Fig. 2
Absolute abundance of Legionella sp. follows sodium concentrations. Quantification by qPCR was done using primers and probes specific to Legionella spp., targeting the 16S rRNA gene. Abundance (Log10 transformed) was analyzed in a water (n = 64) and b biofilm (n = 50) samples, according to spring cluster (left panels) and collection date (right panels). Kruskal–Wallis tests and post hoc via Wilcox with BH corrections are presented in Supplementary Table 9. c Heatmap of Legionella sp. abundance. Samples (rows, n = 22) for each collection date (column, n = 5) were grouped by Hierarchical k-means clustering (Supplementary Fig. 4 and Supplementary Table 8). Sodium concentrations (ranges, in mM) were subsequently added to each sample (red color denotes high sodium concentrations). The heatmap color scale represents high (yellow to red) and low (purple to black) values. Hierarchical k-means clusters are represented by randomly assigned colors to the left of the heatmap. Spring cluster colors are presented to the right of the heatmap. d + e Principal component analysis (PCA) of selected physico-chemical parameters measured at each sampling site, representing the four spring clusters. PCA was performed on d water (n = 55) and e biofilm (n = 39) samples, obtained between January 2018 and January 2019 and included the following environmental variables: Temperature, Ph, Na, Turbidity, TP, SO4, NO3−, and Fe. Size and transparency represent the log + 1 values of qPCR Legionella spp. levels. Loading scores and vectors are presented in Supplementary Table 6 and Supplementary Fig. 3. The environmental variables were chosen following a PCA analysis from a larger subset of environmental factor (see Supplementary Fig. 2 and Supplementary Tables 4 and 5). THS = Tiberias Hot Springs. Significance level; * p ≤ 0.05; ** p ≤ 0.01; *** p ≤ 0.001; **** p ≤ 0.0001