hide
Free keywords:
Arrhenius model, Biotic controls, Ecosystem modelling, Environmental controls, Labile carbon, Soil respiration
Abstract:
Over two-thirds of terrestrial carbon is stored belowground and a significant amount of atmospheric CO2 is respired by roots and microbes in soils. For this analysis, soil respiration (Rs) data were assembled from 31 AmeriFlux and CarboEurope sites representing deciduous broadleaf, evergreen needleleaf, grasslands, mixed deciduous/evergreen and woodland/savanna ecosystem types. Lowest to highest rates of soil respiration averaged over the growing season were grassland and woodland/savanna < deciduous broadleaf forests < evergreen needleleaf, mixed deciduous/evergreen forests with growing season soil respiration significantly different between forested and non-forested biomes (p < 0.001). Timing of peak respiration rates during the growing season varied from March/April in grasslands to July-September for all other biomes. Biomes with overall strongest relationship between soil respiration and soil temperature were from the deciduous and mixed forests (R ≥ 0.65). Maximum soil respiration was weakly related to maximum fine root biomass (R = 0.28) and positively related to the previous years' annual litterfall (R = 0.46). Published rates of annual soil respiration were linearly related to LAI and fine root carbon (R = 0.48, 0.47), as well as net primary production (NPP) (R = 0.44). At 10 sites, maximum growing season Rs was weakly correlated with annual GPP estimated from eddy covariance towersites (R = 0.29; p < 0.05), and annual soil respiration and total growing season Rs were not correlated with annual GPP (p > 0.1). Yet, previous studies indicate correlations on shorter time scales within site (e.g., weekly, monthly). Estimates of annual GPP from the Biome-BGC model were strongly correlated with observed annual estimates of soil respiration for six sites (R = 0.84; p < 0.01). Correlations from observations of Rs with NPP, LAI, fine root biomass and litterfall relate above and belowground inputs to labile pools that are available for decomposition. Our results suggest that simple empirical relationships with temperature and/or moisture that may be robust at individual sites may not be adequate to characterize soil CO2 effluxes across space and time, agreeing with other multi-site studies. Information is needed on the timing and phenological controls of substrate availability (e.g., fine roots, LAI) and inputs (e.g., root turnover, litterfall) to improve our ability to accurately quantify the relationships between soil CO2 effluxes and carbon substrate storage.
