Abstract. Ecosystem shifts related to climate change are anticipated for the next decades to centuries based on a number of conceptual and experimentally derived models of plant structure and function. Belowground, the potential responses of soil systems are less well known. We used geochemical steady state models, soil density fractionation, and soil radiocarbon data to constrain changes in soil carbon based on measurements from detrital (free light), aggregate-bound (occluded) and complexed or chemically bound (mineral associated) carbon pools and for bulk soil. We explored a space-for-time sequence of soils along a cold-to-warm climatic gradient from Alaskan Black Spruce forest soil with permafrost (Gelisols; 50 cm Mean Annual Temperature −1.5 ºC), Alaskan White Spruce forest soil lacking permafrost (Inceptisols; 50 cm MAT +3 ºC ), and Iowa Grassland soil lacking permafrost (Mollisols; 50 cm MAT +9 ºC) developed on similar geologic substrates (wind-blown loess deposits). These temperature ranges were also representative of temperatures at 50 cm soil depth from model output by the Community Land Model for the years 2014, 2100, and 2300 for Interior Alaska. Fitting an exponential equation to depth trends in soil C down to 2 m depths, we found that depth distributions of organic C were related mainly to depths of rooting and changes in bulk density. Using output from the geochemical steady state model, the direction and magnitude of the C loss or gain upon ecosystem shift was dictated by the C stocks of initial and final ecosystems. Radiocarbon measurements specific to each soil fraction (free light, occluded, and mineral associated) allowed us to constrain the timing of the potential loss or gain of C in each fraction driven by climatic shifts. Thawing from the Gelisol to Inceptisol in loess parent materials from present day to year 2100 resulted in small net gains to soil C, reflecting the net balance between loss of detrital and gain into occluded and mineral associated C. Greater warming and shifts from Inceptisol to Mollisol analogous to predicted warming from circa 2100 to 2300 resulted in net C losses from both occluded and mineral associated C, although small gains to the free light C fraction occurred throughout the depth profile. Gains to occluded and mineral associated C post- thaw likely reflect aggregate formation and physical protection of C as well as formation of organo-mineral compounds that accompany microbial processing. Greater warming and shifts from Inceptisol to Mollisol, which are analogous to predicted warming circa 2100 to 2300, resulted in net C losses from both occluded and mineral associated C resulting from enhanced decomposition, small gains to the free light C fraction occurred throughout the transition to Mollisol reflecting deeper rooting of the tallgrass prairie system.