Soil respiration (Rs), the largest carbon emission flux in ecosystems is usually modelled as an empirically parameterized function of temperature (Ts), and sometimes water availability (Ɵ). The likely contribution by other factors, such as carbohydrate substrate supply from photosynthesis hasve been recognized, but modeling capacity to use this information is limited. In the current study we show, using spectral analysis, that the time domains of influence for different driving variables of Rs are discrete, and largely non-overlapping. Using wavelet and cross-wavelet transformations of the residuals of a seasonal Q10 temperature response model, we found that temperature explained variability at multiple time scales (diurnal – synoptic), whereas volumetric soil water content correlated with variability at scales 15-30 days, and photosynthetically active radiation (PAR, as a proxy for and by extension, photosynthetic carbohydrate substrate supply) explained most of the diurnal variation. Notably, the co-spectral peak of PAR and Rs indicated a time delay of 1.5-3.0 hours, which was independent of canopy height, thus lending support to the pressure-concentration-wave model of carbohydrate availability. The temporal domains of influence findings were consistent (1) in young and mature loblolly pine plantations, which have a contrasting balance of autotrophic and heterotrophic contribution to total soil CO2 efflux, and (2) with proposed conceptual models of transient forcing (e.g. rain events) and time delays or memory effects (e.g. CO2 diffusion from point of production to point of measurement).
