Fig. 2
The urban Chinese microbiome is significantly more prevalent in an American cohort than the rural Chinese microbiome. a Heatmap showing the mean log normalized relative abundance from sequencing the 16S rRNA gene at timepoints 1 (T1) and 2 (T2) of taxa significantly different (adjusted p < 0.05) between urban and rural samples after correction for multiple hypothesis testing, based on our mixed linear models with urban/rural and time as fixed effects and subject id as a random effect. Within each taxonomic level, taxa are given from most to least significant between urban and rural samples. No taxa were significant at the class or order level. At the OTU level, the genus and consensus number is given. At all other levels, the name of the taxon is given. 8 taxa were tested at the phylum level, 16 at the class level, 22 at the order level, 40 at the family level, 86 at the genus level, and 703 at the OTU level. Full details for each taxon, including mean and standard deviation, P values and model R 2 are given in Additional file 6: Table S2A–F. b-e We used the consensus sequences from the 80 OTUs higher in rural samples and 91 OTUs higher in urban samples identified from our mixed linear models (unadjusted p < 0.05) as query sequences against different 16S rRNA databases. b OTUs higher in rural and urban communities are present in the SILVA database of full-length 16S rRNA sequences at similar levels of identity (p = 0.08 by unpaired Wilcoxon test). c OTUs higher in urban subjects are better represented in the NCBI collection of finished bacterial genomes than OTUs higher in rural subjects (p = 0.0008). d OTUs higher in rural and urban communities have previously been found in the Human Microbiome Project (HMP), a study of healthy American subjects, at similar levels of identity (p = 0.88), but (e) OTUs higher in urban subjects are seen in a greater proportion of HMP subjects compared to OTUs higher in rural subjects (p = 7.9 × 10−14). ***p < 0.001, ****p < 10−13