In the oceans, inflow of anthropogenic carbon dioxide through air sea gas exchange triggers a series adjustments in bicarbonate and carbonate ion concentrations that causes the pH to decline and net storage in dissolved inorganic carbon (DIC) to be more than a factor of 10 smaller than what would be expected if DIC adjusted directly in proportion to the overlying changes in atmospheric CO2 mole fraction. This effect, known as the Revelle factor, is implicitly built into all state-of-the-art ocean biogeochemistry models through equations that represent the different forms of DIC as a function of temperature, alkalinity, pressure, salinity and the concentration of several other ions. On land, in contrast, carbon storage is often closely regulated by the initial response of gross primary production (GPP) to different forms of global change. The distribution of turnover times in downstream pools (as well as their sensitivity to climate) can modify rates of carbon storage in important ways, yet the donor pool structure of many models means that downstream adjustments to inflows do not offer as much resistance to carbon storage as compared to what occurs in the oceans by means of changes in carbonate chemistry. Here we describe several processes that operate on individual plant, community, stand, and ecosystem levels that may limit carbon storage in terrestrial ecosystems, yet are not well represented in models. These include downregulation of CO2 fertilization effects on net primary production relative to effects on gross primary production, adjustments in tree mortality as a function of tree biomass, changes in stand-level fire disturbance as a function biomass accumulation in live, coarse woody debris, and litter pools, and limits to carbon storage in soils posed by fixed number of binding sites to clay and mineral surfaces in soils. We highlight some of these limits using analysis of earth system models from the 6th Coupled Model Intercomparison Project (CMIP6) and through analysis of global fire and soil carbon age datasets.