Data from: Bumblebee queens are better at olfactory learning and more sensitive to scents than workers
Data files
Mar 23, 2026 version files 360.31 KB
-
Analyses_and_figures_dryad.R
49.08 KB
-
Fieldwork_discrimination_trials_manuscript_data.csv
63.89 KB
-
Fieldwork_learning_trials_manuscript_data.csv
97.04 KB
-
README.md
7.59 KB
-
Y-maze_manuscript_data.csv
142.71 KB
Abstract
Intra-specific variation in cognition can reflect ecological demands, yet cognitive performance can be driven by non-cognitive processes. Bumblebees have been a long-standing model in the study of cognition for their aptitude at learning associations, but nearly all research has been limited to one life stage: foraging workers. Queen bumblebees also forage, and previous work shows that they learn visual associations with fewer errors than workers, offering a useful comparison of the drivers of variation in cognition. Here we tested whether queens’ better learning performance holds across another modality: olfaction, and whether performance could be driven by non-cognitive variables related to their larger size, including peripheral sensitivity. We first assessed olfactory learning in wild Bombus vosnesenskii and found that queens learned with fewer errors than workers. Following additional training and sensitivity tests, queens and workers did not differ, ruling out non-cognitive variables as drivers of learning differences. In a second experiment, we tested scent sensitivity under more controlled conditions in B. impatiens and found that queens were more sensitive. Our findings confirm learning differences between bumblebee castes and show that queens also have greater peripheral sensitivity, hinting at adaptive drivers of these intra-specific differences.
Dataset DOI: 10.5061/dryad.wdbrv1649
Description of the data and file structure
Experiment 1 (field-based) quantified olfactory learning and discrimination sensitivity in wild-caught Bombus vosnesenskii queens (n = 48) and workers (n = 44) tested immediately after collection at desert and montane meadow sites in Nevada and California during their respective foraging seasons (May–July 2024). Bees were captured while foraging and individually housed in ventilated plastic chambers under ambient field conditions. Using a modified Free-Moving Proboscis Extension Response (FMPER) protocol, each individual completed three sequential phases within a single testing session: (1) initial learning (5 training trials followed by 6 unrewarded probe trials at 1:100 scent dilution), (2) additional training to standardize learning (4 training trials followed by a second learning test), and (3) sensitivity testing across decreasing scent concentrations (1:10,000; 1:100,000; 1:1,000,000). Two floral odorants (geraniol and linalool) were counterbalanced as conditioned stimuli (CS+ and CS−). Choice was recorded as a binary response (correct/incorrect) based on the bee’s first proboscis extension or approach. Worker morphology (intertegular distance and flagellum length) was measured as proxies for body size.
Experiment 2 (laboratory-based) measured olfactory detection thresholds under controlled indoor conditions using commercially obtained Bombus impatiens workers (n = 31) and gynes (n = 27) from three colonies. Bees were tested in an acrylic Y-tube olfactometer. Each individual completed three phases within ~2 hours: (1) shaping to explore the apparatus, (2) training to criterion (8 trials pairing linalool at 1:100 dilution with sucrose reward), and (3) sensitivity testing across seven decreasing concentrations (1:1,000 to 1:500,000 dilution). Arm choice (linalool vs. mineral oil control) was recorded as a binary response. Sucrose was provided between sensitivity trials to maintain motivation. Body size measurements (intertegular distance and flagellum length) were obtained for all individuals.
Across both experiments, the dataset includes individual-level binary choice responses across learning and sensitivity trials, associated scent identities and concentrations, bee type (queen/worker/gyne), and morphological measurements where applicable. Each bee was tested in a single session.
Files and variables
File: Analyses_and_figures_dryad.R
Description: see Software section below.
File: Fieldwork_discrimination_trials_manuscript_data.csv
Description: Data from discrimination/sensitivity trials from Experiment 1 (collected in the field)
Variables
- bee_no: individual ID
- date: date of testing
- species: bumblebee species
- type: bee type (queen vs worker)
- caught_on: flower type bee was collected on
- pollen: whether or not individuals had pollen loads
- location: description of testing location
- site_type: meadow vs desert
- GPS: coordinates of site
- time_caught: time individual was initially caught (military time)
- scent: CS+ scent
- time_between_catching_and_testing: hours and minutes between catching and testing; NA = time not recorded
- disc_test: trial number for discrimination/sensitivity test trials
- concentration: scent concentration for both linalool and geraniol during discrimination trials
- response: binary choice during discrimination test: correct = 1; incorrect = 0; NA = did not participate
- avg_IT_mm: average intertegular distance; NA = no measurement taken
- flagellum_mm: flagellum length; NA = no measurement taken
- proportion_correct_disc: proportion of correct choices during discrimination tests; NA = did not participate
File: Fieldwork_learning_trials_manuscript_data.csv
Description: Data from learning trials from Experiment 1 (collected in the field)
Variables
- bee_no: individual ID
- date: date of testing
- species: bumblebee species
- type: bee type (queen vs worker)
- caught_on: flower type bee was collected on
- pollen: whether or not individuals had pollen loads
- location: description of testing location
- site_type: meadow vs desert
- GPS: coordinates of site
- time_caught: time individual was initially caught (military time)
- scent: CS+ scent
- time_between_catching_and_testing: hours and minutes between catching and testing; NA = not recorded
- first_choice_correct: whether or not the individuals' response during the first test trial was correct (correct = 1; incorrect = 0)
- test: test trial number 1-6
- response: whether or not the individual chose the CS+ scent first (correct = 1; incorrect = 0; NA = did not participate)
- no_correct: number of correct choices during testing
- no_incorrect: number of incorrect choices during testing
- total_choices: total number of test trials and individual participated in
- proportion_correct: proportion of correct test trials
- avg_IT_mm: average intertegular distance; NA = no measurement taken
- flagellum_mm: flagellum length; NA = no measurement taken
File: Y-maze_manuscript_data.csv
Description: Data from sensitivity trials conducted in the lab using a y-tube/maze apparatus
Variables
- Bee_ID: individual ID
- Bee_type: caste (gyne vs worker)
- Date: date of testing
- Capture_time: time captured from colony
- Colony: natal colony ID
- Shaping_start: start time for shaping phase
- Shaping_end: end time for shaping phase
- Training_start: start time for training phase
- Training_end: end time for training phase
- Sensitivity_start: start time for sensitivity trials
- Sensitivity_end: end time for sensitivity trials
- Trial_number: trial number
- Trial_type: training vs sensitivity trials
- Linalool_concentration: concentration of linalool
- Scent_side: side linalool was applied
- Choice_side: side the individual chose first
- Correct: whether or not the individual chose the linalool side first (correct = 1; incorrect = 0; NA = did not participate)
- Head_turns: number of times the bee turned their head between sides during sensitivity trials; NA = not recorded or not relevant to trial type
- Choice_latency: time in seconds between start of sensitivity trial and choice; NA = not recorded or not relevant to trial type
- Proportion_correct_training: proportion of correct choices during training trials
- Proportion_correct_sensitivity: proportion of correct choices during sensitivity trials
- Avg_IT_mm: average intertegular distance; NA = no measurement taken
- Flagellum_mm: flagellum length; NA = no measurement taken
- R_L_Flagellum: right or left flagellum measured; NA = no measurement taken
Code/software
All analyses were conducted in R version 4.4.2 (R Core Team, 2024), a free and open-source statistical computing environment available from CRAN (https://cran.r-project.org).
The included script (Analyses_and_figures_dryad.R) reproduces all data processing, statistical analyses, and figures. The script requires the following CRAN packages: glmmTMB, lme4, car, emmeans, effects, MuMIn, performance, DHARMa, dplyr, tidyr, and ggplot2.
To reproduce the analyses, users should place the CSV data files in the working directory, install the required packages, and run the script from top to bottom. No proprietary software is required.
