Arthropods were sampled using pitfall traps, sticky traps, sweep netting, Malaise traps and visual sampling at a national botanic garden, KwaZulu-Natal, South Africa, where the taxonomic impediment is great. The aims were to compare two sites, one of native vegetation and the other of mainly exotic plants, to determine the possible localized extent of biodiversity change across the land mosaic, and to test and compare methodologies and indicator taxa and to make recommendations for ecological landscaping of a botanic garden. Species richness and evenness varied considerably with sampling technique used. From results of a single replicate of data from all trapping methods including 821 arthropod species and 3831 individuals, a number of conclusions could be drawn. Trapping procedures such as sweep netting and pitfall traps, which focus on species with restricted mobility and/or host plant requirements, indicated greatest differences in diversity between two closely located sites. Taxa varied in sensitivity to microlandscape, again depending on the extent of their mobility. Cicindelid and carabid beetles were particularly good indicators of habitat disturbance and type. The management recommendations are that in a species-rich urban botanic garden such as this, as many ecotopes as possible should be preserved or created. These should vary in topography, landscape characteristics and vegetation composition, with as much connectivity as possible. This is a feasible blanket approach to give home to a large number of nameless species and morphs. Patches of different ecotopes should not be separated by more than a few metres by expanses of mown lawn which isolates much of the fauna.
