New global machine-learning estimates of coccolithophore standing stocks and calcification rate accounting for biodiversity

Coccolithophores are a main marine calcifier critical to ocean carbon pumps (via organic matter ballast and the carbonate pump), ultimately controlling atmospheric CO2 and climate. However, their contribution to the global carbon cycle is still very uncertain, limiting our understanding of their impact and response to climate change. One major issue is that most coccolithophore studies rely solely on one outlier species (Emiliania huxleyi), which is relatively small and lightly calcified. Here, we apply novel machine-learning techniques to determine the global distribution of the top 52 species and the total calcite production of coccolithophores. These techniques build predictive models of coccolithophore carbon stocks and calcite production based on newly compiled datasets of coccolithophore abundance and calcification rates, which we combined with environmental data. Our species predictive model shows that a handful of species, including Emiliania huxleyi, are responsible for the global calcite standing stock, with subtropical species being a significant contributor. Our rate predictive model also supports this finding, showing large calcification rates in the subpolar and subtropical regions. This result revisits the traditional view that coccolithophore calcification primarily occurs in sub-arctic bloom-like events and that other species besides Emiliania huxleyi should be considered to resolve coccolithophore’s subtropical contribution.