Data and analyses for: Localised patterns of wild bee abundance indicate woodlands play multiple roles in supporting farmland populations
Data files
May 20, 2025 version files 145.01 KB
-
Complete_analyses_-_Allen_et_al_2025.Rmd
86.58 KB
-
First_season_data1.csv
33.16 KB
-
Floral_indices_by_family.csv
18.99 KB
-
README.md
4.93 KB
-
Relative_open_-_canopy_estimates_all_bees.csv
785 B
-
Site_coverage_of_tree_species.csv
567 B
Abstract
Higher woodland cover has been linked to increased wild bee abundance and diversity in temperate agricultural landscapes. However, our understanding of the roles played by the upper and lower strata of woodlands in supporting on-farm bees throughout the season is lacking.
To explore these roles, we sampled bees (Anthophila) in deciduous woodland canopies and understories, and in an open habitat (at field margins), from May to July, at 12 sites across agricultural landscapes in Norfolk, England.
Before canopy closure in late spring, relative abundance in the two woodland habitats was generally higher, and canopies with flowering Acer pseudoplatanus L. supported more bees. However, throughout the season, open-habitat bee abundance was consistently higher closer to woodlands. After canopy closure, more open areas within woodlands were linked to greater bee abundance, except during a spike in temperatures when understory relative abundance was also at its highest.
These results indicate that deciduous woods on farmland provide floral resources to bees before canopy closure, which can be augmented with nectar-producing canopy trees. They also indicate that more open woodlands likely extend the availability of understory floral resources; and that farmland woods, regardless of management, may continue to provide non-floral resources – including respite from hot weather – throughout the season.
Dataset DOI: 10.5061/dryad.5dv41nshm
Description of the data and file structure
Files and variables
File: Complete_analyses_-_Allen_et_al_2025.Rmd
Description: R code to run all analyses presented in manuscript.
File: First_season_data1.csv
Description: All data contributing to the main analyses presented in the manuscript (Models 1 - 13 in Table 2). For further details on variables, see Methods and Table 1 in the manuscript. Empty cells indicate that the variable is not applicable to the observation in question (except in rows 327 and 404 where they indicate that the observation is missing because of trap failure).
Variables
- period: sampling period (I = 'Early May', II = 'Late May', III = 'Mid-June', IV = 'Mid-July')
- trap: trap location ID
- lat: latitude
- long: longitude
- tot_bees: total bees caught in a trap
- und_bees: total bees caught in an understory trap
- region: sampling region (1 = 'West', 2 = 'South-east')
- site: sampling site ID
- habitat: sampled habitats ('open', 'understory' and 'canopy'; note, 'open' = field margin)
- dist_wood: distance to woodland (near = '60 m', far = '120 m')
- dist_clear: distance to nearest clearing (m)
- dist_edge: distance to nearest woodland edge (m)
- can_open: canopy openness (%)
- flora: floral index (floral cover at trap location)
- syc_si_pr: Sycamore trees (*Acer pseudoplatanus) *(present or absent)
- ches_si_pr: Chestnut trees (Castanea sativa) (present or absent)
File: Relative_open_-_canopy_estimates_all_bees.csv
Description: Extracted model estimates for the correlation analysis (from models 1 - 4 in Table 2; question 2 in manuscript)
Variables
- period: sampling period (I = 'Early May', II = 'Late May', III = 'Mid-June', IV = 'Mid-July')
- region: sampling region (1 = 'West', 2 = 'South-east')
- open: field margin estimate relative to understory estimate
- canopy: canopy estimate relative to understory estimate
- open_low: lower confidence interval limit for 'open'
- open_upp: upper confidence interval limit for 'open'
- cano_low: lower confidence interval limit for 'canopy'
- cano_upp: upper confidence interval limit for 'canopy'
- temp: average temperature over full sampling days (degrees Celsius)
File: Site_coverage_of_tree_species.csv
Description: Woodland canopy coverage (presented in Figure 1 of manuscript).
Variables
- Site: sampling site ID
- Sycamore: canopy coverage (%) of Acer pseudoplatanus
- Chesnut: canopy coverage (%) of Castanea sativa
- Oak: canopy coverage (%) of Quercus robur/petraea
- Ash: canopy coverage (%) of Fraxinus excelsior
- Beech: canopy coverage (%) of Fagus sylvatica
- Other: canopy coverage (%) other tree species
- Canopy gap: canopy gap (%)
File: Floral_indices_by_family.csv
Description: Floral index of each plant family (summarised in figure S1; further details on methodology provided in Appendix S1).
Variables
- period: sampling period (I = 'Early May', II = 'Late May', III = 'Mid-June', IV = 'Mid-July')
- trap: trap location ID
- habitat: sampled habitats ('open', 'understory' and 'canopy'; note, 'open' = field margin)
- flora: total floral index (floral cover at trap location [sum of floral indices across all plant families])
- Lamiaceae: floral index of Lamiaceae
- Rosaceae: floral index of Rosaceae
- Liliaceae: floral index of *Hyacinthoides non-scripta *and Allium ursinum (now classified as different families)
- Caryophyllaceae: floral index of Caryophyllaceae
- Asteraceae: floral index of Asteraceae
- Apiaceae: floral index of Apiaceae
- Brassicaceae: floral index of Brassicaceae
- Ericaceae: floral index of Ericaceae
- Plantaginaceae: floral index of Plantaginaceae
- Fabaceae: floral index of Fabaceae
- Geraniaceae: floral index of Geraniaceae
- Boraginaceae: floral index of Boraginaceae
- Ranunculaceae: floral index of Ranunculaceae
- Adoxaceae: floral index of Adoxaceae
- Orchidaceae: floral index of Orchidaceae
- Cucurbitaceae: floral index of Cucurbitaceae
- Cornaceae: floral index of Cornaceae
- Convolvulaceae: floral index of Convolvulaceae
- Fagaceae: floral index of Fagaceae
- Onagraceae: floral index of Onagraceae
- Hypericaceae: floral index of Hypericaceae
Code/software
The script is formatted for R Markdown in R studio (R version 4.3.2).
Annotations are provided throughout the script which cover library loading; dataset loading and cleaning; analyses (including diagnostics); and figure creation.
This script performs all the analyses presented in "Localised patterns of wild bee abundance indicate woodlands play multiple roles in supporting farmland populations".
