The scent of the stop signal: how honeybees learn to be picky eaters

Methods

Summary

The premise of my research is to apply the methodology utilized by Thom, et al. (2007) to identify the chemicals emitted by a stop signaling honeybee rather than a waggle dancing honeybee. 

To do this, I must collect air samples over individual honeybees that are stop signaling and compare the chemical composition of these samples to that over individuals that are waggle dancing or are foragers, but not dancing. 

Stop signaling occurs when a honeybee forager is recruited to a food site, but finds that it is potentially dangerous. Therefore, I must first train individual foragers to a specific food site at which I can manipulate their behavior. I made an artificial "feeder" and followed standard protocol for training honeybees to an artificial feeder. 

First, I placed it directly outside of the hive. Once a few individuals have found and are feeding from the feeder, I mark their thoraxes and abdomens using paint markers to identify individual honeybees. I then move the feeder several feet back away from the hive at a time until at least fifteen to twenty unique foragers have been marked.

Here are pictures of the artificial feeder apparatus (with enamel paint which I used over the summer and found to be ineffective) and honeybees being marked. The blue bowl is used to provide a bright and distinct color to make it easier for the honeybees to find the feeder. The feeders in the pictures had more grooves added around the plate at a later time than these pictures, allowing the honeybees to feed on individual grooves rather than where the jar meets the plate.

 

Once the honeybees have been trained and are recruiting other foragers to the feeder, I must elicit stop signals. To do this, I simulate predation by pinching a honeybee that is currently feeding on one of its middle legs, closer to the body. I pinch firmly such that the honeybee cannot freely escape and thinks that it is being attacked, but not too hard, as that would injure it. I pinch honeybees that are marked so that I am able to find them again in the hive and observe its stop signal behavior. 

I use a double-sided Plexiglas frame observation hive that is easily opened for direct observation and air sample collection:

I collect air samples using solid phase microextraction (SPME) fibers that, when exposed to the air, absorb the chemicals that are in the air. I hold the exposed SPME fiber over a stop signaling bee and track its movement for five minutes. 

Here is footage of a stop signal in action (credit to Professor James Nieh, UC San Diego). Notice the honeybee following the marked forager and actually landing on top of her at the end:

Once I have collected all of my air samples for a trial (this is limited by the number of SPME fibers I have,) I head to a gas chromatography/mass spectrometry (GCMS) instrument as soon as possible. Here is what the instrument looks like (but with an updated computer):

I manually inject the SPME fiber into the GCMS instrument and utilize TurboMass to draw a chromatogram that identifies what chemicals are removed from the SPME fiber, in which order, and what duration it took to remove. Using the latest National Institute of Standards and Technology (NIST) database that is programmed into Turbomass, I identify which chemicals are associated with the peaks of the chromatogram. I then compare which chemicals are in which kind of sample (normal/waggle dancing/stop signaling).

Here is an example video of manually collecting and analyzing samples with SPME fibers (credit to Sigma-aldrich):

Challenges

There is an implicit safety risk involved with working with honeybees, especially when I am literally harassing individuals by simulating predation and placing my hands in an open hive. When I train foragers to a feeder, the final location for the feeder tends to be at least 50 meters away from the hive. This makes the feeder somewhat far away from the hive, where honeybees tend to be the most aggressive. Additionally, I pinch each individual while they are foraging, during which they tend to be docile (painting them typically elicits no reaction unless it was recently rainy). Still, I will wear nitrile gloves while pinching the bees. It's unclear why, but nitrile gloves seem to dissuade honeybees from crawling on hands. I use nitrile gloves instead of beekeeper gloves while working in the hive for more dexterity and have had very few honeybees crawl on my gloves and have not been stung around my hands (yet). While working in the hive, I don a bee veil and suit or long sleeved and collared shirt. This is typically sufficient for keeping them outside of my clothing. Although I interact with them, I have not yet been attacked when I am not directly harassing or capturing any individuals in the hive (or interacting with the queen). 

One other potential issue is observational bias and the existence of the begging call. A begging call is visually similar to a stop signal; the sender comes into physical contact with a waggle dancer, but there is food exchange between the two individuals. It can be easy to miss this food exchange and mistake a begging call for a stop signal because I expect it to the latter. To combat this, I will record the dance floor while I collect air samples and analyze my footage to be sure if I was observing stop signals. To do so, I must match up the characteristic headbutting motion that conveys a stop signal and a "beep" sound that lasts 150 ms and is around 380 Hz.

Pre Analysis Plan

I hypothesize that the chemical composition of air samples collected over waggle dancing and stop signaling honeybees will be distinctively different. Furthermore, there will be less of the chemicals emitted by waggle dancing bees after being exposed to the stop signal. 

I will be using the NIST database that is programmed into TurboMass to find the identities of each compound that define the peaks of the chromatograms for each air sample. However, searching the NIST database for a compound based on the retention time and mass spectrometry of each peak does not yield only one possible compound, but multiple (in descending order of likelihood,). Therefore, it is possible for their to be variation in which compounds are most represented as what make up the composition of the air sample. I believe that with more samples, I will be more confident in the the compounds that are suggested a majority of the time.

To confirm my results, I will conduct an experiment in which I expose the observation hive with a gaseous solution made with the compounds suggested by NIST to make up a stop signal. I will observe the number of waggle dances occurring before and after exposure to the solution, and expect a significant decrease in waggle dancing. I can also potentially observe changes in recruitment to an artificial feeder before and after exposure to the stop signal-like solution. If I do observe decreases in waggle dancing and recruitment, then the compounds that make up the solution also make up the stop signal.

Although this should stop recruitment to foraging locations on the dance floor, this should not be dangerous to the health of the hive. This is because of two reasons: not all foragers will be on the dance floor at the hive at the same time; there will be many out feeding and will return to provide additional recruitment that isn't hindered by the stop signal. Additionally, because the stop signal is a highly volatile signal, it will not last for long before recruitment starts again.