资源环境科技发展态势分析平台

Atmospheric CO₂ (c _a) rise changes the physiology and possibly growth of tropical trees, but these effects are likely modified by climate. Such c _a × climate interactions importantly drive CO₂ fertilization effects of tropical forests predicted by global vegetation models, but have not been tested empirically. Here we use tree‐ring analyses to quantify how c _a rise has shifted the sensitivity of tree stem growth to annual fluctuations in rainfall and temperature. We hypothesized that c _a rise reduces drought sensitivity and increases temperature sensitivity of growth, by reducing transpiration and increasing leaf temperature. These responses were expected for cooler sites. At warmer sites, c _a rise may cause leaf temperatures to frequently exceed the optimum for photosynthesis, and thus induce increased drought sensitivity and stronger negative effects of temperature. We tested these hypotheses using measurements of 5,318 annual rings from 129 trees of the widely distributed (sub‐)tropical tree species, Toona ciliata . We studied growth responses during 1950–2014, a period during which c _a rose by 28%. Tree‐ring data were obtained from two cooler (mean annual temperature: 20.5–20.7°C) and two warmer (23.5–24.8°C) sites. We tested c _a × climate interactions, using mixed‐effect models of ring‐width measurements. Our statistical models revealed several significant and robust c _a × climate interactions. At cooler sites (and seasons), c _a × climate interactions showed good agreement with hypothesized growth responses of reduced drought sensitivity and increased temperature sensitivity. At warmer sites, drought sensitivity increased with increasing c _a, as predicted, and hot years caused stronger growth reduction at high c _a. Overall, c _a rise has significantly modified sensitivity of Toona stem growth to climatic variation, but these changes depended on mean climate. Our study suggests that effects of c _a rise on tropical tree growth may be more complex and less stimulatory than commonly assumed and require a better representation in global vegetation models.