Tropical and subtropical forests represent the largest terrestrial carbon pool. Elucidating the carbon sources for soil microbial respiration (Rm) in tropical and subtropical forests is of fundamental importance to the global carbon cycle in a warming world. Based on hourly measurements, we quantified Rm of in situ forest soil and soil cores from a subtropical forest. We found recent photosynthates, not soil organic carbon (SOC), contributed 88% ± 12% of the carbon source fueling Rm. The control of recent photosynthates on Rm is also supported by the close relationship between Rm and photosynthetically active radiation as well as literature data synthesis results. These results challenge conventional models based on the tenet that Rm is mainly regulated by soil temperature in all forest ecosystems. The results imply that the widely observed warming-induced Rm increases are largely explained by the enhanced input of recent photosynthates in tropical forests, not SOC consumption.

Three 20 m × 20 m plots randomly distributed in the study forest were set up to conduct the experiment. Within each plot, three randomly selected 1 × 1 m subplots were trenched in April 2010, for a total of 9 trenched subplots, and the trenches were backfilled with soil from the same layer. In this study, nylon mesh sheets with 150 µm mesh size were placed 1 m into the soil to exclude fine root entry while allowing for lateral water movement of photosynthates and ingrowth of mycorrhizal fungi. After the trenching was completed, the subplots were kept free of ingrowth of new seedlings and understory vegetation. One polyvinyl chloride (PVC) collar (20 cm inside diameter × 8 cm height) was inserted into the soil to 4 cm depth in each trenched subplot, and microbial respiration was measured from March 28^th, 2012 until December 31^st, 2014. 

Following the installation of the collar, a Li-8100-103 portable CO₂ infrared gas analyzer (Li-Cor Inc, Lincoln, NE, USA) was used to measure the rate of in situ forest microbial respiration (Rm-f) biweekly. Soil temperature was measured by a portable thermometer (Model SK-250WP, Sato Keiryoki MFG. CO., LTD. Tokyo, Japan) and soil volumetric water content was measured using soil moisture meter (Model TDR-300, Spectrum Technologies, Aurora, IL, USA) at a depth of 8 cm soil, while taking measurements of forest microbial respiration by every two weeks. The rates of forest microbial respiration were monitored from 9:00 am to 12:00 pm based on previously established information that this period encompasses the average rate of daytime respiration (Sheng et al., 2010). After 16 months, a collar with the average microbial respiration rate of each 20 m x 20 m plot was chosen for high frequency measurements using Li-8100-104 automated long-term chamber. Soil temperature and soil water content were monitored using the temperature thermistor and moisture probe from 8100-401 chamber control kit. In these three sub-plots, microbial respiration, soil temperature, and soil water content were measured every 30 minutes from August 1^st, 2013 until December 31^st, 2014. Meanwhile, photosynthetically active radiation (PAR), air temperature, and rainfall were monitored within 50 m distance at a weather station outside the forest. Microbial respiration and all the environmental variables were monitored with the same frequency simultaneously. 

To evaluate the magnitude of SOC decomposition rate and its response to soil temperature in the absence of new C inputs, in January 2015 we excavated 5 soil cores (length × wide × depth: 0.5 × 0.5 × 0.6 m) from the forest and incubated them in the open in a location 1 km from the forest next to the field laboratory building. The extraction of soil cores did not cause changes in soil organic carbon and nitrogen compared to soils in the forest. The respiration rate, soil temperature and soil volumetric water content were monitored simultaneously from January 2015 through December 2016, as described above. However, to minimize the influence of disturbance to the soil structure and decomposition of litter and root residues probably stimulated by core excavation, only data collected from January 2016 to December 2016 were used in this study.

There were no missing values in the high-frequent soil microbial respiration measurement.