棉花苗期棉蚜的二项式抽样设计及其精度分析(英文)

1991年4月—7月在河北省北京农业大学曲周试验站的棉花地进行苗期棉蚜（Aphisgossypii）的田间抽样调查，共收集到24组抽样数据。用泰勒幂法则对数据进行拟合，得到棉花苗期棉蚜为聚集分布。利用每样方（株）虫口不超过数阈值T（分别为0，1，2…9，10，15，20，30）头蚜虫的植株比例（PT）与种群密度（m，头／株）的关系，拟合经验关系式ln（m）=a十bln-ln（PT）]，通过对不同数阈值T的回归决定系数（r2）、种群的回归估计方差（Var（m））和抽样精度（用d估计）等进行综合分析，结果表明该蚜虫在数阈值T为15时，回归估计方差最小，回归决定系数和d值最大，因而T=15为该蚜虫的理想数阈值；而小的T值尤其是T为O时，由于产生太大的回归估计方差，很小的回归决定系数和d值即抽样的精度极低，因而不宜在实际中应用于棉蚜的二项式抽样设计。

BINOMIAL COUNT ANALYSIS BASED ON THE SPATIAL DISTRIBUTION AND POPULATION DENSITY ESTIMATION OF COTTON APHIDS

Based on field population sampling of Aphis gossypii on cotton seedlings in Quzhou Experiment Station of China Agricultural University in Hebei Province in 1991, we obtained a data set consisting of 24 estimates of mean aphid density (m , number of aphids per plant) 9 variance (s2) and the proportion of plants (PT) with no more than T aphids (T=0, 1, 2,..., 8, 10, 15, 20, respectively and defined as tally threshold). Taylor's power law fitted the data well (r2 =0. 958). The resulting slope (1. 515) was significantly greater than 1, indicating that the spatial distribution of thisaphid was in aggregated pattern. An empirical relationship between m and PT was developed for each T value using the parameters from the linear regression In(m) =a bln -ln(PT)]. The importance of the T values in reduction of sampling errors and their application to binomial sampling plans are discussed. Small T values, particularly aphid-free plant (T = 0, conventional binomial sample),could lead to spurious estimates of m from PT. A value of T from 10 to 15 was recommended to develop binomial sampling plans for the aphids on cotton seedlings because of the relatively small sampling errors.