LAUSR

Abstract. Climate variations can have profound impacts on marine ecosystems and the socioeconomic systems that may depend upon them. Temperature, pH, oxygen ( O2 ) and net primary production (NPP) are commonly considered to be important marine ecosystem drivers, but the potential predictability of these drivers is largely unknown. Here, we use a comprehensive Earth system model within a perfect modeling framework to show that all four ecosystem drivers are potentially predictable on global scales and at the surface up to 3 years in advance. However, there are distinct regional differences in the potential predictability of these drivers. Maximum potential predictability ( >10  years) is found at the surface for temperature and O2 in the Southern Ocean and for temperature, O2 and pH in the North Atlantic. This is tied to ocean overturning structures with “memory” or inertia with enhanced predictability in winter. Additionally, these four drivers are highly potentially predictable in the Arctic Ocean at the surface. In contrast, minimum predictability is simulated for NPP (  years) in the Southern Ocean. Potential predictability for temperature, O2 and pH increases with depth below the thermocline to more than 10 years, except in the tropical Pacific and Indian oceans, where predictability is also 3 to 5 years in the thermocline. This study indicating multi-year (at surface) and decadal (subsurface) potential predictability for multiple ecosystem drivers is intended as a foundation to foster broader community efforts in developing new predictions of marine ecosystem drivers.