Fractionating soils so that others do not have to: radiocarbon informs choice of method for scale, ecosystem, or process

Physical separation of soil into various fractions has long been used to address questions concerning mechanisms of soil organic matter stabilization, processes contributing to soil carbon (C) accumulation, and the effects of land use, climate change, and management practice on soil quality and carbon sequestration. However, no published method works well for every soil, ecosystem, or research question. Often a chosen method does not effectively separate soil into C pools that differ in mean residence time (MRT) and sensitivity to change, which can complicate the interpretation of results. Soil C is stabilized by a variety of mechanisms and radiocarbon-based estimates of MRT can reveal the integrated effects of these mechanisms on bulk soil C storage; radiocarbon-based estimates of MRT on isolated soil fractions separated by a carefully chosen method can reveal internal C dynamics invisible to the bulk soil methods. A variety of soils collected around Hawaii (Mollisol, Oxisol, Andisol) were fractionated by several common methods and the radiocarbon-based MRT was estimated for comparison among fractions, soils, and methods. In some cases, depending on the research question of interest, aspects of different methods could be combined to reveal changes in soil C pools on both short and long time frames. For example, for a cultivated, mixed-mineralogy Mollisol, a method that combined a density separation at 1.8 g mL^-1 for free light fraction, then calibrated sonication to disrupt aggregates for an occluded fraction, then further sequential fractionation at 2.0 g ^-1could produce soil C pools with turnover of 3.5 yr (free light fraction), 10 yr (occluded light fraction), 714 yr (1.8-2.0 fraction), and 2090 yr (<2.0 fraction). This method may be ideal for tracking short-term (1-10 year) changes in soil structure due to sustainable agriculture or management practices. For an Andisol, land conversion from old-growth native forest to 80 yr of pasture increased bulk soil C stock in the top 0-15 cm of mineral soil by 26%; however, sequential density separation into 7 fractions revealed 50-69% increases in C within low density fractions with MRT of less than 5 yr but over 300% losses of soil C within dense fraction with MRT of over 1275 yr. In these Andisols, the sequential density fractionation method was highly sensitive to land use change and the range of densities are hypothesized to be associated with different mechanisms for soil C stabilization acting over different time scales, which was confirmed by the radiocarbon-based MRT estimates. Although soil fractionation methods are powerful, other results from similar Andisols suggest that over geologic time scales MRT estimates for bulk soil profiles can be more informative than soil fractions. Careful consideration of the scientific question, study system, and scale is important when choosing a method for fractionating soil. Radiocarbon measurements can provide confirmation that the actual nature of the recovered fractions matches the theoretical one.