森林是陆地生态系统最重要的组成部分之一。分析森林生态系统中树木生长对气候要素的敏感性，是开展应对气候变化的森林管理的基础步骤。油松是我国分布范围第二大的针叶树种，分布区跨干旱区、半干旱区、半湿润区和湿润区，被认为是对气候最为敏感的树种之一。我们在油松分布区内布设了16个采样点，共采集了852棵样树3000余个样芯的油松树木年轮宽度样品。基于这些样品，我们通过树轮年代学方法，从整体生长、极限生长、极端生境中生长三个方面，分析油松生长对气候的响应。 我们建立了16个样点的树轮宽度指数年表，并与文献中提取的65个年表进行汇总，构建油松径向生长序列网络，分析油松生长对气候的整体响应。使用聚类分析对年表进行聚类，可以将所有序列分成4组，分别命名为东部组、南部组、西部组和中部组。各组油松的长期生长趋势类似，1950s后油松生长整体表现为下降趋势，尤其是西部组和南部组中呈显著下降趋势的年表所占比例超过了50%。年表的公共区间分析显示，西部组油松生长的个体差异小；年表的统计参数显示，西部组油松生长的敏感性高。使用标准化降水蒸发指数（SPEI）衡量油松生长面临的湿润状况。相关分析显示，前一年9月的1个月尺度SPEI（SPEI1）和当年6月的3个月尺度SPEI（SPEI3）是与年表相关关系最强的因子。使用线性混合效应模型，以前一年生长量、前一年9月SPEI1和当年6月SPEI3为解释变量，对各组油松生长量进行建模。在东部组、南部组和中部组模型中，前一年生长量是最重要的解释变量，而当年6月SPEI3的重要性高于前一年9月SPEI1；而西部组模型中，前一年9月SPEI1是最重要的解释变量。 我们使用局部丢轮表征油松极限生长状态。通过对16个样点油松样品进行交叉定年，得到了每棵样树的局部丢轮情况。所有样点均出现了局部丢轮，局部丢轮总数占树轮总数的3.0%。油松的局部丢轮情况存在个体差异，大多数样点中树龄越大的样树局部丢轮出现的频率越高。在空间上，分布区西北部的油松局部丢轮率高于南部。使用LASSO-Logistic模型，以树龄和生长年内各月的干旱度作为解释变量，对每个样点局部丢轮出现情况进行建模。根据最佳拟合模型内的变量情况，样点可以划分为干旱主导组和非干旱主导组。干旱主导组包括分布区西北部的8个样点，模型主要含树龄、前一年9月干旱度和当年5~6月干旱度等项。我们使用这三个因子，基于干旱主导组油松局部丢轮情况，建立了区域局部丢轮模型。根据模型预测结果，在未来气候变化条件下，幼龄油松面临的局部丢轮风险较小，而由于干旱加剧，中老龄油松在分布区西北部和东部面临较高的局部丢轮风险。 贺兰山是油松分布区内的极端生境。基于贺兰山油松树木年轮数据，我们在个体尺度上分析了极端条件下树龄对油松生长特征的影响。我们发现两个老龄油松样点的局部丢轮率较高，是全球范围内松属局部丢轮情况最严重的样点之一。年表统计参数显示，油松生长的敏感性随树龄的升高而增强。油松个体生长表现出相似的高频变化，同时相关分析显示个体年表均与前一年9月SPEI1和当年3月SPEI1呈显著正相关关系。前一年9月和当年3月的极端干旱事件与高局部丢轮率年份存在对应关系。针对前一年9月和当年3月同时发生极端干旱的年份，我们评估了油松生长的抵抗力、恢复力和弹性，发现随着树龄的增加，油松生长弹性显著降低。 综合来看，油松径向生长普遍受到前一年生长季末期和当年生长季初期干旱状况的限制。油松生长对气候的响应存在空间异质性，分布区西北部的油松生长对干旱状况最为敏感。同时，油松生长对气候的响应存在个体差异，树龄越大的油松具有更高的丢轮出现概率和更弱的生长弹性。

Forest is one of the most important components of terrestrial ecosystems. Analysis of the sensitivity of tree growth to climatic factors in forest ecosystems is a fundamental step in forest management for climate adaptation. Pinus tabuliformis is the second most widely distributed conifer species in China, which distribution area is at the transition zone of semi-humid and semi-arid area, and thus it is considered as one of the most sensitive tree species to climate. Here, we set 16 sample sites in the Pinus tabuliformis distribution area, and collected over 3,000 tree-ring samples from 852 Pinus tabuliformis, and analyzed the response of Pinus tabuliformis overall radial growth, extreme growth, and growth at extreme habitats to climate. We combined tree-ring 16 chronologies collected in this study and 65 standard chronologies extracted from literatures into a network of Pinus tabuliformis overall radial growth. We divided all the chronologies into 4 groups, the East, the South, the West and the Central, by cluster analysis. The long-term growth trend of all chronologies were similar, and growth showed a decreasing trend after 1950s. Common interval analysis indicated that individual difference of tree growth was large in the West, and chronologies characteristic indicated that tree growth was sensitive in the West. We used the standardised precipitation-evapotranspiration index (SPEI) to measure moisture conditions. Correlation analysis showed that tree growth was mainly related to previous September 1-month SPEI (SPEI1) and current June 3-month SPEI (SPEI3). We used linear mixed-effect models to fit group-level tree growth by growth at the previous year, previous September SPEI1 and June SPEI3. Growth at the previous year was the most importance explanatory variable, and June SPEI3 was more important than previous September SPEI1, in the models of the East, the South, and the Central, while previous September SPEI1 was the most important explanatory variable in the model of the West. We used locally-absent rings (LAR) to represent extreme growth. We identified LAR of each tree through crossdating of collections from 16 sample sites. All the sample sites had LAR, and LAR accounted for 3.0% of all measured rings. There were individual differences in the LAR frequency, that is older trees had higher LAR rate than younger trees. Spatially, LAR rate on the northwest of the distribution area was higher than that in the south. We used LASSO-Logistic models to fit site-level LAR occurrence by tree age and monthly drought index of annual growth period. We divided 16 sample sites into two groups, the Drought-induced and the Drought-free, by explanatory variables in the best-fitting models. The Drought-induced group contained eight sample sites on the northwest of the distribution area, and their models mainly included tree age, the drought index of previous September, and the drought index from current May to June. Thus, we used these three factors to fit region-level LAR occurrence, based on LAR data from the Drought-induced group. The model predicted that the risk of LAR for young pines would be low, while the middle-aged and old pines in the northwest and east of the distribution area would face high risk of LAR in the future. Helan Mountains were the extreme habitat of Pinus tabuliformis. We analyzed tree age effect on the growth characteristics under extreme conditions at the individual level, based on three tree ring collections at Helan Mountains. We found two older pine sites had high LAR rate, and belonged to Pinus sites with the highest LAR rate globally. Chronologies characteristic indicated that growth sensitivity increased with increasing cambial age. Individual growth showed similar high-frequency variation, and was significantly positively correlated with previous September SPEI1 and March SPEI1. Years with drought occurring at previous September and March were related to years with high LAR frequency. We estimated resistance, recovery and resilience of tree growth after synchronous droughts at previous September and March. We found that tree growth resilience significantly decreased with increasing cambial age. To conclude, radial growth of Pinus tabuliformis was generally limited by drought at late previous growing season and early current growing season. There was spatial difference in the growth response of Pinus tabuliformis to climate, i.e. tree growth was most sensitive to drought at the northwest of the distribution area. There was also individual difference in the growth response, i.e. older trees had higher occurrence probability of missing rings and weaker growth resilience.