Our present understanding of the mantle carbon budget is in part built upon measurements of carbon concentrations in olivine hosted melt inclusions. Only a small number of such datasets are thought to have avoided degassing, having been entrapped prior to CO2 vapour saturation, and are therefore able to constrain primary CO2 concentrations. The absence of degassing in melt inclusion datasets has been inferred from the presence of strong correlations between CO2 and trace elements. In this contribution, we demonstrate that partial degassing followed by magma mixing not only retains such positive correlations, but can enhance them.
Simple models of magma mixing and degassing are used to characterise how CO2-trace element systematics respond to CO2 vapour saturation in primary mantle melts entering the crust, followed by magma mixing. Positive correlations are expected between CO2 and most trace elements, and the average CO2/Ba and CO2/Nb ratios are controlled by the pressure of magma storage, rather than the CO2 concentration in the mantle. We find that the best estimates of mantle CO2 are the maximum CO2/Ba ratios observed in melt inclusion datasets, though a large number of analyses are required to adequately characterise the maximum of the CO2/Ba distribution. Using the mixing and degassing models we estimate the number of analyses required to obtain a maximum CO2/Ba observation within 10% of the mantle value. In light of our results, we reassess existing melt inclusion datasets, and find they exhibit systematics associated with partial degassing and mixing. We argue that all the data presently available is consistent with a depleted mantle CO2/Ba ratio of ~140, and there is as yet no evidence for heterogeneity in the CO2/Ba ratio of the depleted mantle.