Fig. 4

Vertical biogeochemical zonation according to patterns of energy metabolism. Total S/N/CH₄ metabolic abundances with seasonal variations (Aa) were predicted with 16S rRNA genes; specific function modules and chemotaxis (Ab–d) were predicted by metagenomic data annotated with KEGG and NCycDB. Sulfate reduction (R² = 0.975, df = 5) and nitrification (R² = 0.854, df = 6) were fitted with exponential model of plateau followed by one phase decay. Methanotrophy (R² = 0.983, df = 4), methanogenesis (R² = 0.999, df = 4), and nitrate reduction (R² = 0.998, df = 4) were fitted piecewise with a linear model followed by exponential one-phase decay/association. Least squares fit was used for all these nonlinear models. Relative abundance of the NC10 class (Ac) serves as proxy for the intensity of N-damo, the nitrate/nitrite-dependent anaerobic methane oxidation. Black dots in plot Ac represent the median value of the seasonal samples labeled by gray dots. OATZ oxic–anoxic transition zone, NATZ nitrate–ammonium transition zone, SMTZ sulfate–methane transition zone, NMTZ nitrate/nitrite–methane transition zone, MGZ methanogenetic zone. B Several ratio indexes relevant to S/N-cycling metabolism. ASR/DSR ratio of assimilatory/dissimilatory sulfate reduction (quadratic fitting). AOA/AOB ratio of ammonia-oxidizing archaea and bacteria reflected by archaeal/bacterial amoABC genes (Mann–Whitney test with P = 0.016*, normality test failed), Denitrification/DNRA ratio of denitrification and dissimilatory nitrate reduction to ammonium (unpaired t test with P = 0.0003***, Shapiro–Wilk normality test passed). C Environmental drivers of energy metabolism. Line color represents Mantel’s p and line width represents Mantel’s r. Red “X” means no significance for Pearson’s correlation