Shaver, Guy (Univ. of California Davis, PO 73023, Davis, CA, 95617; Phone: 530-752- 6216; Email: gshaver@ucdavis.edu)

Exploring Field Scale Variability of Soil Physical Properties, Total Carbon, and Carbond Dioxide Flux in a California Agricultural System

G.G.Shaver *, D.E.Rolston, J.W.Hopmans, C.Van Kessel, J.W.Six, A.King, J.Lee, J.Evatt

The ability of soil to sequester carbon is dependant upon the interactions of physical, chemical, and biological states within the soil system. Scales of dependency between soil variables can range from the microscopic interactions at organo-mineral complexes to local and global patterns of heat distribution. Traditional agricultural-plot experiments generally assume weak stationarity between sample variables thus allowing for ‘independence’ between sampling locations. However, most field-scale studies reveal non-stationary mean structures and dependencies between sampling locations and soil parameters. Thus, accounting for spatial and temporal variability in agricultural systems is necessary to accurately scale-up C sequestration estimates from traditional small-scale plot experiments to local and regional ecosystem analyses. We have sampled and performed an exploratory data analysis on the clay content (% clay), bulk density (BD), total carbon (TC), CO₂ flux, soil water content (swc), and soil temperature (Ts) of a 30 hectare California agricultural field, under two years of minimum-tilled wheat production followed by the division of the field into ‘conventional’ and ‘minimum’ tillage treatments and a growing season of maize. TC, % clay, and BD show non-stationary mean structures and spatial autocorrelations. Spatial-temporal relationships were found between %clay & swc (P less than 0.001), swc & TC (P less than 0.01), and Ts & CO₂ flux (P less than 0.001) throughout the growing season. CO₂ flux tended to have higher central medians within the conventionally tilled treatment; however, no significant difference between treatments was noted. Analyses of the ‘median-polished’ residuals from the above data sets nicely show qualitative trends between field variables at the landscape scale.