Estimating Indez of Population Control (IPC) by Re-sampling Techniques and Its Application to Population Life Table
By the mathematical model of population trend index developed by Watt (1961, 1963), Pang & Liang (1990) conceptualized the index of population control (IPC), proposed a simple and straightforward mathematical formula calculating this index to evaluate the suppressive effect of pest management agents. But to date there is no publication available focused on its variance estimation for this index, which hinders to draw statistically sound conclusion in screening pest management agents to form pest IPM system. Taking the natural population life table of rice leaf roller (Cnaphalocrocis medinalis, Lepidoptera: Pyralidae) published by Wu, Liang, and Pang (1986) as an example, through estimating IPC and its variance via delta method and re-sampling procedures for bio-agent released and chemical insecticides dusted treatments, whether their IPC truly reflect the control effectiveness of these agents are the topics concerned in this paper. By the large sample theory delta method provides a simple approach to estimate the approximate variance for IPC, but the normality of the IPC estimates and the bias of this estimate cannot be tracked. Among four procedures used with resampling technique, the distribution of the estimates from delete-1 Jackknife does not match the normal pattern because of its small estimates available. And there is a big bias in its mean estimate. Thus, the estimates from this procedure are not applied in further analysis. After logtransformation, the distributions of the estimates from delete-d Jackknife, empirical and theoretical Bootstrap procedures match the corresponding normal patterns. Thus, it is possible using Z-test criterion to conduct hypothesis test to judge if the control agent is better than blank control at generation level. If ignoring the normality of the estimates, the low boundaries of the approximate 95％ confidence intervals of delete-d Jackknife and empirical Bootstrap calculated with STD 96 . 1 mean × ± are negative. This shows that the approximate normality is not achieved for IPC estimates from delete-d Jackknife and empirical Bootstrap before logarithm transformation even though 10,000 estimates are produced. Judged by its 95％ confidence intervals and the p-value from Z-test, releasing wasp is much better than blank control since its IPC is significantly less than 1 with p-value<0.05, dusting insecticide is worse than blank control but it is not statistically with p-value > 0.2 although its IPC is bigger than 1. The conclusions drawn in this paper are similar to those drawn only by the mean estimates of IPC, but our conclusions supported by statistic principles are more plausible and dependable.
Indez of Population Control (IPC) Life Table Rice Leaf Roller Delta Method Resampling Technique (Jackknife and Bootstrap)
TAO FangLing WU WeiJian
Laboratory of Insect Ecology, South China Agriculture University, P.R.China
国际会议
2009 International Institute of Applied Statistics Studies(2009 国际应用统计学术研讨会)
青岛
英文
1-10
2009-07-25（万方平台首次上网日期，不代表论文的发表时间）