Before we get too far into the work I did, it would probably be a good idea to see what a turtle ant looks like. I certainly had never seen a picture of one before starting in the Bee Lab.
[1] A turtle ant, Cephalotes varians. Turtle ants are arboreal, typically living in trees.
Now that we have seen a turtle ant, what makes them important to study? Well, one interesting thing about turtle ants is that they form networks. They create connections between nests (something unique to polydomous ants, where single colonies spread across multiple nests) and from nests to food sources, all of which form one network. The thing is, we don’t know a ton about how these networks are formed. While we do know about how other ant species form networks, we don’t know a lot about turtle ants specifically. Networks are created by choices individual ants make, such as whether an ant chooses to explore one part of a tree or another. These choices can be boiled down to which way an ant turns at an intersection. So, to figure out how turtle ant colonies form networks, it is important to know how individual ants make choices when coming to an intersection.
Normally in the Bee Lab, we would be performing actual experiments on actual ants. For example, last year, students made an experimental setup, and ran an experiment to see how ants form networks between nests when presented with bridges between certain nests. This semester, we couldn’t physically run new experiments, so we worked on analyzing data from previous experiments and designing a new experiment to understand how individual ants make decisions.
[2] The setup for last year’s experiments. It has five towers, each with two nests (the red tubes), and a
central tower connecting the other towers.
central tower connecting the other towers.
This semester, I focused on the engineering side of designing that experiment. However, before any engineering could actually get done, we had to figure out what the experimental setup should actually look like and what the goals should be. The goals were fairly apparent; to figure out what factors influence the choices turtle ants make at intersections. But to actually figure out what the setup should look like, I had to learn about how to test ant behavior. Since I couldn’t really look at any ants to see how they behave, I did the next best thing: I read about them.
First, I read about bifurcation mazes. A bifurcation maze is a maze composed of junctions, connected by paths, each with a single entrance and two exits. This seemed to be the best way to test ant decision making, as it provides a sequence of simple binary data: whether each ant chose to turn left or right at a series of junctions. So, the basic structure of this experiment would be a bifurcation maze.
[3] A fork in the road, a bifurcation junction that is common in everyday life.
One potential variable we discussed testing was whether ants are more likely to choose to go on an uphill path or a downhill one. Another variable is whether ants are more likely to go down a path that continues straight, or one that branches off of a main path. Another option would be to test whether ants will turn left or right when presented with a symmetric bifurcation. We could also test whether ants choose to go on wider paths or narrower paths more frequently.
Now that I had some idea of what a potential experimental setup might look like, and what variables might end up being tested, I had to create an actual design for the structure. First, I had to decide whether the paths between the junctions should be flat, or cylindrical, like actual tree branches. I chose cylindrical branches, to create a more realistic environment. I also ended up designing the structure to be elevated and supported, as turtle ants live in the elevated environment of trees.
So, this experimental setup will be a bifurcation maze that can test how angles of elevation, path angles, and path widths impact ant decisions. It will all be elevated, and the junctions will be connected using cylindrical paths, in order to mimic a natural environment. Now, I had to design the junctions. Since I had to easily be able to change the characteristics of the part based on what we wanted to test, I went with something called parametric CAD.
Normal CAD involves creating models with set dimensions, and when dimensions are altered, things can quickly go wrong. Parametric CAD involves designing a model with changing dimensions in mind. When done correctly, the only thing needed to generate a model with a given set of characteristics is to change certain parameters from a list.
[4] A few of the many parameters that can be tweaked on the model.
Here is the first design I created:
[5] Two versions of the first iteration of the junction, each with different parameters. The cylinders represent a point where a path could be connected (the paths wouldn’t be that small).
Unfortunately, this design had some problems. It had very angled corners and edges, which aren’t often present in trees, which could lead to the ants making strange decisions from a lack of familiarity with the environment. It also had unnatural flat sides, odd transitions between the cylindrical branches and the junction, and was very flat, all things that aren’t present in a turtle ant’s normal environment.
Over the course of a few iterations, I was able to address those problems. Here are my next few iterations of the design:
[6] The second iteration of the junction. Although there is less flat space, there are still sharp corners and edges.
[7] The third iteration of the junction. This iteration was a lot more natural, but it was unable to have a straight path with something branching off.
[8] The fourth iteration of the junction. This iteration fixed the straight and branching off paths issue, but was all parallel to the ground, with no sloping in the junction.
And here’s what I ended up with as a “final” iteration:
[9] The fifth iteration of the junction. This iteration allowed for junctions that slope upward and downward.
With a potential design finished, I could start focusing on manufacturing methods. First, for the paths connecting the junctions, I decided to just go with plastic tubes. They are already round and have holes for attachment, and only need to be cut to the correct length.
Now for the junction, this design is pretty weird to manufacture. Because of the odd angles of elevation and path angles, using typical machining techniques won’t work well. This part is best suited for 3D printing. The only problem is figuring out what material will work out best, and the only way to do that is to actually print some parts. Although I didn’t have access to a machine shop or makerspace, like I would at Harvey Mudd, I happen to have a few 3D printers in my basement! So, I printed a few parts using different materials. I printed a few using PLA, the most common 3D printed plastic, a few PLA parts coated in XTC-3D, a special coating that smooths them out, helping to mimic the smoother branch structure, and a few using PLA infused with wood, which could feel more natural to the ants. Now, those parts are being tested to see which material the ants are most comfortable with.
[10] One model of each of the materials. The left is regular PLA, the middle is PLA coated in XTC-3D, and the right is wood infused PLA (which actually smelled like wood for a while!).
And that’s where I am right now with my work in the Bee Lab. Next semester, I’m hoping to finalize the material for the junctions and branches, actually build the experimental structure, and run experiments. I'm also looking forward to the day that I can finally see a turtle ant in person!
Further Reading:
Czaczkes, T. J. 2018. “Using T- and Y-Mazes in Myrmecology and Elsewhere: A Practical Guide.” Insectes Sociaux 65 (2): 213–24. https://doi.org/10.1007/s00040-018-0621-z.
Media Credits:
[1] Photo by Alex Wild. https://www.alexanderwild.com/Ants/Taxonomic-List-of-Ant-Genera/Cephalotes/i-56hwGvx/A
[2] Photo by Matina Donaldson-Matasci.
[3] Photo by Friends San Jacinto. https://flic.kr/p/7PST73
All other figures created by author
No comments:
Post a Comment