Deep Soil Carbon: Quantification and Modeling in Subsurface Layers

Soil is the primary sink for C in forest ecosystems but is often overlooked in ecosystem C budgets. Efforts to quantify C pools often sample soils to a depth of 0.2 m despite observations that deep soil C is neither scarce nor entirely stable. This study examined the systematic sampling depth for ecosystem C analyses in the Pacific Northwest and compared best‐fit models of C in deep soil layers with laboratory measurements. Forest floor samples and mineral soil bulk density samples were collected at regular intervals from the soil surface to depths of 2.5 m from 22 sites across the coastal Pacific Northwest Douglas‐fir [ Pseudotsuga menziesii (Mirb.) Franco] zone. Soil samples were screened to 4.7 mm and analyzed for C content. We found that systematic soil sampling shallower than 1.5 m significantly underestimated total soil C. On average, sampling to 2.5 m compared with 0.5 m increased total C by 156% (85.3–132.7 Mg ha −1 ), and 21% of total C within the depth range sampled was below 1.0 m. A nonlinear mixed model using an inverse polynomial curve form and predicting total C to 2.5 m given only data to 1.0 m was reliable for 20 of 22 sites; the sites that could not be accurately modeled carried the greatest C at depth and contained noncrystalline minerals. Shallow soil sampling at best provides a biased estimate and at worst leads to misleading conclusions regarding soil C. Researchers seeking to quantify soil C or measure change with time should sample deep soil to create a more complete picture of soil pools and fluxes.