Fig. 7

Predicted substrates of PULs and PUL-like clusters in abundant MAGs. Based on 16S rRNA amplicon and corresponding MAG abundance data, we selected 71 MAGs from bacterial clades that were prominent either during particular bloom phases and/or in particular size fractions. A maximum-likelihood tree (computed in anvi’o v7.1 based on protein sequences of 38 universal single-copy genes) is shown to the left. PULs and PUL-like clusters in these MAGs were annotated and corresponding target substrate classes were using the dbCAN3-sub database. The numbers of involved degradative CAZyme genes (GH, PL, CE) are represented by bar charts. The stacked bars to the very right correspond to the total number of CAZyme genes per MAG with and without predicted substrate specificities. In addition, the fractions are indicated where the MAGs were most abundant (FL, 0.2–3 µm; PA3, 3–10 µm; PA10, > 10 µm), with circles representing data obtained from corresponding ASVs and squares representing read-based MAG abundances. Further details are provided in Additional file 1: Table S7