Laboratory of Geo-Information Science and Remote Sensing, Wageningen University and Research, Wageningen, Netherlands.
OpenGeoHub Foundation, Wageningen, Netherlands.
PeerJ. 2023 Jun 23;11:e15593. doi: 10.7717/peerj.15593. eCollection 2023.
The global potential distribution of biomes (natural vegetation) was modelled using 8,959 training points from the BIOME 6000 dataset and a stack of 72 environmental covariates representing terrain and the current climatic conditions based on historical long term averages (1979-2013). An ensemble machine learning model based on stacked regularization was used, with multinomial logistic regression as the meta-learner and spatial blocking (100 km) to deal with spatial autocorrelation of the training points. Results of spatial cross-validation for the BIOME 6000 classes show an overall accuracy of 0.67 and R of 0.61, with "tropical evergreen broadleaf forest" being the class with highest gain in predictive performances (R = 0.74) and "prostrate dwarf shrub tundra" the class with the lowest (R = -0.09) compared to the baseline. Temperature-related covariates were the most important predictors, with the mean diurnal range (BIO2) being shared by all the base-learners (,random forest, gradient boosted trees and generalized linear models). The model was next used to predict the distribution of future biomes for the periods 2040-2060 and 2061-2080 under three climate change scenarios (RCP 2.6, 4.5 and 8.5). Comparisons of predictions for the three epochs (present, 2040-2060 and 2061-2080) show that increasing aridity and higher temperatures will likely result in significant shifts in natural vegetation in the tropical area (shifts from tropical forests to savannas up to 1.7 ×10 km by 2080) and around the Arctic Circle (shifts from tundra to boreal forests up to 2.4 ×10 km by 2080). Projected global maps at 1 km spatial resolution are provided as probability and hard classes maps for BIOME 6000 classes and as hard classes maps for the IUCN classes (six aggregated classes). Uncertainty maps (prediction error) are also provided and should be used for careful interpretation of the future projections.
使用来自 BIOME 6000 数据集的 8959 个训练点和基于历史长期平均值（1979-2013 年）的 72 个环境协变量堆栈，对生物群落（自然植被）的全球潜在分布进行建模。该模型基于堆叠正则化的集成机器学习模型，使用多项式逻辑回归作为元学习者，并采用空间分块（100km）来处理训练点的空间自相关。BIOME 6000 类别的空间交叉验证结果显示，整体准确率为 0.67，R 值为 0.61，其中“热带常绿阔叶林”类别的预测性能增益最高（R=0.74），“匍匐矮灌木苔原”类别的预测性能增益最低（R=-0.09），与基线相比。与温度相关的协变量是最重要的预测因素，平均昼夜范围（BIO2）是所有基础学习者（随机森林、梯度提升树和广义线性模型）共享的。该模型接下来用于根据三种气候变化情景（RCP 2.6、4.5 和 8.5）预测 2040-2060 年和 2061-2080 年期间未来的生物群落分布。对三个时期（现在、2040-2060 年和 2061-2080 年）的预测结果进行比较，结果表明，干旱程度的增加和温度的升高可能导致热带地区（到 2080 年，从热带森林到稀树草原的转变面积高达 1.7×10^5km^2）和北极圈周围的自然植被发生显著变化（到 2080 年，从苔原到北方森林的转变面积高达 2.4×10^5km^2）。还提供了 1km 空间分辨率的全球预测图，作为 BIOME 6000 类别的概率和硬类别图，以及 IUCN 类别的硬类别图（六个聚合类别）。还提供了不确定性图（预测误差），应谨慎用于解释未来的预测。
