Farmland birds are of conservation concerns around the world. In China, conservation management has focused primarily on natural habitats, whereas little attention has been given to agricultural landscapes. Although agricultural land use is intensive in China, environmental heterogeneity can be highly variable in some regions due to variations in crop and non-crop elements within a landscape. We examined how non-crop heterogeneity, crop heterogeneity and non-crop features (non-crop vegetation and water body such as open water) influenced species richness and abundance of all birds as well as three functional groups (woodland species, agricultural land species and agricultural wetland species) in the paddy-dominated landscapes of Erhai water basin situated in Northwest Yunnan, China. Birds, crop, and non-crop vegetation surveys in twenty 1 km × 1 km landscape plots were conducted during the winter season (from 2014 to 2015). The results revealed that bird community compositions were best explained by amounts of non-crop vegetation and compositional heterogeneity of non-crop habitat (Shannon-Wiener index). Both variables also had a positive effect on richness and abundance of woodland species. Richness of agricultural wetland species increased with increasing areas of water bodies within the landscape plot. Richness of total species was also greater in the landscapes characterized by larger areas of water bodies, high proportion of non-crop vegetation, high compositional heterogeneity of non-crop habitat or small field patches (high crop configurational heterogeneity). Crop compositional heterogeneity did not show significant effects neither on the whole community (all birds) nor on any of the three functional groups considered. These findings suggest that total bird diversity and some functional groups, especially woodland species would benefit from increases in the proportion of non-crop features such as woody vegetation and water bodies as well as compositional heterogeneity of non-crop features within landscape.