Abstract
Carbon flux to the deep sea can be dictated by surface ocean phytoplankton community composition, but translating surface ocean observations into quantitative predictions of carbon export requires additional consideration of the underlying ecosystem drivers. Here, we used genetic tracers of phytoplankton detected in surface seawater and within sinking particles collected in the mesopelagic ocean to identify mechanistic links between surface communities and carbon export in the North Pacific and North Atlantic Oceans. Phytoplankton 18S rRNA gene sequences were sampled over a 1-month period in surface seawater and within bulk-collected and individually isolated sinking particles using mesopelagic sediment traps (100-500 m). Nearly all phytoplankton amplicon sequence variants exported from the surface were packaged in large (>300 μm) particles. Individually, each of these particles contained only a few distinct phytoplankton amplicon sequence variants, but collectively, large particles transported about half of the surface taxonomic diversity into the mesopelagic. The relative sequence abundances of the surface community detected within particles were quantitatively related to measured carbon fluxes: a linear model based on the relative sequence abundance of just two pigment-based phytoplankton taxa, diatoms and photosynthetic Hacrobia, was predictive of carbon flux magnitude. These two taxa were also enriched in the ecologically distinct particle classes that had the greatest influence on carbon export magnitude. As global, hyperspectral ocean color satellites begin to quantify these taxonomic groups in the surface ocean, the relationship of these taxa to carbon fluxes demonstrated here may help in developing more accurate algorithms to estimate global carbon export in the ocean.