Department of Ecology, Biogeochemistry and Environmental Protection, University of Wrocław, Maksa Borna Sq. 9, 50-328 Wrocław, Poland.
Institute of Agroecology and Plant Production, Wrocław University of Environmental and Life Sciences, Grunwaldzki Sq. 24A, 50-363 Wrocław, Poland.
Sci Total Environ. 2018 Jun 1;626:1373-1381. doi: 10.1016/j.scitotenv.2018.01.131. Epub 2018 Feb 19.
We compared the effectiveness of explanatory variables representing different environmental spheres on the risk of alien plant invasion. Using boosted regression trees (BRT), we assessed the effect of anthropogenic factors, soil variables, land relief, climate and landscape structure on neophyte richness (NR) (alien plant species introduced after the 15th century). Data on NR were derived from a 2 × 2 km grid covering a total area of 31,200 km of the Carpathian massif and its foreground, Central Europe. Each of the examined environmental spheres explained NR, but their explanatory ability varied more than two-folds. Climatic variables explained the highest fraction of deviation, followed by anthropogenic factors, soil type, land relief and landscape structure. The global model, which incorporated crucial variables from all studied environmental spheres, had the best explanatory ability. However, the explained deviation was far smaller than the sum of the deviations explained by the single-sphere models. The global model showed that the deviation that could be explained by variables representing particular spheres, overlapped. The variables representing landscape structure were not included in the global model as they were found to be redundant. Finally, the climatic variables explained a smaller fraction of the deviation than the anthropogenic factors. The partial dependency plots allowed the assessment of the course of dependencies between NR and particular explanatory variables after eliminating the average effect of all other variables. The relationships were usually curvilinear and revealed some values of environmental variables beyond which NR changed considerably.
我们比较了代表不同环境领域的解释变量对入侵外来植物风险的影响。使用增强回归树（BRT），我们评估了人为因素、土壤变量、地形、气候和景观结构对新生物种丰富度（NR）（15 世纪后引入的外来植物物种）的影响。NR 数据来自一个 2×2 公里的网格，涵盖了喀尔巴阡山脉及其前缘、中欧地区总面积为 31200 平方公里的区域。每个被检查的环境领域都解释了 NR，但它们的解释能力差异超过两倍。气候变量解释了最高的偏差分数，其次是人为因素、土壤类型、地形和景观结构。纳入所有研究环境领域关键变量的全球模型具有最佳的解释能力。然而，解释的偏差远小于单个领域模型解释的偏差总和。全球模型表明，代表特定领域的变量可以解释的偏差是重叠的。代表景观结构的变量没有被纳入全球模型，因为它们被认为是多余的。最后，气候变量解释的偏差比例小于人为因素。偏依赖图允许在消除所有其他变量的平均效应后，评估 NR 与特定解释变量之间的依赖关系。这些关系通常是曲线的，揭示了一些环境变量的值，超过这些值后，NR 会发生显著变化。
