Ecological impacts of tropical forest fragmentation on nonflying mammals and their habitats

There is a dire need to determine the impacts of habitat fragmentation on tropical forest biotas. I used spotlighting and live-trapping techniques to study the distributional ecology of nonflying mammals in fragmented and unfragmented rainforests on the Atherton Tableland in tropical NE Queensland, Australia, in 1984 and 1986/87. A secondary objective was determining the impacts of edge effects and other ecological changes in rainforest fragments.; Data from 160 night-time spotlighting revealed a strong gradient in arboreal mammal extinction-proneness. Most vulnerable were lemuroid ringtail possums (Hemibelideus lemuroides), which declined by {dollar}>{dollar}97% in fragments ranging from 1.4-590 ha, and Lumholtz's tree-kangaroos (Dendrolagus lumholtzii) and Herbert River ringtail possums (Pseudocheirus herbertensis), which declined sharply in small ({dollar}<{dollar}20 ha) fragments. The initial rarity of each species in unfragmented rainforest was a highly misleading indicator of extinction-proneness; the most useful indicator was their relative abundance in secondary vegetation. Arboreal mammal assemblages in fragments comprised a strongly nested subset of higher-diversity faunas in unfragmented rainforest, indicating that small ({dollar}<{dollar}3,000 ha) reserves are not viable strategy for conserving the most vulnerable members of this guild.; Extensive live-trapping surveys ({dollar}>{dollar}22,000 trap-nights) in unfragmented rainforest, forest fragments, and modified habitats demonstrated that the spotted-tailed quoll (Dasyurus maculatus), Atherton antechinus (Antechinus godmani), brown antechinus (A. stuartii), and musky rat-kangaroo (Hypsiprimnodon moschatus) were critically threatened by fragmentation. Fragments had significantly elevated populations of rainforest rodents.; A quantitative analysis demonstrated that interfragment vagility was the most critical determinant of mammal extinction-proneness. Species that utilized and dispersed across pastures and other modified habitats remained stable in fragments, while those that failed to disperse consistently declined or disappeared. Forest fragments were susceptible to a diverse array of edge effects. Mathematical models suggested that these phenomena could alter the ecology of even large (2,000-4,000 ha) fragments. Species that responded positively to forest edges and tree-fall gaps were often successful in fragmented rainforest.