Ortolan Bunting male territory observations from Finland from 2000-2018 and R-code used to analyse the association between population growth rate and various environmental and spatial variables. Data includes individual identification number for 238 singing groups (id), the year of the observation (year), number of observed territories within a singing group at the year of the observation (terri), number of observed territories during the previous year (terpre), proportion of agricultural land use within the singing group landscape (agri5), the mean daily precipitation (mm) (prec) and temperature (ºC) (temp) of the previous year summer, Shannon-Wiener diversity index of crop types within the singing group area (shang), proportion of crop types which provide bare ground in spring (bare), density of buildings (build), rivers (river), streams (stream) and roads (road), the municipality (municipality) and the region (region) where the singing group is located.

The territories of the Ortolan Buntings were mapped annually using a two-visit mapping method. To form a study unit, we aggregated observations of territories from the years 2000–2017 (observations from the year 2018 were added in after the forming of singing groups) based on their proximity and surrounding landscape. Territories that were less than 500 metres away from each other, were considered to belong to the same singing group, i.e., subpopulation. Then, we drew a 250-meter buffer around each territory observation within a singing group, merged the overlapping buffers, and interpreted the outline as delineating the area occupied by that singing group. Some singing groups were split and merged according to field observations and expert opinion. For the statistical analysis, we first delimited our data to only those cases that had data from two consecutive years, i.e., with no gap between annual observations. Then we created a variable describing the number of territories in a singing group during the previous year (terpre) and further removed from the data those observations, where 'terpre' was zero (no territories observed the previous year). From the Finnish topographic database provided by the National Land Survey of Finland, we counted the length of roads, ditches, main drains, and riverbanks (in metres) and the number of buildings within the singing group area. We then divided each of these measures by the area of the singing group to form a variable describing the density of each small-scale element (road, stream, river, building). By combining data from the Finnish land and crop parcel registers by the Finnish Food Authority, we counted the total area of crop species grown within a singing group area. In cases where crop data were untraceable or not available, we completed the data from field notes, when possible, otherwise, the crop was classified as an “unknown species”. Based on plant type or growth form we classified crop species into ten crop types. For each singing group area, we then counted the proportions of each of these crop types and based on those, a Shannon–Wiener diversity index of crop types (shang). Crop types were further classified into two groups based on how much vegetative cover the plants form during the start of the Ortolan Bunting breeding season in early May. Crop types that provide only minor cover and leave substantial amounts of bare ground on the field, mainly as a result of the associated plowing practice, were classified as ‘bare’. We counted the proportion of ‘bare’ crop types within a singing group area (bare). We created a 5-km buffer around the singing group centroid and calculated the proportion of agricultural land cover (agri5) within the buffer using the Finnish Corine land cover data provided by the Finnish Environment Institute. Lastly, we calculated the mean daily temperature (ºC) (temp) and precipitation (mm) (prec) of the previous summer (21. May–15. July). 

Shaping the data for statistical analysis:

For the statistical analysis, we first delimited our data to only those cases that had data from two consecutive years, i.e., with no gap between annual observations. Then we created a variable describing the number of territories in a singing group during the previous year (terpre) and further removed from the data those observations, where terpre was zero (no territories observed the previous year).