Phylogenomic Study of Burkholderia glathei-like Organisms, Proposal of 13 Novel Burkholderia Species and Emended Descriptions of Burkholderia sordidicola, Burkholderia zhejiangensis, and Burkholderia grimmiae

Partial gyrB gene sequence analysis of 17 isolates from human and environmental sources revealed 13 clusters of strains and identified them as Burkholderia glathei clade (BGC) bacteria. The taxonomic status of these clusters was examined by whole-genome sequence analysis, determination of the G+C content, whole-cell fatty acid analysis and biochemical characterization. The whole-genome sequence-based phylogeny was assessed using the Genome Blast Distance Phylogeny (GBDP) method and an extended multilocus sequence analysis (MLSA) approach. The results demonstrated that these 17 BGC isolates represented 13 novel Burkholderia species that could be distinguished by both genotypic and phenotypic characteristics. BGC strains exhibited a broad metabolic versatility and developed beneficial, symbiotic, and pathogenic interactions with different hosts. Our data also confirmed that there is no phylogenetic subdivision in the genus Burkholderia that distinguishes beneficial from pathogenic strains. We therefore propose to formally classify the 13 novel BGC Burkholderia species as Burkholderia arvi sp. nov. (type strain LMG 29317T = CCUG 68412T), Burkholderia hypogeia sp. nov. (type strain LMG 29322T = CCUG 68407T), Burkholderia ptereochthonis sp. nov. (type strain LMG 29326T = CCUG 68403T), Burkholderia glebae sp. nov. (type strain LMG 29325T = CCUG 68404T), Burkholderia pedi sp. nov. (type strain LMG 29323T = CCUG 68406T), Burkholderia arationis sp. nov. (type strain LMG 29324T = CCUG 68405T), Burkholderia fortuita sp. nov. (type strain LMG 29320T = CCUG 68409T), Burkholderia temeraria sp. nov. (type strain LMG 29319T = CCUG 68410T), Burkholderia calidae sp. nov. (type strain LMG 29321T = CCUG 68408T), Burkholderia concitans sp. nov. (type strain LMG 29315T = CCUG 68414T), Burkholderia turbans sp. nov. (type strain LMG 29316T = CCUG 68413T), Burkholderia catudaia sp. nov. (type strain LMG 29318T = CCUG 68411T) and Burkholderia peredens sp. nov. (type strain LMG 29314T = CCUG 68415T). Furthermore, we present emended descriptions of the species Burkholderia sordidicola, Burkholderia zhejiangensis and Burkholderia grimmiae. The GenBank/EMBL/DDBJ accession numbers for the 16S rRNA and gyrB gene sequences determined in this study are LT158612-LT158624 and LT158625-LT158641, respectively.


INTRODUCTION
The genus Burkholderia currently comprises 90 validly named species (Euzeby, 1997) and several uncultured Candidatus species (Van Oevelen et al., 2004;Verstraete et al., 2011;Lemaire et al., 2012) which occupy very diverse niches (Coenye and Vandamme, 2003). Many Burkholderia species have thus far only been isolated as free-living organisms but a growing body of literature reveals that they live in close interaction with numerous plant, animal, fungal or even amoebozoan hosts (Marolda et al., 1999;Van Borm et al., 2002;Kikuchi et al., 2011;Verstraete et al., 2013;Stopnisek et al., 2016;Xu et al., 2016). Burkholderia species may be beneficial to their hosts because some strains can fix nitrogen, produce plant hormones or siderophores, or lower pathogen-related ethylene levels; hence they have been exploited for plant growth promotion and biocontrol of plant diseases (Compant et al., 2008;Vial et al., 2011). Yet, other Burkholderia species are notorious pathogens in plants, animals and humans (Mahenthiralingam et al., 2008). This ecological diversity is likely attributed to their large, multireplicon genomes (typically between 6 and 9 Mb) which also confer a metabolic versatility allowing them to degrade a wide range of recalcitrant xenobiotics (Parke and Gurian-Sherman, 2001;Coenye and Vandamme, 2003).
The present study aimed to perform a phylogenomic study of established and novel species in the B. glathei clade, to formally name the latter and to make reference cultures and whole-genome sequences of each of these versatile bacteria publicly available. The genome sequence-based phylogeny was assessed using the Genome Blast Distance Phylogeny (GBDP) method (Meier-Kolthoff et al., 2013) and an extended multilocus sequence analysis (MLSA) approach. For phenotypic characterization, whole-cell fatty acid profiling and biochemical analyses were performed. Table 1 lists the sources of the 17 studied isolates. Details of type strains of each of the present BGC species were described previously (Zolg and Ottow, 1975;Lim et al., 2003;Lu et al., 2012;Tian et al., 2013;Vandamme et al., 2013;Draghi et al., 2014;Liu et al., 2014;Baek et al., 2015). Strains were grown aerobically on buffered nutrient agar (Oxoid, pH 6.8) and incubated at 28 • C. Cultures were preserved in MicroBank TM vials at −80 • C.

16S rRNA Gene Sequence Analysis
Nearly complete sequences were obtained as described previously (Peeters et al., 2013). gyrB Gene Sequence Analysis Partial gyrB gene sequences were obtained as described previously (Spilker et al., 2009;Peeters et al., 2013). Sequence assembly was performed using BioNumerics v7.5 (Applied Maths). Sequences (589-1182 bp) were aligned based on amino acid sequences using Muscle (Edgar, 2004) in MEGA6 (Tamura et al., 2013). All positions with less than 95% site coverage were eliminated, resulting in a total of 570 positions in the final dataset. Phylogenetic analysis was conducted in MEGA6 (Tamura et al., 2013).

Whole-Genome Sequencing
Genomic DNA of 20 strains (Table 2) was prepared as described by Pitcher et al. (1989). Genomic libraries were prepared using the Nextera kit following the methods introduced by Baym et al. (2015) and the 151 bp paired-end libraries were sequenced on the Illumina HiSeq platform of the University of New Hampshire Hubbard Center for Genomics Studies Burkholderia peredens sp. nov.

Publicly Available Genomes
Twenty three publicly available whole-genome sequences of BGC bacteria were downloaded from the NCBI database (Table 2).
B. gladioli BSR3 (Seo et al., 2011) was used as an outgroup in all phylogenomic analyses. For B. megalochromosomata JC2949 T the whole-genome sequence was not publicly available (February 1st, 2016) and the contig sequences were provided by J. Chun (Baek et al., 2015). For B. sordidicola S170, B. zhejiangensis CEIB S4-3 and B. megalochromosomata JC2949 T no annotation was available and annotation was performed using Prokka as described above.

Phylogenomic Analysis
The latest version of the Genome Blast Distance Phylogeny (GBDP) approach was applied (Meier-Kolthoff et al., 2013) to calculate the intergenomic distance between each pair of genomes (based on the nucleotide data) and included the calculation of 100 replicate distances to assess pseudo-bootstrap support (Meier-Kolthoff et al., 2014a). Distance calculations were conducted under the recommended settings of the Genome-to-Genome Distance Calculator (GGDC 2.1; http://ggdc.dsmz.de), as described earlier (Meier-Kolthoff et al., 2013). The GBDP trimming algorithm and formula d 5 were chosen because of  (Baek et al., 2015).
their advantages for phylogenetic inference (Meier-Kolthoff et al., 2014a) and according distance matrices were prepared (a single matrix for the original distances plus 100 matrices containing the replicates). A phylogenomic tree with branch support (Meier-Kolthoff et al., 2014a) was inferred using FastME v2.07 with tree bisection and reconnection post-processing (Lefort et al., 2015).
Moreover, pairwise digital DNA-DNA hybridization (dDDH) values and their confidence intervals were also determined using GGDC 2.1 under recommended settings (Meier-Kolthoff et al., 2013). The potential affiliation of the novel strains to existing species was determined by clustering using a 70% dDDH radius around each of the 12 BGC type strains as previously applied (Liu Frontiers , 2015). Visualization and annotation of the phylogenetic tree was performed using iTOL (Letunic and Bork, 2011). As an alternative for the GBDP method, an extended MLSA analysis was performed in which a whole-genome phylogeny was calculated based on single-copy orthologous genes as described previously (Pinto-Carbo et al., 2016). In short, singlecopy orthologs were identified using blastp and OrthoMCL v2.0.9 (with e-value cutoff 1e10 −6 and 50% match cutoff; Fischer et al., 2011) and aligned based on their amino acid sequences using MUSCLE. The alignments were trimmed using TrimAl (removing positions with gaps in more than 50% of the sequences) and concatenated to construct a Maximum Likelihood tree using RaXML v7.4.2 (Stamatakis, 2014) with the WAG amino acid substitution model and 100 rapid bootstrap analyses.

Phenotypic Characterization
Phenotypic and cellular fatty acid analyses were performed as described previously (Draghi et al., 2014).

16S rRNA Gene Sequence Analysis
The 16S rRNA gene sequences determined in the present study are publicly available through the GenBank/EMBL/DDBJ accession numbers LT158612-LT158624.
gyrB Gene Sequence Analysis Partial gyrB gene sequences were compared to those of the type strains of the 12 validly named BGC species (Figure 1). The 17 unclassified isolates represented 13 taxa which showed 83.4-96.2% pairwise identity with the gyrB sequences of the type strains of other BGC species. The gyrB gene sequences determined in the present study are publicly available through the GenBank/EMBL/DDBJ accession numbers LT158625-LT158641.

Whole-Genome Sequencing
To further characterize the taxonomic status of these 13 taxa, we determined the whole-genome sequence of one strain per gyrB cluster and of B. sordidicola LMG 22029 T , B. choica LMG 22940 T , FIGURE 1 | Phylogenetic tree based on partial gyrB sequences of the 17 isolates in this study and type strains of phylogenetically related Burkholderia species. The optimal tree (highest log likelihood) was constructed using the Maximum Likelihood method and General Time Reversible model in MEGA6 (Tamura et al., 2013). A discrete Gamma distribution was used to model evolutionary rate differences among sites [5 categories (+G, parameter = 0.5462)] and allowed for some sites to be evolutionarily invariable ([+I], 37.9331% sites). The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) are shown next to the branches if greater than 50%. For B. megalochromosomata JC2949 T the gyrB gene sequence was extracted from the genome sequence. The gyrB sequence of B. kururiensis LMG 19447 T was used as outgroup. The scale bar indicates the number of substitutions per site.
Frontiers in Microbiology | www.frontiersin.org B. humi LMG 22934 T , B. telluris LMG 22936 T , B. terrestris LMG 22937 T , B. udeis LMG 27134 T , and B. cordobensis LMG 27620 T . The assembly of the Illumina HiSeq 150 bp paired end reads resulted in assemblies with 47-657 contigs and a total of 6,166,171-10,051,569 bp ( Table 2). The annotated assemblies of these 20 genomes were submitted to the European Nucleotide Archive and are publicly available through the GenBank/EMBL/DDBJ accession numbers listed in Table 2 and the species descriptions. The genome sequences of the remaining five BGC type strains and of 18 additional strains were publicly available ( Table 2).

DNA Base Composition
The G+C content of all type strains was calculated from their genome sequences and ranged from 62.4 to 64.2 mol% ( Table 2).

Phylogenomic Analysis
The pairwise intergenomic distances and dDDH estimates of the 44 genome sequences are listed in Supplementary Table 1. The phylogenetic tree inferred from the intergenomic distances (Figure 2) was well resolved and most branches showed a very high bootstrap support (average support: 94.8%). Species delineation based on the pairwise dDDH values and a 70% dDDH radius around each type strain yielded 39 species which included the present 12 validly named species as well as the 13 novel species delineated by means of partial gyrB gene sequences (Figure 1).
For the extended MLSA approach, we identified 332 single-copy orthologs that were present in all 44 genomes. The Maximum-Likelihood phylogenetic tree based on the concatenated amino acid alignment (Figure 3) was well resolved and showed a high bootstrap support on almost all branches.
The topologies of the two phylogenomic trees (Figures 2, 3) were very similar and both revealed six clusters of species (A-F). The main difference in tree topology related to the phylogenetic position of the Candidatus species in cluster C. This cluster was supported by a 100% bootstrap value in both analyses but its relative position to cluster D species differed in the two trees (Figures 2, 3). Additionally, the internal branching order of cluster C, E and F species differed minimally between both analyses. Both phylogenomic analyses showed that strain MR1 clustered with B. concitans sp. nov. and that strain RPE67 clustered with B. cordobensis. Finally, the large distances between strains PML1(12) and S170, and the type strains of B. glathei and B. sordidicola, respectively, indicated that both strains were misidentified and wrongly annotated in the NCBI database as B. glathei and B. sordidicola, respectively (Figures 2, 3). Both strains FIGURE 2 | Whole-genome sequence based phylogenomic tree of all BGC genomes inferred by GBDP. The outer column shows the isolation source of the strains. Pseudo-bootstrap support values above 60% are shown. The tree reveals a high average support of 94.8%. Long terminal branches are due to the distinct scaling used by GBDP's formula d 5 . B. gladioli BSR3 was used as outgroup. Red capital letters define subtrees that also occur in the tree depicted in Figure 3. FIGURE 3 | Whole-genome phylogeny based on single-copy orthologs of all BGC genomes. The phylogenetic tree was constructed using the WAG protein substitution model and RAxML and is based on an amino acid alignment with 105,141 positions from 332 single-copy orthologous genes. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (100 replicates) are shown next to the branches if greater than 60%. B. gladioli BSR3 was used as outgroup. Red capital letters define subtrees that also occur in the tree depicted in Figure 2. occupy unique positions in the phylogenomic trees and represent additional novel BGC species.

Cellular Fatty Acid Analysis
The fatty acid profiles of all strains are shown in Table 3. Branched chain fatty acids have not been reported in members of the genus Burkholderia and therefore summed features 2 and 3 very likely represent C 14:0 3-OH and C 16:1 ω7c, respectively (Yabuuchi et al., 1992). The main fatty acid components are C 16:0 , C 18:1 ω7c and summed feature 3 (most probably representing C 16:1 ω7c).

Biochemical Characterization
An overview of biochemical characteristics useful for distinguishing the BGC species is shown in Table 4.

DISCUSSION
While soil is a well-known source of free-living Burkholderia species, these organisms often live in close interaction with plants, animals, fungi, or amoebae (Marolda et al., 1999;Van Borm et al., 2002;Kikuchi et al., 2011;Verstraete et al., 2013;Stopnisek et al., 2016;Xu et al., 2016). The BGC represents a poorly known line of descent within the genus Burkholderia and most of the 12 validly named BGC species have been isolated from soil. Yet, publicly available sequence data indicate that the taxonomic diversity in this clade is severely underestimated (Nogales et al., 2001;Salles et al., 2006;Pumphrey and Madsen, 2008;Draghi et al., 2014;Verstraete et al., 2014;Peeters et al., 2016;Xu et al., 2016). In the present study, gyrB gene sequence analysis was used to screen our strain collection and 17 isolates from human and environmental samples were identified as B. glathei-like bacteria. The gyrB sequence similarity levels toward other BGC species suggested that the 17 isolates

HYDROLYSIS OF
Tween 60   Vandamme et al. (2013). Data for B. grimmiae were extracted from Tian et al. (2013). Data for B. cordobensis and B. zhejiangensis were extracted from Draghi et al. (2014). All other data are from the present study. Test results of the type strains are given first, followed by the remaining strains in the order given above. +, present; −, absent; w, weak reaction; v, variable; ND, not determined; NG, no growth. in this study represented 13 novel species (Figure 1). To further characterize the taxonomic status of these isolates, we analyzed the genome sequence of 13 isolates representative for the 13 gyrB sequence clusters and of 7 BGC type strains and compared those to 23 whole-genome sequences of BGC strains that were publicly available. Additionally, we also studied their chemotaxonomic and biochemical properties to comply with the polyphasic taxonomic consensus approach to bacterial systematics (Vandamme et al., 1996). In this genomics era, state-of-the-art sequencing technologies enable direct access to the information contained in whole-genome sequences and it is no longer adequate to deduce genome relatedness through traditional DNA-DNA hybridization experiments Whitman, 2015). Genomic taxonomy can be studied through various parameters including average nucleotide identity (ANI), GBDP, Maximal Unique Matches index (MUMi), and core gene identity (CGI) (Konstantinidis and Tiedje, 2005;Goris et al., 2007;Deloger et al., 2009;Vanlaere et al., 2009;Meier-Kolthoff et al., 2013). Although, there is a general consensus that genome sequencing could revolutionize prokaryotic systematics (Sutcliffe et al., 2013;Meier-Kolthoff et al., 2014b;  Rossello-Mora and Amann, 2015; Thompson et al., 2015), traditional DDH experiments are still being performed and new genome-based methods are evaluated in terms of their correspondence to the existing classifications which are based on DDH data (Wayne et al., 1987;Stackebrandt et al., 2002). The GGDC implementation of the GBDP method provides a quick and reliable alternative to the wet-lab DDH technique and its dDDH prediction capability (including confidence intervals) produces classifications which correlate better with the traditional DDH values than do any of the ANI implementations (Meier-Kolthoff et al., 2013). Among several advantages, GBDP is independent from genome annotation, is applicable to both nucleotide and amino acid data and is immune against problems caused by incompletely sequenced or low-quality draft genomes.
Finally, GBDP provides branch support values for the resulting phylogenetic trees (Meier-Kolthoff et al., 2013, 2014a. We complemented the results of the GBDP analysis with a whole-genome-based phylogeny based on the sequence analysis of 332 single-copy orthologous genes in all BGC genomes. This extended MLSA approach takes only the coding part of the genomes into account and is therefore not influenced by noncoding sequences or pseudogenes that might have a different evolutionary history than the rest of the genome. It depends however on genome annotation, is unable to cope with problems caused by incompletely sequenced or low-quality draft genomes, and its calculations are more compute-intensive and cannot be carried out incrementally. Although, the GBDP and extended MLSA methods used different algorithms, the conclusions drawn from their phylogenies were consistent thus illustrating the robustness of whole-genome based taxonomic methods (Colston et al., 2014).
The GGDC dDDH values and the application of the 70% dDDH cut-off for species delineation (Supplementary Table 1) demonstrated that the 13 clusters delineated through gyrB sequence analysis (Figure 1) represented 13 novel BGC species and thus confirmed that gyrB gene sequence analysis is a reliable tool for the identification of Burkholderia species (Tayeb et al., 2008;Vandamme et al., 2013). Both phylogenomic analyses identified strain MR1, which was isolated from Florida golf course soil and which was shown to reduce the herbicide methylarsenate, as B. concitans sp. nov. Next to strain YI23, which was previously identified as B. cordobensis by Draghi et al. (2014), the present study also identified strain RPE67, which was isolated from the gut of a stink bug, as B. cordobensis. Finally, both phylogenomic analyses also showed that strain PML1 (12), an ectomycorrhizosphere-inhabiting bacterium with mineralweathering ability (Uroz and Oger, 2015), strain S170, a potential plant growth promoter isolated from coniferous forest soil (Llado et al., 2014), strain RPE64, a bacterial symbiont of the bean bug Riptortus pedestris (Shibata et al., 2013) and strain Leaf177, an Arabidopsis leaf isolate (Bai et al., 2015) all represent novel BGC species.
Burkholderia genomes vary in size from 3.75 Mb (B. rhizoxinica HKI 454) to 11.3 Mb (B. terrae BS001), are characterized by a high G+C content (60-68%) and consist of multiple replicons (Winsor et al., 2008;Ussery et al., 2009). The DNA G+C content of the 13 novel species was calculated from their genome sequences and was in the range of that reported for other BGC species (60-65 mol%). For 10 of the 12 established BGC species, the G+C content was previously calculated by traditional wet-lab methods and the reported values differed by 0.1-3.3 mol% from the values calculated from their genome sequences ( Table 5). As reported by Meier-Kolthoff et al., the G+C content calculations based on genome sequences show a higher precision than calculations based on traditional wet-lab methods because the latter methods do not count nucleotides but estimate the genomic G+C content based on the physical properties of the extracted and/or digested genomic DNA (Meier-Kolthoff et al., 2014b). The difference between literature data (Lim et al., 2003;Lu et al., 2012;Tian et al., 2013) and the genome sequence-based G+C content values of B. sordidicola LMG 22029 T , B. zhejiangensis OP-1 T and B. grimmiae R27 T is larger than 1% and we therefore present emended descriptions of these species. The genome sizes of the type strains of the 13 novel species ranged from 6.2 Mb (B. concitans sp. nov. LMG 29315 T ) to 9.7 Mb (B. arvi sp. nov. LMG 29317 T ) and corresponded with the genome sizes of other free-living BGC species (Table 2). Consistent with reductive genome evolution in obligatory symbionts, the smallest BGC genomes belong to the obligatory leaf endosymbionts (2.4-6.2 Mb; Carlier and Eberl, 2012;Carlier et al., 2015;Pinto-Carbo et al., 2016).
Biochemically, these novel species are similar to their nearest neighbors. However, tests particularly useful for distinguishing BGC species are growth at 37 • C and at pH 8, hydrolysis of tween 60 and 80, nitrate reduction, assimilation of arabinose, caprate and citrate, beta-galactosidase activity and C4 lipase ( Table 4). The most discriminating fatty acids are C 16:0 3-OH, C 17:0 cyclo, C 19:0 cyclo ω8c and summed features 2 and 3 ( Table 3). The overall fatty acid profiles of the novel taxa are similar to those of their nearest neighbors and support their placement in the genus Burkholderia (Yabuuchi et al., 1992).
The present study again underscores the multifaceted nature of Burkholderia bacteria (Coenye and Vandamme, 2003;Mahenthiralingam et al., 2005) and highlights that also BGC species have evolved a broad range of interactions with different hosts. B. cordobensis is a striking example of phenotypic and geographic breadth: it was recovered from agricultural soil in Argentina (strain LMG 27620 T ) (Draghi et al., 2014), from golf course soil in South Korea (strain YI23) (Lim et al., 2012) and from the gut of the bean bug Riptortus pedestris in Japan (strain RPE67) (Takeshita et al., 2014). The two latter strains (YI23 and RPE67) have fenitrothion degrading properties. The former two strains (LMG 27620 T and YI23) were free-living but the latter (RPE67) is an endosymbiont of stink bugs that is not vertically transmitted but acquired from soil by the nymphal insect (Kikuchi et al., 2007). The insecticide resistance to fenitrothion in the pest insects was shown to be established by the endosymbiotic Burkholderia strain in the insect gut (Kikuchi et al., 2012) and was shown to emerge as a consequence of repeated insecticide use (Tago et al., 2015). The Riptortus pedestris-B. cordobensis association thus appears to be a rather young endosymbiosis and contrasts with the symbiosis observed between plant species of the Rubiaceae and Primulaceae families and several Candidatus Burkholderia species. The Candidatus designation is a provisional taxonomic status for organisms that have been characterized but that cannot be cultivated at present (Schleifer, 2009). These obligate leaf endosymbionts are vertically transmitted and represent an obligatory symbiosis which was estimated to originate millions of years ago .
BGC species harbor both beneficial and pathogenic strains. Strains PML1(12) and S170 show biotechnological potential for mineral-weathering and plant growth promotion, respectively, and are exemplary for the metabolic versatility of Burkholderia organisms (Llado et al., 2014;Uroz and Oger, 2015). Mineralweathering bacteria dissolute key nutrients from minerals and thereby increase the bioavailability of chemical nutrients in the environment (Uroz et al., 2009). On the other hand, three strains analyzed in the present study were isolated from human clinical samples, i.e., blood, pleural fluid and lung tissue ( Table 1) and were classified as two novel species (Burkholderia concitans sp. nov. and Burkholderia turbans sp. nov.). They represent, to our knowledge, the first examples of human clinical isolates in the B. glathei clade. Strikingly, strain MR1, which was isolated from Florida golf course soil and shown to reduce the herbicide methylarsenate, was also identified as Burkholderia concitans sp. nov., and this species thus represents yet another human clinical Burkholderia species with interesting biotechnological properties (Coenye et al., 2001;Coenye and Vandamme, 2003;Goris et al., 2004;Mahenthiralingam et al., 2005). This study therefore further underscores that there is no phylogenetic subdivision in the genus Burkholderia that distinguishes beneficial from pathogenic strains (Angus et al., 2014;Sawana et al., 2014;Estrada-de los Santos et al., 2016;Dobritsa and Samadpour, 2016).
Description of Burkholderia arvi sp. nov.
The type strain is LMG 29317 T (=CCUG 68412 T ) and was isolated from agricultural soil in Argentina in 2010 (Draghi et al., 2014). Its G+C content is 62.4 mol% (calculated based on its genome sequence). The 16S rRNA, gyrB and whole-genome sequence of LMG 29317 T are publicly available through the accession numbers LT158615, LT158628, and FCOM02000000, respectively.
Description of Burkholderia hypogeia sp. nov.
The type strain is LMG 29322 T (=CCUG 68407 T ) and was isolated from greenhouse soil in Belgium in 2014 (Peeters et al., 2016). Its G+C content is 63.2 mol% (calculated based on its genome sequence). The 16S rRNA, gyrB and whole-genome sequence of LMG 29322 T are publicly available through the accession numbers LT158620, LT158633, and FCOA02000000, respectively.
Description of Burkholderia ptereochthonis sp. nov.
The type strain is LMG 29326 T (=CCUG 68403 T ) and was isolated from botanical garden soil in Belgium in 2014 (Peeters et al., 2016). Its G+C content is 64.2 mol% (calculated based on its genome sequence). The 16S rRNA, gyrB and whole-genome sequence of LMG 29326 T are publicly available through the accession numbers LT158624, LT158637, and FCOB02000000, respectively.
Description of Burkholderia glebae sp. nov.
The type strain is LMG 29325 T (=CCUG 68404 T ) and was isolated from botanical garden soil in Belgium in 2014 (Peeters et al., 2016). Its G+C content is 62.7 mol% (calculated based on its genome sequence). The 16S rRNA, gyrB and whole-genome sequence of LMG 29325 T are publicly available through the accession numbers LT158623, LT158636, and FCOJ02000000, respectively. An additional strain has been isolated from soil in the Netherlands ( Table 1).
The type strain is LMG 29323 T (=CCUG 68406 T ) and was isolated from greenhouse soil in Belgium in 2014 (Peeters et al., 2016). Its G+C content is 63.0 mol% (calculated based on its genome sequence). The 16S rRNA, gyrB, and whole-genome sequence of LMG 29323 T are publicly available through the accession numbers LT158621, LT158634, and FCOE02000000, respectively. An additional strain has been isolated from the same sample ( Table 1).
The type strain is LMG 29324 T (=CCUG 68405 T ) and was isolated from botanical garden soil in Belgium in 2014 (Peeters et al., 2016). Its G+C content is 62.8 mol% (calculated based on its genome sequence). The 16S rRNA, gyrB, and whole-genome sequence of LMG 29324 T are publicly available through the accession numbers LT158622, LT158635, and FCOG02000000, respectively. An additional strain has been isolated from soil in the Netherlands ( Table 1).
Description of Burkholderia fortuita sp. nov.
The type strain is LMG 29320 T (=CCUG 68409 T ) and was isolated from Fadogia homblei rhizosphere soil in South Africa in 2013 (Verstraete et al., 2014). Its G+C content is 62.9 mol% (calculated based on its genome sequence). The 16S rRNA, gyrB and whole-genome sequence of LMG 29320 T are publicly available through the accession numbers LT158618, LT158631, and FCNX02000000, respectively.
Description of Burkholderia temeraria sp. nov.
The type strain is LMG 29319 T (=CCUG 68410 T ) and was isolated from Fadogia homblei rhizosphere soil in South Africa in 2013 (Verstraete et al., 2014). Its G+C content is 62.7 mol% (calculated based on its genome sequence). The 16S rRNA, gyrB and whole-genome sequence of LMG 29319 T are publicly available through the accession numbers LT158617, LT158630, and FCOI02000000, respectively.
Description of Burkholderia calidae sp. nov.
Burkholderia calidae (ca'li.dae. L. gen. n. calidae from warm water, because this strain was isolated from pond water in a tropical garden).
The type strain is LMG 29321 T (=CCUG 68408 T ) and was isolated from greenhouse pond water in Belgium in 2013 (Peeters et al., 2016). Its G+C content is 62.5 mol% (calculated based on its genome sequence). The 16S rRNA, gyrB and whole-genome sequence of LMG 29321 T are publicly available through the accession numbers LT158619, LT158632, and FCOX02000000, respectively.
Description of Burkholderia concitans sp. nov.
Burkholderia concitans (con.ci'tans. L. fem. part. pres. concitans disturbing, upsetting; because the isolation of this bacterium from human sources, including blood, further disturbs the image of this lineage of Burkholderia species as benign bacteria).
The type strain is LMG 29315 T (=CCUG 68414 T ) and was isolated from human lung tissue in the USA in 2006. Its G+C content is 63.2 mol%. The 16S rRNA, gyrB, and whole-genome sequence of LMG 29315 T are publicly available through the accession numbers LT158613, LT158626 and FCNV02000000, respectively. An additional strain has been isolated from human blood in the USA in 2010 ( Table 1).

Description of Burkholderia turbans sp. nov.
Burkholderia turbans (tur'bans. L. fem. part. pres. turbans disturbing, agitating, because the isolation of this bacterium from human pleural fluid further disturbs the image of this lineage of Burkholderia species as benign bacteria).
The type strain is LMG 29316 T (=CCUG 68413 T ) and was isolated from human pleural fluid in the USA in 2006. Its G+C content is 63.1 mol% (calculated based on its genome sequence). The 16S rRNA, gyrB and whole-genome sequence of LMG 29316 T are publicly available through the accession numbers LT158614, LT158627, and FCOD02000000, respectively.
Description of Burkholderia catudaia sp. nov.
The type strain is LMG 29318 T (=CCUG 68411 T ) and was isolated from Fadogia homblei rhizosphere soil in South Africa in 2013 (Verstraete et al., 2014). Its G+C content is 62.8 mol% (calculated based on its genome sequence). The 16S rRNA, gyrB and whole-genome sequence of LMG 29318 T are publicly available through the accession numbers LT158616, LT158629, and FCOF02000000, respectively.
Description of Burkholderia peredens sp. nov.
The type strain is LMG 29314 T (=CCUG 68415 T ) and was isolated from soil in Japan (Hayatsu et al., 2000). Its G+C content is 63.1 mol% (calculated based on its genome sequence). The 16S rRNA, gyrB and whole-genome sequence of LMG 29314 T are publicly available through the accession numbers LT158612, LT158625, and FCOH02000000, respectively.