小黃薊馬於檬果園內族群發生及其消長因子之分析

Occurrence of Scirtothrips dorsalis Hood (Thysanoptera: Thripidae) and analysis of the factors influencing population dynamics on mango

魏妙楹(Miao-Ying Wei)

指導教授 ： 張念台

國立屏東科技大學/農學院/植物醫學系所/碩士(2011年)

https://doi.org/10.6346/NPUST.2011.00115

摘要

小黃薊馬(Scirtothrips dorsalis Hood)為嚴重危害檬果之纓翅目薊馬
科(Thysanoptera: Thripidae)害蟲，為瞭解其族群於檬果上之發生情形並
探討其消長因子，本研究首先評估其偏好顏色、黏板設置方法及每日活動高峰，進而調查其週年發生情形，並利用資料探勘技術分析影響其族群發生之關鍵因子。小黃薊馬在檬果植株高度平均為280±20cm時，開花期與果實期最適誘集高度分別為離地250cm(29.9±3.3/trap)與150cm(339.1±42.2/trap)
、200cm (155.5 ± 17.6 /trap)。在黃、綠、藍、白等4種顏色裡，其最偏好黃色，其中亮黃色(81.4±7.2/trap)的捕獲效能較橙黃色(66.0±5.4
/trap)為佳。日飛行高峰調查，發現於每日10-12時(日均溫20.1℃)捕獲之族群量為最高(16.0±1.6/trap)較其他時段有顯著差異(p<0.05)。另進行愛文、金煌及柴檨仔等不同檬果品種之小黃薊馬族群動態調查，結果顯示在小果期、檬果抽新梢時為其族群發生之高峰，平均捕獲蟲量最高可達532.5/trap
，而在果實期，小黃薊馬於上述品種的捕獲率分別佔所有薊馬之63.28, 75.76及74.19%，顯著高於各品種生長期與開花期之捕獲率。此外，透過資料探勘平台-懷卡托智慧分析環境(waikato environment for knowledge analysis, WEKA)比較羅吉斯迴歸、決策樹及類神經網路三模式之分類及預測功能的優劣，發現決策樹經10摺交叉驗證後所獲得之整體正確性達83%以上，因此依據小黃薊馬防治基準(17/trap/3day)將資料分為第I類未達防治標準與第II類建議採取防治措施二類別，進一步用決策樹找尋其族群發生歸屬上兩類別的主要影響因子，結果顯示不論檬果品種為何，檬果生長期以溫度 (26.65℃)、果實期以相對濕度(75.75%)、開花期以園中仍有花之月份(2 月)為影響族群是否達防治基準的關鍵因子，另配合其他質與量的影響因素共可產生97項決策規則，其中57項精確率達80%以上，例如愛文檬果於5月果實期之相對濕度>75.75%，降雨量≦26.33mm時，其判定為第II類之精確率
可達82.78%。本研究提供檬果園中以黏板監測小黃薊馬之適宜方法及
快速監測之適當時機，亦利用資料探勘新技術整合並分析質性與量化資
料，探知小黃薊馬族群消長之關鍵因子，並初步建置其與族群發生之關
聯性，可望於未來做為小黃薊馬預警系統之基礎。

關鍵字

小黃薊馬 ； 誘捕效能 ； 族群發生 ； 資料探勘 ； 關鍵因子

並列摘要

Scirtothrips dorsalis Hood (Thysanoptera: Thripidae) is a key pest that cause serious damage on mango. In order to understand their population occurrence and analysis of the key factor influencing population dynamics on mango, this study tested the color perference, sticky trap setting method and daily activity of thrips, in addition to the field survey, the data mining technigue also used to analyze the key factors for the occurrence of S. dorsalis. Result of the height test for traps showed that during blooming stage in December, and fruiting stage in March and April, the optimal trap height are 250 cm (29.9 ± 3.3 /trap), 150 cm (339.1 ± 42.2 /trap), and 200 cm (155.5 ± 17.6 /trap), respectively. The color preference tests with yellow, green, blue sticky trap, indicated S. dorsalis preferred bright yellow color trap mostly (81.4 ± 7.2 /trap) to the orange yellow (66.0 ± 5.4 /trap). The daily rhythm test found that significantly more S. dorsalis population (16.0 ± 1.6 /trap) was caught during AM 10:00-12:00 /day than that in other periods (mean daily temperature was 20.1℃). In the field survey on different cultivar of Irwin, Jin-Hwang, and Tsar-Swain mango, the results showed that the peak of S. dorsalis population occurring in young fruit, and young leave stage when S. dorsalis density can reach as high as 532.5 /trap. The occurrence ratios of S. dorsalis in fruiting stage of Irwin, Jin-Hwang, and Tsar-Swain were 63.28, 75.76 and 74.19%, respectively, and significantly higher than that of other thrips. Furthermore, the data mining platform - waikato environment for knowledge analysis (WEKA) was used to compare logistic regression, decision trees (DT) and artificial neural networks (ANNs) models, the result indicated correctly classified instances of DT is ≧ 83% by 10-fold cross validation. The economic threshold (ET) of S. dorsalis (17 /trap/3 day) was used as the standard and classified the thrips population as I (Below ET) and II (recommended control measures) categories. The analysis of the major population impact factors by DT model showed that the key factor for thrips population reaching ET on vegetative growth stage is temperature (26.65℃), on fruiting stage is the relative humidity (75.75%), and on blooming stage is the month when flowers still bloom in orchard (February). Combining all the quality and quantity factors, there are 57 out of 97 decision rules with higher precision rate (≧ 80%). For example, when relative humidity > 75.75%, precipitation ≦ 26.33 mm during blooming stage (in May), the precise rate is 82.78% on Irwin mango orchard when the population being determined as class II. This study provides the standard operation procedures for sampling of S. dorsalis by sticky trap; integrate and analyze qualitative and quantitative data with new data mining technology to find out key factors influencing the population dynamics of S. dorsalis. The preliminary correlation between those factors and the occurrence of thrips population can help building early warning system to control it in the future.

並列關鍵字

Scirtothrips dorsalis Hood ； catch efficiency ； population occurrence ； data mining ； key factor

參考文獻

Chang CS, Wang NT. 2008. Impact of global warming trend on fruit production in Taiwan. Crop Environ Bioinfo 5: 196-203. (in Chinese)

Chiu YC, Lin FC, Shih HT, Wang CL. 2010a. Toxicity of insecticides to Scirtothrips dorsalis Hood (Thysanoptera: Thripidae) on Mango. J Taiwan Agric Res 59: 134-141. (in Chinese)

Huang KC. 1989. The population fluctuation and trapping of Thrips palmi in waxgourd. Res Bull Taichung DARES 25: 35-41. (in Chinese)

Liao SC, Liao WC. 2002. The control effects of the colored sticky traps, plastic covered, volatile chemicals combined with plant extracts against the Thrips palmi Karny in eggplant field. Res Bull Kaohsiung DARES 13: 14-25. (in Chinese)

Lin FC, Chiu YC, Hsu MY, Wang CL. 2010a. Morphological identification of thrips larvae (Insecta: Thysanoptera). J Taiwan Agric Res 59: 151-164.