Phylogenetic relationships among Staphylococcus species and refinement of cluster groups based on multilocus data

Abstract

Background: Estimates of relationships among Staphylococcus species have been hampered by poor and inconsistent resolution of phylogenies based largely on single gene analyses incorporating only a limited taxon sample. As such, the evolutionary relationships and hierarchical classification schemes among species have not been confidently established. Here, we address these points through analyses of DNA sequence data from multiple loci (16S rRNA gene, dnaJ, rpoB, and tuf gene fragments) using multiple Bayesian and maximum likelihood phylogenetic approaches that incorporate nearly all recognized Staphylococcus taxa.

Results: We estimated the phylogeny of fifty-seven Staphylococcus taxa using partitioned-model Bayesian and maximum likelihood analysis, as well as Bayesian gene-tree species-tree methods. Regardless of methodology, we found broad agreement among methods that the current cluster groups require revision, although there was some disagreement among methods in resolution of higher order relationships. Based on our phylogenetic estimates, we propose a refined classification for Staphylococcus with species being classified into 15 cluster groups (based on molecular data) that adhere to six species groups (based on phenotypic properties).

Conclusions: Our findings are in general agreement with gene tree-based reports of the staphylococcal phylogeny, although we identify multiple previously unreported relationships among species. Our results support the general importance of such multilocus assessments as a standard in microbial studies to more robustly infer relationships among recognized and newly discovered lineages.

Figure 1

Dataset partitioning improves model fit. Shown are log-likelihood plots comparing partitioning strategies used for concatenated BI runs. Error bars represent the mean ± 95% confidence interval.

Figure 2

Bayesian MCMC analysis of the concatenated dataset. Shown is a 50% majority rule phylogram from BI runs under the combined, partitioned dataset in MrBayes. Numbers represent posterior probabilities with grey-filled circles representing a posterior probability of 1.00.

Figure 3

Maximum likelihood phylogram of staphylococcal species. Shown is a ML phylogram obtained from the assessment of the locus-partitioned dataset (similar to MB3) using GARLI v.2.0 [50]. The consensus phylogram was generated from 200 bootstrap replicates with five ML search replicates per bootstrap. Nodes receiving Pp = 1.00 and/or BS = 100% are indicated by grey-filled circles; otherwise, MrBayes posterior probability is shown in red text, and ML bootstrap support is shown in black text. Clades that were not present in MrBayes are indicated by a red §.

Figure 4

Comparison of nodal support between MrBayes and maximum likelihood methodologies. Shown is a scatter plot comparing the differences in MrBayes posterior probabilities (Pp) and maximum likelihood (ML) bootstrap support (BS) for identical nodes (Figure 3). Open circles represent Pp support for discordant nodes present in MrBayes and absent in ML. Open triangles represent BS values for discordant nodes present in ML and absent in MrBayes. Note that MrBayes exhibits heightened overall node support as compared to ML.

Figure 5

Inference of the staphylococcal phylogeny using Bayesian estimation of species trees (BEST) methodology on 16S rRNA anddnaJgene fragments. Shown is a consensus phylogram of the staphylococcal species tree generated under the BEST methodology incorporating only 16S rRNA and dnaJ gene fragments. Each of the two gene fragments were treated as an individual locus for which individual gene trees were estimated. Numbers represent posterior probabilities with grey-filled circles representing a posterior probability of 1.00. Refer to Additional file 7: Figure S5 for the BEST analysis incorporating all four gene fragments.

Figure 6

Staphylococcal species can be combined into six species groups and 15 cluster groups. Shown is a summary phylogram adapted from Figure 2 with clades collapsed to represent staphylococcal groupings. Whenever possible, cluster and species group names were kept consistent with [64]. Cluster groups have been color-coded to represent: blue, species that are novobiocin resistant, coagulase negative, and oxidase positive; green, species that are novobiocin susceptible, coagulase negative, and oxidase negative; orange, species that are novobiocin resistant, coagulase negative, and oxidase negative; purple, species that are novobiocin susceptible, coagulase positive, and oxidase negative; and red, species that are novobiocin susceptible, coagulase variable, and oxidase negative. Color scheme exceptions are: ^#S. schleiferi schleiferi is coagulase negative; *S. simiae is coagulase negative; ^‡S. hominis novobiosepticus is novobiocin resistant; and ^†S. equorum linens is novobiocin susceptible. Members of each cluster group are listed below the cluster group name. Nodes receiving Pp = 1.00 or BS = 100% are indicated by grey-filled circles; otherwise, MrBayes posterior probability is shown in red text, BEST posterior probability is shown in blue text, and ML bootstrap support is shown in black text. Clades that were not present in BEST or ML are indicated by a blue or black §, respectively.