Habitat plays an important role in the life history of birds. Habitat quality is directly related to the bird's survival and reproduction. Since bird species respond to the habitat variation simultaneously across a range of organizational levels, multi-scale approaches can potentially better describe bird species-environment relationships compared to single-scale approaches. Red-crowned crane (Grus japonensis) is a rare, endangered species living in native wetlands. The Yellow River Delta Nature Reserve (YRDNR) in Shandong province is one of most important sites for this species' migration and wintering. Here, we take YRDNR as a case study to explore the main limiting factors and scales of habitat selection of the crane at microhabitat, patch and landscape scales. We collected the red-crowned crane distributional data and microhabitat level factors from two field surveys, and extracted the patch or landscape level factors using remote sensed images. We used hierarchical variance partitioning methods to partition explainable variance by habitat factors in the red-crowned crane distributional data at two hierarchical levels. The first level of analysis partitioned variance explained by microhabitat, patch and landscape level factors alone and in combination. The second level partitioned variance explained by vegetation and water factors independently and jointly at microhabitat scale, landscape composition and configuration factors independently and jointly at landscape scale. By comparing the amount of variance occupied by habitat factors at each hierarchical level, we examined their independent and joint effects on habitat selection of red-crowned crane. Moreover, we used hierarchical partitioning methods to segregate the independent effects of each habitat factors separately, and explored their contributions to the habitat selection. The results showed at the first level, a majority of variance explained by the jointly effect of landscape level factors with microhabitat (10.8%) or patch (25.3%), or microhabitat and patch (28.9%) level factors. Meanwhile, landscape level factors independently accounted for a larger amount of the explained variance (12.7%) than microhabitat (6.0%) or patch (1.6%) level factors. At the second level, the landscape composition was more important than the landscape configuration at landscape level, and the vegetation and water contributed more explainable variation at the microhabitat scale. The results of hierarchical partitioning methods showed that seabilite tidal flat, and water area are the main habitat factors at landscape scale since their Z value of independent effects are statistically significant. Vegetation coverage and water depth contributed more explanatory power at microhabitat scale. At patch scale, patch type was the predominant factor. Besides, the road density also had independent effects on habitat selection of red-crowned crane. This study suggested the landscape scale is the main scale related to the distribution of red-crowned crane. Landscape level factors, together with microhabitat and patch level factors, jointly controlled habitat selection in YRDNR. To protect the wetland habitat for red-crowned crane, we proposed to take effective measures to strengthen the protection and management of the tidal flat, weed marsh or water body, and regulate or control human activities that threaten and destroy the wetland habitat in YRDNR.