Homology-based searches were used to detect marker genes for key metabolic processes in the metagenomic short reads and MAGs. a, Heatmaps showing the abundance of each gene in the metagenomic short reads of the four Australian depth profiles and eight global sites. The percentage of the total community predicted to encode at least one of each gene for a process is shown based on normalization to single-copy marker genes. The genes detected are usually present in single copies in genomes and, where genes performing similar functions are collapsed together, the values are summed up to 100%. b, Dot plot showing the metabolic potential of the 757 MAGs. The size of each point represents the number of genomes in each phylum that encode the gene of interest and the shading represents the average genome completeness. Taxonomy assignment is based on the GTDB51.