Microbial activity is a fundamental component of oceanic nutrient cycles. Photosynthetic microbes, collectively termed phytoplankton, are responsible for the vast majority of primary production in marine waters. The availability of nutrients in the upper ocean frequently limits the activity and abundance of these organisms. Experimental data have revealed two broad regimes of phytoplankton nutrient limitation in the modern upper ocean. Nitrogen availability tends to limit productivity throughout much of the surface low-latitude ocean, where the supply of nutrients from the subsurface is relatively slow. In contrast, iron often limits productivity where subsurface nutrient supply is enhanced, including within the main oceanic upwelling regions of the Southern Ocean and the eastern equatorial Pacific. Phosphorus, vitamins and micronutrients other than iron may also (co-)limit marine phytoplankton. The spatial patterns and importance of co-limitation, however, remain unclear. Variability in the stoichiometries of nutrient supply and biological demand are key determinants of oceanic nutrient limitation. Deciphering the mechanisms that underpin this variability, and the consequences for marine microbes, will be a challenge. But such knowledge will be crucial for accurately predicting the consequences of ongoing anthropogenic perturbations to oceanic nutrient biogeochemistry

A global three‐dimensional marine ecosystem model with several key phytoplankton functional groups, multiple limiting nutrients, explicit iron cycling, and a mineral ballast/organic matter parameterization is run within a global ocean circulation model. The coupled biogeochemistry/ecosystem/circulation (BEC) model reproduces known basin‐scale patterns of primary and export production, biogenic silica production, calcification, chlorophyll, macronutrient and dissolved iron concentrations. The model captures observed high nitrate, low chlorophyll (HNLC) conditions in the Southern Ocean, subarctic and equatorial Pacific. Spatial distributions of nitrogen fixation are in general agreement with field data, with total N‐fixation of 55 Tg N. Diazotrophs directly account for a small fraction of primary production (0.5%) but indirectly support 10% of primary production and 8% of sinking particulate organic carbon (POC) export. Diatoms disproportionately contribute to export of POC out of surface waters, but CaCO3 from the coccolithophores is the key driver of POC flux to the deep ocean in the model. An iron source from shallow ocean sediments is found critical in preventing iron limitation in shelf regions, most notably in the Arctic Ocean, but has a relatively localized impact. In contrast, global‐scale primary production, export production, and nitrogen fixation are all sensitive to variations in atmospheric mineral dust inputs. The residence time for dissolved iron in the upper ocean is estimated to be a few years to a decade. Most of the iron utilized by phytoplankton is from subsurface sources supplied by mixing, entrainment, and ocean circulation. However, owing to the short residence time of iron in the upper ocean, this subsurface iron pool is critically dependent on continual replenishment from atmospheric dust deposition and, to a lesser extent, lateral transport from shelf regions.

An overview of the Community Earth System Model Version 2 (CESM2) is provided, including a discussion of the challenges encountered during its development and how they were addressed. In addition, an evaluation of a pair of CESM2 long preindustrial control and historical ensemble simulations is presented. These simulations were performed using the nominal 1° horizontal resolution configuration of the coupled model with both the “low‐top” (40 km, with limited chemistry) and “high‐top” (130 km, with comprehensive chemistry) versions of the atmospheric component. CESM2 contains many substantial science and infrastructure improvements and new capabilities since its previous major release, CESM1, resulting in improved historical simulations in comparison to CESM1 and available observations. These include major reductions in low‐latitude precipitation and shortwave cloud forcing biases; better representation of the Madden‐Julian Oscillation; better El Niño‐Southern Oscillation‐related teleconnections; and a global land carbon accumulation trend that agrees well with observationally based estimates. Most tropospheric and surface features of the low‐ and high‐top simulations are very similar to each other, so these improvements are present in both configurations. CESM2 has an equilibrium climate sensitivity of 5.1–5.3 °C, larger than in CESM1, primarily due to a combination of relatively small changes to cloud microphysics and boundary layer parameters. In contrast, CESM2's transient climate response of 1.9–2.0 °C is comparable to that of CESM1. The model outputs from these and many other simulations are available to the research community, and they represent CESM2's contributions to the Coupled Model Intercomparison Project Phase 6.

Abstract. Changes in marine net primary productivity (PP) and export of particulate organic carbon (EP) are projected over the 21st century with four global coupled carbon cycleclimate models. These include representations of marine ecosystems and the carbon cycle of different structure and complexity. All four models show a decrease in global mean PP and EP between 2 and 20% by 2100 relative to preindustrial conditions, for the SRES A2 emission scenario. Two different regimes for productivity changes are consistently identified in all models. The first chain of mechanisms is dominant in the low-and mid-latitude ocean and in the North Atlantic: reduced input of macro-nutrients into the euphotic zone related to enhanced stratification, reduced mixed layer depth, and slowed circulation causes a decrease in macronutrient concentrations and in PP and EP. The second regime is projected for parts of the Southern Ocean: an alleviation of light and/or temperature limitation leads to an increase in PP and EP as productivity is fueled by a sustained nutrient input. A region of disagreement among the models is the Arctic, where three models project an increase in PP while one model projects a decrease. Projected changes in seasonal and interannual variability are modest in most regions. Regional model skill metrics are proposed to generate multimodel mean fields that show an improved skill in representCorrespondence to: M. Steinacher (steinacher@climate.unibe.ch) ing observation-based estimates compared to a simple multimodel average. Model results are compared to recent productivity projections with three different algorithms, usually applied to infer net primary production from satellite observations.

A new marine ecosystem model designed for the global domain is presented, and model output is compared with field data from nine different locations. Field data were collected as part of the international Joint Global Ocean Flux Study (JGOFS) program, and from historical time series stations. The field data include a wide variety of marine ecosystem types, including nitrogen-and iron-limited systems, and different physical environments from high latitudes to the mid-ocean gyres. Model output is generally in good agreement with field data from these diverse ecosystems. These results imply that the ecosystem model presented here can be reliably applied over the global domain.The model includes multiple potentially limiting nutrients that regulate phytoplankton growth rates. There are three phytoplankton classes, diatoms, diazotrophs, and a generic small phytoplankton class. Growth rates can be limited by available nitrogen, phosphorus, iron, and/or light levels. The diatoms can also be limited by silicon. The diazotrophs are capable of nitrogen fixation of N 2 gas and cannot be nitrogen-limited. Calcification by phytoplankton is parameterized as a variable fraction of primary production by the small phytoplankton group. There is one zooplankton class that grazes the three phytoplankton groups and a large detrital pool. The large detrital pool sinks out of the mixed layer, while a smaller detrital pool, representing dissolved organic matter and very small particulates, does not sink. Remineralization of the detrital pools is parameterized with a temperature-dependent function. We explicitly model the dissolved iron cycle in marine surface waters including inputs of iron from subsurface sources and from atmospheric dust deposition. r