Canonical Correspondence Analysis (CCA)

Canonical correspondence analysis (CCA) is a constrained ordination that regresses chi-square standardized species abundances on environmental variables with site weights, so that site scores are linear combinations of those variables.

정준대응분석(CCA)은 카이제곱으로 표준화한 종 출현량을 사이트 가중치와 함께 환경변수에 회귀하여, 사이트 점수가 환경변수의 선형결합이 되도록 하는 제약 서열화 기법입니다.

How much of the variation in species composition across my sites is explained by the measured environmental variables?

조사 지점들의 종 구성 변이 가운데 측정한 환경변수가 얼마나 설명하나요?

Canonical correspondence analysis relates a sites-by-species abundance table to a set of environmental variables, extracting ordination axes that are linear combinations of the environmental variables and that best separate species with unimodal responses. The constrained inertia expresses how much of the total chi-square inertia the variables explain, and a permutation test of sites assesses whether that amount is larger than chance.

When to use Canonical Correspondence Analysis (CCA)

• You have non-negative species abundance or frequency data and site-level environmental measurements
• Species are expected to respond unimodally along long environmental gradients
• You want a triplot of sites, species and environmental arrows with a permutation test

When to use something else

• Gradients are short and responses roughly linear: use Redundancy Analysis (RDA) on Hellinger-transformed data
• You only want to test group differences in composition: use PERMANOVA

Assumptions and requirements

• Non-negative abundance data with no empty sites or species (these are removed)
• Environmental variables are not perfectly collinear and fewer than the number of sites
• Sites are exchangeable under the null hypothesis for the permutation test

What the analysis reports

• Total, constrained and unconstrained inertia with constrained eigenvalues
• Species scores (scaling 2), site LC and WA scores, and biplot scores of environmental variables
• Permutation pseudo-F tests for all constrained axes and for the first axis
• Bar chart of constrained eigenvalues and site-score scatter

How to report it

A canonical correspondence analysis of [k] species constrained by [q] environmental variables (N = [n]) explained [value]% of the total inertia; the permutation test of all constrained axes gave pseudo-F = [value], p = [value] ([value] permutations).

Related methods

• Redundancy Analysis (RDA)
• Correspondence Analysis
• PERMANOVA

Run Canonical Correspondence Analysis (CCA) on Quorant