Three scales of temporal resolution from automated soil respiration measurements

Soil respiration (R_s) is a combination of autotrophic and heterotrophic respiration, but it is often modeled as a single efflux process, influenced by environmental variables similarly across all time scales. Continued progress in understanding sources of variation in soil CO₂ efflux will require development of R_s models that incorporate environmental influences at multiple time scales. Coherence analysis, which requires high temporal frequency data on R_s and related environmental variables, permits examination of covariation between R_s and the factors that influence it at varying temporal frequencies, thus isolating the factors important at each time scale. Automated R_s measurements, along with air, soil temperature and moisture were collected at half hour intervals at a temperate forest at Harvard Forest, MA in 2003 and a boreal transition forest at the Howland Forest, ME in 2005. As in other temperate and boreal forests, seasonal variation in R_s was strongly correlated with soil temperature. The organic and mineral layer water contents were significantly related to R_s at synoptic time scales of 2-3 days to weeks, representing the wetting and drying of the soils as weather patterns move across the region. Post-wetting pulses of R_s were correlated with the amount of precipitation and the magnitude of the change from pre-wet-up moisture content to peak moisture content of the organic horizon during the precipitation events. Although soil temperature at 8-10 cm depth and R_s showed strong coherence at a 24-h interval, calculated diel Q₁₀ values for R_s were unreasonably high (6-74) during all months for the evergreen forest and during the growing season for the deciduous forest, suggesting that other factors that covary with soil temperature, such as canopy assimilatory processes, may also influence the diel amplitude of R_s. Lower diel Q₁₀ values were obtained based on soil temperature measured at shallower depths or with air temperature, but the fit was poorer and a lag was needed to improve the fit (peak R_s followed peak air temperature by several hours), suggesting a role for delayed substrate supply from aboveground processes to affect diel patterns of R_s. High frequency automated R_s datasets afford the opportunity to disentangle the temporal scales at which environmental factors, such as seasonal temperature and phenology, synoptic weather events and soil moisture, and diel variation in temperature and photosynthesis, affect soil respiration processes.