Observational constraints on the causes of Holocene CO2 change

The mechanisms that controlled past atmospheric CO2 levels are not directly measurable, hence many proxy data sources are combined when reconstructing past carbon cycling. The accuracy of Holocene modeling reconstructions is checked by seeking consistency between data-based observables and their numerically simulated counterparts. A new framework is presented to evaluate which combinations of observables can best constrain carbon cycle mechanisms with the minimum of uncertainty. We show that when previous studies have combined ocean temperatures, ocean [CO32−], and the δ13C of atmospheric CO2 as observables, uncertainties in the data sources are amplified by over 2 orders of magnitude when reconstructing the mechanisms responsible for CO2 increase. However, incorporating mean δ13C of ocean DIC since 8000 years ago as an additional data source reduces the uncertainties by more than a factor of 5, making this observable a priority for future research. Our analysis indicates that the 20 ppm increase in CO2 between 8000 years BP and preindustrial was caused by significant CaCO3 precipitation and a reduction in the ocean soft tissue pump. Meanwhile, an increase in terrestrial carbon storage opposed the CO2 increase. The methods presented here are useful for investigating a range of paleoclimate events