Changes to particulate versus mineral-associated soil carbon after 50 years of litter manipulation in forest and prairie experimental ecosystems.

Models of ecosystem carbon (C) balance generally assume a strong relationship between NPP, litter inputs, and soil C accumulation, but there is little direct evidence for such a coupled relationship. Using a unique 50-year detrital manipulation experiment in a mixed deciduous forest and in restored prairie grasslands in Wisconsin, combined with sequential density fractionation, isotopic analysis, and short-term incubation, we examined the effects of detrital inputs and removals on soil C stabilization, destabilization, and quality. Both forested sites showed greater decline in bulk soil C content in litter removal plots (55 and 66 %) compared to increases in litter addition plots (27 and 38 % increase in surface soils compared to controls). No accumulation in the mineral fraction C was observed after 50 years of litter addition of the two forested plots, thus increases in the light density fraction pool drove patterns in total C content. Litter removal across both ecosystem types resulted in a decline in both free light fraction and mineral C content, with an overall 51 % decline in mineral-associated carbon in the intermediate (1.85-2.4 g cm) density pool; isotopic data suggest that it was preferentially younger C that was lost. In contrast to results from other, but younger litter manipulation sites, there was with no evidence of priming even in soils collected after 28 years of treatment. In prairie soils, aboveground litter exclusion had an effect on C levels similar to that of root exclusion, thus we did not see evidence that root-derived C is more critical to soil C sequestration. There was no clear evidence that soil C quality changed in litter addition plots in the forested sites; δC and ΔC values, and incubation estimates of labile C were similar between control and litter addition soils. C quality appeared to change in litter removal plots; soils with litter excluded had ΔC values indicative of longer mean residence times, δC values indicative of loss of fresh plant-derived C, and decreases in all light fraction C pools, although incubation estimates of labile C did not change. In prairie soils, δC values suggest a loss of recent C4-derived soil C in litter removal plots along with significant increases in mean residence time, especially in plots with removal of roots. Our results suggest surface mineral soils may be vulnerable to significant C loss in association with disturbance, land use change, or perhaps even climate change over century-decadal timescales, and also highlight the need for longer-term experimental manipulations to study soil organic matter dynamics. [ABSTRACT FROM AUTHOR]