An Overview of Field & Laboratory Tests Conducted on Bees & Their Outcomes
Broadly, experiments with bees as subjects are for assessing learning & memory, decision making, orientation and navigation, social behaviour, social communication, stress and motivation, motor-sensory coordination tasks, amongst others.
This article will give an overview of the experiments used in honey bee research beyond the more-general molecular methods like PCR, imaging & -omic based studies. First, a definition: behavioural and cognitive assays refers to tests that use worker bees to measure learning, memory, decision-making, communication, orientation etc… Experiments can be held under controlled conditions (“laboratory settings”) or out in the field (“field-realistic settings”) and we begin with a laboratory test, the probiscis extension response:
Proboscis Extension Reflex Assesses Associative Learning and Memory
The proboscis extension reflex (or test; “PER”) starts with restraining workers in small harnesses that allow only head movement. When their antennae are touched1 with a sucrose solution, the bee reflexively extends her proboscis towards it (thus serving as the “unconditioned stimulus”). The experiment is then repeated with a stimulus (“the conditioned stimulus”, a neutral odour like linalool or 1-hexanol).
Suppose you want to test the memory formation ability of hives that were used to pollinate rapeseed fields known to contain the neonicotinoid clothianidin. An image of the experimental setup is shown in the image above. The steps are:
- Collect foragers from the test hive’s entrance and chill them to ~6°C to reduce stress. Any injured or spontaneously proboscis-ing bees are excluded
- Also collect foragers from a control hive – one that is known not to forage on plants containing the pesticide
- Place each bee in a test tube containing the harness and leave her head free to move; ensure she can’t crawl back into the tube
- Feed the bees from the sucrose solution to sate them (you don’t want hunger skewing your results!)
- Calibrate the bees by exposing them to clean air and then deploy the conditioned stimulus: stimulating the antennae with a neutral gas like linalool (a fragrant floral compound)
- Any bee that extends her proboscis during the calibration phase should be excluded, as she could have reacted to the linalool, or she could be one that enjoys poking her tongue at things
Now that the baseline is established, it’s time for testing:
- Start clean airflow
- At timepoint 0 (t= 0s from here on): present CS (odour) for 4-6 s
- At t=3s: touch one antenna with 30% sucrose (unconditioned stimulus) to trigger the probiscis extension response, then deliver a droplet of sucrose to the extended proboscis as a reward
- End CS and US together by t=6s
- Record PER occurrence to the odour prior to the sucrose-only contact: did the bee anticipate/extend her probiscis to the odour alone (i.e. before t=3s) or after sucrose was provided?
There is a wait time of 8-10 minutes between trials to allow for learning and for reducing the risk of sensitising the bees to the stimulus; on average a trial is repeated five times.
Responses to the Conditioned Stimulus in the Absence of a Reward Suggests Memory & Learning
To test if the bee was able to form short term memories, a longer wait time (~15 minutes) is held after the last trial. Then, linalool is readministered in the absence of the sugary reward. If the bee sticks her probiscis, that’s a pass, as it means she’s associated the odour with a sugary reward.
For mid- to long-term memories, the bees are removed from their harnesses and housed in a warm & humid box. Small sugar droplets are given every few hours to control hunger and maintain satiety. When it is time for mid-term memory testing (~1-3 hour mark), the bees are reseated in the harness and the CS is reintroduced. A PER indicates that they have formed mid-term memories. The process is repeated for long term memories (~24hrs after the last trial), and if they stick their tongues out again, then- you guessed it – they’ve formed long term memories.
Once the results are logged, the bees are either released (or euthanised), and data analysis begins. If the neonicotinoid treatment group showed a decrease in the number of bees with memory formation and the results were deemed Statistically Significant™, then followup experiments may be conducted such as tissue sampling and brain scans.
Limitation of PER Testing
The first drawback in PER assays is in the methodology: If a bee had extended her probiscis in the absence of stimulus (during the conditioning stage) and was not removed, her data should not be used in the analysis, for she either has a tendency to randomly poke out her tongue, or had detected something in the airflow. The airflow itself may be unclean and contaminated from improper storage or reusage from other experiments – check the equipment and perhaps run air-only trials to see if there’s something in the air.
The artificial restraining of the worker restricts sensory inputs- in nature bees use touch, sight, sounds and motor feedback (amongst others) to guide behaviour; perhaps this sensory-restricted state may engage neural circuits differently than in free-flyers who are exposed to visual stimulus, pheromones and other abiotic feedbacks. Lastly, the binary yes/no nature of the PER does not account for other sublethal effects from pesticides like abnormal waggle dancing (Zhang et al., 2020), or a higher chance of getting lost while foraging (Gill et al., 2012).
Lastly, and perhaps most importantly: it does not reflect colony-level interactions. We can extrapolate that if individual foragers are less sensitive to sucrose, then they will be less successful nectar foragers. It becomes more difficult to apply this PER result in a real-world context, as the colony-level feedback may adapt and adjust to these changes, and since their pollen, water & propolis gathering abilities are potentially unimpacted, the hive can survive and overcome the pesticide toxicity as they move onto other foraging sources.
Free-Flying Assays: Visual Learning, Navigation, Olfactory Learning & More
It’s no surprise that experiments on a freely-flying bee have greater applicability- the bee is bound within a (large) area to fly and choose as if she were on the field, or sometimes she is truly free and inside the hive. We begin with visual learning testing, which measures choices in response to a stimulus:
Visual Learning Tests: Mazes & Flower Patch Arenas
Y-Maze Paradigm is Used to Evaluate Between Two Choices
To test whether bees can learn that one visual cue holds a reward and another does not (e.g. blue vs red flowers), a Y-maze (or T-maze) test is conducted. The experimental setup involves placing the bee at the foot of the Y, with each “arm” holding a visual cue (shapes, colours, obstructions etc…). One arm contains a sugar feeder, whereas the other holds water or a distasteful compound (quinine).
The bees are released one at a time into the maze to make their choice and return to the hive after consuming the syrup. After a while (as above, breaks are required for memory formation [and not just for bees!]), the bee is reintroduced to the maze for another run. Conclusions can be drawn after a dozen or more repeats, such as:
- The accuracy of each bee in selecting the visual cue containing the sugar feeder
- How fast she learns to associate a cue with a reward
- The lag time between each arm entry – how long did bees spend examining their surroundings before making a choice, and did the time with repetition?
- If the cues were swapped (i.e. the cue that held the sugar is swapped with the neutral/negative reward), how many incorrect attempts did the bee make before un-learning or re-learning that the other cue may hold something that is not negative?
The Flower Patch Paradigm Tests Decision Making in the Presence of Multiple Simultaneous Stimuli
In nature, the forager has to determine which of the dozens of flowers contain the most nectar and/or pollen to maximise foraging & energy efficiency. The flower patch test attempts to replicate this using a grid that contains a set number of plastic discs (~3cm in diameter) each bearing a different cue (patterns, colours, shapes, size etc…). Identical small wells are used for rewards or “punishment” (sugar & quinine respectively). The reward-punishment flowers are placed randomly to avoid bees favouring one area.
The bee is released and monitored to assess her learning. Like the Y-maze above, the tests determine her accuracy, the speed at which she learns, the lag times & how soon she switches-to-avoid-errors. A useful assay that replicates the ecological complexity encountered by a forager in the field.
Olfactory Learning is Done at Feeder Stations
A step up from the proboscis extension reflex is using feeding containing either an unconditioned stimulus (water/quinine) or a conditioned one (sucrose). The two feeders are identical in appearance, however each will contain a neutral odour (linalool or 1-hexanol). Bees are then released to feed, where the researchers can measure:
- The time taken to reach the sucrose feeder
- How many bees visit the correct feeder on their first trial
- How many get it wrong
- Whether the bees have learnt to associate the odours with the reward/punishment (assessable when experiment is repeated after hours or days)
As in the maze above, the researchers can also switch the odours so that the punishment feeder now smells like the reward and vice versa to determine how fast the foragers rewire themselves to this reversal. Experimental variations include swapping feeder locations or changing feeder shapes/colour/height to test different aspects of bee cognition, depending on your goal.
The experiment can also measure colony-level memory & foraging bias to certain odours. Here, the colony is exposed a scented sucrose feeder for a few days, and then that feeder is removed. Foragers are taken from the hive and tested in Y-mazes or similar flying assays to determine if they’ll favour stimulus containing the tested odour. How fast they do this, along with the frequency they favour the stimulus is recorded as a sign of colony-level memory2.
and 2.0 M (red). Many bees can be seen feeding at the 2.0 M feeder, about one third as many at the 1.5 M feeder, one at the 1.0 M feeder, and none at the three lower concentrations.”
Orientation, Navigation & Spatial Memory Assays for Landmark Learning, Map Integration & Compass Orientation
The bees’ orientation flights can be manipulated for the purposes of determining how naïve workers learn the local landmarks and establish their frames of reference. One approach involves marking newly emerged bees/young nurses and surrounding the area with artificial landmarks (pots, shapes, poles, panels etc…). Video or radar records their flight path and how their flight radius grows over successive flights, whether they favoured one landmark over the other, as well as the time it took to complete their flight.
Orientation flights can also be used to assess memory & the bees’ map of the terrain. After completing the flights, these bees are trained to visit a feeder at a known distance/direction from the hive, such as 200 metres east. A few days later, these bees are captured and released at a new location, such as 400 metres north, and are tracked to see:
- The problem-solving and path selection involved to align herself with the terrain
- The distance to the hive prior to flight path correction
- Time required to return to the hive
The type of flight tells you the types of memory involved. If the bee flew in a “beeline” (straight line) back to the hive, then she had vector memory. If it was more meandering with turns and adjustments, then her memory was more map-like, using landmarks to guide her way back to the hive.
Tunnel and Map Learning Surveys
Tunnel surveys measure how bees estimate speed and distance and fly through the terrain. One test involves having the bee fly through a short, narrow tunnel lined with stripes to reach a sugary reward, and then analysing the subsequent waggle dance to her nestmates. Variations in the tunnel architecture (vertical vs horizontal stripes, frequency of shapes/stripes etc…) can be implemented based on the parameter being measured. We know that honey bee odometry (measure of speed & positioning) is driven by a vision-based cue (reviewed in Srinivasan, 2021), making such assay useful in investigating the modifications to their perception of distance, flight speed and/or vision.
A map-learning survey involves training foragers to visit a feeder some short distance (~400m) away from the hive, tagging her with paint or RFIDs and then measuring how she adjusts her routes and flight times as the feeder is moved on the daily. The results will tell you the rate of persistence to the old feeder location, how quickly a forager can relocate the feeder, as well as how quickly she corrects her waggle dances to this new location. It can also be a measure of problem solving: does the forager learn that the feeder is moving, and does she optimise her flight paths at the start of a new day to find it faster?
Social Communication Assays Examine Waggle Dances & Trophallaxis to Assess Colony-Level Information Flow
Waggle dances as a source of information have been touched on above, here it is examined directly to see how a stimulant impacts communication accuracy.
One experiment involves the setup of multiple artificial feeders with different sucrose concentrations at equal distance to the hive and then counting the number of waggle dances in favour of each feeder. “But Mister HolyHabibee Entomologically,” you say, “we already know the bees will favour the strongest concentration over the others!” and you are right; this is a given. The experiment can be modified by including an adulterant in the mix (pesticides), or by pairing it with an odour, as shown above (and subsequently reallocating the odour to a different sucrose concentration to determine if the bees were attracted to the feeder because of the odour or because of the higher sucrose concentration).
Something different here are the trophallaxis assay, whereby a fluorescently-labelled feed (or pathogen-associated molecule) is given to a forager and then tracking how it travels inside the hive via the mouth-to-mouth feeding of trophallaxis. Based on your aims, a positive result can either be how fast or how slow it travels. For example, if the compound is a foodstuff, a fast spread would indicate a robust colony that is quick to share information. If the compound was a pesticide, then a slower travel time is ideal, as it indicates a buffering effect whereby bees are slowing its spread in an attempt to neutralise it on their own.
The Stinger Extension Response Assay is Analogous to the Proboscis Extension Reflex
To measure if a stimulant triggers aggressive/defensive behaviours, the stinger extension response assay (SER) is used. The bee is also restrained, this time with the abdomen free rather than the head. The unconditioned stimulus comes as low voltage electric shocks (~2-5 volts) from electrodes placed under her abdomen; the conditioned stimulus remains as the odour.
Instead of repeating the PER methodology, I encourage you to recall/re-read the above and this time replace the sugary reward with electric shocks. The results are likewise similar, measuring the learning rate, the time from odour onset to stinger extension and memory formation at short/mid/long term intervals. The sooner she extends her stinger, the faster the memory formation.
Stings at a Colony Level: Free Moving Target Dummies Coated In Alarm Pheromones
The SER is a test for individual bees. At a colony level, a guarding assay is used whereby a black leather ball or a small dummy (~20cm diameter) is suspended near the hive entrance. This dummy is coated in isopentyl acetate, which is a major component of the alarm pheromone that triggers guarding behaviour. The researcher measures the time to first sting, the intensity of the stings per minute as well as the recruitment of bees to sting the ball and how this changes over time. Useful to assess the effects of pesticides on colonies, or on subspecies comparisons (e.g. the sting response of A.m. scutellata [high] compared to that of A.m. ligustica [lower]).
Other Assays Not Mentioned Above
Beyond the behavioural assays described above, there are also tests on social immunity (e.g. pin- or freeze-killed brood removal), variations on probiscis/stinger extension reflexes, and other free-flying assays. I’d like to think I’ve described the more common experiments done on bees, and the interested reader is directed to the literature for further reading. Bee tissue & bodies are used in more general scientific procedures, as in molecular biology (ELISAs/Microscopies etc…), as well as whole range of -omics: metabolomics, lipidomics, proteomics, glycomics, genomics, epigenomics, toxicogenomics, youshouldgogymomics, pharmacogenomics, monsantosponsoredglyphosateomics, microbiomics… you get the point.
Further reading: Select Papers Who Used These Assays
- For an overview on the proboscis extension reflex there’s Nest, 2007, and its drawbacks: Frost et al., 2012
- Overview of the stinger extension reflex: Vergoz et al., 2007
- Defensive behaviour assays: Guzman-Novoa et al., 2003
- Free flying assays:
- Artificial flowers: Howard et al., 2017
- Navigation and tunnel assays, amongst other cognitive abilities: Srinivasan, 2021
- Foraging & feeder-learning assays: Garbuzov et al., 2020
- Orientation flight tracking: Degen et al., 2016
Read the whole post and found puns, typos mistakes? Voice your displeasure with a strongly-worded message to admin@tgbee.com.au, or by filling this form.
Footnotes
- How the sucrose is delivered is up to the researcher- you can use a toothpick, a length of wire, a glass capillary, a water-repellent surface… anything, so long as it is sterile to prevent contamination. ↩︎
- It’s a sign of colony level memory because during the feeding trial, many foragers would have waggle danced to communicate that the odour-laced feeder is a rich source of sugars, and since bees want what is most energetically efficient, they will associate that odour with an excellent source of forage, thus favouring the scent. ↩︎