According to Tilman et al,

"With 18.5 MJ kg^-1 released upon combustion (S5), the gross prairie energy output is 68.1 and 111.0 GJ ha^-1 yr^-1 for degraded and fertile prairie, respectively." (6)

Since a MJ/kg is the same as a GJ/Mg, the range of prairie biomass production can be computed as 68.1/18.5 = 3.68 Mg to 111.0/18.5 = 6.00 Mg. In the main part of the paper, they Tilman, et al, remark

"Fertile lands yield about 50% more LIHD biomass (and bioenergy) than our degraded soils." (7)

Therefore, I used a range of

"... between 3.7 and 5.6 tonnes of biomass per hectare per year ..."

and took a point in the midrange to be 4.6. Assuming about 50% of the biomass is carbon, I arrived at the figure of 2.3 tonnes of carbon per hectare per year.