---

作 者 ：李玲玲;李凤日;贾炜玮;董利虎

期 刊 ：植物研究 2014年 34卷 03期 页码：349-355

关键词：红松人工林；果实产量；Logistic模型；非线性回归；

Keywords:korean pine plantation, nut yields, Logistic regression, nonlinear regression,

摘 要 ：以黑龙江省佳木斯市孟家岗林场红松人工林为研究对象，基于在12块固定样地中测定的579株红松单株结实量数据，采用Logistic回归模型及非线性回归模型构建了红松单木果实产量预测模型。首先，根据各样木的结实情况，采用SAS9.22统计软件建立了人工红松单株木结实与否的Logistic概率模型，并作为判断林木是否结实的基础模型。然后，通过分析红松人工林单木结实量与林木各调查因子的关系，构建了单株木果实产量的非线性预测模型。结果表明，文中建立的Logistic模型判断红松结实的正确率在65%以上。通过分析拟合效果，选择y=a(D²CW)^b作为红松人工林果实产量预估的最优模型，其预估精度为77%，残差分布均匀，模型的拟合效果较好。最后采用2号样地实测的果实产量数据对两个模型同时进行了检验，结果表明该样地结实量的预估精度为92.78%。本研究为人工红松果实产量的预测提供了可行的方法。

Abstract:Logistic and non-linear models were developed for predicting nut yields for Korean pine (Pinus koraiensis) plantation based on 579 Korean pines measured in 12 sample plots in Mengjiagang forest farm, Jiamusi, Heilongjiang Province. First, the statistical analysis system (SAS 9.22) was employed using data of pine nut yields to establish Logistic model for predicting whether individual pine could seed or not. Second, a non-linear regression was set up for predicting nut yields of individual seeded pines using tree variables. The results showed that the predicting accuracy of whether individual pine could seed or not, was above 65% for Logistic model. The optimal model of predicting nut yields was y=a(D²CW)^b due to the best fitting performance, which included the prediction accuracy of 77% and evenly distributed model residuals. The accuracy assessment was 92.78% for the two models using the observed pine nut yield data of Plot 2, which indicated a good prediction performance. This paper has been provided suitable method for predicting pine nut yields for Korean pine plantation.

---