Spatial variability of soil respiration (R-s) and its controls are subjected to strong seasonality in an even-aged European beech (Fagus sylvatica L.) stand

摘要

Uncertainties arising from the so-far poorly explained spatial variability of soil respiration (R-s) remain large. This is partly due to the limited understanding about how spatially variable R-s actually is, but also on how environmental controls determine R-s's spatial variability and how these controls vary in time (e.g., seasonally). Our study was designed to look more deeply into the complexity of R-s's spatial variability in a European beech even-aged stand, covering both phenologically and climatically contrasting periods (spring, summer, autumn and winter). Although we studied a relatively homogeneous stand, we found a large spatial variability of R-s (coefficients of variation > 30) characterized by strong seasonality. This large spatial variability of R-s suggests that even in relatively homogeneous stands there is a large potential source of error when estimating R-s. This was also reflected by the sampling effort needed to obtain seasonally robust estimates of R-s, which may actually require a number of samples above that used in R-s studies. We further postulate that the effect of seasonality on the spatial variability and environmental controls of R-s was determined by the seasonal shifts of its microclimatic controls: during winter, low temperatures constrain plant and soil metabolic activities and hence reduce R-s variability (temperature-controlled processes), whereas during summer, water demand by vegetation and changes in water availability due to the microtopography of the terrain (i.e., slope) increase R-s variability (water-controlled processes). This study provides novel information on the spatiotemporal variability of R-s and looks more deeply into the seasonality of its environmental controls and the architecture of their causal-effect relationships controlling R-s's spatial variability. Our study further shows that improving current estimates of R-s at local and regional levels might be necessary in order to reduce uncertainties and improve CO2 estimates at larger spatial scales.