The canopy beetle faunas of Gondwanan element trees in Chilean temperate rain forests

Aim To describe the coleopteran fauna occurring in canopies of temperate Gondwanan tree species in terms of their diversity and guild and taxonomic structures, and to test the proposition that this structuring reflects the Gondwanan origins of this fauna. Location The Andes and the coastal cordillera of temperate Chile. Methods Canopy fogging was used to sample beetles from 29 trees. The samples were statistically described using S-chao and the Simpson diversity index D. Cluster analyses and multi-dimensional scaling (MDS) were performed. The taxonomic and guild structures of the Chilean coleopteran fauna were compared quantitatively with those found in other parts of the world using homogeneity chi-square and t-tests. Results A collection of 25,497 beetle specimens was obtained primarily from Nothofagus dombeyi, Nothofagus obliqua and Araucaria araucana. The specimens collected were distributed between 485 morphospecies and included 107 putative, new generic-level taxa and 223 apparently undescribed species. Estimates of the size of the canopy beetle fauna showed that 600+ species were likely to be present. The communities found on a tree species differed markedly between years. MDS plots showed less community divergence between tree species for predators than for phytophages and xylophages. The guild structure was similar to that found on Australian 'Gondwanan' trees but differed significantly from the community structures found on 'Laurasian' tropical and temperate trees in supporting fewer phytophages and saprophages, but more xylophages. The predator guild showed a different pattern, with tropical faunas differing from those of more temperate regions, irrespective of hemisphere, as did the distribution of superfamilies. Main conclusions The beetle fauna found in the canopies of N. dombeyi, N. obliqua and A. araucana was large (600+ species), with about half of the species undescribed. S-chao was found to vary with sample size and to give lower estimates of S than species attenuation curves, raising the possibility that the two methods are estimating the sizes of different statistical communities. It is possible that the attenuation curve is estimating the number of species to be found on a particular tree species, while S-chao is estimating the 'carrying capacity' for beetle species of individual trees, and this varies from tree to tree. Care also needs to be taken in experimental design when monitoring temporal changes in forest insect communities given the difference in communities found between years in this study. The proportions of phytophages, saprophages and xylophages resemble those of a 'Gondwanan' rain forest from Australia and differ significantly from those of tropical and temperate 'Laurasian' forests.