When we extended this diversity analysis to the Tara Oceans samples, we found a mild trend in the SRF layer ( r 2 = 0.37 ), but no correlation in the DCM and MES layers (Fig.
2
—
—
The sentences
We compared the normalized CRISPR repeat abundance of 1,141 human oral samples against Shannon’s diversity index and observed a small but highly significant positive correlation ( r 2 = 0.048 ; Fig. 4A ; Data Set S7).