Study On Rare And Cluster Species Distribution Model And Sampling Technique

The study of relationship between species distribution and environmental, species distribution simulation and simulation of distribution patterns of biodiversity, which play an important role in theoretical and practical for global changing, biodiversity monitoring, conservation and evaluation, species invasive and so on. However, whether the analysis of relationship between species and environment, or the simulation of the potential distribution area and future distribution trends forecast, the most basic and important task is investigated and recorded the status of species distribution. Forest resource survey, especially the forest resource shows scare, cluster distribution. Such as the rare tree species, shrubs, herbs and other endangered animals, the investigation more difficult to perform, and generally not efficient. To improve the accuracy and efficiency of resource survey for sparse species. The adaptive sampling techniques and methods were used for investigating at home and abroad. But adaptive cluster sampling have some problems, e.g, survey people can not know the distribution of species ahead of time. So, There is very important significance for investigate species resources simply, especially the distribution of forest resource.Species distribution models use present species points and absent species points and their corresponding environmental factors to predict all possible species distribution. Adaptive cluster sample combine with species distribution model predict species distribution in advance. The investigation has a target, and greatly improves the efficiency of the investigation. Therefore, the study of species distribution model and its guidance to sampling techniques are extremely important.Based on the population data of four species, Elaeagnus angustifolia, Haloxylon ammodendron, Nitraria tangtorum, Tamarix chinensis, which are rare and clustering. Research area lie at the edge of Wulanbuhe desert at Bayangaolei Town in Dengkuo County in west Inner Mongolia. The spatial pattern of four species were studied first. Actual data and remote sensing data were combined. First the application of MaxEnt and GARP species distribution models were studied. Then the research of species distribution model to guide sampling technique was presented. In the study of MaxEnt species distribution model, the radio of training data and test data, change of environment variable, threshold, sample size, quadrat size and species, which influence on predictive accuracy were main studied. However, about GARP species distribution model, only sample size, quadrat size and species were studied. At last, MaxEnt and GARP models were compared. When study on the guidance of species distribution models to sampling technique, first the probability of present species was studied, then the sample method was presented about rare, cluster species under the guidance of species distribution models. Research conclusions and innovations are mainly as follows:The main innovations are as follows:1. The influence of various factors on predictive accuracy of species distribution was studied.2. The method of sampling techniques and species distribution model to study together was presented, and use the results of species distribution model to guide the sampling.The main research conclutions are as follows:1. Through the study of influence of sample size, quadrat size, species and environment variables,on predictive accuracy, we received that : when the quadrat size is 10×10m, MaxEnt model chooses 100 sample size the accuracy is the best and GARP model choose 150 the accuracy is the best. The predictive accuracy is much large when quadrat size close to resolution of these two models. On the contrary, the predictive accuracy is low.MaxEnt and GARP species distribution models are both better to Elaeagnus angustifolia and Tamarix chinensis, but worse to Haloxylon ammodendron and Nitraria tangtorum. It is proved that different degree of aggregation, received the different predictive accuracy. Different environment variables, different predictive accuracy. And the more variables received higher accuracy.2. Based on the design of study area and 10×10m quadrat size. The probability of present species was presented. When the probability more than 0.55, Species Elaeagnus angustifolia and Tamarix chinensis are shown presence. When the probability more than 0.45, Species Haloxylon ammodendron are shown presence. When the probability more than 0.65, Species Nitraria tangtorum are shown presence. When the probability more than 0.45, Species Haloxylon ammodendron are shown presence. When the probability more than 0.65, Species Nitraria tangtorum are shown presence. When the probability more than 0.7, Species Hedysarum scoparium are shown presence. When the probability more than 0.4, Species Artemisia ordosica are shown presence. Overall, When the species distribution overview was known, if the more point of actual distribution, the smaller probability value was choice; if the smaller point of actual distribution, the larger probability value was choice.3. Though the study of species distribution models together with sampling technique, it is proved that the species of aggregated distribution use based on HT estimator adaptive sampling is the best.