Description: In this study, we investigate the role of macrozooplankton in the biogeochemistry of the Southern Ocean using a three-dimensional global ocean ecosystem model (FESOM- REcoM2). The macrozooplankton group was parameterized according to characteristics of Antarctic krill and a related fast-sinking detritus class (larger particles, e.g. fecal pellets) was introduced in the model. It was then analyzed how the ecosystem structure and major carbon export pathways in the Southern Ocean changed through this extension of the model. The spatial distribution of macrozooplankton biomass in the Southern Ocean was reproduced reasonably well. Preliminary results showed that the zooplankton proportion of living compartments (phytoplankton and zooplankton groups) in the model increased. Thus, zooplankton contribution to the particulate organic carbon (POC) flux increased. The contribution of macrozooplankton to POC export at 100 m depth was 0.12 Pg C per year or 15% of total export in the Southern Ocean. The transfer efficiency of organic carbon nearly doubled and reached up to 50% in regions with high macrozooplankton biomass. These results emphasize the important role of macrozooplankton in the Southern Ocean carbon cycle and have implications for studies of the biological carbon pump.

Description: Low concentrations of iron, an important micronutrient for photosynthetic organisms, limit growth in large parts of the ocean. The solubility and availability of iron is to a large degree determined by organic iron-binding molecules, so-called ligands. While ligands come from a variety of sources, all are produced in autotrophic or heterotrophic production, leading to the possibility of feedbacks between primary production and iron availability in the ocean. The diversity of ligands, reaching from siderophores, small molecules involved in bacterial iron uptake, to breakdown products and long-lived macromolecules like humic substances, means that feedbacks could be both negative and positive. Here we investigate first, how the cycling of this diverse ligand pool can be described simplistically in a model such that it reproduces the observed global distribution of dissolved iron and phosphorus as closely as possible. We show that inclusion of a ligand similar to refractory dissolved organic carbon leads to an improved agreement to observations in our model. Inclusion of a second, shorter-lived siderophore-like ligand does not strongly affect this agreement. In a second step we then study how feedbacks affect how iron distribution and oceanic productivity react to changes in external supply of iron. We show that, to be consistent with present-day iron distribution, the dominant feedback is positive, increasing the sensitivity of global biological productivity and hence carbon cycling to changes in iron supply. The strength of the feedback increases with increasing ligand life-time. The negative feedback associated with siderophore-like ligands has the potential to mitigate the positive feedback, especially at the surface and for global export production, but more research on the production and decay of siderophores is needed for a better quantification. Ocean biogeochemical models that assume a constant ligand concentration and hence neglect possible feedbacks may therefore underestimate the reaction of the global carbon cycle to the strong increase in dust deposition under future or glacial climate conditions.

Description: Mineral dust aerosol constitutes an important component of the Earth’s climate system, not only on short timescales due to direct and indirect influences on the radiation budget but also on long timescales by acting as a fertilizer for the biosphere and thus affecting the global carbon cy- cle. For a quantitative assessment of its impact on the global climate, state-of-the-art atmospheric and aerosol models can be utilized. In this study, we use the ECHAM6.3-HAM2.3 model to perform global simulations of the mineral dust cy- cle for present-day (PD), pre-industrial (PI), and last glacial maximum (LGM) climate conditions. The intercomparison with marine sediment and ice core data, as well as other mod- eling studies, shows that the obtained annual dust emissions of 1221, 923, and 5159 Tg for PD, PI, and LGM, respectively, generally agree well with previous findings. Our analyses fo- cusing on the Southern Hemisphere suggest that over 90 % of the mineral dust deposited over Antarctica are of Australian or South American origin during both PI and LGM. How- ever, contrary to previous studies, we find that Australia con- tributes a higher proportion during the LGM, which is mainly caused by changes in the precipitation patterns. Obtained in- creased particle radii during the LGM can be traced back to increased sulfate condensation on the particle surfaces as a consequence of longer particle lifetimes. The meridional transport of mineral dust from its source regions to the South Pole takes place at different altitudes depending on the grain size of the dust particles. We find a trend of generally lower transport heights during the LGM compared to PI as a con- sequence of reduced convection due to colder surfaces, indi- cating a vertically less extensive Polar cell.

Description: Biogeochemical models are used to project future plankton community composition and biogeochemical fluxes. Phytoplankton growth therein is usually parametrized by sensitivities to the bottom-up factors temperature, light, and nutrient availability. However, many laboratory studies identify the carbonate system as an additional and essential growth-determining factor, especially in light of ongoing ocean acidification. Besides, growth-responses towards one factor are often altered by the level of another factor, and these so-called interactive effects are barely considered in models. In the presented work, model functions for carbonate system dependencies of growth and calcification were developed based on published results of laboratory data and implemented into a biogeochemical model. Using the results of an earlier meta-analysis on dual driver interactions, this new model setup was then substituted by interactive growth effects between the carbonate system, temperature, and light. End-of-century phytoplankton biomass and community composition in a high-emission scenario was projected by using one model version with and one model version without driver interactions. The results reveal that interactive growth effects considerably alter the future community composition compared to the model version without interactions. These alterations are largest in the Southern Ocean. Globally, the model with interactions projects a future phytoplankton community consisting out of more small phytoplankton and fewer diatoms and coccolithophores. Hence, considering interactive growth effects between bottom-up factors can essentially modify the projections of future phytoplankton community composition and related biogeochemical fluxes, and should be considered more closely in biogeochemical models.

Description: Among mechanisms accounting for atmospheric pCO2 drawdown during glacial periods, processes operating in the North Atlantic (NA) and Southern Ocean (SO) have been proposed to be critical. Their individual and synergic effects during a course of glaciation, however, remain enigmatic. We conducted simulations to examine these effects at idealized glacial stages. Under early-glacial-like conditions, cooling in the SO can trigger an initial pCO2 drawdown, while the associated sea ice expansion has little impact on air-sea gas exchange. Under later glacial-like conditions, further cooling in the NA enhances ocean carbon uptake due to a stronger solubility pump, and the SO-induced stronger deep stratification prevents carbon exchange between the deep and upper ocean. Meanwhile, strengthened dust deposition increases the SO contribution to the global biological pump, and CO2 outgassing is suppressed by fully extended sea ice cover. More carbon is then stored in the deep Pacific acting as a passive reservoir.

Description: Marine biogeochemical (BGC) models are highly uncertain in their parameterization. The value of the BGC parameters are poorly known and lead to large uncertainties in the model outputs. This study focuses on the uncertainty quantification of model fields and parameters within a one-dimensional (1-D) ocean BGC model applying ensemble data assimilation. We applied an ensemble Kalman filter provided by the Parallel Data Assimilation Framework (PDAF) into a 1-D vertical configuration of the BGC model Regulated Ecosystem Model 2 (REcoM2) at two BGC time-series stations: the Bermuda Atlantic Time-series Study (BATS) and the Dynamique des Flux Atmosphériques en Méditerranée (DYFAMED). We assimilated 5-day satellite chlorophyll-a (chl-a) concentration and monthly in situ net primary production (NPP) data for 3 years to jointly estimate 10 preselected key BGC parameters and the model state. The estimated set of parameters resulted in improvements in the model prediction up to 66% for the surface chl-a and 56% for NPP. Results show that assimilating satellite chl-a concentration data alone degraded the prediction of NPP. Simultaneous assimilation of the satellite chl-a data and in situ NPP data improved both surface chl-a and NPP simulations. We found that correlations between parameters preclude estimating parameters independently. Co-dependencies between parameters also indicate that there is not a unique set of optimal parameters. Incorporation of proper uncertainty estimation in BGC predictions, therefore, requires ensemble simulations with varying parameter values.