Data from: Species-habitat networks reveal key habitats for landscape-level wild bee conservation
Data files
Mar 06, 2026 version files 102.75 KB
-
Bee_species-habitat_metanetwork.csv
2.26 KB
-
Bees_per_sampling_location.xlsx
59.53 KB
-
Dominant_habitat_type_of_sampling_locations.xlsx
13.79 KB
-
Proportional_cover_of_land_cover_types_in_landscapes.xlsx
11.10 KB
-
README.md
4.32 KB
-
Trait_information_of_bee_species.xlsx
11.75 KB
Abstract
Most agricultural landscapes are composed of a variety of habitats. A landscape perspective is needed to understand biodiversity decline, but many studies focus on single habitat types. In addition, the use of local resources by species within and across habitats implies that species and their habitats are linked in species-habitat networks. However, studies on these networks are scarce. Here we used grid-based sampling to assess wild bees at 224 sampling locations across all major habitat types, i.e., arable land, grassland, forest and orchard, in 14 differently composed agricultural landscapes of Southern Germany. We assigned wild bees to habitat types based on the dominant habitat cover surrounding their sampling location to establish species-habitat networks and assessed how these networks differed in modularity and robustness to habitat loss. Orchards harbored more wild bees than expected based on their proportional cover in the landscape, indicating a preference for this extensively managed but threatened habitat by wild bees. Orchards also supported the highest species richness and proportion of oligolectic wild bees, while forests harbored the lowest richness and more social species. Landscape diversity affected both structure and robustness of bee-habitat networks in response to the simulated loss of habitats. Networks in more diverse landscapes had higher modularity but tended to be less robust, showing that greater landscape diversity and modularity do not necessarily buffer against the effects of habitat loss. However, this effect appeared to be mainly driven by increases in network size, as standardized modularity and robustness (z-scores) were not affected by landscape diversity. We could show that species-habitat networks are a powerful tool to inform ecologists and policy makers about the importance of key habitats and landscape diversity for species conservation. Key habitats for wild bee conservation include extensively managed habitats like traditional orchards. Nevertheless, all habitat types support a similar proportion of endangered species, emphasizing the importance of a diverse landscape. Conserving wild bees requires a variety of complementary habitats at the landscape scale and must consider the management of traditional and intensively managed habitats alike. Policy measures targeting landscape diversity are urgently needed.
Dataset DOI: 10.5061/dryad.s7h44j1p7
Description of the data and file structure
This dataset contains data on wild bee species richness and abundance sampled using pan traps across agricultural landscapes in south-western Germany. The study was conducted in 2023 across fourteen 1-km² study landscapes selected along a gradient of landscape diversity (Shannon diversity index). We applied a grid-based sampling design using 16 sampling locations in each of the 14 study landscapes (= 224 sampling locations in total).
Files and variables
File: Bees_per_sampling_location.xlsx
Description: Data on the abundance of 75 wild bee species sampled at all 224 sampling locations (16 sampling locations in each of the 14 study landscapes of 1 km2).
Variables
sampling_location- name of the sampling location the bees were sampled at. The name is composed of the landscape ID and the position of the sampling location in the landscape. Sampling locations were placed in the landscapes in a grid-based sampling design in 4 columns and 4 rows. The letter (A-D) refers to the identification of the row (N-S axis) and the number to the column (W-E axis). For example, L1_D1 is in the lower left (SW) corner of landscape 1.Andrena_bicolortoXylocopa_violacea- name of the species the abundance refers to. Abundances are summed counts over the two replicates for each sampling location and species.
File: Dominant_habitat_type_of_sampling_locations.xlsx
Description: The name of the dominant habitat type of the 224 sampling locations
Variables
sampling_location- name of the sampling location the bees were sampled at (see also above)dominant_habitat_type- name of the dominant habitat type (arable land, forest, grassland or orchard) in a 100 m buffer around the sampling location. Sampling locations with dominant land cover type 'settlement' were removed from further analyses, but are still represented in all datasets.
File: Proportional_cover_of_land_cover_types_in_landscapes.xlsx
Description: Data on the proportion of each of the seven land cover types (arable land, forest, grassland, orchards, hedges, roads and settlements) in each of the 14 landscapes, each of which covers 1 km² in total.
Variables
landscape- name of landscape / landscape ID (also used in the names of the sampling locations in other datasets)arabletosettlement- proportion of the respective land cover type in the landscape. Landscapes measured 1 km2, thus a land cover proportion of 0.579 refers to an area of 57.9 ha. The proportional cover of all land cover types add up to 1 for all landscapes.
File: Trait_information_of_bee_species.xlsx
Description: Information on the traits of 75 bee species is provided, including mean body size, social behaviour, dietary breadth and endangerment status. For some species (groups), information on certain traits was unavailable and is indicated by 'NA'.
Variables
Species name- names of the 75 bee speciesmean body size- mean body length of female bees given in mmsocial behaviour- social behaviour of the bee species, being 'solitary' (expressed as 0) or 'social' (expressed as 1).dietary breadth- dietary breadth of the bee species, being 'oligolectic' (expressed as 0) or 'polylectic' (expressed as 1)endangerment status- Baden-Württemberg Red List status of the bee species, being 'not endangered' (expressed as 0) or 'endangered' (expressed as 1)
File: Bee_species-habitat_metanetwork.csv
Description: Data table for the meta-network on which Figure 2b in the manuscript is based
Variables
habitat- name of the habitat type that the bee species was assigned toAndrena_bicolortoXylocopa_violacea- name of the species the abundance refers to. Abundances are summed counts over the two replicates and all sampling locations of the same habitat type.
Code/software
All figures and statistical analyses were done in R version 4.4.1 with packages bipartite, effects, emmeans, ggeffects, ggplot2, glmmTMB, iNEXT, lme4, multcomp and vegan.
