Fragmentation-related increases in levels of predation at birds’ nests, and changes to nest predator assemblages, have often been reported in Northern Hemisphere temperate forests. In many cases these changes have been attributed to an influx of matrix-associated avian predators from neighbouring agricultural or urbanised land. Well-documented increases in the abundance of avian nest predators, which are ecologically similar to those causing edge effects elsewhere, occur in Australian agricultural and urban habitats, and near edges within remnant eucalypt forests. Therefore, increases in levels of nest predation, similar to those occurring elsewhere, may be expected to occur near edges and within small remnant patches of Australian eucalypt forests. The same processes would not be expected to occur at the edges of eucalypt forests which border plantations of exotic Pinus tree species, because edge-related nest predation has seldom been reported in Northern Hemisphere temperate forest mosaics. Furthermore, avian nest predators do not generally increase within Australian Pinus plantations, or near edges within remnant forests fragmented by Pinus plantations. The primary aim of this thesis was to assess the potential impacts of habitat fragmentation on levels of predation at birds’ nests, and predator assemblage composition, within subtropical Australian eucalypt forests. Four separate experiments were conducted in south east Queensland’s lowland eucalypt forests. A common methodology was employed throughout, consisting of deploying five or ten artificial open-cup nests along a transect, each baited with a quail egg, and with two imprint-receptive plasticine eggs which allowed assessment of the nest predator assemblages. These nests were exposed for six or seven days, to estimate nest predation levels and assess nest predator assemblages amongst a variety of contexts. In each experiment, treatments consisted of differing fragmentation contexts, in which there were replicate sites which were distributed across a wide area at the landscape scale. Throughout these experiments the unit of replication was a site containing a single transect, and two nest predation measures were used: 1. predation of the quail egg, providing a more conservative predation measure involving larger, stronger predators which would be less likely to be thwarted by parental defence at a real nest; and 2. predation of any egg, which provides a less conservative measure that is likely to include activity by even the smallest potential nest predators. In an initial experiment, effects of distance to edge and type of adjoining land use were tested within large (>400 ha) eucalypt forest remnants. Shrub-nests were deployed at distances of 0-30, 60-90, and 235-265 metres from edges adjacent to areas of urban, pasture, and Pinus plantation. There were eight replicate sites of each edge type, scattered widely across a 30 000 km2 study area. When predation of the quail egg was used to calculate predation levels, predation varied significantly with edge type but not distance to the edge, due to relatively low levels of predation within sites bordering Pinus plantations. When predation of any egg was used to calculate predation levels, predation was not significantly affected by edge type or distance to the edge. Predation levels within eight independent forest interior transects, distributed throughout the study region and located 500-800 m from the nearest edge, were similar to those within transects 0 m from edges. Birds were the most important class of predator within all combinations of site type and distance to edge, and accounted for 92% of predation events for which a predator could be identified. In most of these cases, damage to imprint-receptive eggs was consistent with large-bodied nest-depredating species bird species from the family Artamidae (butcherbirds, currawongs and Australian magpie). In a second experiment, potential effects of edge and type of adjoining land use on predation at artificial nests were again assessed within large (>400 ha) eucalypt forest remnants, using a different spatial layout of sites over a region of about 10 000 km2. Nine sites were selected in each of three contexts: forest/pasture edge, forest/urban edge, and forest interior. Edge sites were 0-30 m from edges, and interiors were located 500-1000 m from the nearest edge. Within each site, nests were placed at each of two heights (ground and shrub), and this experimental design was used yearly for three years at interior and urban edge sites, and two years at pasture edge sites. Averaged across the three years, 62% more ground-nests, and 54% more shrub-nests, were depredated near edges than within forest interiors. Furthermore, complete depredation of all nests within a site was much more common near edges than within forest interiors, particularly for shrub-nests. Predators were identified for 250 of 369 depredated nests, of which 80% had been preyed upon by birds, 12% by mammals, and 7% by reptiles. It is argued that the increase in predation at edges was primarily caused by large-bodied avian predators, particularly species within the family Artamidae. The general occurrence of edge-related nest predation in subtropical eucalypt forests was further supported by a third experiment, which investigated potential impacts of internal fragmentation by picnic areas on nest predation. Bird assemblages and nest predation levels at eight transects abutting picnic areas (‘picnic area edges’) were compared with two types of matched control transect within forest about 150 m away ( ‘forest interiors’), and, ‘context references’, which were similar to the picnic area edges in their proximity to access roads and forest external edges. Within the picnic areas there was an assemblage of relatively large-bodied, aggressive birds, including several avian nest predators (Torresian crow, grey and pied butcherbirds) as well as the Australian magpie and noisy miner, resembling that found in urban areas within the study region. Bird assemblages at picnic area edges, which were dominated by the noisy miner and Australian magpie, clearly differed from those of forest interiors, which were characterised by a number of smaller-bodied insectivorous species. Levels of predation at artificial nests were several-fold greater adjacent to the picnic areas than at forest interiors 150 m from the picnic area edges. Context reference transects were similar to forest interiors when associated with narrow roads, but similar to picnic area edges when associated with roads incorporating wide grassy verges. These results showed that picnic areas exert strong localised edge effects on forest bird assemblages, and appeared likely to cause reduced reproductive success for small-bodied forest bird species if they attempted to nest nearby. A fourth experiment investigated whether levels of nest predation and predator assemblages varied between interiors of small (10-20 ha) urban remnant patches (>50 m from external edges), and extensive tracts (>400 ha) of subtropical Australian eucalypt forest (>500 m from external edges). Artificial nests were deployed both on the ground and in shrubs. Most predation of shrub-nests in both remnants and extensive forests was by birds, although there were some differences in the assemblages involved. Ground-nests were far more likely to be depredated by mammals or lizards in the remnants than in the extensive forests. This may have been caused by mesopredator release, or because small remnants provided more favourable habitat than extensive forests for these predators. However, overall nest predation levels did not differ significantly between remnants and extensive forests, suggesting that any edge-related predation was of insufficient spatial extent to strongly affect the interiors of remnants in this size class. Finally, nest predation results from the second and fourth experiments were reassessed with respect to landscape composition. The relationship between nest predation levels and the proportional cover of pasture, crop, settlement, and forest, within a 5 km radius of each site was investigated. There was no significant association between nest predation levels and surrounding landscape composition at that scale, suggesting that the occurrence of more local scale impacts (such as edge effects) was not influenced by landscape-scale processes. Edge-related nest predation in North American and European forests has often been explained with reference to an influx of relatively large-bodied avian nest predators from the matrix, where they show increases in abundance and diversity compared with unfragmented forests. Species of predator which have often been implicated have included the blue jay and American crow in North America, and the magpie, hooded crow, and jay in Europe. They may proliferate within agricultural and urbanised landscapes because of increased foraging opportunities, and then may penetrate some distance into remnant forests. Fragmentation had been expected to similarly cause increases to nest predation levels within Australian eucalypt forests because bird assemblages within Australian pasture and urban habitats are often dominated by relatively aggressive, medium- to large-bodied native species, including nest predators such as the laughing kookaburra, and species from the families Corvidae and Artamidae. My results supported the general occurrence of such an edge effect of increased predation at birds’ nests within eucalypt forests. However, this effect was caused by sedentary avian species which show increased abundances at edges within eucalypt forests, rather than by an influx of avian predators from the matrix.