The carbon cycle paradox in tropical rubber plantations—Sustained latex carbon export without triggering the expected photosynthetic compensatory enhancement—challenges our understanding of managed forest carbon balance. Using year–round δ 13 C and δ 15 N measurements across three plant organs (leaf, stem, root) and various soil depths in Xishuangbanna rubber plantations, we developed a nonlinear programming framework that simultaneously optimizes isotope source contributions and fractionation values to quantitatively resolve seasonal dynamics of plant and soil carbon–nitrogen sources. Results show: (1) when NLP optimized, time–varying fractionation values were used, nonlinear optimization achieved 2–4 fold higher precision than Bayesian methods, effectively addressing fractionation variability; (2) rubber tree leaf carbon relied primarily on inter–organ reallocation (91%, stem 55.9%, root 35.1%) rather than photosynthetic fixation (9%), challenging the traditional photosynthesis–centric theory; (3) nitrogen acquisition strategies exhibited seasonal vertical shifts, with deep soil contributions reaching 50% in pre–rainy season, reflecting depth–dependent nitrogen processes; (4) carbon–nitrogen responses showed temporal scale differentiation, with 7–day lag for carbon vs. 1–day rapid response for nitrogen systems, and relative humidity > 50% was strongly associated with a switch in nitrogen source acquisition strategies. The study unveils dual adaptive mechanisms in rubber trees under harvest stress–organ carbon reallocation and deep nitrogen acquisition–and provides a new analytical paradigm for isotopic ecology; these insights may inform sustainable management of tropical plantations. • Time–varying fractionation NLP is 2–4 × more precise than MixSIAR. • Rubber tree leaves obtain 91% carbon from internal reallocation. • Deep soil nitrogen increases to ∼50% contribution during pre-rainy season. • Plant carbon lags precipitation by ∼7 days; N responds within ∼1 day to humidity. • Humidity > 50% is associated with shifts in N–source use.
Miao et al. (Fri,) studied this question.