January 2014 Archives

Can I train ants? (Ryan)

Hello Antblog,

I am interested in conducting some experiments with ants and I had a few questions I'd like to ask before I get started and I thought your site would be a good place to start.

1) Can queens of different varieties be trained(forced) to co-exist in one colony if an abundance source of food is present?

2) What is the smartest breed of ants?

3) Could ants in an ant farm learn to farm aphids? How might I go about doing this?

4) Could ants survive on honey alone?

5) Could ants be trained to use little man made tools for things such as cutting or digging through rock?

6) Can ants smell and identify different metals such as iron, nickel, gold, etc.?

I understand these questions are very vague and I do not expect a direct answer for all of them, but any information is appreciated. I'd even like additional links I could go explore myself.
I am only a grade 12 student but I am very interested in ants. My ultimate goal is to create a super ant farm to test the capabilities of ants as well as teach them to advance as a species.

Hi Ryan,

I apologize in advance: this post is probably too long and too speculative. The answers to a few of your questions are simple: (6) I don't think so; (4) some ants can; (3) it's thought that this is not a learned behavior, but some ants do facultatively tend aphids; (1) no, unless they are social parasites (for more stuff on social parasites, click here, here, and here).

Questions #2 and #5 are pretty complicated, philosophically and biologically, so I'm going to spend most of the rest of this post talking about them.

But real quick, before I dive into a heady diatribe about the philosophy of psychometry: ants and honey:

Could ants survive on honey alone? Adult ants don't need very many nutrients, and can survive for at least a week or two (often more) on just sugar water or diluted honey. It is the egg-laying ability of the queen, and the growth of the larva that depends on more protein-rich food sources, like (depending on the ants) seeds or insects. Several groups of ants seem to have bacteria living in their guts that can help them make nitrogen (one of the important building blocks of life that we need to synthesize proteins), but I'm not sure if anyone has quantified how only raising these ants on carbohydrates influences how many eggs the queen can lay, and how fast the larvae can grow. For many ants, we don't know their exact dietary requirements, but by and large, the ants that are most likely to be able to get by on a sugar-water or honey-only diet are arboreal (spend most of their time in trees). Carpenter ants (genus Camponotus) and Acrobat ants (genus Crematogaster) are among the most likely candidates that you'd run into outside of the tropics. Similarly, with respect to your question (3), many of the same ants that are largely herbivorous will occasionally tend aphids or other related phloem-sucking insects. But I'm not aware of a learning component to this behavior: I think it's hard-wired.

Okay. Onwards and upwards.
You know, when I tell people that I study ants, I am often asked (by males especially) if I am working on weaponizing them or figuring out how to harness them for industry. Ants are, after all, famously industrious. There is even an ancient Greek/Indian/Russian legend about ants that mine gold (sort of). I read about it in the book "Shalimar the Clown" by Salman Rushdie (which is a great novel if you're interested in 20th century geopolitics, but doesn't have much useful information about ants).

In general, the major hurdle to training ants for any specific anthropocentric end is the way in which they actually get things done. Deborah Gordon has a really nice book about ant behavior called "Ant Encounters" that I highly recommend. One of the key aspects of ant behavior she does a really good job explaining (not least because she pioneered this research) is "emergence." Her research, and the research of other people who study social insects, is much of the inspiration towards work in "Swarm intelligence" in robotics and computer science. Basically, the problem-solving ability of the colony, or "intelligence" does not arise from the leadership or intelligence of a few ants, but emerges from the trial and error of many non-intelligent individuals. Thus, to train a swarm, you'd really need to understand how voting works in these true, blind democracies. Here are a few links you can check out if you want to learn more. Deb Gordon and Tom Seeley are some of the most eloquent speakers and important researchers on the biological side of swarm intelligence (read/listen/watch more here, here, and here).

The most compelling examples of humans modifying ant behavior are not the results of training, but of trickery. It's easy to get ants to walk where you want them to walk, think a given nestmate is dead, and become alarmed by isolating one of the pheromones they use to communicate, and applying it in the right way. However, I can't think of a way in which you could use pheromones to trick ants into picking up tools, or finding gold, unless you could splice the gene for pheromone production into a bacteria that preferentially grew and expressed that gene in the presence of that metal. This is not completely implausible, and as we learn more about the genomic architecture of ants, and the metabolic pathways involved in pheromone production, using a genetically-modified bacteria-ant system could potentially become economically viable (and probably less ecologically destructive than most current mining techniques). One could aerially spray genetically-modified bacteria over an area that already has high ant densities, using the bacteria to trick the ants into collecting soil particles with high concentrations of metal oxides. Then, a few months later, when the ants have collected the clumps of soil with the most metal oxides, one could find large ant colonies and spray another strain of genetically-modified bacteria directly into the nests, that would make the "dead ant" pheromone in the presence of high concentrations of the metal oxide. After a few days, one could go around collecting the midden piles at the surface of the nest, which would be enriched in the target metal oxide. This sort of strategy would probably work best in desert or grassland environments, where a few species of ants build conspicuous nest mounds, like Pogonomyrmex in the American Southwest. It's possible one might be able to do this chemically, without genetically modified organisms of some sort, but if there was a much simpler way of using ants to work for us, someone (especially in Greece, Russia, or India) probably would have figured it out.

Beyond pheromonal trickery, it is possible that we'll be able to eventually trick ants electrochemically, as you can "currently" (get it? like, electrical current?) do with cockroaches: https://backyardbrains.com/products/roboroach
It's possible that, by controlling one ant with electrical impulses, you could convince the rest of the colony to do your bidding. However, as Tom Seeley found with bees, nestmates might need to be convinces themselves of the validity of the scouts choice, in which case you'd need to coordinate quite a few ants to agree that, for example, a vein of gold is a food source. The trickery arguments above apply primarily to ants that often recruit to a food source; however, there are ants that primarily hunt alone.

As you may have figured out by now, your question "which ant is most intelligent?" depends on what kind of intelligence you're talking about. The types of ants most likely to exhibit high degrees of swarm intelligence are the ants that form the biggest colonies, and need to forage collectively. Leaf cutter ants (especially Atta and Acromyrmex) and army ants (especially Eciton and Dorylus) form colonies of hundreds of thousands of individuals in the wild, and many types of invasive ants (such as the Red Imported Fire Ant, Solenopsis invicta) form sprawling super-colonies that can cover acres in their introduced ranges. I'm not aware of a "swarm intelligence" test (but see Seeley's work on, for example, nest-finding abilities of honey bees in Maine), but the number of worker ants involved, the foraging range of the colony, and the complexity of the environment would all likely factor in. It would be interesting to compare how quickly swarms of different sizes and species could solve a maze, for example. I'm not aware of a study like this: perhaps you could set something up!

Observations of ants in the wild can allow one to make at least cursory qualitative statements about swarm intelligence in ants. The complexity and scale of leaf cutter ant nest architecture, the many tasks involved in maintaining a fungal monoculture, and the hazards of foraging in tropical savannas and forests leads me to suggest that the attines may exhibit the highest degrees of emergent, swarm intelligence, perhaps surpassed only by the fungus-growing termites of the African and Australian tropics. A foraging swath of army or driver ants (Eciton or Dorylus) may be collectively be processing millions of environmental signals per second: scent and vibrational cues about proximity of prey and colony members, light ant dark cues, etc., which is similarly impressive.

Basically everything listed in the paragraph about ants here (and the suspiciously similar list here) is a result of "swarm intelligence," and not at all related to the individual intelligence of any of the colony members. Personally, I don't think there is a clear winner between ants and termites in terms of the complexity that emerges from the swarm intelligence of their respective superorganisms. And I'm somewhat tempted to write a whole 'nuthur blog post on what is wrong with these two articles... perhaps some other time.

So far as individual, organism-level intelligence goes: again, I can only speculate. I am tempted to guess that the ants that have small colonies and have large workers that forage by themselves are most likely be more individually intelligent, because colony fitness will depend to a greater extent on the problem-solving ability of ants acting individually, rather than the emergence of intelligence from trial-and-error and blind democracy. Conversely, I would expect very small ants that live in very large colonies to be the least intelligent, individually, because, as a colony, they can rely on swarm intelligence to effect reproduction. In primates, the "social brain hypothesis" (also, see here) suggests that group size should be positively correlated with brain (neocortex) size. An assumption this hypothesis makes is that intelligence is necessary to maintain group cohesion through diplomacy, strategy, and individual recognition. In most social insects, group cohesion is pheromonally-mediated, so intelligence need not scale with group size. However, in some groups of ants and wasps, there is a shifting dominance hierarchy, which might mean individuals have to remember each other and behave according to past experiences. Elizabeth Tibbetts, to the astonishment of the behavioral and entomological communities, demonstrated face recognition in certain wasps with changing social hierarchies that are reinforced by in-fighting, rather than solitary wasps or eusocial wasps with hormonally-mediated social hierarchies. Thus, the ants that are most individually intelligent (e.g., the most train-able) may be ants that have shifting social hierarchies, and/or spend much of their time hunting by themselves. We're still learning more about the social lives of different species of ants: if you include species that haven't been described or discovered, there are probably more than 20,000 ant species in the world, and many of them have remarkably divergent ways of life. Several species I can think of which have to contend with a reality of changing dominance hierarchies (often due to unconventional reproductive strategies) throughout the lifetimes of individual workers, include ants in the genera Pristomyrmex, Platythyrea, Myrmecia, and Cerapachys. Myrmecia include some of the largest and most visually-oriented ants (two factors often correlated with increased brain size). They are also among the largest of ants, and very effective solitary hunters. So I would guess that some member of the genus Myrmecia might be the smartest ant, on an individual basis. Other strong contenders include ants that forage alone and use visual cues to navigate back to their nests, such as Gigantiops destructor, Cataglyphis, and Melophorus.

"Tandem running," a behavior in which one ant basically takes another ant by the hand and leads them to a food source (actually, they just tap each other's antennae pretty much the whole time) has been proposed as the first example of "teaching" in insects (by their definition, the "waggle dance" of bees is simply "broadcasting").
However, a critique of this article by some of the leading non-human intelligence researchers takes issue with this definition of learning, with implications for how intelligence might be defined (not to mention an implicit critique of the education system). Definitely check out the Lars Chittka's publications page for a veritable treasure trove of readings on insect intelligence.

Not to end on too preachy a note, but your statement "help them advance as a species" is problematic on two counts: (1) ants are a family, the Formicidae, which is comprised of more than 14,000 described species, and (2) "advance" has very strong teleological overtones. Teleology is an understandable (Nietzche was all about it, and would have been really happy with you for using that sort of reasoning). But it's not the way I think biology works. Saying you would like to domesticate ants, or modify them behaviorally or biologically so that they are more useful to humans is one thing, but the advancement of a species is an empty concept. Evolution is change, not change towards a goal.

To wrap it all up: swarm intelligence has many advantages at the colony level, but should not be confused with the individual intelligence of organisms. Without broad, cross-species comparisons of either kinds of intelligence, I can only speculate (perhaps beyond the limits of my own intelligence) about which ant is the smartest. Both types of intelligence would require different approaches to training or other types of behavioral manipulation. If there were an easy way to train ants for industry, someone would have probably figured it out by now. Emerging biotechnologies may make the use of ants (and/or ant cyborgs... cy-ants?) possible, but assembling little robots (equipped with both swarm and individual intelligence) from scratch might be prove to be easier, at least in the short run.

Whew! Was that enough to get you started? :)
Jesse Czekanski-Moir & the AntAsk Team

Hello Experts,

I am wondering how the ants decide who will be the new queen when the old one needs to be replaced. I understand that the workers can feed some larvae a special diet so they will develop into queens, but how do they know which ones to do this with? Is it arbitrary?


Hi Nora,

Thanks so much for this great question! Usually, ant colonies have one queen that has mated with one male. The queen founds the colony and uses the sperm from that one male throughout her entire life. The queen produces fertilized eggs, which develop into females and unfertilized eggs that develop into males (this is called haplo-diploid sex determination). The workers of the colony feed the up-growing larvae and based on the diet, the female eggs either develop into workers or queens. The queens (and males) than usually fly off to mate and find new colonies. They do not stay in their maternal colony. That's the norm in most ant species. However, among the 14,500 described ant species, there are many exceptions to this "standard ant".

With that said, queen replacement is not the norm for ants. However, in honeybees, this it happens (which also have haplo-diploid sex determination). The norm in monogynous ants (ants with one queen in the colony), is for the queen to die, and then the colony then dies, not too long thereafter. Some colonies have multiple queens and likely, the once all queens die, the colony dies (after each individual worker has died).

However, there are also some cases, in which the queen gets replaced. This is the case in ants that can turn workers into queens after they have hatched from the pupation state. This is usually regulated by pheromones. Basically, one ant will produce pheromones that tell the others not to develop into queens. Ants also show behavior called "policing", which means that they eat the eggs produced by workers, but not the eggs produced by the queen.

Here is a link to a related post on the topic.

And here is the link to a great paper that goes more into detail on the subject, "The demise of the standard ant" by Jürgen Heinze.

Research has just started to explore the subject of how, in a "standard" monogyounous colony, workers decide which ants will develop into queens (to fly off and found a new colony) and which will develop into workers. It has to do with many factors, such as season (some ants only produces queens and males in certain seasons), food (the more food, the more reproductives), age of colony (usually only mature colonies produce sexuals).

Hope this helps,
Steffi Kautz, James Trager and the AntAsk Team

Hi AntAsk,

Just wanted to inquire as to why would ants suddenly die ? When I mean ants i mean almost all of the same type... somewhat of a kind of small fire ants... but they are dying everywhere, and to add to that there is another variety , which doesn't bite .. tends to swarm to sweet things, and they seem to be packing up.. as the whole kitchen wall was filled with them carrying the eggs and just still and they kinds disappeared, just wanted to know what could be the cause for this.. as I find the one particular type of ants just dying randomly everywhere in mass quite disturbing.

Looking forward to your response,

Hi Rumesh,

Thanks for your question! Would it be possible that the ants, which you observe dying, have been exposed to some type of poison or insecticide? It almost sounds like that. Alternatively, they might all be from the same batch of eggs and all have reached the end of their life span, which is often relatively short in ants (a couple of months or up to a year, usually). Based on your description that would be my best guess.

All the best,
Steffi Kautz and the AntAsk Team