Data underlying Figures 1, 2, 3, 4 and 5 have been uploaded to Dryad:

Figure 1: Fig1_Data_readable.xls

This figure shows the the lipid content of nitrogen-starved diatoms is dominated by triacylglycerols and free-fatty-acids.

Figure 2: Fig2_Data_readable.xlsx

This figure show that marine bacteria exhibit distinct dietary preferences for different groups of phytoplankton lipids.

Figure 3: Fig3_Data_readable.xlsx

This figure shows that marine bacteria have widely different lipid degradation kinetics.

Figure 4: Fig4_data_readable.xlsx

This figure shows that the lipid degradation kinetics of single bacterial isolates recapitulate their long-term degradation behavior.

Figure 5: Fig5_data_readable.xlsx

This figure shows that pairwise interactions among marine bacteria affect the maximum rate and delay time of lipid degradation.

Supplementary Figures 1, 2, 3, 6, 7 and 10

Data underlying Supplementary Figure 1 is uploaded to Dryad: Supp_Fig_S1_data_readable.xlsx

Data underlying Supplementary Figure 2 and 3 have been uploaded as a Geneious XML file and a comma separated file.

Data_for_Phylogenetic_Tree_Fig_S2.geneious. This file contains the assembled 16s rRNA gene sequences for each of 24 marine bacterial species studied. It also contains a MUSCLE alignment of those sequences. An approximately-maximum-likelihood phylogenetic tree was inferred from the alignment using the FastTree 2 plug-in with default settings and visualization tools in Geneious Prime 2020 (https://www.geneious.com/.

Data_for_Phylogenetic_Tree_Patristic_Distance_Matrix_Fig_S2.csv. The comma separated value (CSV) file contains the patristic distance matrix values.

The code used to generate Supplementary Figure 6 and 7 and to estimate lipid degradation via autofluorescence quenching has been uploaded as ‘Software’ in the form of the following four matlab files:

Chlorophyll_Lipid.mat. This file contains the data on chlorophyll and lipid concentrations.

Ellipse.m. This contains the function to add ellipses to the current plot.

RawFL.mat. This file contains the raw chlorophyll fluorescence data

SuppsFigs.m. This contains the function to generate Supplementary figure 6 and 7.

Instructions to generate Supplementary Figure 10 and the use of the Lipid vertical flux model are available below:

A software repository (https://zenodo.org/records/10809049 and https://github.com/fpeaudecerf/pelagic_lipid_export/tree/v1.0.0) that contains the code associated with the theoretical model of bacterial degradation contains files that are organised as follows:

• oil_model.py contains the chosen numerical parameters from the manuscript (as given in the Supplementary “Full description of the lipid vertical flux model”), all the functions needed to compute the dynamics of degradation and sinking of marine snow particles composed of ballast and oil as presented in the model, and the functions needed to compute transfer efficiencies of lipid fluxes at different depths. These parameters and functions are used at the end of the script to generate a visualisation of results corresponding to Supplementary figure 10 of the manuscript.
• monte_carlo_rates.csv contains results of a Monte-Carlo resampling simulation run from experimental measurements in the lab in order to obtain a range of typical degradation rate constants k_A (see supplementary manuscript “Full description of the lipid vertical flux”). These results are presented in the Supplementary figure 10 and thus are needed by oil_model.py to generate this figure.
• utils.py is a small utilities file.
• sinking_dynamics_example_oil_model.eps is an example output of oil_model.py representing on a figure the depth reached by particles of different initial sizes with time.
• FigSF_full.eps is an example output of oil_model.py corresponding to Supplementary figure 10 of the manuscripted referenced above.
• environment.yml sets the environment for Binder, so that one can directly execute the script contained in oil_model.py in Binder and have it work there, without the need for a local Python installation.

To run the model in Binder without the need of a local Python installation, use the URL below:\
https://mybinder.org/v2/gh/fpeaudecerf/pelagic_lipid_export/HEAD

There, you can select “New” -> “Terminal” and execute the main script from there by typing python oil_model.pyand pressing return.

Supplementary Videos 1, 2, 3, 4, 5, 6, 7 have been uploaded to Dryad:

Movie S1: Supplementary_Video_1.avi

Composite video of Pseudoalteromonas marina (Pm7), Pseudoalteromonas marina (Pm11) *and *Pseudoalteromonas tetradonis (Pt12) *degrading *phytoplankton lipids. All three bacteria belong to cluster I (Fig. 2). Note the loss of lipid autofluorescence over the total video duration of 114 h. Scale bar is 100 µm.

Movie S2: Supplementary_Video_2.avi

Composite video of phytoplankton lipid degradation by Vibrio spp. (Vs8) and P. zhaodongensis (Pz15), two cluster III representatives (Fig. 2). Note the loss of lipid autofluorescence over the total video duration of 114 h. Scale bar is 100 µm.

Movie S3: Supplementary_Video_3.avi

Composite video of phytoplankton lipid degradation by R. pomeroyi DSS-3-B (Rp3), Pseudoalteromonas spp. (Ps19) *and *Pseudoalteromonas mariniglutinosa (Pm22). While Rp3 is a cluster II representative, Ps19 and Pm22 are cluster IV representatives. Note the absence of visible lipid degradation by these three bacteria over the course of 114h of incubation. Scale bar is 100 µm.

Movie S4: Supplementary_Video_4.mp4

Video sequence of the concentration-dependent quenching of chlorophyll fluorescence within a lipid droplet. A droplet containing highly concentrated chlorophyll is visible under transmitted visible light, but invisible under epifluorescence ((Ex.: 628/40 nm HBW, Em.: 692/40 nm HBW) because chlorophyll does not emit fluorescent light due to self-quenching. Upon addition of triacylglycerol (trioctanoylglycerol) with a glass pipette, chlorophyll is diluted and becomes unquenched as a result. 

Movie S5: Supplementary_Video_5.avi

The chemotactic behavior of *P. zhaodongensis *(Pz15) during phytoplankton lipid degradation. Lipid droplets were video recorded for 10 s every 2 minutes. From each individual video, a maximum intensity projection was created for each time point. The video shows the assembled maximum intensity projections over a period of 48 h. Scale bar is 100 µm.

Movie S6: Supplementary_Video_6.avi

The chemotactic behavior of *P. zhaodongensis *(Pz15) during synthetic triacylglycerol degradation. Synthetic TAG droplets (encircled in yellow) were incubated with Pz15::GFP and filmed using fluorescence and phase-contrast microscopy every 30 min for a duration of 60 h. Scale bar is 50 µm.

Movie S7: Supplementary_Video_7.avi

The combined growth of P. zhaodongensis (Pz15::GFP, green fluorescent) and *A. macleodii ATTC 27126 (Am2, not fluorescent) in the presence of a phytoplankton lipid droplet. Isolates were imaged in fluorescence (Ex 470 nm / Em 525 nm) and phase-contrast microscopy every 15 min for 47 h. Scale bar is 50 µm.