由于青藏高原高海拔、低温的特殊环境，使得生态系统呼吸（RE）对气候变化的响应极其敏感，然而对高寒湿地生态系统长时间尺度上的RE动态特征及驱动机制的研究相对薄弱。以青藏高原东北部高寒湿地为研究对象，分析了基于涡度相关系统观测的高寒湿地2004-2016年的CO2通量排放动态及影响机制，对预测高寒湿地碳平衡对未来气候变化的响应具有重要意义。结果表明：高寒湿地在2004-2016年的月平均RE表现为单峰变化趋势，在8月达到峰值；年RE表现为逐年升高的趋势（P<0.05），年RE均值为（608.9±65.6） g C m-2 a-1；生长季RE约是非生长季RE的2.7倍，线性回归分析表明生长季RE （r2=0.66，P=0.001）、非生长季RE （r2=0.47，P=0.01）与全年RE呈极显著正相关。在月尺度上，分类回归树分析和线性回归分析表明土壤温度是月RE的最主要控制因素，暗示高寒湿地的土壤呼吸对整个生态系统的碳排放至关重要。在年际尺度上，生长季积温与生长季RE呈显著正相关（P<0.05），而生长季降水（PPT）与生长季RE呈显著负相关（P<0.05），非生长季气温（P<0.05）、PPT （P<0.05）与非生长季RE呈显著正相关，暗示未来温度的升高将会促进高寒湿地的CO2排放，而生长季、非生长季碳排放对PPT的差异化响应，暗示在分析高寒生态系统CO2排放对未来水热条件响应时需更加谨慎。
It is speculated that the cold and relatively humid climate in alpine ecosystems can facilitate the storage of soil carbon. Besides, these ecosystems play an important role in the carbon cycle on a global scale. Due to the special environment of high altitude and low temperature in the Qinghai-Tibetan Plateau (QTP), the ecosystem respiration (RE) in alpine wetland is sensitive to global climate change. However, previous studies on RE of wetland ecosystem were limited to short time series data, which was difficult to fully explain the dynamic of RE and its influence mechanism. Therefore, to study the carbon budgets from alpine wetland ecosystem has become increasingly important in accurately projecting global carbon cycling in future climatic change. Here, we analyzed the continuous RE measured with the eddy covariance technique over an alpine wetland. The results indicated that monthly RE exhibited a single-peak trend which increased and then decreased from 2004 to 2016, and reached the peak in August. The annual RE was (608.9±65.6) g C m-2 a-1, which showed a gradually increasing trend (P<0.05). The non-growing season RE accounted for a small proportion of annual RE, but it was significantly positively correlated with annual RE (P<0.05). This finding suggested that CO2 emission in alpine wetland during the non-growing season were crucial for annual carbon balance of alpine wetland. At monthly scale, the results of classification and regression tree (CART) and linear regression analysis indicated that soil temperature (Ts) was the predominantly determinant factor that affected the change of monthly RE, suggesting that soil respiration in alpine wetland was critical to the carbon emission. The correlation analysis showed that the growing seasonal RE had significantly positive and negative correlation with growing season degree days (GDD) (P<0.05) and precipitation (PPT) (P<0.05), respectively. Furthermore, linear regression analysis indicated that the non-growing seasonal RE had significantly positive correlation with the non-growing season air temperature (Ta) (P<0.05) and the PPT (P<0.05), respectively. Due to the promoting effect of growing seasonal GDD and non-growing seasonal Ta on RE, we speculate that under the context of global warming, especially in the non-growing seasons, which will exacerbate carbon emission in the alpine wetland of the QTP. However, considering that PPT had different effects on carbon emission of alpine wetland ecosystem in growing season and non-growing season, it is suggested that more caution should be applied in analyzing the response mechanism of carbon emissions of alpine ecosystem to hydrothermal conditions in future.