Marine phytoplankton fix carbon in the sunlit surface ocean, which is the first step in a process called the biological carbon pump that leads to the export and sequestration of carbon in the deep ocean and that exerts strong controls Earth's climates and the chemical state of the biosphere. Phytoplankton communities contain tremendous diversity, which presents challenges to modelers- as shifts in the community composition along environmental gradients complicate the links between environmental drivers such as nutrient concentrations, light levels, or temperature, and the phytoplankton ecosystem functions. Here we show how a trait-based modeling framework that embraces biodiversity can in some cases overcome these challenges. 

We focus on a particular trait of phytoplankton called elemental stoichiometry, which refers to the ratios of carbon, nitrogen, and phosphorus that composes their cells. Phytoplankton stoichiometry substantially modulates the interaction between the biological carbon pump and cycles of other elements such as nitrogen and phosphorus, yet most biogeochemical and Earth System models represent these ratios as constants. Here we show how a trait-based (rather than species-based) model using first principles from physiology, chemistry, and physics can predict how elemental ratios vary in diverse phytoplankton ecosystems across a range of values of environmental drivers and quantify key mechanisms hypothesized to underlie variations in stoichiometric ratios. We then show how this trait-based model can be embedded in a global biogeochemical model framework in order to determine how dynamic stoichiometry alters biogeochemical model predictions and to find the biogeochemical fingerprint of different physiological mechanisms. Our simulation results improve the representation of several key biogeochemical processes and patterns in the Geophysical Fluid Dynamics Laboratories COBALT ocean biogeochemical model, in particular nutrient limitation patterns, nitrogen fixation rates, and carbon export.