Effect of reducing taxonomic resolution on ordinations to detect pollution-induced gradients in macrobenthic infaunal assemblages

Researchers often encounter taxonomic problems when analysing data on changes in macrobenthic assemblages due to pollutlon. In response, some authors have suggested that data at famlly and even phylum level may be sufficient to detect such changes. This hypothesis is commonly tested by visual comparisons of ordinabon plots generated by multidimensional scaling (MDS). Our study assessed this approach using a dataset previously shown to contain patterns in species composition that were related to both habitat type and heavy metal pollution. These data allow us to assess the ability of analyses at reduced taxonomic resolutions to detect 'signals' associated with pollution gradients among the 'noise' associated with habitat gradients. Such situations will arise in most studies of the impacts of pollutlon on inshore macrobenthic assemblages. The patterns associated with both habitat and pollution gradients were visible in ordlnations based on data pooled to hlgher taxonomic levels. The similarity among ordinations based on different sets of pooled data was not fully supported by more detailed analyses. Changing the number of dimensions in ordinations, the taxonomic resolution, the attribute:object ratio and the non-zeros ratio of datasets all affected the matrlx of association measures and the results of ordinations and Procrustes rotations. The effects of these changes were complex. Further work is needed to determine the likely effect of using data with reduced taxonomic resolution affect to detect impacts.