2003年10月行政院農業委員會動植物防疫檢疫局證實在臺灣發現入侵紅火蟻,地點主要為臺北縣、桃園縣及嘉義縣等地。為了降低入侵紅火蟻在臺灣所造成的損失及其族群數量,動植物防疫檢疫局核准了芬普尼、百利普芬及賜諾殺等三種入侵紅火蟻的防治用藥,以供大眾使用。行政院農業委員會所屬桃園農業改良場及臺南農業改良場乃共同進行試驗,期以瞭解推薦藥劑之防治效果,試驗地點分別設於桃園縣八德市及嘉義縣水上鄉。本文目的即根據試驗結果配適模式來估計藥劑的防治率,並比較何種模式的配適較為有利。因為試驗資料類型為縱向資料,亦即對同一樣區在六個不同的時間點重複觀察六次,觀測值間存在群內相關,不適合直接利用一般前提假設為觀測值間為獨立的廣義線型模式配適,故採用廣義估計方程式與廣義線型混合模式對入侵紅火蟻防治用藥的試驗結果進行模式的配適。配適的結果顯示,利用廣義估計方程式與廣義線型混合模式所得到的平均值模式的結果相當類似,後者較前者在求算模式係數估值時具有較佳的資訊利用。且三種藥劑對於入侵紅火蟻均有顯著的防治效果,在施藥後第八週時,防治率至少可達50%以上。在不同地區施用藥劑的結果經檢定後確定,蟻塚數下降的趨勢在桃園與臺南兩地區相同,所以藥劑的藥效不會因在不同地區而有所不同。綜合而言,三種藥劑中以分普尼與賜諾殺兩者之滅蟻藥效相似,並且藥效優於百利普芬。
The aim of this study is to evaluate the effect of three bait-formulated insecticides, including Fipronil, Pyripronxyfen and Spinosyns, on the control rate of red invasive fire ants. Two field experiments were conducted, respectively, in Taoyuan and Chiayi Hsien where the infestations of red fire ant were spotted. Repeated counts of active ant mound in each field plot, with a size 100 m^2, were recorded during the period of eight weeks. Two statistical procedures were employed to analyze these two data sets by using the generalized linear models, i.e., the generalized estimating equation (GEE) model and the generalized mixed-effect model (GLMM). The estimates of control rate resulted from GEE and GLMM are quite similar, though the standard errors (SEs) of slope parameter estimates are different substantially. We recommend that the SEs obtained from GLMM can be used to construct relevant confidence intervals, since variance structure of GLMM does have a better description to the variation of data collected. It is worth noting that all three insecticides show remarkable consistency in control rates at the two experimental sites. Among these three insecticides, Fipronil and Spinosyns have a similar effect in reducing active ant hills, and the control rate is better than that of Pyripronxyfen.