Modern observations indicate that variations in marine phytoplankton stoichiometry correlate with the boundaries of major surface waters. For example, phytoplankton in the oligotrophic subtropical gyres typically have much higher C:N:P ratios (i.e., higher C:P and higher N:P ratios) than those in eutrophic upwelling regions and polar regions. Such a spatial pattern points to nutrient availability as a key environmental driver of stochiometric flexibility. Environmental dependence of phytoplankton C:N:P opens unexplored possibilities for modifying the strength of the biological pump under different climate conditions. Here we present a power law formulation of C:N:P flexibility that is driven by nutrients, temperature, and light. We embed the formulation in a global ocean carbon cycle model with multiple phytoplankton types and explore biogeochemical implications under glacial conditions. We find three key controls on export C:N:P ratio: phytoplankton physiology and community structure as well as the balance in regional production at the global level. Glacial inputs of iron and sea ice expansion are important modifiers of these three controls. We also find that global export C:N:P increases substantially under glacial conditions, and this strongly buffers global carbon export against decrease and draws down approximately 20 μatm of atmospheric CO2. These results point to the importance of including phytoplankton C:N:P flexibility in a mix of mechanisms that drive atmospheric CO2 over glacial-interglacial time scale. Finally, our simulations indicate decoupling of nutrients, which may provide a resolution to the longstanding disagreement regarding nutrient utilization in the glacial Southern Ocean derived from different nutrient proxies.
Bibliographical noteFunding Information:
This work is dedicated to Wally Broecker who passed away in February 2019. His first papers on the biological pump revolutionized paleo and chemical oceanography and spawned many investigations including this work. This work was supported by the U.S. National Science Foundation (OCE‐1827948). K.M. was supported by the Leverhulme Trust Visiting Professorship while on sabbatical at the University of Oxford. R. R. acknowledges support from ERC Consolidator Grant APPELS (ERC‐2015‐COG‐681746) and a Wolfson Research Merit Award. Numerical modeling and analysis were carried out using resources at the University of Minnesota Supercomputing Institute.
© 2020. American Geophysical Union. All Rights Reserved.