Landscape context affects both capture probability and abundance of solitary bees in cities
Data files
Jul 29, 2025 version files 45.09 KB
-
Agapostemon_simple.R
11.03 KB
-
captures_for_Mark_03.01.2024.csv
5.59 KB
-
CMR_04.04.2024.R
13.40 KB
-
Gard_size.csv
167 B
-
Imp_vals_perpot.csv
453 B
-
Mark_recap_raw.csv
7.76 KB
-
README.md
6.69 KB
Abstract
It is intuitive to hypothesize that urban landscapes are broadly hostile to insects, but responses of wild bees to urbanization are inconsistent and often positive. We used a mark-recapture study of the bicolored sweat bee (Agapostemon virescens) to separate two possible causes of this pattern: differences in capture probability (bees are more attractive to sampling sites in low-resource areas) from patterns of true abundance (bees are equally abundant in high- and low-resource landscapes, as long as resources are present locally). To distinguish these possibilities, wee placed experimental patches of Echinacea purpurea, a favored floral resource, in and near community gardens along a landscape gradient of % impervious surface. We collected mark-recapture data of A. virescens visitation to these patches, and analyzed capture probability and abundance in relation to % impervious surface in the surrounding landscape, and local resource context, measured as inside or outside the community garden in low-resource settings (pavement, mulch, or turf grass). This study took place in the cities of Cambridge, Somerville, and Medford, Massachusetts, USA, which are part of the greater Boston metropolitan area. We found that capture probability was higher in low-resource contexts (outside of gardens and/or in landscapes with a high % impervious surface). Fewer individual bees visited patches in non-garden sites and landscapes with high impervious surface, but those bees visited the patches more often.
In addition, all of our experimental patches were used by A. virescens. These data include the capture histories of individual bees, and the characteristics of the sites and community gardens.
Dataset DOI: 10.5061/dryad.80gb5mm2j
Description of the data and file structure
This data set includes two R files and four data files, as described below.
Note that creating Figure 3 also requires use of the t_col function, which we saved as a file called transparent.R. We obtained this function from https://dataanalytics.org.uk/make-transparent-colors-in-r/ (most recently accessed 6/30/2025).
Files and variables
File: Agapostemon_simple.R
Description: This file summarizes basic descriptive statistics from the mark-recapture data, including the number of unique individuals seen at each pot and the number of captures at each pot. It creates Figures 3 a&b in the main manuscript. It also includes analyses of community garden size, and figures of these relationships that were not included in the manuscript.
File: CMR_04.04.2024.R
Description: This file performs mark-recapture analyses of bee captures, and creates Figures 3 c&d from the output of the mark-recapture analysis
File: captures_for_Mark_03.01.2024.csv
Description: This file includes mark-recapture data formatted for formal mark-recapture analysis. In columns 1-4, each row describes an individual bee at an experimental patch. (Extremely few bees moved between experimental patches, and these observations were not included in the mark-recapture analysis; see Dooley et al. 2025 (the manuscript associated with the data set).)
Variables
- ch: Capture history, identifying whether each individual bee was seen at that patch during each of the 16 survey periods of the study. “1” indicates that the bee was observed, “0” indicates that the patch was monitored but the bee was not observed, and “.” Indicates that the patch was not monitored.
- pot: Location of the experimental patch associated with each capture history. “G” = patch in the community garden. “N” = patch outside the community garden.
- site: Abbreviated names for community gardens at which our study took place. These are the same as the “Site” column in the “Imp_vals_perpot.csv” file.
- imperv: Percent of the landscape surrounding each experimental patch that is impervious surface, using a 100-m buffer. These align with the Imp_G and Imp_N columns in the “Imp_vals_perpot.csv” file.
- int: Number of days between each of the 16 survey periods described by the capture history in the “ch” column of this file. There are 15 rows, corresponding to the intervals between these 16 dates.
File: Gard_size.csv
Description: This file contains information about the size and number of floweing units of the community gardens.
Variables
- Site: Abbreviated names for community gardens at which our study took place. These are the same as the “Site” column in the “Imp_vals_perpot.csv” file.
- Area_m2: Area of each community garden, measured in square meters.
- Fl_counts_ingard: Estimated number of flowering units in each garden, measured just prior to our mark-recapture study in late June, 2023.
File: Imp_vals_perpot.csv
Description: This file includes information about the impervious surface surrounding our study sites.
Variables
-
Site: Location in the study area, corresponding to community gardens mapped on Figure 2 of Dooley et al. 2025 (the manuscript associated with these data). In relation to site names in Figure 2, the abbreviations refer to:
TUF – Tufts Park
BIK – Bikeway (74.7%)
SAC – Sacramento St
MCM – McMath Park
POW – Powerhouse Blvd
WAL – Walnut St
LIN – Lincoln Park
WIN – This site was not included in the Figure or analysis because a large proportion of the surrounding area was a river.
-
Imp_park: Impervious surface within 100m of the geographic centroid of each community garden. Units are % of surface area.
-
Imp_G: Impervious surface within 100m of the experimental Echinacea patches located in each community garden. Units are % of surface area. This is the “impervious surface” predictor associated with garden sites in Figure 3 of Dooley et al. 2025 (the manuscript associated with these data).
-
Imp_N: Impervious surface within 100m of the experimental Echinacea patches located outside the community gardens. Units are % of surface area. This is the “impervious surface” predictor associated with non-garden sites in Figure 3 of Dooley et al. 2025 (the manuscript associated with these data).
-
Imp_mid: Impervious surface within 100m of the midpoint of the two experimental Echinacea patches located near each community garden. Units are % of surface area. These are the values used to represent each community garden in Figure 2 of Dooley et al. 2025 (the manuscript associated with these data).
File: Mark_recap_raw.csv
Description: This file includes mark-recapture data for each observation period during our study. This file was used to generate basic statistics such as the number of captures and unique individual bees visiting each experimental patch. It was converted to a capture history (captures_for_Mark 03.01.2024.csv) for formal mark-recapture analysis.
In this data set, rows with 0's in all columns except Site and Date indicate surveys in which sites were visited and no A. virescens were seen.
Variables
- Date: Date on which each capture occurred, in day-month format. The study took place in July 2023.
- Site: Abbreviated names for community gardens at which our study took place. These are the same as the “Site” column in the “Imp_vals_perpot.csv” file.
- Time_start: Starting time of the 30-minute observation period in which each capture occurred. All observations are in the morning (“AM”).
- Capture_time: Time at which each capture occurred. All observations are in the morning (“AM”).
- Pot: Location of the experimental patch at which each capture occurred. “G” = patch in the community garden. “N” = patch outside the community garden.
- Mark_ID: Code representing the unique 3-color combination associated with each individual bee.
- Recap: Identifies whether each capture was an initial capture (“0”) or a recapture of a marked individual bee (“1”).
Access information
Other publicly accessible locations of the data:
Data was derived from the following sources:
- n/a