2.2 STUDY SITE CHARACTERIZATION AND SAMPLING
Sampling activities were performed in July 2019 in 19 sites located in an urbanized area in the central portion of north Italy covering four administrative provinces of northern Italy (i.e., Milano, Monza e della Brianza, Lecco and Como) (Figure 1). In order to avoid non-independence, the distance between sites were at least 2 km, above the foraging range observed for these species (Biella et al., 2022). Sampling sites have been selected to cover a gradient of growing urbanization ranging from semi-natural areas composed of hay meadows close to forest patches to the more urbanized context of the city center of Milan. To select sampling sites impervious surfaces (i.e., concrete, building, and asphalt) have been mapped using regional land-use cartography (2018-DUSAF 6.0;https://www.dati.lombardia.it/Territorio/Dusaf-6-0-Uso-del-suolo-2018/7rae-fng6) as explained in detail in (Tommasi et al., 2022), afterward sampling sites have been chosen along a visible gradient of impervious cover.
In each site, five to six workers per species were collected with random walks within a 50 m × 50 m plot using an entomological net. After collection samples were stored at − 80 °C until further analyses. Overall 192 individuals belonging to the investigated bumblebee species (96 B. pascuorum and 96 B. terrestris ) have been subjected to gut DNA extraction and multi-target parasite screening.
Floral resources were estimated in each site by counting the total number of flowers within six squares 1 m × 1 m placed in flowering spaces within the sampling area (see Tommasi et al., 2022 for a detailed explanation about floral resources availability estimation). The data were also used to estimate the diversity of the available floral resources in each site and was calculated as Shannon diversity Index (Shannon, 1948). The percentage of green habitat (i.e., meadows, forests and urban green spaces) in the 1 km radius buffer surrounding each sampling site have been calculated using the previously mentioned land-use map (see Tommasi et al., 2022 for a detailed explanation about land-use categorization). Specifically, the buffer size has been selected according to the maximum foraging range expected for these species (Knight et al. 2005; Greenleaf et al. 2007). Furthermore, green habitat fragmentation has been estimated for each buffer surrounding the sampling sites by calculating the Euclidean nearest neighbor distance (ENN) of green habitat patches through the package landscape metrics (Hesselbarth et al., 2019) in the software R (version 2022.12.0).
The distance of each sampling site from the closer honey bee hive and the number of hives in the 1 km buffer surrounding the sites have been calculated through the distance matrix function of QGIS (version 3.28.4). Honeybee hive locations have been obtained from the national beekeeping database (BDA) upon request and released by the competence office of each province involved in the present study. Variables calculated for each sampling site are available in supporting information Table S1.
[ FIGURE1]