Carbon (C) allocation among different plant tissues is crucial for maintaining the C balance in forest ecosystems, especially under changing climate conditions. The partitioning of newly assimilated C among plant tissues, interconnected ramets, and soil in forests dominated by giant clonal plants, such as moso bamboo (Phyllostachys edulis), and the influence of drought on this partitioning remains poorly understood. In August 2019, we performed in situ labeling of the entire crown of R0 (ramets that emerged in 2019) of moso bamboo with ¹³CO₂ in plots subjected to a 5–year drought or left untreated (ambient control) in subtropical China. We then traced the ¹³C signatures in the leaves, twigs, and fine roots of R0, R1 (ramets that emerged in 2018 and are connected with R0), and R2 (ramets that emerged in 2017 and are connected with R1), as well as in soil organic C (SOC) and soil respiration over the course of one year post–labeling. Drought reduced leaf ¹³C assimilation and its allocation to sink tissues but did not alter the velocity of C transport from source to sink compared to controls. The peak ¹³C signal was observed on day 15 for SOC and on day 5 for respired CO₂ in both drought and ambient control forests. Labeled ¹³C was detected in R1 ramets on day 3 and in R2 on day 7 post–labeling. This study reveals that new assimilates produced by the 'younger' R0 ramets are preferentially retained within their own tissues to meet their own demands rather than being allocated to interconnected neighboring R1 and R2 ramets. In forests dominated by large clonal plants, such as giant moso bamboo, drought can alter the allocation of newly assimilated C within the tissues of source ramets but may not affect its allocation among interconnected ramets or within plant–soil systems. Our findings highlight the complexity of newly assimilated C partitioning in these forests and suggest that clonal integration may mitigate drought–induced dieback in older ramets through resource sharing under climate change.

Experimental design and sampling

In July 2014, three bamboo forests with similar slopes and altitudes were selected (Table S1). In each forest, two 12 m × 12 m plots were established, separated by a 10–m–wide buffer zone (Fig. S2). One plot out of them was randomly selected to be subjected to drought (i.e., throughfall exclusion), while the other served as a control (natural rainfall) (hereafter referred to as ‘drought’ and ‘control’, respectively). For the drought treatment, we set up transparent PVC waterproof boards suspended at about 1.5 m above the ground surface to exclude 70–80% of precipitation. A trench measuring 12 m in length, 40 cm in width, and 60 cm in depth (with roots concentrated in the 20–40 cm soil depth) was dug to prevent water from flowing into the drought treatment area from the surrounding soil and to restrict root access to areas outside the throughfall exclusion treatment area. PVC waterproof boards were inserted into the trench before the soil was refilled. The throughfall exclusion (drought) treatment started in November 2014, and soil moisture was monitored using an EM 50 data logger (Meter, America) with 5TE sensors (Fig. S3). The outer 1 m of each plot’s four sides was excluded from sampling to avoid edge effects, resulting in an effective subplot area of 10 m × 10 m. The biomass of the bamboo stands was estimated both before (2014) and after the treatments (2019) using the field–based method of Ouyang et al. (2022) (Figs. S4, S5).