Arboreal arthropod biodiversity in woodlands: Effect of collection procedures and geographic distance on estimates of diversity found on two species of Melaleuca



The estimation of the level of biological diversity is critical to conservation planning and management. In this paper, factors affecting arboreal arthropod biodiversity estimates are examined, namely, reproducibility between trees, the proportion of the foliage of a tree that needs to be sampled, the effects of placement position of collectors under the tree and geographical distance between sampled trees of the species Melaleuca linariifolia and M. decora. It was found that arboreal arthropod specimens are not randomly distributed on trees and that, on average, 33% fewer species and 52% fewer specimens occur on the northern (sunny) sides of the trees than on the southern sides. The placement of collectors therefore can radically affect the estimates obtained. The species area curve continued to rise for each host until the entire tree was sampled. The shapes of the curves obtained from the two hosts however, were markedly different and it is concluded that it is not possible to simply transfer protocols from one host species to another, very similar, host species. When the same collecting effort was applied, the same number of species were collected, though M. decora supports twice the number of species as M. linariifolia. Variations of as much as 50% were found between trees of the same species in the number of species present and none of the estimators of biodiversity richness, evenness or information gave consistent results. As a consequence of these observations, the use of changes in standard estimators in decisions related to the management of biodiversity need to be considered carefully. Complementarity, however, was found to be a useful and robust measure of similarity of faunas for both host species. A new descriptor of the geographical distribution of organismal biodiversity, the biodiversity neighbourhood, is described and two methods of measuring it, using either spatial autocorrelation or complementarity, are demonstrated.