Host Manipulation · Parasitology · Ants · Invasive Species · Canada · North America · Evidence BriefRecord to Sep 2026
A parasite that can turn a behaviour off as well as on — and a mechanism nobody has caught
The Switch Nobody Has Found
An ant infected with the lancet liver fluke climbs a flower in the cool of the morning, locks its jaws onto a petal and waits. Hours later it lets go, walks home, and does the whole thing again tomorrow. Almost every other famous case of a parasite steering its host ends with the host dead; this one does not, and the reversibility is what makes it interesting. It is routinely described as a switch the parasite can throw. The evidence assembled here does not support that word — not in the behaviour, which turns out to be a gradient, and not in the biology, where three research groups have named three different candidate molecules and nobody has tested any of them.
What Happened
What the record establishes, and how firmly Graded by standing rather than by importance. The first four rows can be reproduced by anyone with the papers open; the last three are stated by named people who say themselves that they do not yet know.
| Standing | What | On whose record |
|---|---|---|
| Confirmed | The brains of infected ants differ from uninfected ones in 1,349 transcripts, and the difference is concentrated where the behaviour is: 954 at the attached phase, against 164, 162 and 108 at the other three. Sixty ant brains, six replicates per condition, a mean of 48.9 million read pairs each. | Li, Goater and Wasmuth, Molecular Ecology |
| Confirmed | The word “adenosine” does not appear in that paper. Searched across the full text: zero occurrences, including the discussion and the conclusions. The molecule is raised in interview as something worth chasing, not as a result. | The paper itself, read in full |
| Confirmed | Warming releases the ants gradually, not at a threshold. Across 1,264 ants observed on 13 days in Danish forest, each 1 °C of warming released about 11% of those attached — 14% once the date is in the model. Temperature was the only variable that mattered consistently; time of day and sunlight did not. | Gasque and Fredensborg, Behavioral Ecology |
| Confirmed | The behaviour is a near-perfect diagnostic of infection. Of 174 ants collected while biting vegetation, 97% carried the parasite. Of 80 collected at random from the same nests, none did. | Gasque and Fredensborg |
| Confirmed | The researchers tested a rival explanation and could not confirm it. Insect chill coma — a cold-induced, reversible paralysis — would account for reversibility with no parasite switch at all. Potassium-channel gene sets moved in the expected directions, but the authors write that they could not identify a clear pattern to support the hypothesis, and ask for direct ion measurement. | Li, Goater and Wasmuth |
| Confirmed, not verifiable here | The team published a short commentary on reversibility twelve days before the news coverage. Its abstract states the attach-detach-repeat sequence and argues that recognising reversibility improves the general understanding of host manipulation. The full text sits behind a subscription and nothing here is attributed to it beyond the abstract. | Goater, Wasmuth and Li, Trends in Parasitology |
| Unconfirmed | Three different temperatures are given for the same effect. A veterinary reference says jaw tetany sets in below 15 °C; the 2025 paper says ants detach above approximately 18 °C; the news report says they cling below approximately 19 °C. Each is stated plainly by a source that ought to know, and none cites the others. | Merck Veterinary Manual; the 2025 paper; CBC News |
| Unconfirmed | That no other manipulation is both many-sided and reversible. The claim is the authors’ own, phrased in the paper as not being aware of other examples. An absence of known counter-examples is not the same as their absence, and it is the team’s own distinctiveness being asserted. | Li, Goater and Wasmuth |
| Guesswork | That the worm is controlling invertebrate sleep through adenosine. Offered in interview with the reasoning attached — a large rise in adenosine induces sleep in other animals — and with the qualification attached too: “but we don’t actually know.” | Wasmuth, quoted by CBC News |
| Guesswork | That protein tyrosine phosphatase is the regulating factor. A different group’s different candidate, proposed by analogy with baculovirus hyperactivity and fungal death grip, and labelled speculation in the paper that proposes it. | Gasque and Fredensborg |
| Guesswork | That fire suppression in the park is raising snail and ant densities and so helping the parasite. Asserted in interview. Compatible with a separate survey finding infected ants only in aspen stands bordering water, and never on fescue grassland or in pine or spruce — but that survey tests habitat, not fire history. | Goater, quoted by CBC News; Beef Cattle Research Council |
- 1,349Transcripts differing between infected and uninfected ant brains, 954 of them while the ant is attached
- 0Times the word “adenosine” appears in the paper it is attributed to
- 11%Of attached ants released by each degree of warming — a slope, not a threshold
- 3Candidate molecules named, by three groups, none of them tested
Timeline
A parasite that arrived, and a question that has taken sixty years to get to Two of the three earliest entries are ranges rather than dates, and they are written as such. The mechanism has no entry at all, because nothing about it has yet happened that can be dated.
- 1950sThe lancet liver fluke enters eastern North America, on the account of the Canadian beef industry’s own research council.
- After 1985It reaches Cypress Hills, on the Alberta–Saskatchewan border. The same source gives no closer date.
- 2009Hundreds of attached ants begin to be found next to the same three nests, every spring and summer since.
- Feb 2017Sequencing identifies the Alberta intermediate hosts: the snail Oreohelix subrudis and the ant Formica aserva. Neither lives in the fluke’s native European range.
- Jun 2018Micro-CT scanning shows, for the first time, physical contact between the larva and ant brain tissue at the front of the suboesophageal ganglion.
- Mar 2020The larva in the brain is shown to be a clone of the ones in the abdomen it dies for — the genetic case for kin selection.
- 7 Jun 2021Between 06:00 and 09:00, about 120 infected and 120 uninfected ants are collected at the Cypress Hills ‘Staff Camp’ site. These become the transcriptome.
- Aug 2023The Danish field study publishes the number this page turns on: about 11% of attached ants released per degree.
- 9 Sep 2025The brain transcriptome appears in Molecular Ecology: 1,349 differentially expressed transcripts, six candidate systems, and no mechanism.
- 8 Sep 2026The same three authors publish a short commentary in Trends in Parasitology arguing that reversibility itself is the finding worth generalising.
- 20 Sep 2026CBC News reports the work, and the switch metaphor reaches a general audience with the adenosine speculation attached to it.
The Argument
A switch has a threshold. This has a slope. The three temperatures in circulation are not a disagreement to settle. They are three summaries of a curve, and the curve is the finding.
The most-repeated sentence about this parasite is that it has found an on-off switch. It is a good sentence, the researchers use it themselves, and it is the reason the story travels. It is also the thing the evidence is least able to support — because the only study that went out and measured the relationship found no threshold anywhere in it.
- 11%released per degree of warming, steadily, with no step in the curve
- 1,264ants observed over 13 days to establish it
Read that way, the three temperatures stop competing. A reference manual saying tetany below 15 °C is describing the cold end, where nearly every infected ant is locked on. A research paper saying ants detach above about 18 °C is describing the middle, where most have let go. A reporter saying they cling below about 19 °C is describing the same middle from the other side. Every one of the three is a reasonable place to put a pin in a curve that has no corner, and the disagreement between them is an artefact of pinning it at all.
The field data has a second property a switch would not have. On days when the temperature swung widely, individually tagged ants flipped between the two states repeatedly; on cool, steady days many simply stayed attached from dawn to dusk. In October, when the warmest part of the day reached only 11.5 °C, up to a third of them never let go at all. An infected ant is not being commanded to attach and then released; its probability of being attached is tracking the thermometer all day long. That reading is ours and no source states it in those words — but it follows from the published regression rather than around it.
- 96.67%of 30 tagged ants were in the same state, on the same plant, the next morning
What Others Add
The life cycle, the livestock, and three molecules that have not been tested Everything in this section comes from somewhere other than the 2025 paper and the report of it.
The fluke needs three hosts in order. Adults live in the bile ducts of a grazing mammal and their eggs leave in its dung; a land snail eats the eggs and coughs the next larval stage out wrapped in mucus — a slime ball, which is the technical term, carrying 200 to 400 larvae; an ant drinks the slime ball for its moisture and swallows them. Almost all of them settle in the ant’s abdomen and wait. One, occasionally two, goes to the head instead and lodges at the front of the suboesophageal ganglion, where micro-CT imaging has shown it physically touching brain tissue.
The one that goes to the head gives up everything. It cannot survive being eaten into a mammal’s liver the way its siblings can, so it will never reproduce. Genetic work published in 2020 explains why that is not a puzzle: the larva in the brain and the ones in the abdomen are, as a rule, clones of each other, cotransmitted from the same snail. Sacrificing itself to get its own genome into the next host is, in the exact technical sense, not a sacrifice.
- 1larva of the many goes to the brain, and it is a clone of the ones it dies for
Candidate one
Biogenic monoamines
- Six
octopamine receptor beta-2R-liketranscripts are raised in infected ants just before they attach, and they fall in the set specific to starting the behaviour. - Serotonin- and dopamine-synthesis enzymes and a GABA receptor subunit fall at the detaching phases instead.
- Association only. No knockdown, no drug, no functional test.
- Six
Candidate two
Protein tyrosine phosphatase
- Proposed by the Danish group, not the Canadian one, and reached by analogy rather than by measurement.
- The enzyme is implicated in baculovirus-induced hyperactivity in caterpillars and is raised during the fungal death grip in ants.
- Never looked for in this parasite. The authors call it something worth including in future work.
Candidate three
Adenosine
- Raised in interview, absent from the paper, and the one the coverage has carried furthest.
- The biology behind the guess is real: flies carry a single adenosine receptor gene, and overexpressing it in neurons lengthens their sleep.
- Nothing has measured adenosine in an infected ant. It is a direction, and the person who offered it said so.
There is a connection between the second and third columns of that table that nobody appears to have drawn. The 2025 paper’s own fallback explanation is chill coma: a cold-induced, reversible shutdown of the insect nervous system, from which the animal recovers when it warms. Separately, work on locusts has shown that adenosine governs how long recovery from exactly that kind of reversible coma takes, and that caffeine — which blocks adenosine receptors — shortens it. Two candidate mechanisms proposed independently may be one mechanism seen from two ends. This is our inference. No source joins them, and it is a reason to go and measure, not a finding.
| What is at stake | The figure | Where it comes from |
|---|---|---|
| Cattle in the park | 4,000, against 657 elk and 662 deer — so the grazer that matters most for transmission is the managed one | Beef Cattle Research Council |
| Infected livers | 83% of 18 cattle, 68% of 70 elk, 47% of 36 deer — the sheet’s own figures, and it says the samples are small and not randomly drawn | |
| Where the infected ants are | Only in aspen stands bordering water, and only from mid-June to August — never on fescue grassland, pine or spruce | Beef Cattle Research Council |
| What a heavy infection means | Up to 50,000 flukes in a mature sheep, with no immunity acquired, and often minimal visible signs | Merck Veterinary Manual |
| How local it is | Of 500 cattle blood samples taken across western Canada outside the park, 499 were negative and the one positive was judged a false positive | Beef Cattle Research Council |
One more comparison is worth making because it is the one readers already have. The fungus behind the zombie-ant images everyone knows works quite differently: it fills the ant’s muscles and never enters the brain at all, it makes the mandibles hypercontract, the ant climbs to a set height and dies there, and the whole thing happens once. The fluke does enter the nervous system, changes nothing permanently, and needs its ant alive and walking home each evening for the rest of the summer. Similar-looking behaviour, opposite machinery — which is the reason nobody can read a mechanism off the posture.
Conclusion
What to hold firmly, and what to hold loosely
The behaviour is real and it is not a switch
Ants climb, bite, hold on overnight and let go when it warms; 97% of the ones doing it are infected and none of their nestmates are. What is not established is any temperature at which the behaviour flips. The one measurement finds a steady release of about 11% of attached ants per degree, and the three thresholds in circulation are three ways of summarising that. Reach for the metaphor if it helps a reader picture it, then put the number beside it.
Three candidate molecules, no experiment
Monoamines come from differential expression, protein tyrosine phosphatase from analogy with other parasites, adenosine from an interview. Each is a reasonable place to look and none has had a knockdown, a drug or an ion measurement run against it. Until one does, the honest description of the mechanism is that it is unknown — which is what the people doing the work say when asked directly.
The interesting question may not be the parasite
This population is an introduction. It arrived with imported European livestock, and in Alberta it picked up a snail, an ant and a set of grazers that do not exist together anywhere in its native range. Whatever is happening in the ant’s head evolved somewhere else, against other species, and still works here. That is a harder thing to explain than a switch, and it is the part of the story with consequences beyond one park.