The establishment of commercial tree plantations is a common cause of habitat fragmentation globally, yet the influence of this land use on plant species density in fragmented native forests requires further understanding. Theory predicts species density will be highest in large areas of habitat, and should decline as area is reduced, but whether these effects are scale-dependent is largely unknown. We assessed plant species density (total, common and uncommon species) in experimentally fragmented eucalypt forest remnants (0.25, 0.88, 3.06 ha) surrounded by a pine plantation, at three spatial scales using nested quadrats. Specifically, we consider how plant species density varies across three different sized fragments, and whether the response of species density is magnified in common or uncommon species. Species density in small fragments was higher than continuous forest for all species groupings at the smallest spatial scale (1 m2), and for total and common species at the next smallest spatial scale (16 m2). No species groupings responded to reductions in habitat area at the largest spatial scale (144 m2). We suggest that pine plantations may cause higher species density in small fragments via two mechanisms, either by allowing species to infill unoccupied areas within small fragments, or by buffering small fragments from species losses during a severe and prolonged drought. In both cases we suspect reduced moisture stress (e.g. increased soil moisture, higher shading and reduced temperature) in small fragments has led to the observed changes in species density.