Site and treatments
The study was conducted at the Dinghushan Biosphere Reserve in southern China (112°10′ E, 23°10′N). The reserve is characterized by a tropical monsoon humid climate with a mean annual temperature of 21 °C and mean relative humidity of 80%. The mean annual precipitation of 1900 mm has a distinct seasonal pattern, with about 80% of it falling from April to September (warmer wet season) and 20% occurring from October to March (cooler dry season).
The experimental warming site was constructed by translocating the microcosm forest ecosystem from the high-elevation site (control: 600 m) to low-elevation sites (+1.0°C treatment: 300 m; +2.1°C treatment: 30 m) (Fig. S1). In May 2012, three plots (open-top chambers; 3 m length × 3 m width × 0.8 m depth) were established at each of the elevation sites, representing three replicates of each temperature treatment. Each belowground plot was surrounded by concrete brick wall bonding with ceramic tile, with one hole connected with PVC tubing at the bottom and the top of the chamber wall to collect underground water and surface runoff, respectively. All plots were separated by a 3 m buffer zone. Before planting, we collected soil from each of the three horizons (0-20, 20–40, and 40-70 cm) in the adjacent naturally occurring mountain evergreen broad-leaved forest, and homogenized each layer separately, and then placed each soil horizon into each plot. Six one-year old dominant plant species, including arbuscular mycorrhiza-forming species (Schima superba , Itea chinensis , Machilus breviflora , Myrsine seguinii , andArdisia lindleyana ) and only one ectomycorrhiza-forming species (Syzygium rehderianum ) at an elevation of 600 m in the natural forest, were collected and then randomly planted in each plot (six individuals per species). Four 0.2 m × 0.2 m litter traps were placed randomly in each plot. One tree per species was randomly harvested in December 2014, December 2015, and June 2018 to avoid crowding in the chambers.
Soil temperature and soil volumetric water content at 5 cm soil depth were measured in each plot using Campbell 109 constantan-copper thermocouples and time-domain reflectometer probes (CS616, Campbell, USA), respectively. Data were recorded every 30 min using Campbell Scientific (Logan, UT, USA) CR1000 data loggers. Further detailed information regarding the experimental site was described in Lie et al., (2021).