---

作 者 ：金森

期 刊 ：植物生态学报 2007年 31卷 6期 页码：1154-1160

关键词：Huffman编码；Shannon-Wiener指数；Rényi熵；结构复杂性；有序结构复杂性；无序结构复杂性；

Keywords:complexity, structural complexity, Huffman code, Shannon-Wiener index, Rényi entropy, organized complexity, disordered complexity,

摘 要 ：Anand和Orlóci（1996）提出用Huffman编码的平均码长来衡量群落结构总复杂性， 用12阶Rényi熵测度群落无序结构复杂性，用两者之差测度
群落有序结构复杂性。这是一种与基于生物多样性的复杂性测度完全不同，至少在思路上不同的结构复杂性测度，但对于这种测度的性质研究
还不多。该文模拟建立了具有代表性的各种结构的群落400多万个，计算了这些群落的复杂性测度，对其性质进行了研究。结果表明：1）群落
结构总复杂性、无序结构复杂性和有序结构复杂性受群落多样性、变异程度、优势种组成等影响。其中，群落结构总复杂性与ShannonWiener
指数高度相关（决定系数>0.99），完全可由群落组成的多样性决定。群落无序结构复杂性与群落优势种对数比例或变异系数关系最大，与群落
优势种对数比例之间的决定系数>0.99。2）该群落有序结构复杂性可表示为群落生物多样性和优势种对数比例的线性组合。在生态意义上可看
作主要由群落中非优势种的组成比例决定，而组成多样性能解释41%～46%的群落有序结构复杂性。3）群落组成物种总数的增加会导致群落复杂
性的测度有所增加，但没有像文献（Anand & Orlóci，1996）中描述的那么大，且组成总数对结构复杂性与多样性和优势种的关系等没有影响
。

Abstract:Aims Anand & Orlóci (1996) proposed a method measuring the total, disordered and organized complexity of community structure
using the mean length o f Huffman code, the 12th order of Rényi entropy, and the difference between the se two measures,
respectively. This method is quite different from measures based on community biodiversity and is potentially of great
significance. Although some properties of the measures were discussed by Anand & Orlóci (1996), a thorough study has not been
conducted, especially on their ecological meaning. The purpose of the paper is to reveal: 1) factors affecting the measures
and their ecological meaning, 2) relationship between complexity and biodiversity, and 3) effect of species number on
structural complexity.
Methods First, >4 000 000 simulated communities with different structures were constructed by two methods: evenly
distributed mode and dominated mode. Then the three complexity measures were computed. Correlation analyses were conducted
between these measures and potentially related factors such as Shannon-Wiener index, coefficient of variation and logarithmic
proportion of most dominated species. Multiple linear regressions between the complexity measures and combination of these
factors were carried out to explore the relationship between complexity and biodiversity. Communities with different species
number were then constructed and tested to explore the effect of species number on complexity measures.
Important findings The three complexity measures are strongly correlated with the Shannon-Wiener index, coefficient of
variation and logarithmic proportion of the most dominated species. For total complexity, the correlation coefficient between
it and the Shannon-Wiener index is nearly 1 (0.996), so is that between the 12th order of Rényi entropy and logarithmic
proportion of the most dominated species. The determining coefficient of multiple linear regression between organized
complexity and Shannon-Wiener index and logarithmic proportion of the most dominated species is also near 1 (0.97). These
findings suggest that all three measures of community structural complexity are affected by the Shannon-Wiener index,
coefficient of variation and logarithmic proportion of the most dominated species. For organized structural complexity, it
can be expressed as a linear combination of Shannon-Wiener index and logarithmic proportion of the most dominated species.
Biodiversity can explain 41～46% of variation of organized structural complexity. Although the three complexity measures
increase with number of species, they do not increase as much as described by Anand & Orlóci (1996), and it does not affect
the relationship between the complexity measures and related factors.

---