While coupling carbon and nitrogen processes is critical for earth system models to accurately predict future climate and land biogeochemistry feedbacks, it has not yet been analyzed how numerical methods that land biogeochemical models apply to couple soil mineral nitrogen mobilizing and immobilizing processes affect predicted ecosystem carbon and nitrogen cycling. These effects were investigated here by using the E3SM land model and an evaluation of three plausible and widely used numerical couplings: (1) the mineral nitrogen based limitation scheme, (2) the net uptake based limitation scheme, and (3) the proportional nitrogen flux based limitation scheme. It was found that these three schemes resulted in large differences (with a range of 316 Pg C) in predicted cumulative land-atmosphere carbon exchanges under the RCP4.5 atmospheric CO2 simulations. This large uncertainty is without accounting for the different representations of the many land biogeochemical processes, but is about 73% of the range (434 Pg C) reported for CMIP5 RCP4.5 simulations. These results help explain the large uncertainty found in various model inter-comparison experiments and suggest that more robust numerical implementations are needed to improve carbon-nutrient cycle coupling in terrestrial ecosystem models.