不同程度火烧迹地兴安落叶松延迟性死亡特征

在2018年汗马国家级自然保护区火烧迹地按不同火烧程度设立9个标准样地(30 m×30 m),以样地内760株延迟性死亡兴安落叶松为全部样本,划分为7个径级,分析不同程度火烧迹地不同径级的兴安落叶松(Larix gmelinii(Rupr.)Kuzen)延迟性死亡特征。将全部样本的80%(610株)作为建模样本,20%(150株)作为检验样本,使用Logistic回归方程对兴安落叶松延迟性死亡率进行预测。结果表明:轻度火烧样地中,兴安落叶松延迟性死亡率随着径级的增加而逐级降低,降低幅度分别为8.5%、28.5%、17.4%,胸径大于20 cm的树木延迟性死亡率为0;在所有延迟性死亡树木中,胸径小于10 cm的树木占76.8%。中度火烧样地中,兴安落叶松延迟性死亡率随径级的增加也逐级降低,但降低幅度没有轻度火烧样地明显,胸径大于25 cm的树木延迟性死亡率为0;在所有延迟性死亡树木中,胸径6~15 cm的树木占86.8%。重度火烧样地中,兴安落叶松延迟性死亡率明显升高,最低为66.7%,最高为100%,胸径为26~35 cm的树木没有发生延迟性死亡。火烧程度与径级对兴安落叶松延迟性死亡的影响均极显著(P<0.01),火烧程度与径级的优势比值分别为3.208和0.725。预测发生延迟性死亡的准确率为71.7%,预测为未发生延迟性死亡的总样本的准确率为60.4%。总样本A_U,C值为0.764,说明预测模型准确率较高,拟合效果良好。

Delayed Death of Xing'an Larch in Burns of Different Degrees

At the burns ofHanma National Nature Reserve in 2018 according to different degree of fire,we set up 9 standard sample(30 m by 30 m),a total of 760 delayed death Xing’an Larch as the whole sample,DBH was divided into seven levels to study the different degree burns different DBH grade Xing’an Larch delayed mortality characteristics.Taking 80%of the total sample(610)as the model sample,20%(150)as the test sample,we used Logistic regression equation to forecast the Xing’an Larch delayed mortality.The results showed that the delayed mortality of Xing’an Larch decreased gradually with the increase of DBH,with the decreases by 8.5%,28.5%and 17.4%,respectively.Of all delayed death trees,76.8%had a DBH of less than 10cm.In moderately burned plots,the delayed mortality of Xing’an Larch decreased with the increase of DBH,but the decrease was not as significant as that of moderately burned plots.Of all delayed death trees,86.8%were 6-15 cm in DBH.In the severe fire,the delayed mortality of Xing’an Larch was significantly increased,with a minimum of 66.7%and a maximum of 100%,and no delayed death occurred in the trees with a DBH of 26-35 cm.Degree of Logistic model for predicting the fire with DBH class two impact factors were extremely significant(P<0.01),with the ORsof 3.208 and 0.725,respectively.The accuracy range of predicting the occurrence of delayed death was 59.1%-84.6%,and the total sample accuracy was 71.7%.The total sample accuracy was 60.4%.The A_U,C of the total sample was 0.764,indicating that the prediction model has a high accuracy and a good fitting effect.