The population of the Yellow-breasted Bunting Emberiza aureola, a formerly widely distributed and abundant songbird of northern Eurasia, suffered a catastrophic decline and a strong range contraction between 1980 and 2013. There is evidence that the decline was driven by illegal trapping during migration, but potential contributions of other factors to the decline, such as land-use change, have not yet been evaluated. Before effects of land-use change can be evaluated, a basic understanding of the ecological requirements of the species is needed. We therefore compared habitat use in ten remaining breeding regions across the range, from European Russia to Japan and the Russian Far East. We also assessed large-scale variation in habitat parameters across the breeding range.

We found large variation in habitat use, within and between populations. Differences were related to the cover and height of trees and shrubs at Yellow-breasted Bunting territories. In many regions, Yellow-breasted Buntings occupied early successional stages, including anthropogenic habitats characterized by mowing, grazing or fire regimes. We found that the probability of presence can be best predicted with the cover of shrubs, herbs and grasses. Highest probabilities were found at shrub cover values of 40 to 70 %.

Differences in habitat use along a longitudinal gradient were small, but we found strong differences across latitudes, possibly related to habitat availability. We conclude that the remaining Yellow-breasted Bunting populations are not limited to specific habitat types. Our results provide important baseline information to model the range-wide distribution of this critically endangered species and to guide targeted conservation measures.

We compiled data from a network of collaborators that had monitored Yellow-breasted Buntings during the breeding season (May to July) in 2017, 2018 and 2019, following the report of strong declines in 2015 (Table 1). All contributors searched for singing males and used the individual song post as the centre of a 10 by 10 meters plot for habitat mapping (“presence plots”). We visually estimated the total vegetation cover [in %] and the cover [in %] of trees, shrubs, dwarf shrubs, grass, herbs and litter by standing at each of the four corners of the plot. We defined trees as a woody plant with a single trunk, and a shrub as a woody plant with more than one trunk. Low-growing shrubs (< 1 m height) were defined as dwarf shrubs (e.g. Salix myrtilloides). Reed was classified as grass. Woody parts of Artemisia were classified as herbs. If burned and dead parts were still attached to a plant, they were treated as part of it. We estimated the mean height [in cm] of trees, shrubs, dwarf shrubs, grass (, herbs and litter. Additionally, the cover of bare soil [in %] was estimated. Furthermore, we noted signs of fire, grazing or mowing from the current breeding season (0 = no, 1 = yes). Moisture was estimated in four categories (0 = completely dry, 1 = moist or wet, 2 = waterlogged and 3 = standing open water or flooded soil).

To analyse habitat use we additionally recorded all habitat parameters at pseudo-absence points, which were placed 100 m to the east of song posts (“absence plots”). We expected these absence plots to be outside of the territory, as distances between nests of 40-100 m were found (Rejmers 1966 in Glutz von Blotzheim & Bauer 1993). This allowed us to establish preferences for certain habitat features, as opposed to the descriptive notion of habitat use employed when only observational data at presence points is included. Additional pseudo-absence points were randomly selected in the Amur region, to get a representative sample (n > 30) for both burned and unburned wetlands.