B21G-0146:
Carbon - Bulk Density Relationships for Highly Weathered Soils of the Americas

Tuesday, 16 December 2014
Lucas E Nave, University of Michigan, Ann Arbor, MI, United States
Abstract:
Soils are dynamic natural bodies composed of mineral and organic materials. As a result of this mixed composition, essential properties of soils such as their apparent density, organic and mineral contents are typically correlated. Negative relationships between bulk density (Db) and organic matter concentration provide well-known examples across a broad range of soils, and such quantitative relationships among soil properties are useful for a variety of applications. First, gap-filling or data interpolation often are necessary to develop large soil carbon (C) datasets; furthermore, limitations of access to analytical instruments may preclude C determinations for every soil sample. In such cases, equations to derive soil C concentrations from basic measures of soil mass, volume, and density offer significant potential for purposes of soil C stock estimation. To facilitate estimation of soil C stocks on highly weathered soils of the Americas, I used observations from the International Soil Carbon Network (ISCN) database to develop carbon – bulk density prediction equations for Oxisols and Ultisols. Within a small sample set of georeferenced Oxisols (n=89), 29% of the variation in A horizon C concentrations can be predicted from Db. Including the A-horizon sand content improves predictive capacity to 35%. B horizon C concentrations (n=285) were best predicted by Db and clay content, but were more variable than A-horizons (only 10% of variation explained by linear regression). Among Ultisols, a larger sample set allowed investigation of specific horizons of interest. For example, C concentrations of plowed A (Ap) horizons are predictable based on Db, sand and silt contents (n=804, r2=0.38); gleyed argillic (Btg) horizon concentrations are predictable from Db, sand and clay contents (n=190, r2=0.23). Because soil C stock estimates are more sensitive to variation in soil mass and volume determinations than to variation in C concentration, prediction equations such as these may be used on carefully collected samples to constrain soil C stocks. The geo-referenced ISCN database allows users the opportunity to derive similar predictive relationships among measured soil parameters; continued input of new datasets from highly weathered soils of the Americas will improve the precision of these prediction equations.