Partitioning the soil surface CO₂ flux (R_S) flux is an important step in understanding ecosystem-level carbon cycling, given that R_S is poorly constrained and its source components may have different sensitivities to climate change. Trenched plots are an inexpensive but labor-intensive method of separating the R_S flux into its root (autotrophic) and soil (heterotrophic) components. This study tested if various methods of plant suppression in trenched plots affected R_S fluxes, quantified the R_S response to soil temperature and moisture changes, and estimated the heterotrophic contribution to R_S. It was performed in a boreal black spruce (Picea mariana) plantation, using a randomized complete block design, during the 2007 and 2008 growing seasons. Trenched plots had significantly lower R_S than control plots, with differences appearing ∼100 days after trenching; spatial variability doubled immediately after trenching but then declined throughout the experiment. Most trenching treatments had significantly lower (by ∼0.5 μmol CO₂ m⁻² s⁻¹)R_S than the controls, and there was no significant difference in R_S among the various trenching treatments. Soil temperature at 2 cm explained more R_S variability than did 10-cm temperature or soil moisture. Temperature sensitivity (Q₁₀) declined in the control plots from ∼2.6 (at 5 °C) to ∼1.6 (at 15 °C); trenched plots values were higher, from 3.1 at 5 °C to 1.9 at 15 °C. We estimated R_S for the study period to be 241 ± 40 g C m⁻², with live roots contributing 64% of R_S after accounting for fine root decay, and 293 g C m⁻² for the entire year. These findings suggest that laborious hand weeding of trenched plot vegetation may be replaced by other methods, facilitating future studies of this large and poorly-understood carbon flux.