Modelling and Experimental Studies Revealed Strong Hydrological Control on Soil Carbon Budget in a Subtropical Montane Cloud Forest

The patterns of soil carbon dynamics might have some particularity in tropical montane cloud forests, since high precipitation and humidity condition could result in low decomposition rate and high leaching loss of dissolved organic carbon (DOC).  Most ecosystem process models that simulate soil carbon budget have ignored the process of DOC leaching as this component of carbon export is usually negligible for most of the ecosystems. Further challenge in simulating TMCF carbon budget is to deal with the lowered microbial activities in the decomposition process under anoxia soil condition. Given these limitations, the ecosystem models e.g. BIOME-BGC, will overestimate soil respiration and soil carbon storage of TMCFs. The objective of this study was thus to improve the carbon budget simulation of BIOME-BGC by (1) experimental determination of the moisture control of soil respiration, and (2) adding/modifying the routines related to soil carbon processes of the BIOME-BGC model.

The study was done at the Chi-Lan Mountain (CLM) site (24°35'N, 121°25'E) in northern Taiwan. The dominant specie is yellow cypress (Chamaecyparis obtusa var. formosana). CLM site is characterized by frequent fog and high humidity in the air and soil. Soil moisture manipulation experiment was conducted for one year. The permanently high soil moisture was artificially reduced by building roofs that prevented the entrance of throughfall to the soil. Three roof plots and three control plots were selected and the soil respiration (R_s) in each plot was measured by an automatic soil respiration system (LI-8100, LI-COR Inc., USA). The measured R_s data from the condition of higher soil temperature (higher than 18 degrees Celsius) were selected to establish an equation relating R_s to soil water content. The equation was used to modify the decomposition module in BIOME-BGC model. A routine for DOC leaching was added to the model based on a previous study at the site. Four simulation schemes were implemented: scheme of modified decomposition module (D), scheme of added DOC leaching module (L), scheme of both D and L (DL), and scheme of the unmodified BIOME-BGC (O). All modified schemes improved the ability to simulate R_s and soil carbon storage. The original BIOME-BGC had 371.9% overestimation of R_s compared to measurement data, while schemes D, L and DL were 157.0%, 66.4% and 34.4% overestimation, respectively. Linear correlation between simulated and measured R_s gave R² values of 0.22, 0.24, 0.44 and 0.50 for schemes O, D, L, and DL, respectively. Simulated R_s of scheme D and DL exhibited better agreement with observed data, but scheme D over-predicted it. Scheme L significantly decreased the overestimation of R_s but with poor R² value. The modification of decomposition module increased the organic carbon accumulation in the soil. R_s dramatically rose at scheme D when soil was dry. Scheme DL that performed DOC leaching simulation simultaneously could remove excess soil carbon by fixed proportion, and therefore could decrease the sharp fluctuation of R_s in line with the field measured data. Simulated ecosystem respiration was compared with data from eddy covariance method. Simulated ecosystem respirations of Scheme DL were close to the measurement values with 11.5% overestimation (R²=.65), while scheme L agreed better with measurement data (31.2% overestimated, R²=.76). Organic carbon leaching plays an important role of soil carbon budget at CLM site. Although BIOME-BGC didn't contain soil anaerobic processes, the modification as we implemented here improved the ability to simulate soil carbon budget under moist soil condition.