摘要
桔小实蝇Bactrocera dorsalis(Hendel)是一种多食性害虫,明确其可能适生的区域对该虫的科学监测及防治意义重大。利用桔小实蝇在我国的已知分布点数据和亚洲地区的14个环境地理变量图层,运用GARP生态位模型结合GIS空间分析模块预测了该虫在亚洲的地理分布。结果表明桔小实蝇可分布在中国、日本、菲律宾、马来西亚、泰国北部、越南、柬埔寨、老挝、缅甸、尼泊尔、巴基斯坦、孟加拉国和斯里兰卡,这与EPPO报道的分布区域一致。将拟合过程中获得的生态位运算法则投影到我国,并考虑模型间的一致性,预测桔小实蝇在我国各省及市县范围的分布:云南大部、四川南部和东部、贵州大部、重庆大部、广西、广东、台湾、香港、澳门、海南、福建、江西、浙江大部、湖南大部、湖北大部、上海、江苏南部、河南局部及安徽部分地区为桔小实蝇的适生区。次适生区沿适生区周围分布,为四川、贵州、重庆、湖北北部、河南南部和江苏南部的一些零星地区。适生区和次适生区大多有较高密度的寄主果树,为桔小实蝇的生存提供了条件。预测结果经独立验证数据的适合性测验表明,选择的最优模型具有显著的统计学意义,显示了很好的预测能力。GARP生态位模型可以解决生态学、生物地理学和环境保护方面的一系列问题,具有广泛的应用前景,为物种已知基础分布点资料的综合分析以及有害生物的适生性分析、监测和防治提供了技术平台。
Oriental fruit fly, Bactrocera dorsalis （Hendel）, is a polyphagous pest. In order to monitor and control B dorsalis effectively, it is necessary to investigate the potential distributional areas of this pest. In this study, an ecological niche modeling method, machine-learning algorithm for nonparametric prediction of species distribution （the Genetic Algorithm for Rule-set Prediction, or GARP）, combined with Spatial Analysis function in GIS （ Geographic Information System） program, was applied to predict potential geographic distribution of B. dorsalis in Asia, especially in China. On the basis of the data about current distribution of B. dorsalis in China and 14 environmental and geological covariates in Asia, the resuhing GARP geographic predictions were exported as ASCII raster grid files into ArcGISS. 3 for further analysis. Projection of the distribution of B. dorsalis in Asia indicated that the pest could be distributed to some Asian countries and regions including China, Japan, Philippines, Malaysia, northern Thailand, Vietnam, Cambodia, Laos, Burma, Nepal, agreement with Plant Protection Pakistan, Bangladesh and Sri Lanka, broadly across eastern and southern Asia. The resuhs were in the known worldwide distribution regions of B dorsalis reported by the EPPO （ European and Mediterranean Organization）. Using the same sets of ecological dimensions and taking the model agreement into account, the ecological niche model was projected onto the landscape of provinces, municipalities and counties in China. The results showed that the most part of Yunnan Province, eastern and southern part of Sichuan Province, most part of Guizhou Province and Chongqing Municipality, Guangxi Zhuang Autonomous Region, Hong Kong, Macao, Guangdong, Taiwan, Hainan, Fujian and Jiangxi provinces, most part of Zhejiang, Hunan and Hubei provinces, Shanghai Municipality, southern part of Jiangsu, and some parts of Henan and Anhui provinces were suitable establishment areas of B. dorsalis＇ invasion, surrounded by sub-suitable establishment areas, which included some scattered areas in Sichuan Province, Guizhou Province, Chongqing Municipality, northern part of Hubei province and southern part of Jiangsu and Henan provinces. Host fruit plants have been cultivated at high densities in both suitable and sub-suitable establishment areas. The remaining test points were overlaid with the distributional maps to determine the numbers falling within the areas of predicted present and test the validity of this approach, and the one-tailed goodness-of-fit χ2-statistics were employed to assess the significance of the model. All of our predictions were highly statistically significant, indicating an excellent predictive ability. The success of this test demonstrated that the ecological niche modeling approaches provide a promising tool for addressing synthetic analyses technical platform based on primary sets of present-day point occurrence information. Moreover, this methodology provides a for the suitability analysis, monitoring, prevention and control of the pest.
出处
《生态学报》
CAS
CSCD
北大核心
2007年第8期3362-3369,共8页
Acta Ecologica Sinica
基金
上海市重大科研资助项目(03DZ19315)
国家重点基础研究发展规划"973"资助项目(2002CB111405)~~
关键词
桔小实蝇
GARP生态位模型
GIS
潜在分布预测
适合性验证
Bactrocera dorsalis
GARP ecological niche modeling
GIS
prediction of potential distribution
one-tail goodness-of-fit test
