砧木对梨叶芽休眠的影响及其需冷量估算方法的统计评价
Effects of Rootstocks on Dormancy of Pear Leaf Buds and Statistic Evaluation for Estimating Methods of Chilling Requirement

以杜梨（Pyrus betulaefolia）和豆梨（Pyrus calleryana）实生树，以及嫁接在杜梨和豆梨2种砧木上的砂梨品种‘丰水’（Pyrus pyrifolia）为试材，利用3种需冷量计算模型，对以杜梨和豆梨为砧木的丰水梨叶芽休眠进程及需冷量变化进行线性回归和正态累积函数分析。结果表明：供试材料叶芽的休眠进程与3种模型估算的需冷量间均存在极显著的线性回归关系，≤7.2 ℃模型的估计值最大，而犹他模型的估计值最小；≤7.2 ℃模型计算的离回归标准差（Sy/x）最小，其精度和可靠度均优于0~7.2 ℃模型和犹他模型；砧木的需冷量与接穗‘丰水’叶芽的休眠进程有着密切的关系，低需冷量砧木有利于加速接穗品种的休眠进程，而高需冷量砧木延缓接穗品种的休眠进程；利用线性回归与正态累积函数估算需冷量的平均差异约为30 h，2种统计方法均能较好地减小需冷量估算过程中的随机误差。
The Bunge (P.betulaefolia Bunge), the Decne (P.calleryana Decne), and the P.pyrifolia cv. Housui grafted on Bunge and Decne were used as experimental materials in this paper to examine the effects of rootstocks on the dormant process of Hosui pear leaf buds and to analyze the changes in chilling requirement which were estimated by the three kinds of calculation models in linear regression and accumulation normal distribution function. The results showed that there were significant linear regression relationships between the dormant process of leaf buds in tested materials and the chilling requirements calculated by the three kinds of estimating models. The estimation value of chilling requirement by ≤7.2 ℃ model was the biggest and that by Utah model was the smallest in the three kinds of models.Since the standard error of the linear regression (Sy/x) in ≤7.2 ℃ model was the smallest among the three kinds of models, the estimation accuracy and reliability of chilling requirement seemed superior to the 0-7.2 ℃ and the Utah models.There was a close relationship between the chilling requirement of the rootstocks and the dormant process of leaf buds in Housui scion. The rootstock of low chilling requirements was helpful to accelerate the dormant process of the scion varieties, or it could delay the dormant process of the scion varieties.The average differences of chilling requirements which were calculated by linear regression and accumulation normal distribution function were about 30 h.It might be thought that the two kinds of statistic methods could decrease the random error effectively in estimation of chilling requirements.