广义模型及分类回归树在物种分布模拟中的应用与比较

比较3个应用较广的模拟物种地理分布模型:广义线性模型(GLM)、广义加法模型(GAM)与分类回归树(CART)对中国树种地理分布模拟的优劣,以提出更为合适的模拟物种地理分布模型,并用于预测气候变化对物种地理分布的影响。3个模型对中国15种树种地理分布的模拟研究表明:除对油松、辽东栎分布的模拟精度稍差外,对其余树种分布的模拟精度均较高,其中以GAM模型最好。结合地理信息系统(GIS),比较分析了这3个模型对青冈、木荷、红松和油松4种树种的地理分布模拟效果,结果亦表明:这3个模型均能很好模拟青冈和木荷的地理分布,而GLM模型对红松分布的模拟结果不太理想,3个模型对油松分布的模拟结果均不甚理想,其中以GLM模型最差。基于3个模型对未来气候变化下青冈与蒙古栎地理分布的预测表明:GLM模型与GAM模型对青冈分布的预测结果较为接近,青冈在未来气候变化情景下向西和向北扩展,而CART模型预测青冈在未来气候变化情景下除有向西、向北扩展趋势外,广东和广西南部的青冈分布区将消失;3个模型均预测蒙古栎在未来气候变化情景下向西扩展,扩展面积的大小为:模型的模拟面积>模型>模型。

Application and comparison of generalized models and classification and regression tree in simulating tree species distribution

Comparison of three models: Generalized Linear Models(GLM),Generalized Additive Models(GAM) and Classification and Regression tree(CART) in simulating tree species distribution in China were done in this paper, in order to select a suitable model for simulating and predicting tree species distributions under climate change in the future. The simulating 15 tree species distributions by three models indicated that three models could simulate the tree species geographical distributions very well except Pinus tabulaeformis and Quercus liaotungensis. GAM model is the best one among them. Combining with geographical information system(GIS), the simulating effects for geographical distributions of Cyclobalanopsis glauca, Schima superba,Pinus koraiensis, and Pinus tabulaeformis were compared. The results showed that three models could simulate the geographical distributions of of Cyclobalanopsis glauca and Schima superba very well; GLM model could not simulate the geographical distribution of Pinus koraiensis very well; and three models could not simulate the geographical distribution of Pinus tabulaeformis. The relative performance of different models was discrepant among species, suggesting that the most accurate model varies between species. Moreover, the geographical distributions of typical tree species: Cyclobalanopsis glauca and Quercus mongolica under climate change were studied based on GLM model, GAM model and CART model. The simulated results of Cyclobalanopsis glauca geographical distributions by GLM model and GAM model were similar, and showed its geographical distribution would shift to the west and the north under climate change, however CART model indicated that the Cyclobalanopsis glauca geographical distributions in Southern Guangdong province and southern Guangxi province would disappear. The simulated results of Quercus mongolica geographical distributions by three models were similar, and they would shift to the west, and the expanding areas ranged from GAM model, GLM model to CART model.