Long-term effects of agriculture on soil carbon pools and carbon chemistry along a Hawaiian environmental gradient

Intensive agriculture has the potential to reduce soil carbon stocks in the years following initial cultivation, although the magnitude and direction of the effect can vary with ecosystem and management factors. Agriculture can also shift the carbon chemistry of soils via changes in crop plant chemistry, decomposition, and/or soil amendments e.g. black carbon (i.e. charcoal)]. It is possible that soil carbon levels can recover if intensive cultivation ends, but the factors driving the extent and quality of this recovery are not well understood. Here, we examined soil carbon pool sizes and carbon chemistry >200 years after intensive cultivation by early Hawaiians. We compared soils from an extensive pre-European-contact agricultural field system with reference sites under similar modern management. Sites were selected along a climate and soil weathering gradient to investigate interactions between historic land use and ecosystem properties, such as soil mineralogy, in driving soil carbon recovery. Soil carbon content was measured from 0 to 30 cm depth, and carbon chemistry was assessed using ¹³C nuclear magnetic resonance spectroscopy. Overall, we found significantly lower soil carbon stocks in pre-contact agricultural sites compared to reference sites. Radiocarbon dating of bulk soil carbon showed a trend toward older carbon in agricultural versus reference soils, suggesting decreased retention of newer C in agricultural sites. Radiocarbon dating of macroscopic charcoal particles from under agricultural field walls indicated that there were black carbon inputs concurrent with pre-contact agricultural activity. Nonetheless, black carbon and carbonyl carbon levels were lower in agricultural versus reference soils, suggesting decreased retention of specific carbon groups in cultivated sites. Proteins were the only biomolecule higher in abundance in agricultural versus reference sites. Finally, there was an interacting effect of soil mineralogy and historic land use on soil carbon stocks. Whereas short range order (SRO) minerals were positively associated with total soil carbon overall, differences in soil carbon between agricultural and reference soils were largest in soils with high concentrations of SRO minerals. Our results indicate that the negative effect of agriculture on soil carbon stocks can be long-lived, may be associated with persistent changes in soil carbon chemistry, and can vary with soil mineralogical properties.