Valid inferences about soil carbon in heterogeneous landscapes

Abstract

Answering research and policy questions about soil organic carbon (SOC) sequestration requires reliably quantifying SOC stocks and changes in stocks. SOC measurement accuracy is limited by the natural spatial heterogeneity of SOC concentrations and soil bulk density, variations in sample preparation, variability of laboratory assays, unmet assumptions in statistical analyses, and the relatively small magnitude of SOC changes over time. Simulations and data from two intensive field sampling campaigns on one rangeland site (662 samples) and seven cropland sites (455 samples) in California, USA, show (1) spatial heterogeneity is a primary driver of uncertainty; (2) dry combustion assays have relatively low uncertainty, although inorganic C can increase error; (3) Student’s t-test and its relatives can be unreliable for drawing inferences about changes in SOC stocks; and (4) in heterogeneous agricultural landscapes, common sample sizes of 10–30 cores cannot reliably detect the modest changes in SOC that a few years of common management interventions are expected to produce. To inform policy decisions, investigators should: (1) identify the measurement goal, including the consequences of error; (2) use random sampling with sample sizes adequate to meet that goal, accounting for anticipated spatial heterogeneity, e.g., by conducting pilot surveys; (3) composite samples less than is currently common; (4) assay with dry-combustion instruments; and (5) use nonparametric statistical te…