Biogeography of American Northwest Hot Spring A/B '-Lineage Synechococcus Populations

Abstract

Previous analyses have shown how diversity among unicellular cyanobacteria inhabiting island-like hot springs is structured relative to physical separation and physiochemical differences among springs, especially at local to regional scales. However, these studies have been limited by the low resolution provided by the molecular markers surveyed. We analyzed large datasets obtained by high-throughput sequencing of a segment of the photosynthesis gene psaA from samples collected in hot springs from geothermal basins in Yellowstone National Park, Montana, and Oregon, all known from previous studies to contain populations of A/B'-lineage Synechococcus. The fraction of identical sequences was greater among springs separated by <50 km than among springs separated by >50 km, and springs separated by >800 km shared sequence variants only rarely. Phylogenetic analyses provided evidence for endemic lineages that could be related to geographic isolation and/or geochemical differences on regional scales. Ecotype Simulation 2 was used to predict putative ecotypes (ecologically distinct populations), and their membership, and canonical correspondence analysis was used to examine the geographical and geochemical bases for variation in their distribution. Across the range of Oregon and Yellowstone, geographical separation explained the largest percentage of the differences in distribution of ecotypes (9.5% correlated to longitude; 9.4% to latitude), with geochemical differences explaining the largest percentage of the remaining differences in distribution (7.4-9.3% correlated to magnesium, sulfate, and sulfide). Among samples within the Greater Yellowstone Ecosystem, geochemical differences significantly explained the distribution of ecotypes (6.5-9.3% correlated to magnesium, boron, sulfate, silicon dioxide, chloride, and pH). Nevertheless, differences in the abundance and membership of ecotypes in Yellowstone springs with similar chemistry suggested that allopatry may be involved even at local scales. Synechococcus populations have diverged both by physical isolation and physiochemical differences, and populations on surprisingly local scales have been evolving independently.

Keywords: biogeography; ecotype; microbial species; population genetics; thermophilic Synechococcus.

Figure 1

Approximate locations and geographic separation (km) among the 7 geographic regions studied across the Northwest United States by Papke et al. (2003). Satellite imagery insets show the approximate locations and geographic separation (m) among springs within Yellowstone National Park and Oregon basins. The satellite imagery was produced using Google Maps. Colors used for basins correspond between the map, satellite imagery, and later figures.

Figure 2

Percentage of shared A-like Synechococcus psaA gene segment HFS2 variants relative to separation between pairs of springs sampled. Each circle represents a pair of springs, with colored circles representing pairs from within the same basin. Gray circles represent pairs of springs from different basins within and around Yellowstone, while open circles represent comparisons between springs in Oregon with springs in and around Yellowstone National Park. The black trend line was calculated from all comparisons.

Figure 3

Phylogenies of 16S rRNA defined A′ (PEA1-PEA17), A (PEA18-PEA66), and B′-like Synechococcus (gray bars) based on partial psaA gene HFS10 variants from Yellowstone National Park, Montana, and Oregon Springs, with putative ecotype (PE) demarcations based on Ecotype Simulation 2 (ES2). Shaded regions of the trees mark branches that are endemic to a single basin. Regions with a diagonal hatch mark branches predominantly found in a single basin (>90%). The ES2 PE demarcations are displayed next to each tree as vertical black and colored bars, with colored bars representing predominant PEs analyzed in detail. Demarcation colors are reused between the A- and B′-like phylogenies and are the same in later figures. Labels are skipped for most PEs with only a single member.

Figure 4

Abundance of HFS10 psaA sequence segments in predominant Synechococcus A- and B′-like putative ecotypes (PEs) for springs within Yellowstone National Park. Poorly sampled springs are shaded gray.

Figure 5

Abundance of HFS10 psaA sequence segments in predominant Synechococcus A- and B′-like putative ecotypes (PEs) for springs outside of Yellowstone National Park. Poorly sampled springs are shaded gray.

Figure 6

Canonical correspondence analyses of Synechococcus A- (A) and B′-like (B) psaA sequence segment HFS10 diversity recovered from Yellowstone National Park, Montana, and Oregon hot springs. Larger symbols represent sequences described previously by Becraft et al. (2015). Synechococcus strains JA-3-3Ab, 60AY4M2, and JA-2-3B'a(2-13) share psaA sequence segments with high-frequency sequences (HFSs) in putative ecotypes (PEs) PEA21 (A1), PEA30 (A14), and PEB20 (B′12-1), respectively, and are labeled on each plot. Small gray dots represent HFSs from lower-abundance PEs or from the other lineage. Directional arrows represent the vector of influence of each of the significant parameters on the ordination space. PE names in the legend are followed by the number of HFSs making up the PE in parenthesis and a p-value that represents the probability that the observed PE cluster is randomly produced.

Figure 7

Canonical correspondence analyses of Synechococcus A- (A) and B′-like (B) psaA sequence segment HFS10 diversity recovered from Yellowstone National Park and Montana hot springs. Larger symbols represent sequences described previously by Becraft et al. (2015). Synechococcus strains JA-3-3Ab, 60AY4M2, and JA-2-3B'a(2-13) share psaA sequence segments with high-frequency sequences (HFSs) in putative ecotypes (PEs) PEA21 (A1), PEA30 (A14), and PEB20 (B′12-1), respectively, and are labeled on each plot. Small gray dots represent HFSs from lower-abundance PEs or from the other lineage. Directional arrows represent the vector of influence of each of the significant parameters on the ordination space. PE names in the legend are followed by the number of unique HFSs making up the PE in parenthesis and a p-value that represents the probability that the observed PE cluster is randomly produced.

Figure 8

Canonical correspondence analyses (CCA) relative to pH of Synechococcus A- (A) and B′-like (B) HFS10 psaA sequence segments within predominant putative ecotypes (PEs) recovered from Yellowstone National Park and Montana hot springs. Weighted density of predominant A- (C) and B′-like (D) PEs in the ordination space defined by the pH vector. Larger symbols in A and B represent sequences described previously by Becraft et al. (2015). Synechococcus strains JA-3-3Ab, 60AY4M2, and JA-2-3B'a(2-13) share psaA sequence segments with high-frequency sequences (HFSs) in putative ecotypes (PEs) PEA21 (A1), PEA30 (A14), and PEB20 (B′12-1), respectively, and are labeled in A and B. Small gray dots in (A,B) represent HFSs from lower-abundance PEs or from the other lineage. Directional arrows represent the vector of influence of pH on the ordination space in (A,B). PE names in the legend are followed by the number of unique HFSs making up the PE in parenthesis and a p-value that represents the probability that the observed PE cluster is randomly produced.