A GIS-based Multi-scale Approach To Ecogeographical Invasive Species Modeling

Invasive exotic species pose a growing threat to the economy, public health, and ecological integrity of nations worldwide. Explaining and predicting the spatial distribution of invasive exotic species is of great importance to prevention and early warning efforts. But there are several unresolved problems during the prediction modeling including the uncertainty in biodiversity data, the uncertainty in model selection and the uncertainty in niche cross the geographic space. As the same time, ecogeographical GIS modeling has the limitation of the spatial scale and classification level and it is difficulty to analyze in multi-scales. This paper makes pilot study on the GIS-based multi-scale ecogeographical GIS invasive species modeling, coupled with ecology, geography, RS, which is supported by the National Science Foundation of China (NSFC) under grant No.40371084 named "Research on biotic invasion potential in China and U.S. by using geospatial modeling". The main research work includes the following parts:(1) The paper proposes the various steps of predictive modeling from the conceptual model formulation to prediction and application, and discusses a selection of important conceptual aspects. Aiming to the problems of vague concepts in predictions of species distribution, this paper develops the conception of the uncertainty that are relevant to the GIS-based multi-scale ecogeographical GIS invasive species modeling from the point of view of the quality of spatial data and the spatial analysis. This paper proposes that the uncertainty in models of species distribution falls broadly into four main groups:uncertainty in biodiversity data, the uncertainty in model selection, the uncertainty in niche cross the geographic space and the uncertainty in niche cross the spatial scales. A global precedent approach for decreasing the uncertainty is proposed.(2) It is a long-and-hard task for the collection and extraction of the geographical situation information, which is a crucial part of habitat-suitability modeling. The author gathers the herbarium specimen for several months in the herbarium of the Kunming Institute of Botany and the herbarium of the Institute of Botany of the Chinese Academy of Sciences. The herbarium specimen pictures from the museum of United States and Canada are distinguished and processed. With the help of the toponym database and internet map tool, the location information (longitude/latitude) of the invasive species herbarium and literature records during 30 years are extracted.(3) This paper defines three general types of ecogeographical pattern GIS invasive species modeling and relates these to the kinds of data required. These types of patterns differ in the space in which shapes are formed: principally geographic space, and environmental space. These three methods including geographic space modeling methods based on the Convex Hull, Global environmental space modeling methods based on the logistic regression and local environmental space modeling method based on GARP (Genetic Algorithm for Rule-set Production) are applied to the case of ragweed. and the advantage and shortage are analyzed and compared Presented with the challenge, this paper develops an improved logistic regression approach using Information Theory. Information-theoretic approach is used to compute and assesse the modeling choice as well as produced a weighed-average model based on the multiple-models rather than using the sole model with the lowest AIC value or the highest Akaike weight. This multiple-model inference is useful to reduce model selection bias. In order to reduce the sampling bias, the paper develops a new approach specifically to compartmentalize the relative habitat-suitability for invasive species using a frequency statistic approach. A Self-adaption and more robust method are proposed when using the habitat-suitability model developed in native ranges to "project" onto exotic ranges.(4) This paper reviews the issues on multi-scale in ecogrographical modeling systematically and the scale effect in the process of the GIS-based ecogeographical invasive species modeling. A multi-hierarchy mixed modeling method for predicting the distribution of species based on the Hierarchy Theory is proposed. In order to illustrate this approach, this paper develops multi-scales GIS protection modeling including macroscopical scales(national level), medi-macroscopical scales(provincial level and regional level) and microscopical scales(countial level) through a habitat-suitability analysis of Ageratina adenophora and analysis the importance of the GIS variables in different scales.