Proposing a resolution to debates on diversity partitioning
Proposing a resolution to debates on diversity partitioning
复制标题
DOI：
10.1890/11-1817.1
复制
发表时间：
2012-09-01
期刊：
影响因子：
4.8
通讯作者：
Hsieh, T. C.
中科院分区：
文献类型：
--
作者：
Chao, Anne;Chiu, Chun-Huo;Hsieh, T. C.
关于将区域多样性（gamma）分解为群落内成分（alpha）和群落间成分（beta）一直存在激烈的争论。虽然最近的生态学论坛在使用“数量当量”（希尔数）作为适当的多样性度量方法方面达成了共识，但三个相关的主要问题仍未解决。(1) α多样性和β多样性的“独立性”或“统计独立性”的确切含义是什么？(2)对于给定的应用程序，应该使用哪种分区（加法还是乘法）？(3)由于文献中有两个公式，alpha多样性的合适公式是什么？本文对这些问题提出了一个可能的解决方案。对于第一个问题，我们从两个角度厘清了“独立性”和“统计独立性”的定义，以澄清这个问题的困惑。我们还讨论了依赖的原因，以便在两种划分方案中任意两个多样性组分之间的依赖关系可以通过理论严格地证明，也可以通过模拟直观地理解。对于第二个问题，基于希尔数的乘法和加性beta多样性都是有用的测量方法，可以量化群落的不同方面。然而，这两种方法都不能直接用于比较具有不同群落数量的多个区域的相对成分相似性或分化，因为乘法β多样性取决于群落数量，而加性β多样性还取决于α（相当于γ）。这些依赖应该被移除。我们提出了消除这些依赖关系的转换，并表明转换后的乘法β和加性β都导致相同类别的度量，这些度量总是在[0,1]的范围内，因此可以用于比较多个地区社区之间的相对相似性或差异。这些相似性度量包括Sorenson、Jaccard、Horn和Morisita-Horn度量的多群落概括。对于第三个问题，我们提出了一些观察结果，包括关于哪个α公式产生独立的α和β成分的发现。这些可能有助于解决选择合适的alpha多样性公式的问题。并对相关问题作了简要讨论。
There have been intense debates about the decomposition of regional diversity (gamma) into its within-community component (alpha) and between-community component (beta). Although a recent Ecology Forum achieved consensus in the use of "numbers equivalents" (Hill numbers) as the proper choice of diversity measure, three related major issues were still left unresolved. (1) What is the precise meaning of the "independence" or "statistical independence" of alpha diversity and beta diversity? (2) Which partitioning (additive vs. multiplicative) should be used for a given application? (3) What is the proper formula for alpha diversity, as there are two formulas in the literature? This paper proposes a possible resolution to each of these issues. For the first issue, we clarify the definitions of "independence" and "statistical independence" from two perspectives so that confusion about this issue can be cleared up. We also discuss the causes of dependence, so that the dependence relationship between any two diversity components in both partitioning schemes can be rigorously justified by theory and also intuitively understood by simulation. For the second issue, both multiplicative and additive beta diversities based on Hill numbers are useful measures and quantify different aspects of communities. However, neither can be directly applied to compare relative compositional similarity or differentiation across multiple regions with different numbers of communities because multiplicative beta diversity depends on the number of communities, and additive beta diversity additionally depends on alpha (equivalently, on gamma). Such dependences should be removed. We propose transformations to remove these dependences, and we show that the transformed multiplicative beta and additive beta both lead to the same classes of measures, which are always in a range of [ 0, 1] and thus can be used to compare relative similarity or differentiation among communities across multiple regions. These similarity measures include multiple-community generalizations of the Sorenson, Jaccard, Horn, and Morisita-Horn measures. For the third issue, we present some observations including a finding about which alpha formula produces independent alpha and beta components. These may help to resolve the choice of a proper formula for alpha diversity. Some related issues are also briefly discussed.