Global average temperatures are forecast to increase by 4 °C by 2100 under the Intergovernmental Panel on Climate Change SSP5–8.5 scenario. This warming could accelerate soil organic carbon (SOC) mineralization, net loss of soil carbon to the atmosphere, and consequently exacerbate global warming through a positive feedback loop. It is generally assumed that mineral–associated organic matter (MAOM) is less sensitive to warming compared to particulate organic matter (POM), especially in subsoil; yet more empirical data investigating the whole–soil response to warming is still required to test this assumption.</p> <p>Our study was conducted in a whole–soil field warming experiment in a temperate mixed–conifer forest at Blodgett Forest Research Station, University of California, Berkeley, which had been subjected to 10 years of warming. Soils taken at three depths (10–20, 40–50, and 80–90 cm) were separated into three density fractions, and we then investigated the SOC concentration (elemental analysis) and composition of bulk soil and fractions with diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy. We found that a decade of experimental warming shifted subsoil bulk SOC composition towards lignin and C–H aromatic bonds. warmed plots had significantly lower mass of the POM fractions relative to control plots in the subsoil (80–90 cm), but there was no difference in the topsoil, which could occur if higher decomposition losses of POM were obscured by fresh plant inputs. In contrast, the mass of MAOM and its chemical composition was not different between warmed and control treatments, but did shift along the depth gradient. This study thus supports the assumption that POM is more responsive to warming than MAOM, which was particularly evident in the subsoil at Blodgett Forest.