Soil respiration was measured over four growing seasons (2007-2010) at the Borden Forest Research Station, a deciduous mixedwood forest on the lands of the Canadian Forces Base Borden, in southern Ontario (44°19'N, 79°56'W). An automated single-chamber system (LI-COR Inc., model LI-8100) with a chamber 20 cm in diameter, was located in a primarily deciduous area thought to be representative of the bulk of the forest vegetation, and measurements were made at 15-minute intervals. The automated measurements were supplemented with occasional manual spatial measurements conducted on 30 additional soil collars in the primarily deciduous area, and on six soil collars in a primarily coniferous part of the forest. The spatial measurements used a survey chamber of identical size to the automated chamber.
The behaviour of soil respiration was evaluated with respect to soil temperature and soil moisture. Relationships with temperature were similar to those reported for other forest sites; Q10 ~ 3.2 under non-moisture stressed conditions. Interannual variability was strongly controlled by precipitation patterns through changes in soil moisture. Drying of the soil resulted in reductions in the observed CO2 fluxes from the soil. The relationship between respiration and soil moisture was more stable when respiration values were normalized with respect to temperature and a significant moisture control was evident as soil moisture decreased below 10% by volume. Algorithms for modelling respiration with respect to temperature and soil moisture are compared.
The automated chamber data show slightly larger respiration values than the survey measurements but the difference varies from year to year. Interannual variability is spatially variable, possibly related to soil moisture patterns, in particular to soil moisture at depths greater than 50 cm which varies significantly from year to year depending on long-term precipitation patterns. The results suggest that our ability to model the response of soil respiration to a changing climate is also dependent on the realism of the modelled water balance (i.e. precipitation and evaporation).