Objective: To study the prediction model of O. hupensis in the lake and marshland regions in order to provide methodological basis for quantitative study of O. hupensis.
Methods: The research sites were randomly selected from the bottomlands along Qiupu River in the Guichi District, Anhui Province. A random and stratified sampling method was administrated according to the type of vegetation; the frame size of snail survey was 0.11 m2. Snail data was collected by crosscheck-random sampling inspection survey. Elevation, soil temperature and air temperature, height of vegetation, soil humidity and types of vegetation were measured through GPS machine, T&D Recorder for Windows, tape measure and attemperator. All the data were doubly inputted into the computer and checked. The final dataset for developing the prediction model was set up after necessary data preprocessing, such as, recoding the variable of elevation. The generalized linear models were used to develop the prediction model, and the statistics of deviance and AIC were used to determine the best model structure. Model diagnostics and model evaluation of efficiency were performed with the determined best model structure.
Results: The sample size was 162, and there were 6 explanatory variable including 2 categorical variables and 4 quantitative variables. A complicated relationship was observed among all the variables. Snail was positively associated with height of vegetation (r = 0.36), while negatively associated with soil humidity (r = - 0.22), and the air temperature had a close positive relations with soil temperature (r = 0.59), and the soil temperature was negatively associated with height of vegetation (r = - 0.36), the soil humidity had negative relations with the soil and air temperature (r = -0.34 and -0.12). The best structure fitting for the liner model selected in gamma distribution was the error distribution, reciprocal as the conjunction function in mathematics, and the mean square as the variance function. The results showed that the elevation, soil humidity, soil temperature, types and the height of vegetation were statistically significant to predict the O. hupensis, while t-values were -3.202, 3.124, -1.989, 2.668 and -2.371, respectively, and P-values were 0.00166, 0.00214, 0.04849, 0.00846 and 0.01897 respectively.
Conclusion: The generalized linear models can be used to develop the predictive model, which could broaden the area of quantitative study for O. hupensis.