Using similarity measures in benthic impact assessments

Area of Science:

• Ecology
• Environmental Science
• Marine Biology

Background:

• Ecological impact assessments often rely on similarity measures and cluster analysis to evaluate environmental changes.
• Previous studies have used various coefficients and algorithms to interpret community data, but results can be inconsistent.

Purpose of the Study:

• To evaluate the objectivity of cluster analysis in ecological impact studies.
• To assess the impact of clam digging on intertidal infauna using different similarity measures.
• To re-analyze power plant effluent impact data with various similarity coefficients.

Main Methods:

• Employed eleven similarity measures to analyze infauna community data from a mud-flat impacted by clam digging.
• Utilized a divisive polythetic algorithm for clustering similarity matrices and generating dendrograms.
• Re-analyzed existing power plant effluent data using five different similarity coefficients.

Main Results:

• Clustering dendrograms produced conflicting conclusions regarding the impact of clam digging, with 3 coefficients indicating impact and 8 not.
• Re-analysis of effluent data showed that 3 out of 5 coefficients indicated a difference in benthic communities, while 2 did not.
• The choice of similarity coefficient and clustering algorithm significantly influenced the interpretation of ecological impacts.

Conclusions:

• The objectivity of cluster analysis in environmental impact studies is questionable due to inherent subjective choices in methods and interpretation.
• There is a need for objective criteria, based on intrinsic and ecological properties, to guide the selection of similarity measures and algorithms.
• Analysis of variance of similarity matrices is proposed as a more objective alternative to clustering for impact assessments.