In comparison with other tropical forest land uses such as selective logging, little is known of the impacts on wildlife of the many forms of small-scale agriculture practised across the tropics. We present density estimates, derived using a point count distance sampling method, for 31 bird species in primary forest, old abandoned gardens and active/recently abandoned gardens at two altitudes in the Crater Mountain Wildlife Management Area (CMWMA), Papua New Guinea. There were clear habitat differences between the six habitat/altitude categories, with, for example, clines in tree sizes and canopy cover from highest values in primary forest to lowest values in current gardens. At lower altitudes, primary forest held highest densities of most species, whereas at higher altitudes, old abandoned gardens had greater densities of many birds, especially insectivores. CANOCO was used to ordinate bird species with respect to major habitat gradient axes. Major axes were associated with differences in bird responses to forest conversion as well as altitudinal differences in species composition. Most important was that several insectivores (especially monarchs, fantails, etc.) formed a cluster of species associated with intact, high-biomass forest. We suggest that most species reacted moderately to habitat changes currently occurring, and this may be due in part to the fact that only a small proportion of the landscape at CMWMA has been converted to agriculture (around 13% may be current or recently abandoned gardens). There were, however, species with comparatively low densities in agricultural habitats and these included several insectivores, the terrestrial Blue Jewel-babbler Ptilorrhoa caerulescens, and three out of four birds of paradise.
Shifting cultivation (including slash-and-burn and gardening) is a major land use and cause of deforestation in tropical regions (Fujisaka et al. 1996, Raman 2001). In fact, Myers (1991) described the landless peasants ('shifted cultivators') practising shifting cultivation as the main agent of tropical forest loss, accounting for at least 60% of deforestation. Despite the large area of the tropics over which it is practised, the great diversity of agricultural systems, and the debate as to the degree to which such land uses contribute to biodiversity loss (Myers 1991, Halladay & Gilmour 1995), few papers have assessed the impact on wildlife of the myriad small-scale agricultural systems practised (for birds see the partial review by Dunn 2004; see also Blankespoor 1991, Thiollay 1995, Raman 2001, Naidoo 2004).
Here, we present one of the first multi-species bird studies examining differences in bird abundance between primary forest and small-scale 'garden' agriculture plots. We present population density estimates for 31 forest bird species in primary forest, current and recently abandoned gardens, and old gardens within two altitude bands (432–650 m and 651–935 m) at Crater Mountain Wildlife Management Area (CMWMA), a hillforest region of eastern New Guinea. The last two habitat categories represent small-scale mixed agriculture/agroforestry, the main generic form of forest alteration across much of Papua New Guinea (Levett & Bala 1994). In CMWMA, agricultural plots are small and exist within an extensive matrix of little-disturbed forest and this mosaic of low-intensity and diverse gardens is rather different from the agricultural landscapes of other studies (Thiollay 1995, Estrada et al. 1997). It is also, however, the type of traditional low-intensity agricultural system that is under threat as a result of agricultural expansion and homogenization in many areas of the tropics (Allen 1985, Thiollay 1995).
Furthermore, the density estimates we present are, in almost every species, the first indications of absolute abundance for birds on New Guinea, so we compare bird densities and reactions to habitat change with those recorded on nearby islands and draw some conclusions as to the likely impact of habitat change on the avifauna as a whole.