We Make Trees… For Ants

Trees come in many shapes, sizes, thicknesses, porousness, etc. There are a lot of parameters for any given tree, so tree structures become quite complicated as they get large. There’s a reason why trees grow slowly! My labmate and I don’t have a growing season to cultivate a tree, so we need to make a tree ourselves.

So, why do we need to make a tree in the first place? Well, the answer lies in what we study. We are studying how Cephalotes varians, i.e. turtle ants, form their transportation networks. Unlike ants that live on the ground which can form networks by simply walking somewhere, turtle ants form networks in tree branches (they are arboreal ants, meaning that they live in trees). Turtle ant networks are highly constrained by the tree network that they live in, which means that having a tree structure to explore is crucial for observing network formation.

[1] A mangrove tree, home to wild turtle ants in the Florida Keys

We will need a planar, branch-like structure that is easily observable by eye and camera. But it can’t just be a flat cutout; for the ants to like our tree, we need round branches to more closely mimic a natural tree branch’s curvature and a friendly texture that they like to walk on. By mimicking their natural habitat as much as we reasonably can, we hope to observe natural ant decisions.

[2] A snapshot of James’ junction model, which we build off of

Past bee lab work can help us out. Our tree builds upon the work of James Clinton, a student in the HMC Bee Lab who designed parameterizable junctions for a 3d tree structure. It would be ideal to use this system to build our tree, but this system has fatal flaws. While its 3D-printed junctions would ensure a consistent product, the work required to connect each junction with the plastic tubes that serve as branches would be very tedious, and printing each junction needed would take too long. Harnessing the precision of 3D printing is crucial to consistency, but reducing the wait time by reducing printing amounts makes our work faster.

Additionally, this system has many supports and abrupt transitions between junctions and branches. Avoiding both of these issues would help to improve the quality of our tree, as smoothness and strength are key to mimicking a tree branch.

[3] An egg carton, inspiration for our moldable tree concept

Looking into ways to adapt mass-produced products with similar material properties to our lab-scale experiment would help reduce the time needed for fabrication. But many of these products, like paper egg cartons and various plastics, are spray or injection-molded in industrial settings. Adapting a molding process for paper to the lab scale would unlock the consistency of molding and make a higher-quality tree. To attempt this, we design a mold to dry paper pulp into a rigid structure.

[4] An early mold negative prototype

Adapting the 3d models of James’ junctions into a negative for our molds is the first step. Adding drafts, which connects the junction form to a baseplate, allows the curved structure to be intact on one side of the molded part while also allowing the 3d printed mold negative to be pulled out of the molds that we cast with it. To limit the amount of 3d printed parts, we can reduce the number of junction types to a few standardized types and then cast several quick-curing molds that are modularly assembled into a tree structure.

[5] Wet pulp in a plaster mold to test modularity of plaster molds and [6] dry paper tree taken out of the mold

While the paper pulp concept seemed feasible in a draft phase, our first prototypes were difficult to unmold and were very weak. Additionally, the plaster molds were brittle and had many bubbles. The low quality of our early tests indicated the need to try new materials.

Moving forwards, we want a stronger paper pulp product and a flexible mold. Using silicone instead of plaster allows us to demold the trees immediately after casting, allowing them to air dry quickly and stay intact. Different types of reinforced paper pulp were tested with this system to see if any were strong enough.

[7] Paper branches cast in a silicone mold and [8] an early resin test

The flexible molds make casting much easier, but no paper products tested are strong enough for our liking. They also take a long time to dry, making the process slow and tedious. Using a fast-curing resin in the flexible silicone molds instead of paper provides us with more strength and makes molding easier since it is pourable.

Now that we have the material and construction process down, we can apply them to the decided-upon branching structure. Our turtle ants often encounter new branching networks via leaves touching or connections between branches of different trees, as opposed to climbing up the trunk of a new tree each time, so we are having them enter midway along the branch itself (marked with a pink arrow) from the effective trunk (marked in blue). 

[9] An early representation of our tree structure

Utilizing 8 standardized junction types, we devised a structure that has nests and food platforms equally far away from the entry point. This prevents any distance bias from impacting where the ants go. With our modular system ready, we assemble the molds and pour a tree to mirror this structure design.

[10] A whole tree’s mold and [11] resin poured into the tree mold

Despite its leakages, our modular design is very advantageous because of how we are able to assemble the molds into any possible tree structure. With the structure above, if a change is required, we simply replace one junction brick with another and repour.

[12] Final tree structure in the arena

So, after all of the trials and errors, the failures and slow successes, we have learned how to make a tree branch suitable for turtle ants. With its ant-friendly design, we hope that our tree can extract the most natural exploration patterns possible from our ants. We just hope that they feel as comfortable exploring our beautiful tree as they do in the wild!

Images:
[1] Photo by Andrew Tappert. https://en.wikipedia.org/wiki/File:Red_mangrove-everglades_natl_park.jpg.
[3] Public domain image. https://commons.wikimedia.org/wiki/File:Eierdoosmet10eierengevuld2010.jpg.
All other images by the author

Further Reading:

Chang, Joanna, Scott Powell, Elva J. H. Robinson, and Matina C. Donaldson-Matasci. 2021. “Nest choice in arboreal ants is an emergent consequence of network creation under spatial constraints.” Swarm Intelligence 15 (April): 7-30. https://link.springer.com/article/10.1007/s11721-021-00187-5.

Latty, Tanya. “Nature’s traffic engineers have come up with many simple but effective solutions.” The Conversation. 6 June, 2018. https://theconversation.com/natures-traffic-engineers-have-come-up-with-many-simple-but-effective-solutions-94818.