Département de Sciences Biologiques, Université de Montréal, C.P. 6128, succursale Centre-ville, Montréal, Québec H3C 3J7, Canada.
Ecology. 2009 Dec;90(12):3566-74. doi: 10.1890/08-1823.1.
Ecologists often face the task of studying the association between single species and one or several groups of sites representing habitat types, community types, or other categories. Besides characterizing the ecological preference of the species, the strength of the association usually presents a lot of interest for conservation biology, landscape mapping and management, and natural reserve design, among other applications. The indices most frequently employed to assess these relationships are the phi coefficient of association and the indicator value index (IndVal). We compare these two approaches by putting them into a broader framework of related measures, which includes several new indices. We present permutation tests to assess the statistical significance of species-site group associations and bootstrap methods for obtaining confidence intervals. Correlation measures, such as the phi coefficient, are more context-dependent than indicator values but allow focusing on the preference of the species. In contrast, the two components of an indicator value index directly assess the value of the species as a bioindicator because they can be interpreted as its positive predictive value and sensitivity. Ecologists should select the most appropriate index of association strength according to their objective and then compute confidence intervals to determine the precision of the estimate.
生态学家经常面临研究单一物种与代表栖息地类型、群落类型或其他类别的一个或多个组的站点之间的关联的任务。除了描述物种的生态偏好外，关联的强度通常在保护生物学、景观制图和管理以及自然保护区设计等应用中引起了很大的兴趣。评估这些关系最常用的指标是关联的φ系数和指示值指数（IndVal）。我们将这两种方法放入一个更广泛的相关度量框架中进行比较，其中包括几个新的指标。我们提出了置换检验来评估物种-站点组关联的统计显著性，并使用自举方法获得置信区间。相关性度量，如φ系数，比指示值更依赖于上下文，但允许专注于物种的偏好。相比之下，指示值指数的两个组成部分直接评估物种作为生物指标的价值，因为它们可以解释为其阳性预测值和敏感性。生态学家应根据自己的目标选择最合适的关联强度指标，然后计算置信区间以确定估计的精度。
