Native arthropods on exotic sand dune flowers: consideration of sample size and number for investigating rare species and sparse communities

When studying arthropod visitors of flowers, the sampling unit (individual flowers, groups of flowers, areas of plants, timed visits, etc.) depends upon the aim of the study and the sampling method employed. In this study, arthropods using flowers of the ice plant, Carpobrotus edulis, were recorded on the sand dunes at New Brighton in the south island of New Zealand. Of 3600 flowers, only 10% contained invertebrates and only 478 specimens were recorded in total. Of 32 arthropod species observed on this exotic plant, we consider at least 20 to be native to New Zealand and five species are probably New Zealand endemics. Based on an occupation rate of individual flowers of 10%, a binomial model indicated that a sample of 100 flowers would have <0.003% chance of containing no specimens, and 96% chance that 5–16 flowers would contain some animals. Species accumulation models (e.g. bootstrap, Chao and rarefaction) and models examining the likelihood of recording rare species indicated that after examining 2000 flowers, 80% of arthropod species would be recorded, and that only the rarest species in our study would fall below an 80% statistical power of detection. The results suggest that for this flower–invertebrate system, a scheme that involved 20 independent samples, each consisting of 100 flowers, would provide a good chance of (1) avoiding totally empty samples (2) collecting a high proportion of the total species present and (3) recording all but the very rarest species that occur in this system.