi CTAW11T, B. rifense CTAW71T, and Bradyrhizobium sp. strain ORS278). For 214 other Bradyrhizobium strains from diverse legumes in eight regions of North and Central America, data for six gene loci were reported in Parker (2012) and Koppell and Parker (2012): Panama (28 strains), Costa Rica (25), Oaxaca (35), Chihuahua (22), south Texas (21), Washington State (25), North Carolina (29), and northeastern U.S. (29). We sequenced the recA marker in all of these strains in order to obtain complete 7-locus data comparable to that for the Guadeloupe strains. To supplement these results, we also obtained complete 7-locus sequence data on 45 additional Bradyrhizobium strains from two other regions (Table S1): Puerto Rico (23 strains) and the Philippines (22 strains; Andam and Parker, 2008). The primary sampling objective was to obtain symbionts from a maximum diversity of legume taxa native to each region. Regions varied in the number of legume species that could be sampled within a single habitat, so the spatial scale of sampling varied across regions (Table 1). To compare nucleotide diversity among eleven regional populations, DnaSP (Librado and Rozas, 2009) was used to estimate pT, the average pairwise number of nucleotide differences per site, for the 7-locus concatenated data set. NeighborNet analysis using SplitsTree v4.11.3 (Huson and Bryant, 2006) indicated that concatenation of the six housekeeping gene loci resulted in a network that was only marginally more reticulated than networks for individual loci. The network for nifD sequences had a largely bifurcating structure with modest reticulation. However, concatenation of nifD with the housekeeping loci resulted in a network with extensive reticulation, indicating that the phylogenetic signal in nifD sequences conflicted greatly with the other loci. To summarize bacterial relationships, we therefore present one tree for concatenated sequences of the six housekeeping loci, and a separate tree for nifD sequences. Trees were inferred using MrBayes (Ronquist and Huelsenbeck, 2003) with protein coding loci partitioned by codon position and separate estimates of rates and nucleotide composition for each codon position and locus. A HKY substitution model was used. Some alternative models were tested but all yielded very similar trees. Analyses were run for 4 million generations, sampling every 400 generations, with the last 1000 samples saved for tree analysis. Replicate runs yielded highly congruent consensus tree topologies. Trees were rooted using Azorhizobium caulinodans ORS571 (NC009937) and Xanthobacter autotrophicus Py2 (NC009720), and Methylobacterium strain 4-46 (CP000943.1) as outgroups. Phylocom 4.0.1 (Webb et al., 2008) was used to analyze whether bacteria were phylogentically clustered according to geographic source region. Clustering analyses were performed separately for the nifD tree and the housekeeping gene tree. For each regional group of bacterial strains, MPDsample (Webb et al., 2008) was calculated as the average across all pairs of strains of the branch length separating them on the phylogenetic tree. Significance was assessed by comparing MPDsample values to the null distribution inferred from 1000 random permutations of isolate names across tips of the observed tree. The Net Relatedness Index
112
M.A. Parker, A. Rousteau / Molecular Phylogenetics and Evolution 77 (2014) 110–115
Table 1 Nucleotide polymorphism and sample attributes. Host Region Guadeloupe Puerto Rico Philippines Panama Costa Rica Oaxaca Texas Chihuahua North Carolina Northeastern U.S. Washington State a b c D e f
a
N 39 23 22 28 25 35 21 22 29 29 25
b
p ± 1 SD 0.085 ± 0.004 0.085 ± 0.004 0.078 ± 0.003 0.085 ± 0.006 0.079 ± 0.006 0.074 ± 0.005 0.079 ± 0.004 0.074 ± 0.005 0.043 ± 0.008 0.060 ± 0.006 0.022 ± 0.008
Species 19 7 3 20 16 19 7 6 17 13 5
Host Genera 18 7 3 16 14 17 7 3 13 8 2
Latitude O
( N) 16.2 18.3 14.1 9.1 9.2 16.4 26.7 27.5 35.5 42.1 47.4
Sampled localitiesc 44 km 38 kmd 15 kmd 5 kme 110 kme 165 kme 180 kme 33 kme 430 kmf 410 kmf 440 kme
Number of isolates studied per region. Mean fraction of nucleotide differences among isolate pairs within a region, across seven gene loci. Maximum distance among sampling localities within region. Table S1. Koppell and Parker (2012). Parker (2012).
(NRI), a standardized measure of the magnitude of clustering, was calculated as the difference between MPDsample and the MPD value for the same sample size calculated from the null distribution (MPDnull), divided by the standard deviation of MPDnull (Webb et al., 2008). GenBank accession numbers for the 833 new sequence records obtained in this study, and for the 1323 other existing sequence records utilized, are provided in Table S2. 3. Results 3.1. Patterns of nucleotide diversity Almost every Guadeloupe Bradyrhizobium isolate had a unique multilocus genotype (haplotype) when data from all seven marker loci were combined. The only exceptions involved six isolates from three legume trees (Abarema jupunba, Inga ingoides, and Ormosia krugii) that all shared a single haplotype, as did two isolates from the tree Pterocarpus officinalis, and two isolates from the tree Swartzia caribaea. Nucleotide diversity in the Guadeloupe Bradyrhizobium sample was among the highest of any of the eleven regional populations analyzed (Table 1). Across all eleven populations, nucleotide diversity was inversely related to source region latitude (Spearman rank correlation, rs = 0.76, P < 0.005). Nucleotide diversity was not significantly correlated with the number of legume host species sampled in a region or the number of legume genera sampled (rs = 0.38 and 0.50, respectively; P > 0.05; this result was not altered when diversity data were analyzed separately for symbiotic [nifD] vs. nonsymbiotic loci). 3.2. Origins of Guadeloupe Bradyrhizobium
Fig. 1. Bayesian tree for six concatenated housekeeping loci (3370 bp) in 308 Bradyrhizobium isolates. Outgroups were deleted before plotting to save space, and strains in terminal clades were grouped (see Supplementary Data Fig. S3 for complete topology, strain identity and clade posterior probability values). Withinclade disparity in branch length is indicated by triangle size. Filled triangles indicate clades with Guadeloupe strains (listed in bold, along with the number and geographic affinities of non-Guadeloupe clade members: Phil = Philippines; Oax = Oaxaca, Mexico; CR = Costa Rica; Pan = Panama; NC = North Carolina; neU.S.=northeastern United States; Chih = Chihuahua, Mexico; TX = Texas, PR = Puerto Rico). The common, broad host-range neotropical lineage described in Parker (2008) is marked with an asterisk.
Bayesian analysis yielded a well-resolved tree for the concatenated housekeeping gene data (Fig. 1; 83% of the clades had a posterior probability of 0.95 or higher). The 308 Bradyrhizobium strains fell into three major clades. The most basally diverging clade (group 1) included the photosynthetic Aeschynomene symbiont ORS278 and three strains from Panama. The remainder of the strains split into two groups that each included several reference taxa: B. diazoefficiens/B. liaoningense/B. yuanmingense/B. canariense/B. cytisi/B. rifense (group 2), and B. elkanii/B. pachyrhizi/B. jicamae (group 3). If two Guadeloupe Bradyrhizobium strains are placed into distinct well-supported clades with different non-Guadeloupe relatives, the most parsimonious inference would often be that their
establishment on Guadeloupe involved two separate colonization events. Thus, an upper limit on colonization events can be estimated by counting how many distinct terminal clades encompassed both Guadeloupe and non-Guadeloupe strains. For each Guadeloupe strain, the smallest terminal clade was identified that included at least one non-Guadeloupe relative and that was supported by a Bayesian posterior probability of 0.95 or higher (Table S3). This analysis partitioned the 39 Guadeloupe isolates into 21 distinct terminal clades. Most terminal clades were very homogeneous. The median nucleotide divergence among strain pairs within a clade was 0.01 (Table S3). These results imply that the Guadeloupe Bradyrhizobium community is descended from
113
M.A. Parker, A. Rousteau / Molecular Phylogenetics and Evolution 77 (2014) 110–115
multiple colonization events, involving ancestors from several source regions. One Guadeloupe strain (deg3.5) proved to be identical at all six housekeeping loci to four Puerto Rican isolates from the same host legume (Dalbergia ecastaphyllum). Such cases of close relationship for bacterial strains from the same legume host in nearby regions are hardly surprising. Seven other legume species were also sampled both in Guadeloupe and elsewhere (Canavalia rosea, Centrosema virginianum, Chamaecrista nictitans, Desmodium axillare, Desmodium incanum, Galactia striata, Macroptilium lathyroides). For two of these legumes (C. rosea, M. lathyroides), certain Bradyrhizobium isolates from Guadeloupe and non-Guadeloupe populations were very similar. However, most of these legume species harbored somewhat divergent symbionts in different regions. In a few cases, bacterial isolates from unrelated host legumes in distant locations exhibited surprising genetic similarity. Four Guadeloupe strains were more similar to certain Philippine Bradyrhizobium strains than to any of the Western Hemisphere strains included in the study (Fig. 1). For each terminal clade having one or more Guadeloupe strains, its geographic affinity was represented by the proportion of included strains from each region (Table S3). Pooled across all 39 Guadeloupe isolates, this analysis indicated that the top two inferred source regions for Guadeloupe bradyrhizobia were Texas (23%) and Oaxaca (16%), followed by Puerto Rico (15%) Costa Rica (14%) and Panama (14%). Only a very few Guadeloupe bradyrhizobia had geographic affinities to samples from temperate North America (Washington State, North Carolina, or the northeastern U.S.). BLAST searches with single genes found only a few cases where Bradyrhizobium strains from indigenous legumes in other regions had higher similarity to Guadeloupe strains than did any of the 269 reference strains in the current sample. Guadeloupe strain dbd20 had a gyrB sequence identical to that of Bradyrhizobium iriomotense (from Okinawa, Japan), and Guadeloupe strain ork1 had a recA sequence that differed at only one nucleotide from that of Bradyrhizobium strain STM7334 from Brazil. Data from only one or two loci were available for these strains, so it was not possible to include them in the concatenated housekeeping gene tree analysis. However, these results suggest that other regions beyond the ten locations of the current study (Table 1) are likely to harbor strains with close affinities to some of the Guadeloupe Bradyrhizobium lineages. Bayesian analysis of nifD sequences yielded a well-resolved tree (Fig. 2; 75% of clades had a posterior probability of 0.95 or higher). However, the nifD tree topology conflicted substantially with the housekeeping gene tree. The most basally diverging strains (group 1) were the same as in the housekeeping gene tree (Fig 1). However, strains belonging to the two other large, well-differentiated housekeeping-gene clades (groups 2, 3) were intermingled in the nifD tree. Phylogenetic incongruence of the two trees is consistent with much other research suggesting that lateral transfer of SIregion loci such as nifD has been a common event in Bradyrhizobium evolution (Moulin et al., 2004; Stepkowski et al., 2012; Parker, 2012). Colonization events were estimated in the same way as for the housekeeping gene tree, by counting the number of distinct terminal nifD clades that encompassed both Guadeloupe and non-Guadeloupe strains (Table S4). This analysis indicated that Guadeloupe nifD sequences may be derived from at most 20 colonization events, which was quite similar to the estimate for the housekeeping gene tree (21). However, the top two inferred source regions for Guadeloupe bradyrhizobial nifD sequences were not the same as for housekeeping genes. Most Guadeloupe nifD sequences had affinities to Bradyrhizobium reference strains from Puerto Rico (47%) and the Philippines (14%). Texas ranked as the third most
Fig. 2. Bayesian tree for nifD (822 bp) in 308 Bradyrhizobium isolates. Outgroups were deleted before plotting to save space, and strains in terminal clades were grouped (see Fig. S4 for complete topology, strain identity and clade posterior probability values). Within-clade disparity in branch length is indicated by triangle size. Filled triangles indicate clades with Guadeloupe strains (listed in bold along with the number and geographic affinities of non-Guadeloupe clade members; location abbreviations as in Fig. 1 legend). nifD clades that were uniform for one of the three major housekeeping gene clades (1, 2 or 3; Fig. 1) are marked with that number at their base.
important source (10%), and all other regions had inferred contributions below 10% (Table 2). Across the two trees, geographic affinities conflicted significantly for seven Guadeloupe strains (deg2.1, deg3.11, dbd20, isg8, pbt1c, rhr14, rhr28, pog17). Each of these strains grouped into a strongly supported clade (Tables S3 and S4) with relatives from different regions in the two trees. These can be interpreted as cases of geographic mosaic ancestry (Andam and Parker, 2008), where lateral gene transfer subsequent to arrival in Guadeloupe may have generated strains with chromosomal portions from more than one source region.
Table 2 Phylogenetic clustering of Bradyrhizobium isolates in relation to geographic source region.
Region: Guadeloupe Puerto Rico Philippines Panama Costa Rica Oaxaca Texas Chihuahua North Carolina Northeastern U.S. Washington State Combined a
Na 39 23 22 28 25 35 21 22 29 29 25 295
nifD tree
6-locus housekeeping gene tree
NRI b 1.65 (ns) 0.80 (ns) 1.80 (ns) 3.41** 2.95** 5.12*** 5.76*** 5.80*** 12.23*** 10.67*** 15.24*** 44.42***
NRI 1.16 (ns) 0.86 (ns) 2.05** 2.13*** 2.51*** 2.24*** 0.13 (ns) 1.47* 5.55*** 2.98*** 7.36*** 24.39***
Number of isolates. NRI = (MPDnull–MPDsample)/(standard deviation of MPDnull), where MPDsample is the mean pairwise patristic distance among strains from one region; positive values of NRI indicate that strains from the same region tend to be phylogenetically close, relative to the null distribution generated by random permutation of strains across the tree. Probability of NRI inferred by comparison to null distribution based on 1000 random permutations of strains across tree branch tips (***P < 0.001; **P < 0.01; * P < 0.05; (ns), P > 0.05). b
114
M.A. Parker, A. Rousteau / Molecular Phylogenetics and Evolution 77 (2014) 110–115
3.3. Differentiation of regional populations Tree permutation analyses (Webb et al., 2008) indicated that there was substantial phylogenetic clustering of bacteria according to geographic source region in both trees (Table 2 and Figs. S1 and S2). Thus, bacteria from most individual locations had genes that were significantly more similar, on average, relative to the null distribution generated by permuting strains across tips on each tree. However, phylogenetic clustering was not significant in either tree for the bacterial populations from Guadeloupe or Puerto Rico (Table 2). These two populations (along with Texas in the housekeeping gene tree, and the Philippines for the nifD tree) had so many diverse lineages that pairwise genetic distance for these samples was not distinguishable from the null distribution representing strain pairs selected at random across the entire sample. For the Guadeloupe sample, the lack of phylogenetic clustering (Table 2) is consistent with the terminal clade analysis (Figs. 1 and 2; Tables S3 and S4) showing that this bacterial population is a composite of lineages drawn from many source locations.
4. Discussion Our main finding was that the Bradyrhizobium community of Guadeloupe is strikingly diverse, and originated from numerous colonists from multiple source regions. Despite the relatively small size of this island and its modest age, it harbored bacterial symbionts with a genetic diversity that equaled or exceeded that found in older and larger continental areas (Table 1). Much work remains to be done in characterizing potential source regions for Caribbean populations. In particular, there has been no comprehensive sampling in northern South America, a likely source for symbiotic bacteria now resident in Guadeloupe, given its proximity and the fact that it shares many legume host species (Table S1). Although more detailed sampling of continental habitats may modify inferences about geographic affinity for particular clades, it is not likely to alter the basic conclusion that Guadeloupe bacteria are derived from many Bradyrhizobium clades from diverse localities. Our results indicate that evolution of the Guadeloupe Bradyrhizobium population has been affected not only by recurrent migration but also by substantial horizontal gene transfer for a SI region locus (nifD). The impact of SI region horizontal transfer on community assembly merits more detailed study. Loci in the SI region appear to be important in determining which legume hosts a bacterial strain can successfully interact with (Parker, 2012). A bacterial strain may be perfectly well adapted to physical features of an island habitat, but there is no guarantee that its SI region will be optimally suited to the particular legumes that are abundant in any given site. Horizontal transfer of the SI region is a process that may facilitate adaptation in a novel ecosystem, by reassembling traits from different ancestral environments. This likely contributed to the high overall genetic diversity detected among Guadeloupe Bradyrhizobium (Table 1). Mechanisms enabling long-distance migration in Bradyrhizobium are not well understood. These bacteria have no resistant spore stage readily transported in the atmosphere, and are not particularly salt-tolerant, as would be needed to reach an island on a floating raft of vegetation debris. Raza et al. (2001) reported that strains of Bradyrhizobium varied in salinity tolerance, but for even the most tolerant strains, only about one out of a thousand cells survived a week of exposure to the salinity of seawater. However, Bradyrhizobium populations are known to have the capacity to survive in the soil for decades in the absence of legumes hosts (Norman, 1942). Thus, any mechanism that could transport soil particles (e.g., a bird’s muddy feet, prevailing wind currents (Yamaguchi et al., 2012), or extreme weather events such as
hurricanes) could potentially introduce rhizobia into an island habitat. It is still difficult to account for the finding of some closely similar strain pairs in Guadeloupe and the Philippines, which are half a world apart (16,600 km). Both samples came from indigenous legumes in forested non-agricultural habitats, so there is no obvious explanation involving human agency. An earlier study also detected a few cases of close genetic similarity between Bradyrhizobium strains in the Philippines and in eastern North America (Andam and Parker, 2008). It remains an open problem to explain why certain Bradyrhizobium lineages have achieved such wide geographic ranges (and why others have not). In principle, a clade that is distributed both in Guadeloupe and another location could be explained by a migration event in either direction. However, the small area of Guadeloupe means that it is much more likely to be a recipient than a source of migrants. For example, suppose that the probability of microbial transfer between Guadeloupe and the much larger island of Puerto Rico depended strictly on their relative areas (1440 km2 vs. 9104 km2), and not on other factors such as the prevailing direction of hurricane storm tracks. In this case, a six-fold asymmetry in migrant exchange would be expected. To the extent that Bradyrhizobium lineages in other locations came from Guadeloupe ancestors, the method used in Tables S3 and S4 will overestimate migrant arrival into Guadeloupe. Thus, a count of terminal clades including both Guadeloupe and non-Guadeloupe strains (20–21) should be viewed simply as an estimated upper bound on the probable number of colonization events that populated this island. Three of the Guadeloupe legumes are trees that belong to genera whose symbionts have not been previously characterized in any detail: Abarema (tribe Ingeae in the subfamily Mimosoideae), Swartzia (tribe Swartzieae) and Ormosia (tribe Sophoreae), both in the subfamily Papilionoideae (Lewis et al., 2005). All three of these trees utilized Bradyrhizobium strains related to a common lineage (marked with an asterisk in Fig. 1) that is widespread in central America and that apparently has a very broad host range (Parker, 2008). In Panama and Costa Rica, this lineage has been sampled from fifteen legume genera of diverse life form (herbs, lianas, and trees). In the current sample, it was also detected in Puerto Rico on an additional genus of legume trees (Andira; Table S1). Our results thus provide further data suggesting that some Bradyrhizobium lineages that lack narrow host specificity may be widely distributed in the Neotropics. This lineage is relatively homogeneous at housekeeping loci (mean pairwise nucleotide divergence (p) ±1 SD = 0.013 ± 0.002). However, strains within this group show much more diversity for the nifD locus in the SI region (p = 0.061 ± 0.010), perhaps indicative of diversifying selection associated with varied host utilization (Parker, 2012). Current sampling is far too incomplete to know whether these patterns are broadly characteristic of legume-bacterial relationships in this region. In the future, wider analysis of the striking diversity of Neotropical legume taxa will help to resolve how Bradyrhizobium lineages have spread geographically and across host plant clades. Acknowledgments We are grateful to C. Andam, F. Muller, R. Rivas, E. Velazquez and P. Vinuesa for help with collecting or for providing strains. Financial support was provided by the National Science Foundation grant MCB-0640246. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ympev.2014. 04.011.
M.A. Parker, A. Rousteau / Molecular Phylogenetics and Evolution 77 (2014) 110–115
References Andam, C.C., Parker, M.A., 2008. Origins of Bradyrhizobium nodule symbionts from two legume trees in the Philippines. J. Biogeography 35, 1030–1039. Aserse, A.A., Räsänen, L.A., Aseffa, F., Hailemariam, A., Lindström, K., 2012. Phylogenetically diverse groups of Bradyrhizobium isolated from nodules of Crotalaria spp., Indigofera spp., Erythrina brucei and Glycine max growing in Ethiopia. Mol. Phylogenet. Evol. 65, 595–609. Delamuta, J.R.M., Ribeiro, R.A., Ormeño-Orrillo, E., Melo, I.S., Mart´ınez-Romero, E., Hungria, M., 2013. Polyphasic evidence supporting the reclassification of Bradyrhizobium japonicum group Ia strains as Bradyrhizobium diazoefficiens sp. nov. Int. J. Syst. Evol. Microbiol. 63, 3342–3351. Feuillet, N., Papponnier, P., Manighetti, I., Villemant, B., King, G.C.P., 2004. Differential uplift and tilt of pleistocene reef platforms and quaternary slip rate on the Morne-Piton normal fault (Guadeloupe, French West Indies). J. Geophysical. Res. 109, B02404. Fournet, J., 2002. Flore illustrée des phanérogames de Guadeloupe et de Martinique. CIRAD, Gondwana éditions, La Trinite, Martinique. Guerrouj, K., Ruíz-Díez, B., Chahboune, R., Ramírez-Bahena, M.H., Abdelmoumen, H., Quiñones, M.A., El Idrissi, M.M., Velázquez, E., Fernández-Pascual, M., Bedmar, E.J., Peix, A., 2013. Definition of a novel symbiovar (sv. retamae) within Bradyrhizobium retamae sp. nov., nodulating Retama sphaerocarpa and Retama monosperma. Syst. Appl. Microbiol. 36, 218–223. Gyaneshwar, P., Hirsch, A.M., Moulin, L., Chen, W.-M., Elliot, G.N., Bontemps, C., Estrada-de los Santos, P., Gross, E., dos Reis, F.B., Sprent, J.I., Young, J.P., James, E.K., 2011. Legume-nodulating betaproteobacteria: diversity, host range, and future prospects. Mol. Plant-Microbe Int. 24, 1276–1288. Horn, K., Parker, I., Małek, W., Rodríguez-Echeverría, S., Parker, M.A., 2014. Disparate origins of Bradyrhizobium symbionts for invasive populations of Cytisus scoparius (Leguminosae) in North America. FEMS Microbiol. Ecol. (in press). http://dx.doi.org/10.1111/1574-6941.12335. Howard, R.A., 1988. Flora of the Lesser Antilles, Dicotyledoneae, Part 1. Arnold Arboretum of Harvard University, Jamaica Plain, Massachusetts Huson, D.H., Bryant, D., 2006. Application of phylogenetic networks in evolutionary studies. Mol. Biol. Evol. 23, 254–267. Iturralde-Vinent, M.A., MacPhee, R.D.E., 1999. Paleogeography of the Caribbean region: implications for Cenozoic biogeography. Bull. Am. Mus. Nat. Hist. 238, 1–95. Kaneko, T., Nakamura, Y., Sato, S., et al., 2002. Complete genomic sequence of nitrogen-fixing symbiotic bacterium Bradyrhizobium japonicum USDA110. DNA Res. 9, 189–197. Koppell, J.H., Parker, M.A., 2012. Phylogenetic clustering of Bradyrhizobium symbionts on legumes indigenous to North America. Microbiology 158, 2050– 2059. Lewis, G., Schrire, B., Mackinder, B., Lock, M., 2005. Legumes of the world. Royal Botanic Gardens, Kew, UK. Librado, P., Rozas, J., 2009. DnaSP v5: A software package for comprehensive analysis of DNA polymorphism data. Bioinformatics 25, 1451–1452. Moulin, L.G., Bena, G., Boivin-Masson, C., Stepkowski, T., 2004. Phylogenetic analyses of symbiotic nodulation genes support vertical and lateral gene cotransfer within the Bradyrhizobium genus. Mol. Phylogenet. Evol. 30, 720–732. Leary, J.K., Singleton, P.W., Borthakur, D., 2004. Canopy nodulation of the endemic tree legume Acacia koa in the mesic forests of Hawaii. Ecology 85, 3151–3157. Morton, J.F., 1989. Creeping indigo (Indofera spicata Forsk.) (Fabaceae) – a hazard to herbivores in Florida. Econ. Bot. 43, 314–327.
115
Nandasena, K.G., O’Hara, G.W., Tiwari, R.P., Howieson, J.G., 2006. Rapid in situ evolution of nodulating strains for Biserrula pelecinus L. through lateral transfer of a symbiosis island from the original mesorhizobial inoculant. Appl. Env. Microbiol. 72, 7365–7367. Norman, A.G., 1942. Persistence of Rhizobium japonicum in soil. Amer. Soc. Agron. J. 34, 499. Nzoue, A., Miche, L., Klonowska, A., Laguerre, G., de Lajudie, P., Moulin, L., 2009. Multilocus sequence analysis of bradyrhizobia isolated from Aeschynomene species in Senegal. Syst. Appl. Microbiol. 32, 400–412. Parker, M.A., 2001. Mutualism as a constraint on invasion success for legumes and rhizobia. Diversity Distrib. 7, 125–136. Parker, M.A., 2008. Symbiotic relationships of legumes and nodule bacteria on Barro Colorado Island, Panama: a review. Microbial Ecol. 55, 662–672. Parker, M.A., 2012. Legumes select symbiosis island sequence variants in Bradyrhizobium. Molec. Ecol. 21, 1769–1778. Parker, M.A., Lafay, B., Burdon, J.J., van Berkum, P., 2002. Conflicting phylogeographic patterns in rRNA and nifD indicate regionally restricted gene transfer in Bradyrhizobium. Microbiology 148, 2557–2565. Parker, M.A., Malek, W., Parker, I.M., 2006. Growth of an invasive legume is symbiont-limited in newly occupied habitats. Div. Distrib. 12, 563–571. Raza, S., Jørnsgård, B., Abou-Taleb, H., Christiansen, J.L., 2001. Tolerance of Bradyrhizobium sp. (Lupini) strains to salinity, pH, CaCO3 and antibiotics. Lett. Appl. Microbiol. 32, 379–383. Ronquist, F., Huelsenbeck, J.P., 2003. MRBAYES 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19, 1572–1574. Samper, A., Quidelleur, X., Lahitte, P., Mollex, D., 2007. Timing of effusive volcanism and collapse events within an oceanic arc island: Basse-Terre. Guadeloupe archipelago. Earth Planet. Sci. Let. 258, 175–191. Saur, E., Carcelle, S., Guezennec, S., Rousteau, A., 2000. Nodulation of legume species in wetlands of Guadeloupe (Lesser Antilles). Wetlands 20, 730–734. Stepkowski, T., Hughes, C.E., Law, I.J., Markiewicz, L., Gurda, D., Chlebicka, A., Moulin, L., 2007. Diversification of lupine Bradyrhizobium strains: evidence from nodulation gene trees. Appl. Environ. Microbiol. 73, 3254–3264. Stepkowski, T., Watkin, E., McInnes, A., Gurda, D., Gracz, J., Steenkamp, E.T., 2012. Distinct Bradyrhizobium communities nodulate legumes native to temperate and tropical monsoon Australia. Mol. Phylogenet. Evol. 63, 265–277. Sullivan, J.T., Patrick, H.N., Lowther, W.L., Scott, D.B., Ronson, C.W., 1995. Nodulating strains of Rhizobium loti arise through chromosomal symbiotic gene transfer in the environment. Proc. Natl. Acad. Sci. USA 92, 8985–8989. Vinuesa, P., Leon-Barrios, M., Silva, C., Willems, A., Jarabo-Lorenzo, A., PerezGaldona, R., Werner, D., Martınez-Romero, E., 2005. Bradyrhizobium canariense sp. nov., an acid-tolerant endosymbiont from the nodules of endemic woody legume species (Papilionoideae: Genisteae) growing in the Canary Islands along with B. japonicum bv. genistearum, Bradyrhizobium genospecies a and Bradyrhizobium genospecies b. Int. J. Syst. Evol. Microbiol. 55, 569–575. Webb, C.O., Ackerly, D.D., Kembel, S.W., 2008. Phylocom: software for the analysis of phylogenetic community structure and trait evolution. Bioinformatics 24, 2098–2100. Wei, G., Chen, W., Zhu, W., Chen, C., Young, J.P.W., Bontemps, C., 2009. Invasive Robinia pseudoacacia in China is nodulated by Mesorhizobium and Sinorhizobium species that share similar nodulation genes with native American symbionts. FEMS Microbiol. Ecol. 68, 320–328. Yamaguchi, N., Ichijo, T., Sakotani, A., Baba, T., Nasu, M., 2012. Global dispersion of bacterial cells on Asian dust. Sci. Rep. 2, 525; DOI:10.1038/srep00525.
Contents lists available at ScienceDirect
Molecular Phylogenetics and Evolution journal homepage: www.elsevier.com/locate/ympev
Mosaic origins of Bradyrhizobium legume symbionts on the Caribbean island of Guadeloupe Matthew A. Parker a,⇑, Alain Rousteau b a b
Department of Biological Sciences, State University of New York, Binghamton, NY 13902, USA Laboratoire de Biologie et Physiologie Vegetales, Universite des Antilles et de la Guyane BP 592, 97159, Pointe-a-Pitre Cedex, Guadeloupe, France
a r t i c l e
i n f o
Article history: Received 14 February 2014 Revised 4 April 2014 Accepted 8 April 2014 Available online 18 April 2014 Keywords: Lateral gene transfer Legume symbiosis Multilocus sequence analysis Symbiont colonization Tropical community assembly
a b s t r a c t To analyze geographic affinities of Bradyrhizobium sp. symbionts associated with the diverse legume flora on the Caribbean island of Guadeloupe, 39 isolates from 18 legume genera were compared to a reference set of 269 Bradyrhizobium strains from North America, Central America, Puerto Rico and the Philippines. A multilocus sequence analysis (4192 bp) showed that nucleotide diversity in Guadeloupe equaled or exceeded that found in all other regional Bradyrhizobium populations examined. Bayesian phylogenetic tree analysis grouped the Guadeloupe Bradyrhizobium strains into clades with at least 20 distinct sets of non-Guadeloupe relatives, implying that the island was colonized numerous times from multiple source regions. However, for 18% of the Guadeloupe isolates, inferred geographic affinities for the nifD locus, in the symbiosis island region of the Bradyrhizobium chromosome, conflicted with the source region deduced from a tree based on six concatenated housekeeping genes. Geographic mosaic ancestry was therefore evident among Guadeloupe bradyrhizobia. Horizontal gene transfer subsequent to island colonization appears to have generated strains that carry combinations of genes from disparate source regions. Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction The ecological success of many legumes depends on symbiosis with nitrogen-fixing bacteria, and for such plants, island colonization poses some interesting challenges (Parker, 2001). No mechanism exists to ensure coordinated dispersal of legumes and their root-nodule bacteria. Thus, the absence of coadapted bacterial symbionts may act as an initial barrier to colonizing legumes when they arrive in a novel habitat (Parker et al., 2006). However, such barriers are expected to be transient, because establishment of a foothold plant population enhances the opportunity for subsequent colonization by symbionts (Parker, 2001). Bacterial horizontal gene transfer is another important factor affecting colonization dynamics. Two studies have shown that when exotic legumes and rhizobia were jointly introduced into a new geographic region, resident bacteria rapidly evolved the capacity to nodulate the novel host by acquiring symbiotic genes from the inoculant rhizobia (Sullivan et al., 1995; Nandasena et al., 2006). Similar outcomes have been reported for symbiont acquisition by invasive legume weeds that have spread into new
⇑ Corresponding author. Fax: +1 607 777 6521. E-mail address: [email protected] (M.A. Parker). http://dx.doi.org/10.1016/j.ympev.2014.04.011 1055-7903/Ó 2014 Elsevier Inc. All rights reserved.
regions without any deliberate human-mediated bacterial transport. In two case studies, exotic legumes acquired rhizobia that appeared to be indigenous to the invaded habitat based on housekeeping gene loci. However, these rhizobia possessed symbiotic genes from other bacterial lineages that were associated with the exotic legume in its ancestral range (Wei et al., 2009; Horn et al., 2014). Although legume nodule symbionts have been studied in many continental regions (e.g., Nzoue et al., 2009; Gyaneshwar et al., 2011; Stepkowski et al., 2012; Aserse et al., 2012; Koppell and Parker, 2012; Guerrouj et al., 2013), there has been little research specifically focused on the diversity and origins of legume symbionts on oceanic islands (Leary et al., 2004; Vinuesa et al., 2005). In this study, we analyzed Bradyrhizobium sp. symbionts associated with the rich legume flora of Guadeloupe, the largest oceanic island in the Lesser Antillean island arc (1440 km2). Guadeloupe consists of two islands separated by a narrow (50 m) sea channel. One is covered by marine limestone deposited in the past 400,000 years (Feuillet et al., 2004) and the other is largely volcanic with the oldest surface igneous rocks dated at 2.8 million years (Samper et al., 2007). The antiquity of terrestrial communities on Guadeloupe remains an open issue, because some organisms may have arrived from older islands in the region that are now below sea level (Iturralde-Vinent and MacPhee, 1999). Nevertheless, it appears
M.A. Parker, A. Rousteau / Molecular Phylogenetics and Evolution 77 (2014) 110–115
that Guadeloupe itself has been continuously habitable by terrestrial organisms for no more than 3 million years. To characterize Guadeloupe Bradyrhizobium, we sequenced seven genetic markers in strains sampled from a broad suite of legume hosts. To identify possible source regions that may have contributed to establishment of this population, we constructed a parallel data set for Bradyrhizobium strains from ten other regional populations (4 in the U.S., 2 in Mexico, and one each from Costa Rica, Panama, Puerto Rico, and the Philippines). We then used phylogenetic methods to search for the closest relatives of the Guadeloupe strains. To analyze the possible role of horizontal gene exchange in the evolution of the Guadeloupe bacteria, we compared phylogenetic tree structure for one symbiosis-related gene and for six non-symbiotic (housekeeping) loci. In Bradyrhizobium, symbiosis-related genes are clustered in a portion of the chromosome termed the symbiosis island (hereafter SI) region, which has hallmarks of being a mobile genetic element (Kaneko et al., 2002). Markers in the SI region commonly have a different genealogical history from other genes (Moulin et al., 2004; Stepkowski et al., 2007). Comparison of SI and non-SI loci can identify likely cases of lateral transfer of symbiosis-related traits, and may help to understand how bacteria have adapted to novel hosts (Parker, 2012).
2. Materials and methods 2.1. Sampling of nodule isolates The flora of Guadeloupe contains about 75 nodule-forming legume species in 42 genera (Howard, 1988; Saur et al., 2000; Fournet, 2002). Seven genera that associate primarily or exclusively with rhizobial lineages other than Bradyrhizobium were excluded from sampling, and nodules were sought from all other Guadeloupe legume species encountered. Samples were obtained from a total of 19 species in 18 legume genera at eight sites (Supplementary Data Table S1): Plage de Clugny (nodules from Canavalia, Chamaecrista, Dalbergia, Indigofera, Machaerium, Macroptilium, Stylosanthes), Chauffours, Pointe-a-Pitre (Lonchocarpus, Pterocarpus), Canal de la Belle Plaine (Inga, Pterocarpus), Bras David Riviere (Abarema, Desmodium, Inga, Ormosia, Swartzia), Vieux Fort near the southern tip of Basse-Tierre (Lonchocarpus), Soufrière Volcano (Rhynchosia), west coast near Bouillante (Piscidia), and Quai de la marina, Pointe-a-Pitre (Centrosema, Galactia). All of the sampled legume species appear to be indigenous to Guadeloupe except for Indigofera spicata, which is native to the Old World tropics (Morton, 1989). Only one of the legume species is endemic to the Lesser Antilles (Swartzia caribaea; Supplementary Data Table S1). Three others are restricted to northern South America plus the Lesser Antilles (Abarema jupunba, Inga ingoides, Lonchocarpus violaceus), but most are distributed throughout the Caribbean, Mexico, Central America and South America (Table S1). One bacterial isolate was purified from each nodule as described (Parker et al., 2002). Genomic DNA was obtained by heating cells at 95 °C for 5 min in a lysis buffer containing 1% Triton X-100, followed by chloroform extraction (Parker et al., 2002). Seven DNA regions were sequenced in 39 isolates chosen at random from among the available stains for each host legume. For 36 isolates, inoculation tests with the promiscuous legume Macroptilium atropurpureum verified that all were capable of inducing root nodule development. The three remaining isolates had multilocus haplotypes identical to other Guadeloupe strains that formed nodules on M. atropurpureum. For a marker in the SI region, we analyzed an 822 bp portion of nifD gene (which codes for dinitrogenase a subunit) as described (Parker et al., 2002). Six markers outside of the SI region were also
111
sequenced as described (Stepkowski et al., 2007; Parker, 2012; Koppell and Parker, 2012): dnaK (603 bp), gyrB (781 bp), a 50 portion of the 23S rRNA gene (497 bp), recA (513 bp), rplC (432 bp), and rpoB (537 bp). 2.2. Phylogenetic analyses Guadeloupe bacteria were compared to 269 Bradyrhizobium strains from other locations. Ten strains represented well-known taxa (B. elkanii USDA 76, B. diazoefficiens USDA 110 (formerly B. japonicum USDA 110; Delamuta et al., 2013), B. liaoningense USDA 3622, B. yuanmingense LMG 21827, B. canariense BTA1, B. jicamae PAC68T, B. pachyrhizi PAC48T, B. cytisi CTAW11T, B. rifense CTAW71T, and Bradyrhizobium sp. strain ORS278). For 214 other Bradyrhizobium strains from diverse legumes in eight regions of North and Central America, data for six gene loci were reported in Parker (2012) and Koppell and Parker (2012): Panama (28 strains), Costa Rica (25), Oaxaca (35), Chihuahua (22), south Texas (21), Washington State (25), North Carolina (29), and northeastern U.S. (29). We sequenced the recA marker in all of these strains in order to obtain complete 7-locus data comparable to that for the Guadeloupe strains. To supplement these results, we also obtained complete 7-locus sequence data on 45 additional Bradyrhizobium strains from two other regions (Table S1): Puerto Rico (23 strains) and the Philippines (22 strains; Andam and Parker, 2008). The primary sampling objective was to obtain symbionts from a maximum diversity of legume taxa native to each region. Regions varied in the number of legume species that could be sampled within a single habitat, so the spatial scale of sampling varied across regions (Table 1). To compare nucleotide diversity among eleven regional populations, DnaSP (Librado and Rozas, 2009) was used to estimate pT, the average pairwise number of nucleotide differences per site, for the 7-locus concatenated data set. NeighborNet analysis using SplitsTree v4.11.3 (Huson and Bryant, 2006) indicated that concatenation of the six housekeeping gene loci resulted in a network that was only marginally more reticulated than networks for individual loci. The network for nifD sequences had a largely bifurcating structure with modest reticulation. However, concatenation of nifD with the housekeeping loci resulted in a network with extensive reticulation, indicating that the phylogenetic signal in nifD sequences conflicted greatly with the other loci. To summarize bacterial relationships, we therefore present one tree for concatenated sequences of the six housekeeping loci, and a separate tree for nifD sequences. Trees were inferred using MrBayes (Ronquist and Huelsenbeck, 2003) with protein coding loci partitioned by codon position and separate estimates of rates and nucleotide composition for each codon position and locus. A HKY substitution model was used. Some alternative models were tested but all yielded very similar trees. Analyses were run for 4 million generations, sampling every 400 generations, with the last 1000 samples saved for tree analysis. Replicate runs yielded highly congruent consensus tree topologies. Trees were rooted using Azorhizobium caulinodans ORS571 (NC009937) and Xanthobacter autotrophicus Py2 (NC009720), and Methylobacterium strain 4-46 (CP000943.1) as outgroups. Phylocom 4.0.1 (Webb et al., 2008) was used to analyze whether bacteria were phylogentically clustered according to geographic source region. Clustering analyses were performed separately for the nifD tree and the housekeeping gene tree. For each regional group of bacterial strains, MPDsample (Webb et al., 2008) was calculated as the average across all pairs of strains of the branch length separating them on the phylogenetic tree. Significance was assessed by comparing MPDsample values to the null distribution inferred from 1000 random permutations of isolate names across tips of the observed tree. The Net Relatedness Index
112
M.A. Parker, A. Rousteau / Molecular Phylogenetics and Evolution 77 (2014) 110–115
Table 1 Nucleotide polymorphism and sample attributes. Host Region Guadeloupe Puerto Rico Philippines Panama Costa Rica Oaxaca Texas Chihuahua North Carolina Northeastern U.S. Washington State a b c D e f
a
N 39 23 22 28 25 35 21 22 29 29 25
b
p ± 1 SD 0.085 ± 0.004 0.085 ± 0.004 0.078 ± 0.003 0.085 ± 0.006 0.079 ± 0.006 0.074 ± 0.005 0.079 ± 0.004 0.074 ± 0.005 0.043 ± 0.008 0.060 ± 0.006 0.022 ± 0.008
Species 19 7 3 20 16 19 7 6 17 13 5
Host Genera 18 7 3 16 14 17 7 3 13 8 2
Latitude O
( N) 16.2 18.3 14.1 9.1 9.2 16.4 26.7 27.5 35.5 42.1 47.4
Sampled localitiesc 44 km 38 kmd 15 kmd 5 kme 110 kme 165 kme 180 kme 33 kme 430 kmf 410 kmf 440 kme
Number of isolates studied per region. Mean fraction of nucleotide differences among isolate pairs within a region, across seven gene loci. Maximum distance among sampling localities within region. Table S1. Koppell and Parker (2012). Parker (2012).
(NRI), a standardized measure of the magnitude of clustering, was calculated as the difference between MPDsample and the MPD value for the same sample size calculated from the null distribution (MPDnull), divided by the standard deviation of MPDnull (Webb et al., 2008). GenBank accession numbers for the 833 new sequence records obtained in this study, and for the 1323 other existing sequence records utilized, are provided in Table S2. 3. Results 3.1. Patterns of nucleotide diversity Almost every Guadeloupe Bradyrhizobium isolate had a unique multilocus genotype (haplotype) when data from all seven marker loci were combined. The only exceptions involved six isolates from three legume trees (Abarema jupunba, Inga ingoides, and Ormosia krugii) that all shared a single haplotype, as did two isolates from the tree Pterocarpus officinalis, and two isolates from the tree Swartzia caribaea. Nucleotide diversity in the Guadeloupe Bradyrhizobium sample was among the highest of any of the eleven regional populations analyzed (Table 1). Across all eleven populations, nucleotide diversity was inversely related to source region latitude (Spearman rank correlation, rs = 0.76, P < 0.005). Nucleotide diversity was not significantly correlated with the number of legume host species sampled in a region or the number of legume genera sampled (rs = 0.38 and 0.50, respectively; P > 0.05; this result was not altered when diversity data were analyzed separately for symbiotic [nifD] vs. nonsymbiotic loci). 3.2. Origins of Guadeloupe Bradyrhizobium
Fig. 1. Bayesian tree for six concatenated housekeeping loci (3370 bp) in 308 Bradyrhizobium isolates. Outgroups were deleted before plotting to save space, and strains in terminal clades were grouped (see Supplementary Data Fig. S3 for complete topology, strain identity and clade posterior probability values). Withinclade disparity in branch length is indicated by triangle size. Filled triangles indicate clades with Guadeloupe strains (listed in bold, along with the number and geographic affinities of non-Guadeloupe clade members: Phil = Philippines; Oax = Oaxaca, Mexico; CR = Costa Rica; Pan = Panama; NC = North Carolina; neU.S.=northeastern United States; Chih = Chihuahua, Mexico; TX = Texas, PR = Puerto Rico). The common, broad host-range neotropical lineage described in Parker (2008) is marked with an asterisk.
Bayesian analysis yielded a well-resolved tree for the concatenated housekeeping gene data (Fig. 1; 83% of the clades had a posterior probability of 0.95 or higher). The 308 Bradyrhizobium strains fell into three major clades. The most basally diverging clade (group 1) included the photosynthetic Aeschynomene symbiont ORS278 and three strains from Panama. The remainder of the strains split into two groups that each included several reference taxa: B. diazoefficiens/B. liaoningense/B. yuanmingense/B. canariense/B. cytisi/B. rifense (group 2), and B. elkanii/B. pachyrhizi/B. jicamae (group 3). If two Guadeloupe Bradyrhizobium strains are placed into distinct well-supported clades with different non-Guadeloupe relatives, the most parsimonious inference would often be that their
establishment on Guadeloupe involved two separate colonization events. Thus, an upper limit on colonization events can be estimated by counting how many distinct terminal clades encompassed both Guadeloupe and non-Guadeloupe strains. For each Guadeloupe strain, the smallest terminal clade was identified that included at least one non-Guadeloupe relative and that was supported by a Bayesian posterior probability of 0.95 or higher (Table S3). This analysis partitioned the 39 Guadeloupe isolates into 21 distinct terminal clades. Most terminal clades were very homogeneous. The median nucleotide divergence among strain pairs within a clade was 0.01 (Table S3). These results imply that the Guadeloupe Bradyrhizobium community is descended from
113
M.A. Parker, A. Rousteau / Molecular Phylogenetics and Evolution 77 (2014) 110–115
multiple colonization events, involving ancestors from several source regions. One Guadeloupe strain (deg3.5) proved to be identical at all six housekeeping loci to four Puerto Rican isolates from the same host legume (Dalbergia ecastaphyllum). Such cases of close relationship for bacterial strains from the same legume host in nearby regions are hardly surprising. Seven other legume species were also sampled both in Guadeloupe and elsewhere (Canavalia rosea, Centrosema virginianum, Chamaecrista nictitans, Desmodium axillare, Desmodium incanum, Galactia striata, Macroptilium lathyroides). For two of these legumes (C. rosea, M. lathyroides), certain Bradyrhizobium isolates from Guadeloupe and non-Guadeloupe populations were very similar. However, most of these legume species harbored somewhat divergent symbionts in different regions. In a few cases, bacterial isolates from unrelated host legumes in distant locations exhibited surprising genetic similarity. Four Guadeloupe strains were more similar to certain Philippine Bradyrhizobium strains than to any of the Western Hemisphere strains included in the study (Fig. 1). For each terminal clade having one or more Guadeloupe strains, its geographic affinity was represented by the proportion of included strains from each region (Table S3). Pooled across all 39 Guadeloupe isolates, this analysis indicated that the top two inferred source regions for Guadeloupe bradyrhizobia were Texas (23%) and Oaxaca (16%), followed by Puerto Rico (15%) Costa Rica (14%) and Panama (14%). Only a very few Guadeloupe bradyrhizobia had geographic affinities to samples from temperate North America (Washington State, North Carolina, or the northeastern U.S.). BLAST searches with single genes found only a few cases where Bradyrhizobium strains from indigenous legumes in other regions had higher similarity to Guadeloupe strains than did any of the 269 reference strains in the current sample. Guadeloupe strain dbd20 had a gyrB sequence identical to that of Bradyrhizobium iriomotense (from Okinawa, Japan), and Guadeloupe strain ork1 had a recA sequence that differed at only one nucleotide from that of Bradyrhizobium strain STM7334 from Brazil. Data from only one or two loci were available for these strains, so it was not possible to include them in the concatenated housekeeping gene tree analysis. However, these results suggest that other regions beyond the ten locations of the current study (Table 1) are likely to harbor strains with close affinities to some of the Guadeloupe Bradyrhizobium lineages. Bayesian analysis of nifD sequences yielded a well-resolved tree (Fig. 2; 75% of clades had a posterior probability of 0.95 or higher). However, the nifD tree topology conflicted substantially with the housekeeping gene tree. The most basally diverging strains (group 1) were the same as in the housekeeping gene tree (Fig 1). However, strains belonging to the two other large, well-differentiated housekeeping-gene clades (groups 2, 3) were intermingled in the nifD tree. Phylogenetic incongruence of the two trees is consistent with much other research suggesting that lateral transfer of SIregion loci such as nifD has been a common event in Bradyrhizobium evolution (Moulin et al., 2004; Stepkowski et al., 2012; Parker, 2012). Colonization events were estimated in the same way as for the housekeeping gene tree, by counting the number of distinct terminal nifD clades that encompassed both Guadeloupe and non-Guadeloupe strains (Table S4). This analysis indicated that Guadeloupe nifD sequences may be derived from at most 20 colonization events, which was quite similar to the estimate for the housekeeping gene tree (21). However, the top two inferred source regions for Guadeloupe bradyrhizobial nifD sequences were not the same as for housekeeping genes. Most Guadeloupe nifD sequences had affinities to Bradyrhizobium reference strains from Puerto Rico (47%) and the Philippines (14%). Texas ranked as the third most
Fig. 2. Bayesian tree for nifD (822 bp) in 308 Bradyrhizobium isolates. Outgroups were deleted before plotting to save space, and strains in terminal clades were grouped (see Fig. S4 for complete topology, strain identity and clade posterior probability values). Within-clade disparity in branch length is indicated by triangle size. Filled triangles indicate clades with Guadeloupe strains (listed in bold along with the number and geographic affinities of non-Guadeloupe clade members; location abbreviations as in Fig. 1 legend). nifD clades that were uniform for one of the three major housekeeping gene clades (1, 2 or 3; Fig. 1) are marked with that number at their base.
important source (10%), and all other regions had inferred contributions below 10% (Table 2). Across the two trees, geographic affinities conflicted significantly for seven Guadeloupe strains (deg2.1, deg3.11, dbd20, isg8, pbt1c, rhr14, rhr28, pog17). Each of these strains grouped into a strongly supported clade (Tables S3 and S4) with relatives from different regions in the two trees. These can be interpreted as cases of geographic mosaic ancestry (Andam and Parker, 2008), where lateral gene transfer subsequent to arrival in Guadeloupe may have generated strains with chromosomal portions from more than one source region.
Table 2 Phylogenetic clustering of Bradyrhizobium isolates in relation to geographic source region.
Region: Guadeloupe Puerto Rico Philippines Panama Costa Rica Oaxaca Texas Chihuahua North Carolina Northeastern U.S. Washington State Combined a
Na 39 23 22 28 25 35 21 22 29 29 25 295
nifD tree
6-locus housekeeping gene tree
NRI b 1.65 (ns) 0.80 (ns) 1.80 (ns) 3.41** 2.95** 5.12*** 5.76*** 5.80*** 12.23*** 10.67*** 15.24*** 44.42***
NRI 1.16 (ns) 0.86 (ns) 2.05** 2.13*** 2.51*** 2.24*** 0.13 (ns) 1.47* 5.55*** 2.98*** 7.36*** 24.39***
Number of isolates. NRI = (MPDnull–MPDsample)/(standard deviation of MPDnull), where MPDsample is the mean pairwise patristic distance among strains from one region; positive values of NRI indicate that strains from the same region tend to be phylogenetically close, relative to the null distribution generated by random permutation of strains across the tree. Probability of NRI inferred by comparison to null distribution based on 1000 random permutations of strains across tree branch tips (***P < 0.001; **P < 0.01; * P < 0.05; (ns), P > 0.05). b
114
M.A. Parker, A. Rousteau / Molecular Phylogenetics and Evolution 77 (2014) 110–115
3.3. Differentiation of regional populations Tree permutation analyses (Webb et al., 2008) indicated that there was substantial phylogenetic clustering of bacteria according to geographic source region in both trees (Table 2 and Figs. S1 and S2). Thus, bacteria from most individual locations had genes that were significantly more similar, on average, relative to the null distribution generated by permuting strains across tips on each tree. However, phylogenetic clustering was not significant in either tree for the bacterial populations from Guadeloupe or Puerto Rico (Table 2). These two populations (along with Texas in the housekeeping gene tree, and the Philippines for the nifD tree) had so many diverse lineages that pairwise genetic distance for these samples was not distinguishable from the null distribution representing strain pairs selected at random across the entire sample. For the Guadeloupe sample, the lack of phylogenetic clustering (Table 2) is consistent with the terminal clade analysis (Figs. 1 and 2; Tables S3 and S4) showing that this bacterial population is a composite of lineages drawn from many source locations.
4. Discussion Our main finding was that the Bradyrhizobium community of Guadeloupe is strikingly diverse, and originated from numerous colonists from multiple source regions. Despite the relatively small size of this island and its modest age, it harbored bacterial symbionts with a genetic diversity that equaled or exceeded that found in older and larger continental areas (Table 1). Much work remains to be done in characterizing potential source regions for Caribbean populations. In particular, there has been no comprehensive sampling in northern South America, a likely source for symbiotic bacteria now resident in Guadeloupe, given its proximity and the fact that it shares many legume host species (Table S1). Although more detailed sampling of continental habitats may modify inferences about geographic affinity for particular clades, it is not likely to alter the basic conclusion that Guadeloupe bacteria are derived from many Bradyrhizobium clades from diverse localities. Our results indicate that evolution of the Guadeloupe Bradyrhizobium population has been affected not only by recurrent migration but also by substantial horizontal gene transfer for a SI region locus (nifD). The impact of SI region horizontal transfer on community assembly merits more detailed study. Loci in the SI region appear to be important in determining which legume hosts a bacterial strain can successfully interact with (Parker, 2012). A bacterial strain may be perfectly well adapted to physical features of an island habitat, but there is no guarantee that its SI region will be optimally suited to the particular legumes that are abundant in any given site. Horizontal transfer of the SI region is a process that may facilitate adaptation in a novel ecosystem, by reassembling traits from different ancestral environments. This likely contributed to the high overall genetic diversity detected among Guadeloupe Bradyrhizobium (Table 1). Mechanisms enabling long-distance migration in Bradyrhizobium are not well understood. These bacteria have no resistant spore stage readily transported in the atmosphere, and are not particularly salt-tolerant, as would be needed to reach an island on a floating raft of vegetation debris. Raza et al. (2001) reported that strains of Bradyrhizobium varied in salinity tolerance, but for even the most tolerant strains, only about one out of a thousand cells survived a week of exposure to the salinity of seawater. However, Bradyrhizobium populations are known to have the capacity to survive in the soil for decades in the absence of legumes hosts (Norman, 1942). Thus, any mechanism that could transport soil particles (e.g., a bird’s muddy feet, prevailing wind currents (Yamaguchi et al., 2012), or extreme weather events such as
hurricanes) could potentially introduce rhizobia into an island habitat. It is still difficult to account for the finding of some closely similar strain pairs in Guadeloupe and the Philippines, which are half a world apart (16,600 km). Both samples came from indigenous legumes in forested non-agricultural habitats, so there is no obvious explanation involving human agency. An earlier study also detected a few cases of close genetic similarity between Bradyrhizobium strains in the Philippines and in eastern North America (Andam and Parker, 2008). It remains an open problem to explain why certain Bradyrhizobium lineages have achieved such wide geographic ranges (and why others have not). In principle, a clade that is distributed both in Guadeloupe and another location could be explained by a migration event in either direction. However, the small area of Guadeloupe means that it is much more likely to be a recipient than a source of migrants. For example, suppose that the probability of microbial transfer between Guadeloupe and the much larger island of Puerto Rico depended strictly on their relative areas (1440 km2 vs. 9104 km2), and not on other factors such as the prevailing direction of hurricane storm tracks. In this case, a six-fold asymmetry in migrant exchange would be expected. To the extent that Bradyrhizobium lineages in other locations came from Guadeloupe ancestors, the method used in Tables S3 and S4 will overestimate migrant arrival into Guadeloupe. Thus, a count of terminal clades including both Guadeloupe and non-Guadeloupe strains (20–21) should be viewed simply as an estimated upper bound on the probable number of colonization events that populated this island. Three of the Guadeloupe legumes are trees that belong to genera whose symbionts have not been previously characterized in any detail: Abarema (tribe Ingeae in the subfamily Mimosoideae), Swartzia (tribe Swartzieae) and Ormosia (tribe Sophoreae), both in the subfamily Papilionoideae (Lewis et al., 2005). All three of these trees utilized Bradyrhizobium strains related to a common lineage (marked with an asterisk in Fig. 1) that is widespread in central America and that apparently has a very broad host range (Parker, 2008). In Panama and Costa Rica, this lineage has been sampled from fifteen legume genera of diverse life form (herbs, lianas, and trees). In the current sample, it was also detected in Puerto Rico on an additional genus of legume trees (Andira; Table S1). Our results thus provide further data suggesting that some Bradyrhizobium lineages that lack narrow host specificity may be widely distributed in the Neotropics. This lineage is relatively homogeneous at housekeeping loci (mean pairwise nucleotide divergence (p) ±1 SD = 0.013 ± 0.002). However, strains within this group show much more diversity for the nifD locus in the SI region (p = 0.061 ± 0.010), perhaps indicative of diversifying selection associated with varied host utilization (Parker, 2012). Current sampling is far too incomplete to know whether these patterns are broadly characteristic of legume-bacterial relationships in this region. In the future, wider analysis of the striking diversity of Neotropical legume taxa will help to resolve how Bradyrhizobium lineages have spread geographically and across host plant clades. Acknowledgments We are grateful to C. Andam, F. Muller, R. Rivas, E. Velazquez and P. Vinuesa for help with collecting or for providing strains. Financial support was provided by the National Science Foundation grant MCB-0640246. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ympev.2014. 04.011.
M.A. Parker, A. Rousteau / Molecular Phylogenetics and Evolution 77 (2014) 110–115
References Andam, C.C., Parker, M.A., 2008. Origins of Bradyrhizobium nodule symbionts from two legume trees in the Philippines. J. Biogeography 35, 1030–1039. Aserse, A.A., Räsänen, L.A., Aseffa, F., Hailemariam, A., Lindström, K., 2012. Phylogenetically diverse groups of Bradyrhizobium isolated from nodules of Crotalaria spp., Indigofera spp., Erythrina brucei and Glycine max growing in Ethiopia. Mol. Phylogenet. Evol. 65, 595–609. Delamuta, J.R.M., Ribeiro, R.A., Ormeño-Orrillo, E., Melo, I.S., Mart´ınez-Romero, E., Hungria, M., 2013. Polyphasic evidence supporting the reclassification of Bradyrhizobium japonicum group Ia strains as Bradyrhizobium diazoefficiens sp. nov. Int. J. Syst. Evol. Microbiol. 63, 3342–3351. Feuillet, N., Papponnier, P., Manighetti, I., Villemant, B., King, G.C.P., 2004. Differential uplift and tilt of pleistocene reef platforms and quaternary slip rate on the Morne-Piton normal fault (Guadeloupe, French West Indies). J. Geophysical. Res. 109, B02404. Fournet, J., 2002. Flore illustrée des phanérogames de Guadeloupe et de Martinique. CIRAD, Gondwana éditions, La Trinite, Martinique. Guerrouj, K., Ruíz-Díez, B., Chahboune, R., Ramírez-Bahena, M.H., Abdelmoumen, H., Quiñones, M.A., El Idrissi, M.M., Velázquez, E., Fernández-Pascual, M., Bedmar, E.J., Peix, A., 2013. Definition of a novel symbiovar (sv. retamae) within Bradyrhizobium retamae sp. nov., nodulating Retama sphaerocarpa and Retama monosperma. Syst. Appl. Microbiol. 36, 218–223. Gyaneshwar, P., Hirsch, A.M., Moulin, L., Chen, W.-M., Elliot, G.N., Bontemps, C., Estrada-de los Santos, P., Gross, E., dos Reis, F.B., Sprent, J.I., Young, J.P., James, E.K., 2011. Legume-nodulating betaproteobacteria: diversity, host range, and future prospects. Mol. Plant-Microbe Int. 24, 1276–1288. Horn, K., Parker, I., Małek, W., Rodríguez-Echeverría, S., Parker, M.A., 2014. Disparate origins of Bradyrhizobium symbionts for invasive populations of Cytisus scoparius (Leguminosae) in North America. FEMS Microbiol. Ecol. (in press). http://dx.doi.org/10.1111/1574-6941.12335. Howard, R.A., 1988. Flora of the Lesser Antilles, Dicotyledoneae, Part 1. Arnold Arboretum of Harvard University, Jamaica Plain, Massachusetts Huson, D.H., Bryant, D., 2006. Application of phylogenetic networks in evolutionary studies. Mol. Biol. Evol. 23, 254–267. Iturralde-Vinent, M.A., MacPhee, R.D.E., 1999. Paleogeography of the Caribbean region: implications for Cenozoic biogeography. Bull. Am. Mus. Nat. Hist. 238, 1–95. Kaneko, T., Nakamura, Y., Sato, S., et al., 2002. Complete genomic sequence of nitrogen-fixing symbiotic bacterium Bradyrhizobium japonicum USDA110. DNA Res. 9, 189–197. Koppell, J.H., Parker, M.A., 2012. Phylogenetic clustering of Bradyrhizobium symbionts on legumes indigenous to North America. Microbiology 158, 2050– 2059. Lewis, G., Schrire, B., Mackinder, B., Lock, M., 2005. Legumes of the world. Royal Botanic Gardens, Kew, UK. Librado, P., Rozas, J., 2009. DnaSP v5: A software package for comprehensive analysis of DNA polymorphism data. Bioinformatics 25, 1451–1452. Moulin, L.G., Bena, G., Boivin-Masson, C., Stepkowski, T., 2004. Phylogenetic analyses of symbiotic nodulation genes support vertical and lateral gene cotransfer within the Bradyrhizobium genus. Mol. Phylogenet. Evol. 30, 720–732. Leary, J.K., Singleton, P.W., Borthakur, D., 2004. Canopy nodulation of the endemic tree legume Acacia koa in the mesic forests of Hawaii. Ecology 85, 3151–3157. Morton, J.F., 1989. Creeping indigo (Indofera spicata Forsk.) (Fabaceae) – a hazard to herbivores in Florida. Econ. Bot. 43, 314–327.
115
Nandasena, K.G., O’Hara, G.W., Tiwari, R.P., Howieson, J.G., 2006. Rapid in situ evolution of nodulating strains for Biserrula pelecinus L. through lateral transfer of a symbiosis island from the original mesorhizobial inoculant. Appl. Env. Microbiol. 72, 7365–7367. Norman, A.G., 1942. Persistence of Rhizobium japonicum in soil. Amer. Soc. Agron. J. 34, 499. Nzoue, A., Miche, L., Klonowska, A., Laguerre, G., de Lajudie, P., Moulin, L., 2009. Multilocus sequence analysis of bradyrhizobia isolated from Aeschynomene species in Senegal. Syst. Appl. Microbiol. 32, 400–412. Parker, M.A., 2001. Mutualism as a constraint on invasion success for legumes and rhizobia. Diversity Distrib. 7, 125–136. Parker, M.A., 2008. Symbiotic relationships of legumes and nodule bacteria on Barro Colorado Island, Panama: a review. Microbial Ecol. 55, 662–672. Parker, M.A., 2012. Legumes select symbiosis island sequence variants in Bradyrhizobium. Molec. Ecol. 21, 1769–1778. Parker, M.A., Lafay, B., Burdon, J.J., van Berkum, P., 2002. Conflicting phylogeographic patterns in rRNA and nifD indicate regionally restricted gene transfer in Bradyrhizobium. Microbiology 148, 2557–2565. Parker, M.A., Malek, W., Parker, I.M., 2006. Growth of an invasive legume is symbiont-limited in newly occupied habitats. Div. Distrib. 12, 563–571. Raza, S., Jørnsgård, B., Abou-Taleb, H., Christiansen, J.L., 2001. Tolerance of Bradyrhizobium sp. (Lupini) strains to salinity, pH, CaCO3 and antibiotics. Lett. Appl. Microbiol. 32, 379–383. Ronquist, F., Huelsenbeck, J.P., 2003. MRBAYES 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19, 1572–1574. Samper, A., Quidelleur, X., Lahitte, P., Mollex, D., 2007. Timing of effusive volcanism and collapse events within an oceanic arc island: Basse-Terre. Guadeloupe archipelago. Earth Planet. Sci. Let. 258, 175–191. Saur, E., Carcelle, S., Guezennec, S., Rousteau, A., 2000. Nodulation of legume species in wetlands of Guadeloupe (Lesser Antilles). Wetlands 20, 730–734. Stepkowski, T., Hughes, C.E., Law, I.J., Markiewicz, L., Gurda, D., Chlebicka, A., Moulin, L., 2007. Diversification of lupine Bradyrhizobium strains: evidence from nodulation gene trees. Appl. Environ. Microbiol. 73, 3254–3264. Stepkowski, T., Watkin, E., McInnes, A., Gurda, D., Gracz, J., Steenkamp, E.T., 2012. Distinct Bradyrhizobium communities nodulate legumes native to temperate and tropical monsoon Australia. Mol. Phylogenet. Evol. 63, 265–277. Sullivan, J.T., Patrick, H.N., Lowther, W.L., Scott, D.B., Ronson, C.W., 1995. Nodulating strains of Rhizobium loti arise through chromosomal symbiotic gene transfer in the environment. Proc. Natl. Acad. Sci. USA 92, 8985–8989. Vinuesa, P., Leon-Barrios, M., Silva, C., Willems, A., Jarabo-Lorenzo, A., PerezGaldona, R., Werner, D., Martınez-Romero, E., 2005. Bradyrhizobium canariense sp. nov., an acid-tolerant endosymbiont from the nodules of endemic woody legume species (Papilionoideae: Genisteae) growing in the Canary Islands along with B. japonicum bv. genistearum, Bradyrhizobium genospecies a and Bradyrhizobium genospecies b. Int. J. Syst. Evol. Microbiol. 55, 569–575. Webb, C.O., Ackerly, D.D., Kembel, S.W., 2008. Phylocom: software for the analysis of phylogenetic community structure and trait evolution. Bioinformatics 24, 2098–2100. Wei, G., Chen, W., Zhu, W., Chen, C., Young, J.P.W., Bontemps, C., 2009. Invasive Robinia pseudoacacia in China is nodulated by Mesorhizobium and Sinorhizobium species that share similar nodulation genes with native American symbionts. FEMS Microbiol. Ecol. 68, 320–328. Yamaguchi, N., Ichijo, T., Sakotani, A., Baba, T., Nasu, M., 2012. Global dispersion of bacterial cells on Asian dust. Sci. Rep. 2, 525; DOI:10.1038/srep00525.
