FRAGMENTATION, DISTURBANCE, AND PLANT-DISTRIBUTION - MISTLETOES IN WOODLAND REMNANTS IN THE WESTERN-AUSTRALIAN WHEAT-BELT

Spatial heterogeneity and patchy distributions of species in intact landscapes are likely to lead to complex and unpredictable distribution patterns in remnants following fragmentation. We examined this proposition in relation to the mistletoe Amyema miquelii, which exhibits a clumped distribution in Eucalyptus salmonophloia woodlands in the Western Australian wheatbelt. We sampled mistletoe distribution and abundance in 14 woodland fragments ranging from 2.4 to 60.5 ha and in 14 sections of roadside corridors. These sites represent all known fragments and corridors containing E. salmonophloia in a 1680-km2 study area. We found that large fragments were more likely to have mistletoes than small fragments, but that small fragments either contained many or few to no mistletoes, reflecting the way fragmentation ''samples'' the pre-existing distribution. Superimposed on this sampling effect is the influence of disturbance. Fragments subjected to stock grazing contained no mistletoes. This indicates that grazing modified the habitat either for the mistletoe itself, through changed water relations, or for the frugivorous birds which may disperse mistletoe fruit, through removal of the shrubby understory. Only one A. miquelii plant was found on 26.3 km of roadside corridor, despite tree densities in corridors being similar to those in fragments. Roadside areas are generally considered good habitat for mistletoes, and their absence suggested that fruit-dispersing birds either did not use the corridors or did not stay in them long enough to deposit mistletoe seeds. These results indicate that, in order to predict biotic responses to fragmentation, information on distribution patterns and scales of patchiness in the prefragmentation landscape is required and the effects of fragmentation per se are likely to be confounded by other factors such as disturbance. Furthermore, quantifying fragmentation effects is difficult because of the small sample sizes typical of highly fragmented landscapes.