Table 2.

Environmental variables of the CityScapeLab Berlin (von der Lippe et al. 2020) used for statistical analyses of predictors

Variable	Unit	Explanation	Method/equipment/software	Data source/reference
Urban matrix
[1] Urbanisation	%	Percentage of impervious surface in a 500-m buffer around the biotope patch in which the plot is located	Software: QGIS version 2.18.0; tool: Zonal statistics	Berlin Environmental Atlas/Actual Use of Built-up Areas, Inventory of Green and Open Spaces 2010 (SenUDH 2011)
[2] 3D connectivity	0-…/#	Hanski’s connectivity index (modified): considers area size, distance to other dry grassland patches and buildings height	Software: ArcGIS 10.3.1 for Desktop; tool: Generate Near Table (Analysis)	Berlin Environmental Atlas/Biotope Types (SenUDH 2014a), Biotope mapping Brandenburg (Landesamt für Umwelt Brandenburg 2009); Berlin Environmental Atlas/Building and Vegetation Heights (SenUDH 2014b) Hanski (1994, 1999)
Local habitat scale
[3] Patch size	m2	Size of the dry grassland patch in which the plot is located	Software: QGIS version 2.18.0	Berlin Environmental Atlas/Biotope Types (SenUDH 2014a), Biotope mapping Brandenburg (Landesamt für Umwelt Brandenburg 2009)
[4] Cover of herbaceous plant species	%	Estimated cover of herbaceous plants on the plot (no grasses, no coniferous trees) herbaceous plants potentially form flowers and thus represent the main resource source for many pollinators	Estimation flowering ability derived from Biolflor (Klotz et al. 2002)
[5] Cover of non-native herbaceous plant species	%	Estimated cover of non-native herbaceous plants on the plot (no grasses, no coniferous trees)	Estimation floristic status derived from Biolflor (Klotz et al. 2002)
[6] Mean air temperature	°C	Mean maximum air temperature of the plot measured at 2 m between April and September (month selection was based on the pollinator activity periods)	Data logger: EasyLog EL-USB-2 + , Lascar Electronics