My Colored Sand Box
Seeing below ground
Here’s something everyone knows: a lot of ants live in underground nests that they dig, worker ants making piles of dirt around the nest opening as they carry and deposit little lumps of soil. Everyday stuff… but I wondered, what is the shape of the cavities those ants are digging? It wasn’t that hard to figure out. I simply filled the nest void with casting material (dental plaster) and dug out the hardened cast. I guess I’m simplifying it a bit, but in principle, that’s how it works. Lost wax casting, lost sand casting, pretty much the same, really. We can skip over the reconstruction and the issues of supporting the reconstructed cast, and we see the result of one of my early casts of a harvester ant nest below.
This cast reveals the shape of the void that ants created underground--- large, complex, closely spaced chambers just below the surface, with increasingly simple chambers and greater spacing with depth. All these chambers are connected by four helical shafts down to over 2 meters in depth. But how does the colony, composed of about 5000 workers, each weighing about 4 mg and totaling about 20 g, achieve the astounding feat of moving 20 kg of sand in only 4 to 6 days, and then moving the entire colony into the new nest? Is the process like humans mining, with distinct phases: breaking material into transportable sizes at the mine face; transporting this loose material along the tunnels and shafts; dumping it at the surface? Does the colony have ants that specialize in each of these activities, or is this a catch-as-catch-can with any ant likely to do any part of this? The last seems unlikely, for it has long been known that worker ants are stratified within the nest by their age--- the youngest workers mostly in the bottom of the nest along with the brood (from whom they arose and who they care for), and the oldest workers in the uppermost regions of the nest, from which they depart as foragers during the final two to four weeks of life.
A simple experiment showed that excavating nests is predominately something that older workers do. I separated workers from the bottom, middle, and top of mature nests (and therefore young, mixed ages, and old), and marked them with different colors, then released 400 into each screen-bottom cage through whose bottom they could dig a nest, which they did quite rapidly. After a week or so, I recaptured as many workers as appeared on the surface, then made a plaster cast of the nest they had dug. The old workers dug much larger nests than the young workers, and more of them were recaptured on the surface, mostly while dumping sand from excavation. In mixed groups, old workers were much more likely to appear at the surface dumping sand. But while the labor of excavation falls more upon the narrow shoulders of old workers, the shape of the resulting nest does not differ greatly between these groups. They always have large chambers at the top, tapering strongly with depth.
For sure, this is not much detail, but so far, so good. But how does all this play out underground, a region we cannot see? The sand is dumped at the surface in the form of pellets, each composed of dozens to hundreds of grains. Does each worker carve out a pellet from the active mine face, then carry it to the surface to dump it? Or do different ants, or groups of ants, carry out the different parts of mining and transport? How is labor divided among the tasks, or is it? What we see on the surface cannot answer these questions.
Answering these questions gives us a chance to play with colored sand, an activity that most of us enjoyed in our toddlerhood. Here’s what we are going to do. We will dig a two meter deep hole, and then fill that hole with 12 layers of colored sand (each color is diluted with three parts of native sand to reduce costs), each layer 10 cm thick (the bottom two will be 50 cm thick), and then we will plant a full-sized harvester ant colony in the top of this “layer cake” and let them dig a nest while we keep daily track of the amount of each color of sand they dump on the surface. The colored sand will tell us the current depth and chamber volume of the nest, as well as the rate of deepening and chamber excavation.
At this point, I need to come clean by reporting that this experiment was carried out in collaboration with an archeologist and a geologist who wanted to know if ants had the potential for causing problems with age estimation of soil strata by “optically stimulated luminescence” (OSL). The longer quartz sand has been buried (deeper equals longer), the greater the amount of UV light it is able to emit when stimulated by a short flash of visible light. This ability reverts to zero when the sand is exposed to light. Thus, for example, grains from deeper, having been buried longer, emit more UV light, but if the ants deposit them at another level, they muddy the dates determined from that level. Our plan was to detect such below-ground deposition when sand of one color resided in a host layer of a different color. The amount of such deposition was an estimate of the risk of an incorrect age determination. But for us ant biologists, such deposition told us a lot about how the ants allocate the labor to dig those beautiful nests.
Setting aside the OSL issue (you can find out more about it here and here), the colors in the daily sand collections showed that the ants were digging deeper and deeper day by day and simultaneously increasing the size of the chambers from the top down, with increasing intervals between deeper chambers. Not surprisingly, there was frenzied digging when we first plunked the colony into the enclosure, and this decreased with time until there was no activity in the winter between late November and mid-March. Once they resumed digging in the spring, they did so with a vengeance for a few days.
From the colors in the graph below, it’s clear that digging during October-November established the colony to its full depth and size. It only took three days to dig to the blue layer at half a meter, then another five days to reach the violet, orange, and yellow layers at about a meter and finally, the clay layer at over 1.6 m by the end of October (18 days). The colony then spent the first half of November deepening the nest (red sand), with modest enlargement before taking their winter siesta until mid-March. Overall, the colony excavated a lot more sand from the upper and mid layers (chocolate, green, violet, blue), reflecting the relative volume of chambers visible in the cast at the top of this essay.
When spring brought on another digging spree, the colony did not deepen the nest more, but concentrated mostly on enlarging the upper chambers (native and chocolate layer, 0-10; 10-20 cm). Apparently, in their opinion, their nest was deep enough, but they needed more room at the top. This seems to coincide with the resumption of foraging, with foragers dropping goodies in the uppermost chambers, and transfer workers moving them down to the appropriate level.
By the end of the 7-month experiment, the colony had excavated a nest much as in the schematic above (right), with chamber volume decreasing from top to bottom. This story was told entirely by the color and amount of sand without excavating the nest. But the story was very general. Can we deduce more detail from the sand?
Workers that carried sand to the surface never carried it as single grains, but always as pellets made up of an average of 160 grains of which 60 were colored (recall that each color was diluted with native sand for cost reasons). In the video below, more than one color of sand is visible in many pellets, and this tells us a lot about how sand is mined and transported to the surface. If each worker mined a pellet from the “mine face” and carried it to the surface, all pellets would contain only a single color of sand. But they don’t. A great many are mixtures of colors, suggesting that these pellets are dumped at an intermediate layer of a different color, then reformed into a new pellet for further transport toward the surface.
With very light forceps, it is possible to collect intact individual pellets directly from the mandibles of workers exiting the nest, and to analyze these for the colors they contained. On average, pellets were about 35% colored grains, with an average of between four and five colors (range from 0 to 10). Examples of unmixed, moderately mixed, and highly mixed pellets are shown below. In general, the more colors, the more even their proportion, suggesting that these pellets had been reformed several times in different color host layers. This suggests that a substantial fraction of sand excavated below is not carried directly to the surface but is deposited and formed into pellets multiple times en route, leading to the observed mixture of colors in the pellets.
When we excavated and mapped the nest at the end of seven months, the below-ground deposition was obvious in the multi-colored linings of many tunnels and shafts, for much of the dropped sand had been smoothed into the walls and floors of chambers and shafts, a veritable designer's pallet of colors, ready to gussy up any client’s room. Here are a few examples of what we found.
From all this deposition, we see that the nest is not simply a cavity dug in the sand, but is a dynamic entity constantly being remodeled as the workers drop and reform pellets, thus adding or removing sand. This is particularly apparent in the uppermost chambers in which backfilling and re-excavation seem to occur frequently and on a large scale. However, these results are from the final nest at the end of seven months. When I say “dynamic”, just how dynamic is this remodeling? Does it play out on a time scale of months, weeks, or days? The final experiment in this series established a time scale.
The experiment was simple. We dug a meter-deep pit next to harvester ants, then cut in toward the nest until we just nicked a chamber. Through this opening, we spooned in a dollop (that’s a cubic dollop, a unit in the metric system) of pink sand, closed the opening again, and filled in the pit. After one to three days, we excavated and mapped these nests to look for pink sand pellets or deposits.
The ants didn’t like the mess we had made in one of their chambers, and within less than a day, almost all nests had pink sand on the nest disc. More than 60% of the chambers, especially upper ones, contained pink sand, ranging from backfilling, to scattered grains, to intact pellets. Thus, such "contamination" of nest levels by sand from below is an integral part of excavation--- some fraction of the transported sand is dropped and picked up more than once before reaching the surface. The work of excavation and transport, like most work in ant colonies, is organized in a “series-parallel” way in which individual ants need not take any task to completion. Rather, any ant coming upon the incomplete task can pick it up and take it closer to completion. There are no skilled carpenters, no TIG welders, no backhoe operators, no roofers, and no long-distance truckers among the ants. Every worker can do most of the common jobs that present themselves in the life of a colony. As a result, labor in ant nests is so redundant and abundant that tasks get completed efficiently in pieces by an unskilled work force. This is why the work of ants appears so chaotic to many people.
Thus, deposition and reformation are immediate and continuous processes, and result from the way that social insects organize their work. Had more time elapsed before excavating the layer cake nest, the non-host deposits, while still present, would have been different in composition and location. Like so much in life, the nest is not a static entity, and no matter how we visualize it, the nest, like all living things, is always a snapshot in time.
Well, what can one say? A simply brilliant experiment from conception to execution to conclusion. It would be interesting to see how different ant colonies of the same species treat an identical set up.
To readers unfamiliar with professor Tschinkel's remarkable work, I recommend his book, Ant Architecture, Princeton University Press, 2021., especially Chapter 7. A wonderful read through out.
love using the forceps to take bits from their mandibles!!! I remember your telling me about the kind of planning that went into your dissertation experiment - that it was set up so that the actual implementation (in the field?) took place over just 3 weeks or so. What struck me is how skilled and efficient the overall research called for so it would work as planned and not crash and burn the years of prep! i may not have this exactly accurate, but the main point of having sound methodology and an awful lot of patience (and maybe luck?) has always stayed with me! thanks so much - and this article, pix, text, etc are just wonderful. I know parts are published here and there, but wish it was in one handy overgrown "pamphlet" with all the images, too.