Influence of molecular resolution on sequence-based discovery of ecological diversity among Synechococcus populations in an alkaline siliceous hot spring microbial mat

Abstract

Previous research has shown that sequences of 16S rRNA genes and 16S-23S rRNA internal transcribed spacer regions may not have enough genetic resolution to define all ecologically distinct Synechococcus populations (ecotypes) inhabiting alkaline, siliceous hot spring microbial mats. To achieve higher molecular resolution, we studied sequence variation in three protein-encoding loci sampled by PCR from 60°C and 65°C sites in the Mushroom Spring mat (Yellowstone National Park, WY). Sequences were analyzed using the ecotype simulation (ES) and AdaptML algorithms to identify putative ecotypes. Between 4 and 14 times more putative ecotypes were predicted from variation in protein-encoding locus sequences than from variation in 16S rRNA and 16S-23S rRNA internal transcribed spacer sequences. The number of putative ecotypes predicted depended on the number of sequences sampled and the molecular resolution of the locus. Chao estimates of diversity indicated that few rare ecotypes were missed. Many ecotypes hypothesized by sequence analyses were different in their habitat specificities, suggesting different adaptations to temperature or other parameters that vary along the flow channel.

FIG. 1.

Neighbor-joining phylogenetic trees of A-like and B′-like Synechococcus diversity in Mushroom Spring at 60°C (open circles) and 65°C (closed circles) for the 16S rRNA gene (A) and the ITS region (B), with putative ecotypes (PEn) demarcated by ecotype simulation. Replicate sequences of the dominant variant within a demarcated putative ecotype are highlighted in the shaded box for Synechococcus sp. A-like ITS. Bootstrap values of >50% are shown. Bars represent the number of fixed point mutations per sequence position.

FIG. 2.

Neighbor-joining phylogenies for Synechococcus A-like population sequences for apcAB, aroA, and rbsK genes, with putative ecotypes demarcated by ecotype simulation and habitat-specific clusters demarcated by AdaptML. Open and closed circles demarcate sequences from 60°C and 65°C samples, respectively. Replicate sequences of the dominant variant within a demarcated putative ecotype are highlighted in shaded boxes. Putative subdominant variants in aroA PE1 are highlighted in the unshaded box. Superscripts M60 and M65 next to AdaptML cluster demarcation indicate habitat specificity. Bootstrap values of >50% are shown. Bars represent the number of fixed point mutations per sequence position.

FIG. 3.

Neighbor-joining phylogenies for Synechococcus B′-like population sequences for apcAB, aroA, and rbsK genes, with putative ecotypes demarcated by ecotype simulation. Replicate sequences of the dominant variant are highlighted in shaded boxes for each demarcated ecotype. Putative subdominant variants of apcAB PE1 and rbsK PE1 are highlighted in unshaded boxes. Bootstrap values of >50% are shown. Bars represent the number of fixed point mutations per sequence position.

FIG. 4.

Demarcated putative ecotype prediction of ecotype simulation as a function of molecular resolution (i.e., average evolutionary divergence) for Synechococcus A-like (closed circles) and B′-like (open circles) variants used in ecotype simulation analyses, as shown in Fig. 1 to 3.