Searching for species in haloarchaea

Abstract

Prokaryotic (bacterial and archaeal) species definitions and the biological concepts that underpin them entail clustering (cohesion) among individuals, in terms of genome content and gene sequence similarity. Homologous recombination can maintain gene sequence similarity within, while permitting divergence between, clusters and is thus the basis for recent efforts to apply the Biological Species Concept in prokaryote systematics and ecology. In this study, we examine isolates of the haloarchaeal genus Halorubrum from two adjacent ponds of different salinities at a Spanish saltern and a natural saline lake in Algeria by using multilocus sequence analysis. We show that, although clusters can be defined by concatenation of multiple marker sequences, barriers to exchange between them are leaky. We suggest that no nonarbitrary way to circumscribe "species" is likely to emerge for this group, or by extension, to apply generally across prokaryotes. Arbitrary criteria might have limited practical use, but still must be agreed upon by the community.

Fig. 1.

Venn diagrams of allele and ST distributions. Individual circles represent 36% and 22% salinity Spanish saltern ponds and the Algerian site. Values inside the circles reflect number of alleles or STs in each set.

Fig. 2.

Concatenated gene phylogeny of Halorubrum STs. Color bars indicate where strains representing STs were obtained. STs with multiple color bars have individual strains cultivated from different sites. Phylogroups are defined as relatively tight clades on this tree. Phylogroup Z was chosen as outgroup based on knowledge of the 16S rRNA gene phylogeny of the haloarchaea. Only bootstrap support values >70% are shown.

Fig. 3.

Maximum likelihood mapping analysis of phylogenetic signal within individual loci. STs (Fig. 2) were divided into four groups corresponding to phylogroups A, B, and C and the rest of the taxa, and three possible phylogenetic relationships (T₁, T₂, or T₃) among these groups were evaluated by using all possible quartets. Support for each relationship and each locus is summarized in barycentric coordinates. In this coordinate system, topologies are placed at the vertices of an equilateral triangle whose area is divided into regions. Values inside each region show the percentage of quartets that fall into the region (and the quartet location is determined by the likelihood of supporting each topology). Values in the regions closest to a vertex show percentage of quartets that strongly support the tree topology at the vertex. Values along edges indicate the percentage of quartets unable to discriminate between the two possible topologies located at the adjacent vertices. Values in the middle indicate the percentage of quartets that did not discriminate between the three topologies. When more than one topology is strongly supported (e.g., 16S rRNA), we interpret this as evidence for highly supported conflicting relationships within a locus, as opposed to noise, which is reflected in the values in the middle region. Notably, different loci support different relationships among the tested groups.