Map of tree floristic patterns in Peruvian Amazonia

Floristic patterns of trees at genus-level were predicted throughout Peruvian Amazonia using Landsat satellite imagery, climate, elevation and soil layers. Genus-level tree floristic variation from 157 inventory plots was summarized through non-metric multidimensional scaling (NMDS). Random forest regression was used to predict each of the floristic gradients (NMDS axes 1, 2 and 3) at 450-m resolution over all Peruvian Amazonia. The model predicted 63%, 61% and 40% of the variation along the NMDS axes 1, 2 and 3 respectively. The map is limited to Peruvian Amazonia. Non-forest areas were masked out as well as areas with more than 2000 meters of altitude. The pixel values of each floristic gradient (NMDS axes) represent the position along the floristic ordination axes.

本研究借助陆地卫星（Landsat）影像、气候因子、高程数据与土壤图层，对秘鲁亚马逊全域的乔木属级植物区系格局开展预测。研究从157个野外调查样地获取乔木属级植物区系变异数据，通过非度量多维标度法（non-metric multidimensional scaling, NMDS）完成汇总与排序；随后采用随机森林回归模型，以450米的空间分辨率，对秘鲁亚马逊全域的各植物区系梯度（即NMDS轴1、2、3）进行空间预测。该模型对NMDS轴1、2、3沿线上的物种组成变异的解释度分别为63%、61%与40%。本研究生成的空间分布图仅覆盖秘鲁亚马逊区域，已对非森林区域以及海拔超过2000米的区域实施掩膜剔除。各植物区系梯度（NMDS轴）的像素值，代表其在植物区系排序轴上的对应位置。