Global coccolithophore diversity: Drivers and future change

Abstract We use the MAREDAT global compilation of coccolithophore species distribution and combine them with observations of climatological environmental conditions to determine the global-scale distribution of coccolithophore species diversity, its underlying drivers, and potential future changes. To this end, we developed a feed-forward neural network, which predicts 78% of the observed variance in coccolithophore diversity from environmental input variables (temperature, PAR, nitrate, silicic acid, mixed layer depth, excess phosphate (P∗) and chlorophyll). Light and temperature are the strongest predictors of coccolithophore diversity. Coccolithophore diversity is highest in the low latitudes, where coccolithophores are a relatively dominant component of the total phytoplankton community. Particularly high diversity is predicted in the western equatorial Pacific and the southern Indian Ocean, with additional peaks at approximately 30°N and 30°S. The global, zonal mean pattern is dominated by the Pacific Ocean, which shows a clear latitudinal gradient with diversity peaking at the equator, whereas in the Atlantic Ocean diversity is highest in the subtropics. We find a unimodal relationship between coccolithophore diversity and biomass, as has previously been observed for total phytoplankton assemblages. In contrast, diversity shows a negative relationship with total chlorophyll. Applying our diversity model to projections from the CMIP5 climate models, we project an increase in the diversity of coccolithophore assemblages by the end of this century.