Microbial distribution pattern based on the complete pyrosequencing dataset at the 97% OTU level.

<p><b>A.</b> Microbial cluster tree using jackknifed unweighted pair group method with arithmetic averages (UPGMA); <b>B.</b> Non-metric multidimensional scaling (NMDS) ordination for the community structure for all the samples. Ordination is based upon the Bray-Curtis similarity of the square-root-transformed abundances. The lower the 2D stress is, the better the ordination is; <b>C.</b> PCoA scatter plot that begins with a table of distances between the samples. The first two factors P1 and P2 can explain 26.7% and 23.7% variations, respectively. All three plots show the same pattern: the soil cluster, the low-temperature cluster, and the moderate-temperature cluster. The samples are primarily grouped by the sample type (hot spring sediments vs. adjacent soil); among the hot spring sediments, there are two groups separated by temperature (low vs. moderate temperature). Sample codes consist of sample ID and temperature for that spring. For example, GL13.4_66 represents the 4th site along an outflow channel of the 13th hot spring in the Gulu area with a temperature of 66°C. Different symbols are used to differentiate the different clusters.