On the global scale, soils contain four-times more carbon (C) than the atmosphere and the magnitude of annual soil CO₂ efflux (R_s) is nearly an order of magnitude higher than anthropogenic C emissions. However, considerable uncertainties still exist concerning the biophysical controls of R_s and its possible feedback to climate change. We report continuous measurements of R_s made with automated chambers, and its biophysical controls in two different aged (20 and 60 years old, hereafter known as HDF88 and DF49, respectively) Douglas-fir stands on Vancouver Island, Canada. The two stands, located within similar biogeoclimatic units of the dry maritime Coastal Western Hemlock subzone, CWHxm, are on humo-ferric podzol soils with similar concentrations of C and N measured in the surface organic and mineral soil layers with a somewhat thicker surface organic layer at HDF88. HDF88 with an average tree height of about 8 m has a dense, mainly deciduous, understory, whereas understory at DF49 is sparse. The mean annual temperature and precipitation are 9.6 and 8.6°C and 1610 and 1470 mm at HDF88 and DF49, respectively, and soil water deficits frequently occur in August, September and October.

Our multi-year (2004-2009 at DF49 and 2006-2009 at HDF88) continuous measurements showed that both daily and hourly R_s values in the two stands showed different types of responses to shallow-depth soil temperature (T_s) and volumetric soil water content (q). While daily R_s followed a quadratic relationship with T_s at HDF88 with a maximum at a T_s of 13 °C, it increased exponentially at DF49 throughout the range of measured T_s. Similarly, daily R_s at HDF88 increased with q, reached a maximum at a q of 0.11 m³ m^-3 and declined thereafter, whereas it linearly decreased with q at DF49 over the entire range of q. Time series analysis revealed that a photosynthetic signal (from eddy-covariance-measured gross ecosystem photosynthesis, GEP) in chamber-measured R_s appeared at different temporal scales in the two stands. On the annual scale, daily R_s lagged behind daily GEP by about 3 weeks at DF49 with no lag at HDF88. However, on the seasonal scale, daily R_s lagged behind daily GEP by 12 days at both sites but only in the cool and wet season (November-March). These lags corresponded with the 12-day lag of air temperature (T_a) behind solar irradiance (S_o) during the cool and wet season at both the sites. On the other hand, daily R_s values showed no lags with respect to GEP during the warm and moist (April-July) and warm and dry (August-October) seasons. On the diurnal scale, half-hourly R_s values lagged behind half-hourly GEP by 4 hours, which corresponded closely with the diurnal lag of T_a behind S_o. Furthermore, R_s was highly correlated to ecosystem respiration (R_e) and GEP but followed slightly different types of relationships at the two sites. The annual ratio of R_s to R_e was 0.94 and 0.62 at HDF88 and DF49, respectively. Implications of the results with respect to the possibility and usefulness of incorporating GEP in soil respiration models are discussed.