Ocean productivity is one of the most important processes on the planet, as it fuels marine food webs and helps to regulate the global carbon cycle. Phytoplankton, unicellular photosynthetic organisms that live in the sunlit surface waters of the ocean, perform photosynthesis to drive this global process. Yet, despite the importance of ocean productivity, it is unclear how it will respond to a changing climate.

Future projections of phytoplankton productivity from climate models are critical because they are used to forecast impacts on ecosystems and biogeochemical cycles. However, at present we lack a clear answer, with some models predicting an increase in future ocean productivity while others suggest a decline. This disagreement amongst models increases uncertainty and reduces confidence in future model projections of ocean productivity. These uncertainties propagate through different ecosystem models, which rely on ocean productivity projections to assess how fisheries and other key ecosystem services will be impacted by climate change.

Using a satellite-record dating back to 1998 we examine how ocean productivity is changing utilising a suite of different algorithms that quantify productivity from optical properties. We find that across the globe there is a general decline in ocean productivity. These declines are associated with increasing ocean temperatures and parallel changes in water column structure that dictate the nutrient and light availability for phytoplankton.

We exploited the interlinked relationships between ocean productivity and environmental drivers to build a ranking system that assesses how well different climate models represent the relationships observed from satellites. Models which project declines by the end of the century were ranked highest, whilst those that projected increases were not. However, even in the best ranked models their predicted declines in production were not as severe as those estimated from satellites over the past 26 years.

Our study suggests that the reason for the mismatch in the magnitude of productivity declines between models and satellites is linked to the model’s inability to correctly reproduce the relationship between declining ocean productivity and increasing ocean temperatures. This link in itself likely integrates a range of mechanisms associated with the direct effect of temperature, but also the co-varying impacts of resource availability and other ecological factors. This implies that any improvement to climate models, which can more accurately represent this sensitivity, may project even greater declines in future ocean productivity, even under high mitigation scenarios that currently predict relatively stable productivity. This would have important ramifications for the ocean carbon cycle and marine ecosystems, affecting the trophic system that underpins biodiversity, fisheries and marine resources.