Global Ecology and Conservation (Jun 2020)
Climatic, soil, and vegetation controls of the temperature sensitivity (Q10) of soil respiration across terrestrial biomes
Abstract
Understanding how climatic factors, soil properties, and vegetation characteristics influence the inter-site and inter-annual variations in Q10 of soil respiration across biomes may help in understanding the mechanisms of soil carbon (C) cycle and its feedback to global warming. We compiled in-situ measured temperature sensitivity (Q10) of soil respiration and analyzed its relationship with potential controls of climate factors, soil properties, and vegetation characteristics across various biomes at the global scale. Results indicate that the Q10 of soil respiration at 5 cm depth varies from 1.100 to 13.464 across various biomes, with the coefficient of variation of 47.4%. Q10 across all measurement sites increases significantly (P < 0.001) with latitude (LAT) but is correlated negatively (P < 0.001) with mean annual temperature (MAT) and annual precipitation (AP). LAT and MAT explained 20.9% (R2 = 0.209) and 16.0% (R2 = 0.160) of the variations in Q10, respectively, across measurement sites. The soil property of BD shows the highest significantly (P < 0.001) and negatively exponential relationship (R2 = 0.302) with Q10 among the examined soil properties. BD has higher explanation of variations in Q10 than climatic factors and vegetation characteristics at the global scale. The explanatory ability of vegetation characteristics of litter fall (LF), tree height (TH), and leaf area index (LAI) is lower than that of soil properties and climatic factors. The R2 for the relationship between Q10 and vegetation variables of LF, TH, and LAI was 0.130, 0.074, and 0.117, respectively. Two models including geographical, climatic and soil factors can explain more than 50% [R2 = 0.509 for the model including MAT, BD, soil total nitrogen (STN), ratio of soil carbon to nitrogen (C/N), and soil organic carbon (SOC) and R2 = 0.542 for the model including LAT, MAT, BD, STN, C/N, and SOC, respectively] of the variation in Q10 across all measurement sites. The two models can well simulate the variations in Q10, as the measured Q10 is significantly (P < 0.001) and positively correlated with the modeled Q10 and the slope of the regression line is closely corresponding to the 1:1 line.
