Resource information
Relationships among biological indicators of soil quality and organic matter characteristics were evaluated across a continuum of long-term agricultural practices in Missouri, USA. In addition to chemical and physical soil quality indicators, dehydrogenase and phenol oxidase activity were measured, ¹³C nuclear magnetic resonance (¹³C NMR) and diffuse reflectance Fourier transform (DRIFT) spectra of soil organic matter were collected, and visible, near-infrared reflectance (VNIR) spectra of whole soil were collected. Enzyme activities were positively correlated with several soil quality indicators and labile fractions of soil organic matter (r� =� 0.58–0.92), and were negatively correlated with DRIFT indices of decomposition stage and recalcitrance (r� =� −0.62 to −0.76). A comparison of vegetative and land management practices was scored using the soil management assessment framework (SMAF)—a soil quality index. Perennial vegetation (i.e., native prairie, restored prairie, and timothy) plots exhibited the greatest soil quality (SMAF scores 93.6–98.6 out of 100), followed by no-till and conventionally cultivated plots, with wheat outranking corn. Among fertilization practices, soil quality followed the order: manure� >� inorganic fertilizer� >� unamended soil. Finally, in the estimation of soil properties, VNIR spectra generally outperformed DRIFT spectra using partial least squares regression� (PLSR) and multiple, linear regression (MLR). The strongest estimates of dehydrogenase and phenol oxidase activity were found using MLR models of VNIR spectra (R²� >� 0.78, RPD� >� 2.20). Overall, this study demonstrates the potential utility and versatility of enzymes in modeling and assessing changes in soil organic carbon fractions and soil quality, and emphasizes the benefits of maintaining long-term agricultural experiments.
