Using Global Climate Models to Inform Multi-Driver Biological Ocean Experiments

Henry C Holm , Sonya T Dyhrman, Abby P Shaum, Amanda M Ellis, Benjamin Acosta, Anna Y Chen, Boswell Wing, Galen A McKinley, Sarah Hurley

This century, widespread shifts in temperature and CO2-driven acidification are projected for the surface ocean. Determining how ocean biota will respond to these changes requires scaling multiple drivers together because biological responses to one driver often depend on the conditions of the others. However, ocean temperature and carbonate chemistry vary across regions, seasons, and emissions scenarios, creating more combinations than laboratory experiments can test. While best-practice literature recommends designing experiments around climate projections, translating complex model output into a tractable set of realistic multi-driver treatments requires simplifying substantial spatial, seasonal, and scenario-dependent variation. To address this, we synthesized surface-ocean model projections into relevant multi-driver scalings of temperature and the full carbonate system across three decades (1850s, 2000s, and 2090s). These scalings use public data from the Coupled Model Intercomparison Project Phase 6 (CMIP6) which we group into 14 open-ocean biomes with annual and monthly mean conditions. The biome means reveal that most projected temperature changes are smaller than spatial and seasonal variation, whereas possible future carbonate system changes exceed present day variability. We then compared these projections with 2,456 treatments reported in 196 phytoplankton studies spanning the last two decades. We find that the majority of treatments (57%) fall within the modeled condition space, but many studies examine conditions unrealistic for the past or future open ocean. Finally, we applied North Atlantic conditions in batch cultures of two eukaryotic phytoplankton taxa. The achieved carbonate-system conditions fell within modeled biome variability, and the temperature increase matched projected warming. Together, this framework provides regional and emissions-scenario context for selecting tractable multi-driver treatments and comparing conditions across experiments.