Biological sources and sinks of dimethylsulfide disentangled by an induced bloom experiment and a numerical model

Dimethylsulfide (DMS) is a climatically active trace gas promoting cloud formation. The biochemical precursor of DMS, dimethylsulfoniopropionate (DMSP), is a phytoplankton metabolite and a source of reduced sulfur for many microbial species. Because of the complex interactions between their many producers and consumers, the dynamics of DMSP and DMS in the ocean are still poorly constrained. In this study we measure particulate DMSP, dissolved DMSP (DMSPd$$ {\mathrm{DMSP}}_{\mathrm{d}} $$), and DMS concentrations in seven mesocosms where two consecutive phytoplankton blooms (first, pico- and nano-algae; second, Emiliania huxleyi) were induced by nutrient addition, and we build a mechanistic numerical model to identify the sources and sinks that best account for the observations. The mesocosms were designed as replicates but differ from each other by their E. huxleyi virus abundance due to stochastic differences in initial conditions. The model shows that heterotrophic bacteria cannot be the only consumers of DMSPd$$ {\mathrm{DMSP}}_{\mathrm{d}} $$. A fraction of dissolved DMSPd$$ {\mathrm{DMSP}}_{\mathrm{d}} $$ must be consumed by phytoplankton to avoid excessive DMSPd$$ {\mathrm{DMSP}}_{\mathrm{d}} $$ accumulation during the first bloom. The induced blooms increase DMS concentration by 220% on average, until an increase in the abundance of DMS-consuming bacteria brings DMS concentration back to its pre-bloom value, after 3 weeks of experiment. Therefore phytoplankton blooms can increase DMS emission to the atmosphere, but only during a transient regime of a few weeks. The model also shows that the DMS yield, production and emission are increased when the coccolithophore bloom is terminated by a viral infection, but decreased if the infection occurs several days before the bloom can reach its maximum.