THE EFFECTS OF HABITAT FRAGMENTATION AND LIVESTOCK-GRAZING ON ANIMAL COMMUNITIES IN REMNANTS OF GIMLET EUCALYPTUS-SALUBRIS WOODLAND IN THE WESTERN-AUSTRALIAN WHEAT-BELT .1. ARTHROPODS

1. The effects of habitat fragmentation (remnant biogeographic charact eristics) and livestock activity (grazing, trampling) on arthropod com munities were examined within 26 remnants of gimlet Eucalyptus salubri s woodland in the central wheatbelt of Western Australia. 2. Significa nt correlations between remnant biogeographic variables were between r emnant area and connectivity (+ ve, P < 0.001), and between connectivi ty and distance from the study remnant to the nearest patch of native vegetation (- ve, P < 0.001). 3. Remnant disturbance indices (sheep fa ecal pellet density, percentage cover of weeds) were significantly cor related with remnant biogeographic characteristics (P < 0.01). Small a nd poorly connected remnants showed significantly higher intensities o f disturbance than larger and better connected remnants. 4. When distu rbance indices were used to categorize study remnants into groups with high, moderate or low disturbance, remnants with high intensities of disturbance had significantly less lichen, litter and shrub cover. Hig hly disturbed remnants were associated with (i) fewer scorpion species , lower termite and ant diversity (Shannon-Wiener), (ii) a lower abund ance of scorpions, termites and mygalomorph spiders, (iii) more beetle species and higher beetle diversity, and (iv) greater abundance of ea rwigs and beetles. Cockroach, earwig and ant species richness showed n o significant response to disturbance. Species richness of termites, a nd the abundance of lycosid and idiopid (mygalomorph) spiders, isopods , cockroaches and ants, was highest under moderate disturbance. 5. Sig nificant biogeographic covariates were area (abundance of araneaomorph spiders, 'associated subordinate' and 'opportunistic' ants; richness of carabid, scarabid and 'other' beetle species, subordinate and oppor tunistic ants), connectivity (richness of termites, scarabid beetles) and distance to the nearest native vegetation (richness of 'dominant' ants). 6. When disturbance and biogeographic effects were combined, to tal termite richness, and the richness of termite functional groups, d eclined markedly in highly disturbed, small and poorly connected remna nts. 7. Termite communities in relatively undisturbed remnants were mo re similar in species composition to communities in moderately disturb ed quadrats than to communities in highly disturbed quadrats. Communit y similarity values for ants and beetles were similar across the study quadrats with different degrees of disturbance. 8. Arthropod communit ies were also examined by canonical variate analyses across remnants w ith different degrees of disturbance, using total abundance and richne ss, and abundance and richness of predators (scorpions, spiders, carab id beetles, ants) and detritivores/herbivores (termites, isopods, earw igs, cockroaches, weevil and scarab beetles). Effective site separatio n into the three disturbance categories was found for abundance and ri chness of all arthropods, and for predators alone. Abundance and richn ess of detritivores/herbivores separated into two groups of sites: hig h disturbance sites, and sites with low or moderate disturbance with n o separation. 9. In stepwise regression analyses, lichen cover, weed c over and sheep faecal pellet density were the most significant indicat ors of faunal abundance, richness and diversity. Remnant biogeographic variables explained a low percentage of variation in faunal character istics. 10. Habitat disturbance was the major influence on the arthrop od communities, with remnant biogeographic factors consistently explai ning low variations in the abundance or diversity of the fauna. 11. Th e implications of our findings for the management of remnant vegetatio n are discussed.