Invasive species are widely recognized as a major threat to global diversity and an important factor associated with global change. Species distribution models (SDMs) have been widely applied to determine the range that invasive species could potentially occupy, but most examples focus on predictive variables at a single spatial scale. In this study, we simultaneously considered a broad range of variables related to climate, topography, land cover, land use, and propagule pressure to predict what areas in the southeastern United States are more susceptible to invasion by 45 invasive terrestrial plant species. Using expert-verified occurrence points from EDDMapS, we modeled invasion susceptibility at 30-m resolution for each species using a maximum entropy (MaxEnt) modeling approach. We then analyzed how environmental predictors affected susceptibility to invasion at different spatial scales. Climatic and land-use variables, especially minimum temperature of coldest month and distance to developed areas, were good predictors of landscape susceptibility to invasion. For most of the species tested, human-disturbed systems such as developed areas and barren lands were more prone to be invaded than areas that experienced minimal human interference. As expected, we found that landscape heterogeneity and the presence of corridors for propagule dispersal significantly increased landscape susceptibility to invasion for most species. However, we also found a number of species for which the susceptibility to invasion increased in landscapes with large core areas and/or less-aggregated patches. These exceptions suggest that even though we found the expected general patterns for susceptibility to invasion among most species, the influence of landscape composition and configuration on invasion risk is species specific.

中文翻译：

---

评估预测的植物入侵热点的景观特征

入侵物种被广泛认为是对全球多样性的主要威胁，也是与全球变化相关的重要因素。物种分布模型 (SDM) 已被广泛应用于确定入侵物种可能占据的范围，但大多数示例都集中在单个空间尺度上的预测变量上。在这项研究中，我们同时考虑了与气候、地形、土地覆盖、土地利用和繁殖压力相关的广泛变量，以预测美国东南部的哪些地区更容易受到 45 种陆地入侵植物物种的入侵。使用来自 EDDMapS 的专家验证的发生点，我们使用最大熵 (MaxEnt) 建模方法对每个物种在 30 米分辨率下的入侵敏感性建模。然后，我们分析了环境预测因素如何影响不同空间尺度的入侵敏感性。气候和土地利用变量，特别是最冷月份的最低气温和与发达地区的距离，是景观易受入侵的良好预测指标。对于大多数测试的物种，人类干扰的系统，如发达地区和贫瘠的土地，比人类干扰最小的地区更容易受到入侵。正如预期的那样，我们发现景观异质性和繁殖体传播走廊的存在显着增加了大多数物种的景观对入侵的敏感性。然而，我们还发现，在具有大核心区域和/或较少聚集斑块的景观中，一些物种的入侵敏感性增加。