Beyond species-specific assessments: an analysis and validation of environmental distance metrics for non-indigenous species risk assessment
Beyond species-specific assessments: an analysis and validation of environmental distance metrics for non-indigenous species risk assessment
复制标题
DOI：
10.1007/s10530-015-0970-8
复制
发表时间：
2015-12-01
影响因子：
2.9
通讯作者：
Leung, Brian
中科院分区：
文献类型：
--
作者：
Bradie, Johanna;Pietrobon, Adam;Leung, Brian
环境距离度量使用多个环境变量来量化位置之间的环境相似性。它们通常应用于水生非本土物种风险评估，以评估不同地点对之间物种转移的相对风险。尽管这些指标在全球政府风险评估中得到应用，但尚未得到经验验证。我们使用从全球入侵物种信息网络数据库中获得的419种物种的经验数据来验证环境距离指标。我们探讨了环境距离的能力，区分存在缺席的水生和陆生环境。我们研究变量选择（包括变量的数量和类型）和不同的度量（欧氏距离，马氏距离，和加权版本的每个）对度量性能的影响。总体而言，使用未加权欧氏距离计算的环境距离表现最佳。当应用适当的变量时，它能够区分高达93%的物种的存在和不存在距离。变量的选择显着影响指标的性能，包括较少的，相关的变量优于应用程序，其中包括许多变量。我们的研究结果支持在水生和陆生环境中使用环境距离度量。
Environmental distance metrics quantify environmental similarity between locations using a number of environmental variables. They are commonly applied in aquatic non-indigenous species risk assessments to assess the relative risk of species transfer between different location pairs. Despite their application in governmental risk assessments globally, these metrics have not yet been empirically validated. We use empirical data for 419 species obtained from the Global Invasive Species Information Network database to perform a validation of environmental distance metrics. We explore the ability of environmental distance to discriminate presences from absences in both aquatic and terrestrial environments. We examine the effect of variable choice (both the number and types of variables included) and different metrics (Euclidean distance, Mahalanobis distance, and weighted versions of each) on metric performance. Environmental distance calculated using unweighted Euclidean distance performed best overall. When applied with appropriate variables, it was able to discriminate between presence and absence distances for up to 93 % of species. Variable choice significantly influenced metric performance, and including fewer, relevant variables outperformed applications where many variables were included. Our results support the use of environmental distance metrics in both aquatic and terrestrial environments.