Figure 2

The first 15 MDS axes were regressed against sample type (swab or stool), participant ID and time point. The −log10 (p-value) for the null hypothesis that sample type, participant ID and time point have no impact on the MDS axes are all shown. While there are significant differences in the first MDS axis in stool vs swab samples, the MDS axes thereafter are significantly different between the participants. Taxonomic calls were based on QIIME closed-referenced OTU picking against GreenGenes database.