There is major concern internationally, and specifically in the Ethiopian afromontane biodiversity hotspot, about the impact of forest habitat fragmentation on biodiversity conservation. This study assessed the effect of land-use change on land-cover pattern, forest patch spatial structure and consequent effects on plant species richness and composition in two areas of southwest Ethiopia: Illubabor and the Gamo highlands of Gamo Gofa. Land-use change analysis was conducted on three sites, I (1500-2000 m altitude) and II and III (1878-2422 m altitude), each of 47,648 ha, in Illubabor and one area of 66,765 ha in the Gamo highlands. Two Landsat images from the years 1986 and 2000 and one Spot image from 2007 covering Illubabor and two Landsat images from 1995 and 2010 covering most of the Gamo highlands study area were analysed. The change in area, number, shape and edge density of patches of each land-cover class were quantified between the years. Field sampling of woody plants was conducted in plots within 30 forest patches in Illubabor stratified between the three sites, and ten in the Gamo highlands stratified between sacred groves and non-sacred forests. Information on local knowledge, cultural association, institutions, practices and threats of the sacred groves was obtained by interviewing 24 of their traditional custodians. The landscape pattern in both Illubabor Zone and Gamo highlands has changed rapidly over 20 years. In highland areas there has been a rapid conversion of forest to farm, settlement and grazing land. At lower altitude forest cover has been retained but it has been degraded by its use for understorey coffee cultivation. The area and number of forest patches has decreased while patch shape and edge density has increased. In Illubabor forests’ woody plant species richness was higher in the lower altitude site I (70.8±9.2) than in sites II (50.9±6.3) and III (54.3±4.9), with little difference in the shape of their species accumulation curves. Species composition also differed between site I and sites II and III; altitude and disturbance were more strongly associated with this difference than were patch size, shape and edge density. There was little difference in tree density or basal area amongst the sites. Within forest patches, tree basal area was higher in the patch interior (96.8±9.4 m2 ha-1) than in the edge zone (77.2±15.3 m2 ha-1), however total tree density did not differ significantly. The interior forest had twice the density of trees taller than 22 m and a higher density of small trees (5-14.9 cm DBH) than the edge. Tree species richness did not differ significantly between the two habitats, however including shrubs and vines total woody species richness was higher in the edge (69.3±5.9) than interior (52±3.5) forest. While the upper canopy of interior forest was dominated by species with a wide habitat distribution range, it also had a higher abundance of forest-habitat specialist species than the edge. Species with a distribution associated with forest-margins were, as expected, more abundant in the edge habitat. No association was found between tree density, basal area or height in both habitat types and any fragmentation variables (patch size, shape or edge density) or environmental variables (rainfall, altitude or cumulative disturbance). However, edge habitat basal area was negatively associated with disturbance. Shrub, vine and geophytic angiosperm herb species composition differed between forest edge and interior habitats. Species richness of vascular epiphytes was higher in interior (28.9±1.8) than edge (13.6±1.4) habitat, as was their individual density (114.5±6.5 and 42.7±3.7 respectively), which was associated with the density of large DBH trees. Epiphyte density was not associated with any fragmentation variables, rainfall or altitude, however in the edge habitat it was negatively associated with disturbance. Species composition varied between the two habitats, with forest-habitat distribution species, which tended to have herbaceous stems and leaves, being more abundant in the interior habitat. However, the same three epiphyte species were dominant in both habitats, and species with succulent stems and leaves or woody stems had similar abundance in both. Geophytic fern species richness was higher in interior (29.4±1.8) than edge (22.1±1.4) habitat, as was individual density (104±22.3 and 59.8±13.7 respectively). Species with forest-habitat distribution, those with creeping or erect rhizomes and those with tufted fronds were more abundant in the interior, while generalist distribution species and those with spaced fronds showed no difference. In the Gamo highlands, sampled woody plant species richness and diversity were higher in sacred groves than in non-sacred forests, however, the species accumulation curve showed no difference between the two categories of forest. Their species composition differed and the sacred groves had a higher proportion of species endemic to Ethiopia (12.5%) than the non-sacred forests (9.2%). Two national conservation priority species, Cordia africana and Hagenia abyssinica, were only recorded in the sacred groves, and one IUCN red list species, Prunus africana, was more abundant there. Mean basal area was significantly higher in the sacred groves (1.55±0.45 m2 ha-1) than the non-sacred forests (1.28±0.41) as was tree seedling density (1111.7±393.2 ha-1 and 476.8±87.3 ha-1 respectively) but not sapling density. In conclusion, forests in southwest Ethiopia are undergoing a high rate of fragmentation and degradation. The resulting loss of forest interior core habitat is associated with a loss of biodiversity, especially of vulnerable forest-specialist species of woody plant, vascular epiphyte and geophytic fern. Nonetheless, the remaining small forest patches do still have high biodiversity value and they should be made a high conservation priority. In the Gamo highlands these small patches are generally sacred sites with high cultural value, however they have recently come under high levels of threat which risks loss of biocultural diversity. Their conservation through strengthening of traditional community institutions is a high priority.