Why lose a lung?
Lungless tadpoles and life in fast water
In the last post I told you that most tadpoles breathe air with lungs, starting almost as soon as they hatch. That’s true, and it’s the rule. But like most rules in biology, it has exceptions, and the exceptions are the interesting part.
Some tadpoles never develop working lungs at all. Not “not yet,” not “small and forming,” but functionally absent: tiny, non-vascularized nubs that never inflate, sitting where a real lung would be. And here’s the part that makes it strange rather than just a fun fact: those same tadpoles grow up into completely normal, fully lunged, air-breathing adult frogs. The loss is specific to the tadpole stage. Whatever machinery it takes to build a lung is still in there. It just doesn’t get switched on until later.
Same organ, two very different amounts of it Top: Notaden nicholsi, with a large, richly vascularized lung, typical of an ordinary lunged tadpole. Bottom: Odontobatrachus natator, a lungless, suctorial species, whose “lung” is a non-functional thread, outlined in yellow both times so you know what you’re looking at.
A lung is a balloon, and balloons are bad news in a current
So who are these tadpoles, and why would they do this?
Almost all of them live in fast water: mountain streams, rapids, torrents, the kind of current that will sweep away anything that lets go for a moment. And a lung, mechanically speaking, is a bag of air. Bags of air float. An animal trying to stay pinned to a rock in a torrent does not want to be buoyant, and it definitely does not want to leave that rock every few minutes to go refill the bag at the surface.
Tadpoles that live this way tend to look the part. Most have flattened bodies, and many have evolved a sucker-like mouth, essentially a suction cup where a normal tadpole would have a soft, fleshy disc, that lets them cling to rock faces in current strong enough to knock a person over. There’s a wonderful word for this: gastromyzophorous, from Greek roots meaning belly-sucker-bearing. Some gastromyzophorous tadpoles can hold on to, and even climb, wet vertical rock in the spray zone next to a waterfall.
A mouth built to not let go Otophryne, one of the sand-dwelling, sucker-mouthed lineages. That ringed structure on the left is the oral disc face-on: a suction cup with rows of tiny keratinized teeth, standing in for the soft, ordinary tadpole mouth entirely.
Other lungless tadpoles solve the current a different way, by not being out in it at all: they wedge themselves into sand, gravel, or leaf litter at the stream bottom instead of clinging to the surface of a rock. Either way, the habitat is the same, and a lung is the same kind of liability.
So how often does this actually happen?
To find out whether that story actually holds up, my coauthors and I put together the largest dataset of its kind: larval lung presence or absence across 529 frog species in 44 families, mapped onto a modern, time-calibrated family tree of frogs.
We found at least 28 independent origins of tadpole lunglessness scattered across the anuran tree. That’s not one weird lineage doing something strange. That’s dozens of unrelated groups arriving at the same solution on their own. And the pattern was exactly what the mechanical story predicts: lung loss is overwhelmingly concentrated in stream specialists, both the sucker-mouthed climbers and the sand-and-gravel dwellers, plus a handful of fully terrestrial tadpoles that never experience standing water long enough to need buoyancy control at all.
All 28 losses, in one tree Purple branches are lotic specialists (the sucker-mouthed climbers and sand-burrowers), red is lentic (still water), and brown, underlined names are terrestrial. Seventeen, three, and eight, respectively, and every one of them arrived at lunglessness on its own.
The exceptions are almost more interesting than the rule. A few lungless lineages live in still water, where none of the current-and-buoyancy logic applies, and those species have each independently evolved their own strange workaround for getting oxygen without lungs. The African forest toad Mertensophryne has a bizarre, doughnut-shaped crest on its head. The red toad Schismaderma has evolved a fold of skin across the top of its head, dense with blood vessels, that appears to work like an external gill for pulling oxygen straight from the air. Neither one needed to reinvent the lung. They just needed a patch of well-vascularized skin in the right place.
Schismaderma carens, crest pressed against the air That red, branching pattern is a dense network of blood vessels, sitting right against the underside of the surface film. More on exactly what this tadpole is up to in a later post.
Two ghost frogs, same river, very different lungs
Here’s a case that makes the whole picture messier, in the best way.
South Africa’s ghost frogs (family Heleophrynidae) are about as torrent-adapted as tadpoles get: flattened bodies, big sucker mouths, life spent gripping rock in cold, fast mountain streams. The family has two genera, Heleophryne and Hadromophryne, and they diverged from each other around 64 million years ago. By the logic above, you’d expect both to have ditched their lungs long ago.
Only one of them did. When we dissected tadpoles of both genera, Heleophryne turned out to be essentially lungless, just like the pattern predicts. But Hadromophryne natalensis, living in the same kind of stream, doing the same sucker-mouthed rock-clinging, has small, thin-walled lungs that are clearly inflated and functional.
FIGURE PLACEHOLDER: side-by-side dissection comparison of a lungless Heleophryne tadpole and a lunged Hadromophryne tadpole (your Figure 2/3 material from the ghost frog note).
Why would one torrent specialist keep its lungs while its closest relative lost them entirely? We don’t know for certain, but there’s a suggestive clue: Hadromophryne tadpoles are known to climb right out of the water onto wet, vertical rock faces, something Heleophryne tadpoles have never been observed doing. Out of the water, gills stop working, but a lung would still function just fine. It’s possible that willingness to leave the stream, even briefly, is exactly what has kept the lungs of Hadromophryne in business.
Maybe it’s not really about the air
The standard explanation for lung loss has always been about gas exchange: streams are cold and full of dissolved oxygen, so a tadpole living there doesn’t need lungs to survive, and eventually stops making them. That’s a perfectly reasonable story, but it doesn’t quite explain why lung loss is so narrowly concentrated in the suckered and burrowing specialists, rather than showing up broadly across every tadpole that lives in a well-oxygenated stream. Plenty of ordinary, lunged tadpoles share those same streams without any apparent problem.
We think there’s a piece missing, and it isn’t about oxygen levels at all. It’s about how dangerous or costly it is to actually go breathe.
For a sucker-mouthed tadpole gripping a rock in strong current, letting go to swim up for a breath means risking getting swept downstream, and the return trip against the current is not guaranteed. For a tadpole buried in sand or gravel, coming up to the surface means becoming visible to everything that would like to eat it. In both cases, the behavior of air-breathing itself, not the need for oxygen, is what got expensive.
That distinction matters because a lung isn’t a passive structure that just sits there ready to go. It needs to actually be used, inflated with real breaths, to develop properly. Experiments have shown that if you prevent a tadpole from air-breathing, even one that would normally be a strong, frequent breather, its lungs fail to develop normally. So if a lineage stops finding it worthwhile to come up for air, whether because it’s dangerous, exhausting, or simply unnecessary, the lungs don’t need to be actively selected against to disappear. They just stop being built.
Does losing a lung ever cost you anything?
All of this raises an obvious question: if lungless tadpoles are doing fine without lungs, does it ever come back to bite them?
To test this, we caught wild tadpoles representing 15 genera and 10 families, some lunged and some naturally lungless, some from ponds and some from streams, and dropped them into a low-oxygen environment to see how well they coped.
Pond-living lungless tadpoles, like the toads with their weird head ornaments, handled low oxygen just fine. That tracks: they evolved their lunglessness alongside a backup plan for exactly this scenario. Stream-living tadpoles were a different story entirely. Both lunged and lungless stream-dwellers were more strongly affected by low oxygen than their pond relatives, and the lungless stream tadpoles fared the worst of every group we tested.
FIGURE PLACEHOLDER: simplified version of the responsiveness-under-hypoxia comparison across lunged/lungless and pond/stream tadpoles (your Figure 1 material from the hypoxia paper).
That makes sense once you think about it from the tadpole’s perspective. Streams and torrents are almost never low in oxygen, so a stream-living species has never needed to evolve a way to cope when oxygen drops, lungs or no lungs. That’s a fine bet to make, right up until something changes the water itself. Dam construction, groundwater pumping, and drought can all reduce or interrupt streamflow, and stagnant water gets hypoxic fast. The critically endangered Table Mountain ghost frog, Heleophryne rosei, has declined sharply alongside reservoir construction that reduced flow in the streams it depends on. A tadpole that lost its lungs because breathing air was once dangerous may now be stuck without one, in a stream that no longer behaves the way it did when that trade was made.
Which leaves an obvious follow-up question: once a tadpole’s lineage has gone this route, is there any way back? That turns out to be a much bigger question than it sounds, and it’s worth a post of its own.
Phillips, JR, PH Dias, and MC Womack (2025). Lungless tadpoles breathe fresh air into hypotheses for tetrapod lung loss and trait regains. Evolution 79(12), 2776-2790.
Phillips, JR, GK Nicolau, SS Ngwenya, EA Jackson, and MC Womack (2024). Habitat and respiratory strategy effects on hypoxia performance in anuran tadpoles. Integrative and Comparative Biology 64(2), 336-353.
Phillips, JR, J Reissig, and GK Nicolau (2023). Notes on lung development in South African ghost frogs (Anura: Heleophrynidae). African Journal of Herpetology 72(1), 81-90.