Data to accompany the paper ‘Redundancy analysis reveals complex den use patterns by eastern spotted skunks, a conditional specialist’, Ecosphere

Wildlife
managers tasked with understanding habitat and resource selection at the
population level attempt to characterize patterns in nature that aid and inform
conservation. Resource selection functions (RSF), such as discrete choice
analyses, are the standard convention to characterize the effects of habitat
attributes on resource selection patterns. RSF are invaluable tools for in
wildlife management and conservation and have proven successful in numerous
studies. However, the analysis of small datasets using RSF becomes problematic
when attempting to account for for complex sources of variation and inclusion
of factors such as weather or intrinsic variation on target species’ response
may produce models with poor predictive ability. We compared the application of
generalized linear mixed modeling (GLMM) and redundancy analysis (RDA) on
Appalachian spotted skunk (Spilogale putorius putorius) den selection data at
four study sites within national forest and surrounding private lands in the
Appalachian Mountains of western Virginia and northeastern West Virginia. We
assessed the need for the inclusion of alternative sources of variation, (i.e.,
weather conditions and individual intrinsic variation) in addition to standard
habitat attributes to better identify sources of variation in den selection.
The RDA elucidated complex and opposing relationships whereby den type use was
based on reproductive status or weather condition, that were not evident in the
GLMM model that relied solely on habitat measures. Our results demonstrate the
importance of examining resource selection data using multivariate techniques
in addition to conventional discrete choice analyses to better understand
intricate habitat–species relationships, especially for small data sets.
Further, from our analyses, we propose that spotted skunks are neither a true
generalist nor specialist species. We introduce and define the term
‘conditional specialist’ to represent a species that is conditionally selective
of a given resource in response to one or more current environmental or
intrinsic conditions.

野生动物管理者在理解种群层面的栖息地及资源选择时，致力于描述有助于和指导保护的天然模式。资源选择函数（RSF），例如离散选择分析，已成为表征栖息地属性对资源选择模式影响的规范方法。RSF 在野生动物管理和保护中极为宝贵，已在众多研究中证明其有效性。然而，当试图解释复杂变异来源并包含如天气或目标物种反应的内生变异等因素时，使用 RSF 对小数据集进行分析便会变得复杂，可能导致预测能力较差的模型。我们比较了广义线性混合模型（GLMM）和冗余分析（RDA）在阿巴拉契亚斑点臭鼬（Spilogale putorius putorius）洞穴选择数据中的应用，这些数据来自位于西弗吉尼亚州和东北部西弗吉尼亚州阿巴拉契亚山脉国家森林及其周边私人土地上的四个研究地点。我们评估了在标准栖息地属性之外，纳入如天气条件和个人内在变异等替代变异来源的必要性，以更好地识别洞穴选择中的变异来源。RDA 阐明了复杂且对立的关系，其中洞穴类型的利用基于繁殖状态或天气条件，这些关系在仅依赖于栖息地测量的 GLMM 模型中并不明显。我们的结果表明，使用多元技术来分析资源选择数据，除了传统的离散选择分析外，对于更好地理解栖息地与物种之间的复杂关系至关重要，尤其是在小数据集的情况下。此外，从我们的分析中，我们提出斑点臭鼬既不是真正的泛化物种，也不是专门化物种。我们引入并定义了“条件专门化”这一术语，用以表示一种物种在响应一种或多种当前环境或内在条件时，对特定资源具有条件选择性的特性。