论文部分内容阅读
目的探讨并建立湖沼地区钉螺数量的预测模型,为钉螺的定量化研究提供方法学依据。方法在安徽省贵池区秋浦河沿岸随机数字表法随机选择滩地作为研究现场,根据植被类型分层随机抽样,以交叉复核随机抽检法(随机数字表法)调查钉螺,分别用差分 GPS 仪、T&D Recorder forWindows、卷尺及烘干称重法收集高程、土壤温度(简称土温)和气温、植被高度、土壤湿度(简称土湿)和植被类型6个环境变量信息。对数据进行预处理,通过偏差量和 Akaike 信息准则比较不同误差分布和连接函数组合的广义线性模型,确定最佳模型结构,建立并验证预测模型效果。结果建模样本量为162框,变量之间存在着复杂的相关性,钉螺数与植被高度呈正相关(r=0.36),与土湿呈负相关(r=-0.22),气温与土温呈正相关(r=0.59),土温与植被高度呈负相关(r=-0.36),土湿与土温和气温均呈负相关(r=-0.34和-0.12)。广义线性模型的最佳结构是以 gamma 分布为误差分布、倒数为连接函数和均数平方为方差函数的模型结构。模型拟合结果显示高程、土湿、土温、植被类型和植被高度对于预测钉螺的数量有统计学意义,t 值分别为-3.202,3.124,-1.989,2.668和-2.371,P 值分别为0.00166,0.00214,0.04849,0.00846和0.01897,而气温的作用被土温取代没有进入模型。结论广义线性模型可用于建立钉螺的预测模型,为钉螺的定量化研究提供了广阔的研究前景。
Objective To explore and establish a prediction model for the number of snails in lakes and mountains, and provide a methodological basis for the quantitative study of snails. Methods The random selection of beach land as the research site in the Qiu River area of Guichi District, Anhui Province was conducted randomly. Stratified random sampling was conducted according to vegetation types. The snails were investigated by cross-check random sampling method (random number table method) T & D Recorder forWindows, tape measure and drying weighing method to collect the information of six environmental variables including elevation, soil temperature (temperature) and temperature, vegetation height, soil moisture (referred to as soil wet) and vegetation type. The data is preprocessed, and the generalized linear model with different error distribution and connection function combination is compared by the deviation and Akaike information criterion to determine the best model structure and establish and verify the prediction model. Results The sample size was 162 frames. There was a complicated correlation between variables. The number of snails was positively correlated with vegetation height (r = 0.36), negatively correlated with soil moisture (r = -0.22), and the temperature and soil temperature were positive (R = 0.59). Soil temperature was negatively correlated with vegetation height (r = -0.36), soil moisture was negatively correlated with soil temperature (r = -0.34 and -0.12). The optimal structure of the generalized linear model is the model structure with gamma distribution as error distribution, reciprocal connection function and mean square as variance function. The model fitting results showed that elevation, soil wetness, soil temperature, vegetation type and vegetation height were statistically significant for the prediction of snails, with t values of -3.202, 3.124, -1.989, 2.668 and -2.371, respectively, P values of 0.00166 , 0.00214, 0.04849, 0.00846 and 0.01897, respectively, while the temperature effect was replaced by soil temperature without entering the model. Conclusion The generalized linear model can be used to establish the prediction model of Oncomelania snails, which provides a broad prospect for the quantitative study of Oncomelania snails.