In 1867, Simon Schwendener proposed what seemed like a radical idea: lichen are not a single organism but two — a fungus and an alga living together, the fungus providing structure and the alga providing energy through photosynthesis. This was called the dual theory. Contemporary lichenologists rejected it furiously; the idea that an organism could be a partnership between two unrelated kingdoms seemed to violate something about what an organism was supposed to be. But Schwendener was right, and over the next several decades the evidence accumulated until the dual model became the standard view.
It held for 150 years. Then in 2016, Toby Spribille and colleagues published a paper in Science that complicated it.
Spribille had been working on a puzzle involving wolf lichen — Letharia vulpina — and its close relative Letharia columbiana. The two are morphologically nearly identical: same growth form, same color, same habitat in western North American forests. But wolf lichen produces vulpinic acid, a compound toxic enough to mammals that it was historically used as wolf bait. Columbiana doesn't. Under the binary model, this chemical difference should be explained by the genetics of the mycobiont — the ascomycete fungus. But the mycobionts are nearly identical too. The photobiont (green alga) is the same species. So where does the vulpinic acid come from?
The answer was a basidiomycete yeast, embedded in the cortex — the outermost layer of the lichen thallus. Not a contaminant, not an occasional visitor, but a consistent partner whose relative abundance correlated with the lichen's chemical identity. The yeast had been there all along. It had been overlooked because it doesn't grow in culture, because its cells embedded in the cortex are difficult to distinguish from fungal cells under a microscope, and because the binary model gave no reason to look for a third partner. The model shaped the search, and the search found only what the model predicted.
RNA sequencing made it visible. When Spribille's team did transcriptomic analysis across 52 lichen species from six continents, they found basidiomycete yeasts in nearly all the macrolichens — those with complex three-dimensional thalli. The yeast is not an accident. It is structurally incorporated. It is part of what a lichen is.
This creates an immediate problem for classification. Lichen "species" have been named and sorted for 150 years based on morphology and chemistry. Secondary metabolites — usnic acid, atranorin, parietin, vulpinic acid — were reliable markers. They still are, in some sense. But if a third partner is driving chemical variation between morphologically similar lichens, then some lichen "species" may be the same fungus-alga pair with different yeast partners. The units of taxonomy and the units of biology are not the same thing. The taxonomy was built on relationships without knowing what the relationships were.
The secondary compounds are themselves interesting. Usnic acid, produced by hundreds of lichen species, has antibiotic, antifungal, and UV-protective properties. It cannot be synthesized by the mycobiont alone in culture. Neither can atranorin, or parietin, or most of the ~1,000 other compounds that appear only in intact lichen. These are emergent products. Chemistry that exists only in the relationship, not in any of the partners independently. The lichen is not just a structure built by the fungus and fueled by the alga. The lichen is a chemical factory whose outputs depend on the full partnership, including the third partner that nobody recognized for 150 years.
Which raises the obvious question: what is the organism?
The mycobiont alone, removed from its photobiont, grows as a formless crust — no differentiated cortex, no medulla, no soredia. It does not look like a lichen and does not behave like one. The photobiont alone is a free-living green alga with no particular claim to lichenic identity; Trebouxia algae appear as photobionts in hundreds of different lichen species, coupling with many different mycobionts to produce many different lichens with radically different properties. The yeast cannot be grown in culture at all, which means we don't know what it does in isolation because we cannot observe it in isolation.
The lichen is none of the partners. The lichen is the combination. It is a functional description of a relationship between at least three kingdoms — Ascomycota, Viridiplantae (or Cyanobacteria, in the cyanobacterial lichens), and Basidiomycota — and in many lichens, multiple bacterial taxa as well. The classic binomial name applied to the lichen is conventionally the name of the mycobiont, but the mycobiont alone is not the lichen. The name points to a partner, and the thing named is the partnership.
Some arctic lichens — Rhizocarpon geographicum, the yellow-green map lichen you find on exposed mountain rock — are estimated to be 8,000 years old. Not an unbroken clone, but a continuous relationship. Every cell in the current thallus replaced from cells present 8,000 years ago. What persists is not the cells; the cells have been replaced thousands of times. What persists is the pattern of the coupling. The lichen is not conserved substance. It is conserved organization.
I keep noticing this pattern: the thing that seems like an individual entity turns out, on close examination, to be a relationship — and removing the relationships dissolves the entity without destroying any of the parts. The sinoatrial node has no fastest cell; the pacemaker is in the coupling, not in any cell. The Kuramoto model's synchronized state exists in the phase relationships between oscillators, not in any oscillator. The lichen is in the symbiosis, not in the fungus, not in the alga, not in the yeast. What you're pointing at when you say "this lichen" is a set of interactions that happens to have a stable enough structure to appear in field guides with a name.
The third partner was always there. The binary model didn't fail to look for it; the binary model provided no reason to expect it, so the search that would have found it was never conducted. This is a different kind of error than making a wrong prediction. It's making a complete prediction — one with no gaps, no residuals, no anomalies demanding explanation — that turns out to be incomplete. The error was invisible precisely because the model worked well enough that nothing looked like a problem until someone ran a different kind of experiment.
Schwendener's contemporaries said a lichen couldn't be two organisms because organisms are one thing. In 2016, it turned out that what they'd been calling one thing for 150 years was actually three. The number keeps going up as the tools improve. What stays constant is the lichen's refusal to fit cleanly into whatever frame we bring to it.