Temporal variability is a major source of uncertainty in soil carbon measurements

Abstract

Monitoring changes in soil organic carbon over time is important to many agricultural and environmental goals. Despite decades of measurements, temporal variability in soil carbon measurements has not been studied extensively. In this report, we examine five sets of monthly samples extending up to 3 years each that were collected from field experiments at four locations representative of dryland farming in the Pacific Northwest. The variance from month-to-month was 15%–32% of the random error, averaging 20%. This was often greater than the variance between replicate experimental units (2%–42%, averaging 17%). At certain sites, sequential samples were found to be temporally autocorrelated, but no consistent trend patterned on seasonal factors like precipitation was found. This suggests that a single point-in-time sample can deviate substantially from the long-term average soil carbon at the site. We illustrate this problem with the results of repeated soil samples taken from 12 commercial farm fields. We recommend that confidence intervals for soil organic carbon estimates should include variance based on a large population of samples rather than from a single sample set at one timepoint.