Root excavating and identification
Root sampling was conducted in early summer from April to June in 2016 by randomly selecting 625 locations in the 50-ha plot as described in detail in Luo et al. (2020). In brief, at each location, three 300-cm3 soil cubes (5 × 6 × 10 cm) were sequentially extracted with a depth increment of 10 cm using a 5 × 6 × 40 cm steel spade. To further explore vertical root distributions in deeper soils, 12 cubes were sampled to a depth of 100 cm. An additional 300-cm3 soil cube of soil next to each root-sampling location was collected for further analyses of soil chemical and textural properties.
In the laboratory, each soil cube was soaked in water for c . 40 min and intact and all well-branched woody fine root segments (≤ 2 mm in diameter) with length ≥ 10 cm were selected for further analysis. Root segments generally included 2-4 orders of roots (Valverde‐Barranteset al. 2015). For each soil cube, the selected root segments were further sorted into morphotypes based on root branching patterns, color, distinctive odor, diameter and the presence of nodules or latex (Luoet al . 2020). As reported in Luo et al. (2020), some root morphotypes could be confirmed to species based on their identifiable and conservative root characteristics. All other morphotypes were sequenced and compared with four DNA barcodes (rbcLa ,ITS2 , matK and trnH-psb ) generated from leaf collections of 237 tree canopy species in the plot (Zhou 2014). Following these procedures, we identified c . 94% of the morphotypes that in combination contained 143 species. 109 species with occurrences ≥ 5 cubes were selected for the present study.