Fig. 1

Lack of microbial consensus and loss of biogeographical distinction in ICU patients. Principal coordinate analysis (PCoA) ordination using the Bray-Curtis dissimilarity metric between the ICU and healthy cohorts demonstrate that samples collected from the healthy cohort tend to cluster per body sites (PERMANOVA, p < 0.001, R² = 0.529) whereas the samples from different anatomical sites tend to overlap in the ICU cohort (PERMANOVA, p < 0.001, R² = 0.082; a). The ordination plot of group dispersions within body site demonstrates a lack of compositional homogeneity within anatomical sites in the ICU cohort (p < 0.001; b). The overlaying lines on the scatter plot show the median distances between the cluster’s centroid displayed with a black circle and each samples within the group and the interquartile range of each site. UPGMA dendogram showing the Bray-Curtis dissimilarity between specimens displays a perfect segregation of samples in the healthy cohort based on collections sites. This is not observed in the ICU cohort (c)