Abstract
1. Weekly samples of Malaise trapped Coleoptera from regenerating shrubland and an adjacent mature, podocarp‐broadleaved forest, in the North Island, New Zealand, were compared over one season. Three traps were monitored in each site.2. Diversity indices did not group samples by site whereas the divisive polythetic classification analysis TWINSPAN showed that the two sites possessed clearly different communities. In the classification, site was of primary importance, with time of year, and trap position within site of secondary and tertiary importance respectively.3. Samples were compared at different taxonomic levels and using different subsets of the database. Determination of the main families, morphotyped to species or species complex, was found to be sufficient to classify most samples to the appropriate community group.4. The divisive classification procedure applied to four consecutive weekly Coleoptera samples over early summer, is suggested as a means of describing and identifying terrestrial arthropod communities characteristic of site and year of collection. This approach provides a potentially sensitive tool for monitoring terrestrial ecosystems.