The relationship between the distribution of halobios and environmental variables has been a focus of present research in marine ecology. Species distribution models (SDMs) have been widely employed to predict the distribution patterns and potential habitat suitability assessments of marine species and provide an efficient approach for marine biodiversity conservation, invasive species prevention, and fishery management. SDMs mainly include habitat suitability index models, mechanism models, and statistical models. In this paper, the theoretical basis of SDMs was firstly concluded and summarized. Next, the exploitation and case-studies of SDMs, especially the statistical models, for predicting potential distribution of marine species were reviewed. Then, comparisons of various methods for variable selection and model validation were made. Conclusions could be drawn that Akaike information criterion showed excellent performance when it came to variable selection, while Kappa coefficient and Area Under receiver operating character Curve (AUC) were widely used in relation to model validation. Finally, problems and prospects of SDMs were discussed. With the development of research on physiological characteristics, using mechanism models to predict potential habitats of halobios would become a trend.
海洋生物的栖息分布与环境要素的关联性一直是海洋生态学研究的热点之一.近年来,物种分布模型被广泛应用于预测海洋物种分布、潜在适宜性生境评价等研究,为保护海洋生物多样性、防治外来物种入侵及制定渔业管理措施等提供了一条有效途径.物种分布模型主要包括生境适宜性指数模型、机理模型和统计模型.本文对物种分布模型的理论基础进行了归纳和总结,回顾了物种分布模型在预测海洋物种潜在地理分布研究中的开发与应用,重点介绍了不同类型统计模型在海洋物种潜在分布预测中的研究实例.比较各种选取变量和模型验证方法,认为赤池信息准则对于选取模型变量具有优势,Kappa系数和受试者操作特征曲线下面积在验证模型精度中应用最广泛.阐述了物种分布模型存在的问题及未来发展趋势,随着海洋生物生理机制研究的进一步深入,机理模型将是今后物种分布模型发展的重点.
Keywords: model evaluation; marine species; mechanism model; species distribution model; statistical model.