麦长管蚜的序贯二项式分类抽样设计与管理决策优化：样本单位虫口数阈的重要性

OPTIONAL PLANS OF BINOMIAL SEQUENTIAL CLASSIFICATION SAMPLING FOR MANAGEMENT DECISION MAKING OF THE ENGLISH GRAIN APHID (HOMOPTERA: APHIDIDAE):VALUE OF TALLY THRESHOLDS OF APHIDS IN SAMPLE UNITS

依据１９８８～１９９１年间对麦长管蚜（Ｓｉｔｏｂｉｏｎａｖｅｎａｅ）种群的抽样数据和新的抽样理论而设计了用样本单位（分蘖株）虫口数阈Ｔ（每株不超过Ｔ头蚜虫，Ｔ≥０）进行样本分类的序贯二项式分类抽样方案，用于蚜害的管理决策。负载不超过Ｔ头蚜虫的植株比例（ＰＴ）与种群密度（ｍ，头／株）的关系通过拟合经验式ｌｎ［－ｌｎ（ＰＴ）］＝ａ＋ｂｌｎ（ｍ）和负二项分布的概率模型而建立。一系列Ｔ值（０～２５）和３个干扰阈限即经济危害水平（２．５和１０头／株）用于抽样方案的设计和比较，以优化麦作不同生育期控制蚜害的决定。抽样设计的可靠程度和经济性状分别由运行特征（ＯＣ）函数和平均样本数（ＡＳＮ）函数判别。结果显示，各干扰阈限的抽样设计随Ｔ值的增大而愈趋合理。传统的０－ｌ抽样（Ｔ＝０）被证明错误决策的风险很大，所需样本数最多，应当摒弃。考虑到Ｔ过大会增加处理单位样本的时间，作者建议使用略高于干扰阈限的Ｔ值进行麦蚜的分类抽样和田间管理决策。

Binomial sequential classification sampling plans were developed for use in classifying populations of the English grain aphid, Sitobion avenae, below or above an intervention threshold density(mIT) for management decision making. Based on field data from sampling the aphid populations on wheat during the 1988～1991 summers, the proportion of tillers with no more than T(tally threshold) aphids(pT) was related to mean aphid density(m)with an empirical model in the form of ln[ -ln (PT)] = a+bln(m)or the negative binomial distribution (NBD) model. Three mean intervention threshold(mIT) values(2, 6,and 10 aphids per tiller) with T values ranging from 0 to 25 were used to meet the needs for control of the aphid pest varying in density at the flowering,milky and mid-dough stages, respectively. The sampling plans were evaluated using operating characteristic(OC) and average sample number (ASN) functions computed using the empirical PT-m model or the NBD model. The robustness of the sampling plans improved with increasing T values. When empty tiller(T= 0)was used to classify sample units, the resulting sampling plans were not robust for all mIT values con- sidered. This indicated a high risk for use of conventional binomial sampling plans. Based on the OC and ASN functions,the sampling plans recommended for use were those developed with T values slightly higher than the intervention thresholds at the corresponding crop stages.Sampling stop lines for each mIT and T were provided along with a discussion of their applica-tion to aphid sampling and management decision making.