Six experimental treatments were set up the following day for each season as described in Sánchez et al 2020). Briefly, the treatments consisted on: (i) unfiltered seawater in light/dark cycles (CL) and in the dark (CD), (ii) seawater prefiltered through a 1 µm filter to remove large predators while preserving most bacteria in light/dark cycles (PL) and in the dark (PD), (iii) unfiltered seawater diluted 1/4 with 0.2-µm-filtered seawater to reduce predators and increase nutrient availability for bacteria in light/dark cycles (DL), and (iv) unfiltered seawater diluted 1/4 with 30 kDa-filtered seawater to reduce predators, viruses and increase nutrient availability, in light/dark cycles (VL). The different treatments were incubated in triplicated 9 L Nalgene bottles for 48 h at in situ temperature (see Table 1 in Sánchez et al 2020) in a water bath with circulating seawater. Light treatments were limited to photosynthetically active radiation, and dark treatments were covered with several layers of dark plastic. Samples were taken for community DNA, bacterial isolation, flow cytometry, inorganic nutrient concentration and other ancillary data (reported in Sanchez et al 2020) at times 0 h, 12 h, 24 h, and 36 h in summer and winter or 48 h in the fall and spring experiments. For isolation, 1 mL seawater subsamples were mixed with 75 µL dimethyl sulfoxide (DMSO) in cryovials that were stored at -80° C in triplicates. Community DNA extraction and sequencing. Samples were prefiltered through a 20 µm mesh to remove large particles and microbial biomass was concentrated onto 0.2 µm polycarbonate filters using a peristaltic pump. About 2-4 L were filtered from each replicate of all treatments. We extracted the DNA from the filters as described in Massana et al 1997, which was purified and concentrated using Amicon 100 columns (Millipore) and quantified in a NanoDrop-1000 spectrophotometer (Thermo Scientific). DNA was stored at -80° C. For sequencing of metagenomes, an aliquot from each sample was processed in a Novaseq6000 machine (Centre Nacional d'Anàlisi Genòmica, CNAG) with paired-end fragments of 150 bp. A total of 66 samples were sequenced with an average 115 million reads (min = 67M, max = 238M) each. The winter and summer experiments presented 2 replicates for the final times, whereas the spring experiment presented 3 replicates. The fall experiments were not sequenced due to budget limitations. We used illumina-utils for quality filtering the short reads from the metagenomes with the iu-filter-quality-minoche function (default parameters), which removes noisy reads.