Colony-level pollen collection reflects visitation of managed bumble bees (Bombus impatiens) in strawberry fields and surrounding landscapes without reducing pollen limitation dataset.
Data files
Jun 12, 2026 version files 13.43 KB
-
fruitdata.csv
4.26 KB
-
msdata.csv
1.58 KB
-
README.md
7.59 KB
Abstract
Managed bees are frequently used to supplement pollination of a range of crop plants. Yet, their effectiveness varies, and much of their foraging can be outside of the focal crop. Bee foraging behavior is highly sensitive to the availability of alternative floral resources which depend on the composition of the surrounding landscape. To understand how foraging and crop pollination dynamics of the managed common eastern bumble bee (Bombus impatiens) are influenced by landscape composition, we assessed their visitation and pollination in cultivated strawberry fields along a landscape gradient. We also used pollen extracted from bumble bee (Bombus impatiens) colony wax to understand whole-colony foraging over the blooming period. Bumble bee visits to strawberry crop flowers were very low and pollen limitation was not reduced at sites with greater visitation; however, colonies in fields with higher visitation rates had more pollen from the Rosaceae family (including strawberry). We found that landscape composition affected foraging effort on strawberry at the colony level, as a lower proportion of rosaceous pollen was collected in landscapes with more surrounding pasture, while a greater proportion of rosaceous pollen was collected at sites with larger strawberry fields. Overall, our results suggest that managed bumble bees do forage on strawberry crop flowers, but they predominantly forage elsewhere, especially in surrounding pastures when strawberry fields are small. Therefore, the addition of bumble bee colonies may not be economically beneficial for strawberry growers at least in diverse agricultural landscapes similar to that of our study region.
Dataset DOI: 10.5061/dryad.05qfttfhv
Description of the data and file structure
To understand how foraging and crop pollination dynamics of the managed common eastern bumble bee (Bombus impatiens) are influenced by landscape composition, we assessed their visitation over four sampling periods and pollination in cultivated strawberry fields along a landscape gradient. We also used pollen extracted from bumble bee (Bombus impatiens) colony wax to understand whole-colony foraging over the blooming period.
Files and variables
File: msdata.csv
Description: contains all the data related to bee visitation, pollen collection, and landscape surrounding each site.
Variables
- Site: name of each site
- BBave: refers to the average number of bumble bees recorded over the four sampling periods per site
- WBave: refers to the average number of other bees recorded over the same sampling periods
- TotAve: refers to the average total number of bees recorded over these sampling periods
- rosaceae: proportion of pollen counted from the wax that belonged to the family rosaceae
- PAST_1000: proportion pasture surrounding 1000m around a site
- AG_1000: proportion agriculture surrounding 1000m around a site
- AG_500: proportion agriculture surrounding 500m around a site
- AG_750: proportion agriculture surrounding 750m around a site
- NAT_OPEN_1000: proportion natural area surrounding 1000m around a site
- NAT_OPEN_500: proportion natural area surrounding 500m around a site
- NAT_OPEN_750: proportion natural area surrounding 750m around a site
- PAST_500: proportion pasture surrounding 500m around a site
- PAST_750: proportion pasture surrounding 750m around a site
- Size: strawberry fields size (acres)
File: fruitdata.csv
Variables
- Site: name of each site
- Plant: identifier for each plant
- Weight_diff: at each "Site" a "Plant" had two fruits weighed: one fruit that was open-pollinated ("open" by bees and the site and one fruit that was hand-pollinated ("hand") to ensure complete pollination. The column "Weight_diff" refers to our metric of pollen limitation: (hand-open)/hand
Code/software
All analyses were done with R version 4.4.2 (R Core Team, 2024), and checked for violations of model assumptions (package: DHARMa; Hartig 2024). Scripts are not included here. To test if bumble bee visitation rates differed from visitation rates by all other bees to strawberry flowers we used a linear mixed effect model (package: lme4; Bates et al., 2015) to assess if the average number of bees recorded was predicted by type of bee (bumble bee vs other bees) with field site as a random factor.
1. Does managed bumble bee visitation rate to strawberry depend on surrounding land cover?
We evaluated if landscape composition surrounding each field influenced bumble bee visitation to strawberry flowers by running three individual linear models, one for each land cover category (agriculture, pasture, and semi-natural) as a predictor with the transformed (log + 1) average number of bumble bee visits as our response variable. We also evaluated if field size influenced bumble bee visitation with linear models. The average number of bumble bee visits was log + 1 transformed in all analyses to aid in model fit. We estimated predictor significance with ANOVA likelihood ratio tests. We use the landscape 1000 m surrounding the fields because this scale has previously been shown to be most relevant to bees foraging on strawberry in this region (Connelly et al. 2015) and relevant to bumble bee foraging distances (mean: 842 m (McArt et al., 2017b)), but also evaluated the 750 m and 500 m scales (Table 1).
2. Do fields with greater visitation by managed bumble bees have reduced pollen limitation?
To then assess if fields with greater observed bee visitation had reduced pollen limitation we ran a linear mixed effects model (package: lme4; Bates et al., 2015) with the difference between the open and hand-pollinated fruit divided by the hand-pollinated fruit as the response and the transformed (log + 1) average number of bumble bee visits as the predictor, we also ran these models with average number of other bee visits (log + 1) and with average total number of bee visits (log + 1) as predictors to understand the role of increased visitation by all bees. Last, we ran a model with field size as a predictor to understand the role that the amount of cultivated strawberry around the colonies had on pollen limitation. We included site as a random factor to account for the multiple plants from each field. We estimated predictor significance with ANOVA likelihood ratio tests.
3. Is visitation rate by managed bumble bees reflected in colony Rosaceae pollen collection?
To understand the relationship between observed bumble bee visitation at our focal crop and the proportion of rosaceous pollen we recovered from the colonies we ran a beta regression model (package: betareg; Cribari-Neto & Zeileis, 2012) with the transformed (log + 1) average number of bumble bee visits as the predictor and the proportion of rosaceous pollen as the response. We estimated predictor significance for the beta regression with a likelihood ratio test (package: lmtest; Zeileis & Hothorn, 2002).
4. Is pollen collection by managed bumble bees dependent on landscape context?
We then evaluated if landscape composition surrounding each field influenced the proportion of rosaceous pollen from the colony by running three models (package: betareg; Cribari-Neto & Zeileis, 2012), one for each land cover category (agriculture, pasture, and semi-natural) as a predictor with the proportion rosaceous pollen as the response variable. We also ran a model with field size as a predictor to understand the role that the amount of cultivated strawberry around the colonies had on rosaecous pollen collection. As above, we estimated predictor significance with likelihood ratio tests (package: lmtest; Zeileis & Hothorn, 2002).
Access information
Other publicly accessible locations of the data:
- NA
Data was derived from the following sources:
- NA