Nonmetric multivariate analysis in community‐level ecotoxicology

摘要: Community-level data, typically in the form of abundances over 100 species, are widely collected context environmental monitoring, e.g., effects disposal drilling muds offshore oil operations. The statistical properties resulting abundance arrays preclude use “classical” multivariate analyses, such as principal components and analysis variance. One alternative is nonparametric displays tests, nonmetric multidimensional scaling (MDS) variations Mantel tests on similarity matrices. These do not require restrictive assumptions parametric techniques possess a conceptual simplicity, facilitating their understanding by managers regulators. A monitoring example discussed from Norwegian fields, for which analyses have had significant impact practice. equally applicable to assessing outcomes community-level laboratory experiments bioassays. paper exemplifies approach taken at Plymouth Marine Laboratory (and encapsulated PRIMER software) through (1) observational studies pollution gradients North Sea sediments heavy metal pollutants Fal estuary, UK; (2) an experimental study differential metals marine nematode communities; (3) bioassay employing microcosm experiment estuary sediments.