Distance‐based multivariate analyses confound location and dispersion effects

摘要: Summary 1. A critical property of count data is its mean–variance relationship, yet this rarely considered in multivariate analysis ecology. 2. This study considers what being implicitly assumed about the relationship distance-based analyses – based on a matrix pairwise distances and effect any misspecification relationship. 3. It shown that make implicit assumptions are typically out-of-step with observed real data, which has major consequences. 4. Potential consequences are: confounding location dispersion effects ordinations; misleading results when trying to identify taxa an expressed; failure detect unless it expressed high-variance taxa. 5. Data transformation does not solve problem. 6. A solution use generalised linear models their recent generalisations, here have desirable properties.