A Link Between CMIP5 Phytoplankton Phenology and Sea Ice in the Atlantic Southern Ocean

Earth system models from the Climate Model Intercomparison Project phase 5 (CMIP5) are strongly biased in Southern Ocean phytoplankton phenology, especially in the marginal ice zone. In this study we describe the mechanisms driving CMIP5 models to misrepresent seasonal primary production in the Atlantic marginal ice zone during late winter. We link subsurface light availability during this period to simulated early growth, arguing that a combination of ice cover and deep winter mixing prevent biomass accumulation in the real ocean, while in models this combination of factors is not present. Furthermore, we find a statistically significant correlation across the model ensemble between vertical stratification and the location of the ice edge; whereby the more equatorward the ice edge is, the closer to the surface stratification occurs. We argue that models may be grouped according to how strongly they express two major controls on their phenology, namely, the location of the ice edge and the degree of stratification present in the water column in late winter. We find that models with small biases in just one of these controls (but large biases in the other) are able to simulate bloom initiation close to observations, while models with significant biases in both controls initiate growth 2-4 months early. Plain Language Summary Our study is focused on the Coupled Model Intercomparison Project Phase 5, an initiative which seeks to better understand the Earth's climate by bringing together some of the most sophisticated and realistic climate models available. The aim of the study is to understand why Coupled Model Intercomparison Project Phase 5 models are unable to simulate the seasonal growth and decline of microscopic marine algae (phytoplankton) in the ocean surrounding Antarctica. Our findings suggest that the sea ice plays a key role in explaining the discrepancy between the models and what we observe with satellites. In particular, models tend not to have sea ice at the right time and place, which strongly affects the amount light available for phytoplankton to grow. This is important for global climate because phytoplankton growth can potentially reduce atmospheric CO2 concentrations, with changes in the timing and intensity of growth affecting how efficiently they are able to do this. This makes it essential that these cycles are captured by climate models and that these issues are addressed in the currently underway CMIP6.