Formation and mechanics of fire ant rafts as an active self-healing membrane

The unique ability of fire ants to form a raft to survive flooding rain has enchanted biologists as well as researchers in other disciplines. It was established during the last decade that a three-dimensional aggregation of fire ants exhibits viscoelasticity with respect to external compression and shearing among numerous unusual mechanical properties. Continuing these works, we will study the ant raft in its natural form, i.e., composing no more than two layers. This allowed us to focus on the cracks that are unique to membranes and see how their patterns are influenced by the fact that these ants are mobile and can self-repair the damage to keep their raft from disintegration. In the beginning, we show that vertical and horizontal shaking can also prompt fire ants to aggregate. The canonical view that the stability of ant raft relies on the Cheerios effect and a combination of other parameters is tested. The force-displacement experiment is performed to show that two distinct mechanical responses and fracture patterns, characteristic of ductile and brittle materials, can be elicited, depending on the magnitude of the pull speed. During the process, we counted the number of ants that actively participated in the stress-strain relation and used this information to roughly sketch out the force chain. The latter information reveals that the pull force expedites the alignment of fire ants, in analogy to the effect of an electric field on liquid crystal polymers. To highlight the self-healing nature, we employ the creep experiment to study how the length and Young's modulus of the raft change or relax with time. One major finding is that the raft can exhibit zero Poisson's ratio without resorting to specific geometry structures. This is enabled by the active recruitment of ants from the top layer to the bottom layer to keep the raft from disintegrating.

Figure 1

(a) Schematic experimental setup for measuring the mechanical and creep properties of fire ant rafts. (b) The displacement of the laser point on the ruler allows us to determine the bending angle of the optical lever and hence the resistance force of ants from this bottom view of (a).

Figure 2

A shaker is employed to test whether flooding is the only means to induce the action of aggregation for fire ants. Note that ants never touch the wall when jiggled horizontally. The container is coated with Teflon (Chemours DISP 30) to prevent fire ants from gripping and escaping.

Figure 3

The pull force versus length of ant raft under fast (1.7 mm/s) and slow (0.25 mm/s) pull speeds. Representative snapshots for each region are shown, where red curves are used to highlight the cracks that developed in the interior and at the edge of the ant raft. In contrast to being stationary in region II, these cracks expand irreversibly upon entering region III. Regions I, II, and III show the elastic, plastic, and rupture behaviors common to ductile materials [48]. The data for high pull speed are characterized as being brittle due to its lack of region II. The larger error bars for slow pulling are likely due to the fact that the ants are mobile, which accumulates more errors when given a longer time duration.

Figure 4

In order to capture how the number of aligned ants $N$ is correlated with the pull force $F$, they are plotted in double $y$-axis graphs as a function of time. Both quantities are rescaled by their maximum value to facilitate comparison. (a) and (b) are respectively for low and high pull speed. Alignment is defined by a deviation less than ${20}^{\circ }$ between the ant orientation and the pull direction. These ants are marked by red lines in the photos, otherwise they are in green.

Figure 5

Visualized by connecting the blue lines as defined in the text, force chains gradually develop from regions I and II in (a) and (b) to III in (c) for low pull speed.

Figure 6

Length $L$ of the ant raft relaxes under a constant $F=2.46$, 13.43, and 8.93 mN for Slow Reg. I, II, and III, and 3.4 mN for Fast Reg. I. The blue dashed line denotes the transition for Slow Reg. II-III data, i.e., for a slow pull speed, and from region II to III, when the ant raft ruptures irreversibly. Likewise, black is for Fast Reg. I-II, and green is for Slow Reg. I-II for holes appearing in the ant raft. The inset shows that the change of morphology for yellow and red lines is accompanied by an approximate saturation of raft area $A$.

Figure 7

(a) $K$ vs $L$ for the same samples in Fig. 6. Except for Slow Reg. II-II, the other four data are fitted respectively by solid ($1/L$) and dashed ($L$) green lines for slow speed, and the same in black for fast speed. The inset shows the log-log plot for Slow and Fast Reg. I-I that can be fit by the linear line. Although roughly observing the increasing trend as Slow Reg. I-II, the data of Slow Reg. II-II fluctuate wildly because we are forced to adopt a smaller $\mathrm{\Delta }F$ to avoid hastening the raft into region III. This increases the uncertainty in $\mathrm{\Delta }L$. Snapshots of Slow and Fast Reg. I-II at $t=1600$ s are shown in (b) where the primary neck, the one with the smallest width $W$, is identified and highlighted by the red rectangular box. (c) The $W$ for the Fast Reg. I-II sample is not only narrower than that of Slow Reg. I-II, but also shrinks faster.

Figure 8

Procedures for tailoring the width $W$ and length $L$ of an ant raft. (a) tTwo rods separated by a distance $L$ with the aid of a ruler underwater. (b) After being gently deposited on water, the ball-shaped ants spread out and grip the rods. (c) A tweezer is used to remove the ants that exceed the designed range of $W$ and $L$ into the eventual raft sample in (d).

Figure 9

To test the effect of physiological change of dead ants to raft formation on water, the ratio of freshly dead ants to those of 1-day mortality is modified for roughly 500 dead fire ants. Their ratios are (a) 1:0, (b) 1:1, (c) 0:1.