Abstract. During the Last Glacial Maximum (LGM), atmospheric CO2 was around 90 ppmv lower than during the pre-industrial period. The reasons for this decrease are most often elucidated through factorial experiments testing the impact of individual mechanisms. Due to uncertainty in our understanding of the real system, however, the different models used to conduct the experiments inevitably take on different parameter values and different structures. In this paper, the objective is therefore to take an uncertainty-based approach to investigating the LGM CO2 drop by simulating it with a large ensemble of parameter sets, designed to allow for a wide range of large-scale feedback response strengths. Our aim is not to definitely explain the causes of the CO2 drop but rather explore the range of possible responses. We find that the LGM CO2 decrease tends to predominantly be associated with decreasing sea surface temperatures (SSTs), increasing sea ice area, a weakening of the Atlantic Meridional Overturning Circulation (AMOC), a strengthening of the Antarctic Bottom Water (AABW) cell in the Atlantic Ocean, a decreasing ocean biological productivity, an increasing CaCO3 weathering flux and an increasing deep-sea CaCO3 burial flux. The majority of our simulations also predict an increase in terrestrial carbon, coupled with a decrease in ocean and increase in lithospheric carbon. We attribute the increase in terrestrial carbon to a slower soil respiration rate, as well as the preservation rather than destruction of carbon by the LGM ice sheets. An initial comparison of these dominant changes with observations and paleoproxies other than carbon isotope and oxygen data (not evaluated directly in this study) suggests broad agreement. However, we advise more detailed comparisons in the future, and also note that, conceptually at least, our results can only be reconciled with carbon isotope and oxygen data if additional processes not included in our model are brought into play.

Abstract. During the Last Glacial Maximum (LGM), atmospheric CO 2 was around 90 ppmv lower than during the preindustrial period. Despite years of research, however, the exact mechanisms leading to the glacial atmospheric CO 2 drop 15 are still not entirely understood. Here, a large (471-member) ensemble of GENIE-1 simulations is used to simulate the equilibrium LGM minus preindustrial atmospheric CO 2 concentration difference (∆CO 2 ). The ensemble has previously been weakly constrained with modern observations and was designed to allow for a wide range of large-scale feedback response strengths. Out of the 471 simulations, 315 complete without evidence of numerical instability, and with a ∆CO 2 that centres around -20 ppmv. Roughly a quarter of the 315 runs predict a more significant atmospheric CO 2 drop, between ~30 and 90 20 ppmv. This range captures the error in the model's process representations and the impact of processes which may be important for ∆CO 2 but are not included in the model. These runs jointly constitute what we refer to as the "plausible glacial atmospheric CO 2 change-filtered (PGACF) ensemble".Our analyses suggest that decreasing LGM atmospheric CO 2 tends to be associated with decreasing SSTs, increasing sea ice 25 area, a weakening of the Atlantic Meridional Overturning Circulation (AMOC), a strengthening of the Antarctic Bottom Water (AABW) cell in the Atlantic Ocean, a decreasing ocean biological productivity, an increasing CaCO 3 weathering flux, an increasing terrestrial biosphere carbon inventory and an increasing deep-sea CaCO 3 burial flux. The increases in terrestrial biosphere carbon are predominantly due to our choice to preserve rather than destroy carbon in ice sheet areas. However, the ensemble soil respiration also tends to decrease significantly more than net photosynthesis, resulting in relatively large 30 increases in non-burial carbon. In a majority of simulations, the terrestrial biosphere carbon increases are also accompanied by decreases in ocean carbon and increases in lithospheric carbon. In total, however, we find there are 5 different ways of 2Comparison of the PGACF ensemble results against observations suggests that the simulated LGM physical climate and biogeochemical changes are mostly of the right sign and magnitude or within the range of observational error, except for the change in global deep-sea CaCO 3 burial flux -which tends to be overestimated. We note that changing CaCO 3 weathering flux is a variable parameter (included to account for variation in both the CaCO 3 weathering rate and the un-modelled CaCO 3 shallow water deposition flux), and this parameter is strongly associated with changes in global CaCO 3 burial rate. The 5 increasing terrestrial carbon inventory is also likely to have contributed to the LGM increase in deep-sea CaCO 3 burial flux via the process of carbonate compensation. However, we do not yet rule out either of these processes as causes of ∆CO 2 since missing processes such as Si fertilisation, Si leakage and the effect of d...