Identifying performance indicators of the effects of forest management on ground-active arthropod biodiversity using hierarchical partitioning and partial canonical correspondence analysis

Biodiversity conservation in Australian native hardwood production forest i s a central aim of sustainable forest management. However, the effects of t imber extraction and fuel-reduction burning on the biodiversity of forest i nvertebrates are poorly understood. In this study, we use hierarchical part itioning and partial canonical correspondence analysis to examine the indep endent effects of these management practices on the richness and compositio n of ground-active ant and beetle assemblages. The study makes use of surve y data from two types of production forest widespread in north-east New Sou th Wales, New England Blackbutt (Eucalyptus campanulata, R. Baker and H.G. Smith) and Messmate/Forest Ribbon Gum (E. obliqua, L'Her and E. nobilis, L. Johnson and K. Hill). Pitfall-trapped ants, and beetles from the families Carabidae, Scarabaeidae and Pselaphidae, were sorted to morphospecies and t hese survey data, together with habitat data, were used to explore two ques tions: (1) how much variation in the richness and composition of the arthro pod assemblages at survey sites could be accounted for (independently) by l ogging and fire history? and (2) were other habitat variables better able t o (independently) account for variation in arthropod richness and compositi on? Answers to these questions were used to identify potential performance indicators of the effects of forest management practices on ground-active a rthropod biodiversity. Our analyses found that in New England Blackbutt for est sites, where selective logging was at low intensity, logging history ex plained little variation in arthropod richness or composition. By contrast, a strong influence of fire history and correlated habitat variables was ev ident. Fire history also explained significant variation in Messmate/Forest Ribbon Gum forest, but in this forest type, where logging intensity was hi gher, the influence of logging history was also apparent. This study finds support for the use of measures of ant richness, percent cover of sub-canop y (5-15 m above ground level) and ground-layer characteristics (litter dept h, % cover of litter and bare ground) as performance indicators, because th ese attributes are relatively easy to measure, are affected by forest manag ement, and are indicative of the effects of forest management on ground-act ive arthropod biodiversity, (C) 2000 Elsevier Science B.V. All rights reser ved.