Bayesian Analysis of Tree Distributions Across Space and Time in Eastern North America 2010-2011

The distributions of many organisms are spatially autocorrelated, but it is unclear whether including spatial terms in species distribution models (SDMs) improves projections of future species distributions. We provide the first comparative test of a purely spatial SDM, a purely non-spatial SDM, and an SDM that combines spatial and environmental information. Spatial SDMs provided better fits to the calibration data, more accurate predictions of a hold-out validation data set of modern trees, and lower false positive rates at all time periods than non-spatial SDMs. Hindcasted projection of spatial SDMs had higher variance than those of non-spatial SDMs. Overall predictive performance of non-spatial and spatial SDMs varied temporally and as a function of niche overlap. Ecological modelers should include spatial terms in SDMs used for projecting future distributions of species.

诸多生物的分布均存在空间自相关性，但目前尚不明确在物种分布模型（Species Distribution Models，SDMs）中纳入空间项，是否能够提升未来物种分布的预测精度。本研究首次针对纯空间物种分布模型、纯非空间物种分布模型，以及融合空间与环境信息的物种分布模型开展对比测试。相较于非空间物种分布模型，空间物种分布模型对校准数据的拟合效果更优，对现代树木留出验证数据集的预测精度更高，且所有时间段下的假阳性率更低。空间物种分布模型的回溯预测结果，其方差高于非空间物种分布模型的对应结果。非空间与空间物种分布模型的整体预测性能，会随时间推移以及生态位重叠程度的变化而发生改变。生态建模研究者在构建用于预测物种未来分布的物种分布模型时，应纳入空间项。