The Many Origins

Chapter 1 — The Bush

STATUS: v0 (all seven parts drafted) Documentary parallel: chapters/00-cold-open.md (cold open) + chapters/01-the-bush.md (Beats 1.1–1.10) Last updated: 2026-05-20

What this chapter does

The documentary’s first chapter draws a picture: the tree of life with the origins of multicellularity scattered across all three of its branches. The image is the documentary’s central deliverable; every subsequent chapter returns to it.

This chapter does the same work as that first chapter, in writing, with mechanism. By the end of it, you should understand:

  1. What the three branches of the tree of life actually are, and where you live on it.
  2. Why “multicellularity was invented many times” is a specific empirical claim, not a generality.
  3. What the bacterial cases look like at the level of cells doing things — the heterocyst trick, Streptomyces, myxobacteria, biofilms.
  4. Why the archaeal cases were unsolved until 2023, and what changed.
  5. The eukaryotic cases — animals, plants, algae, fungi, slime moulds — and why the count of origins escalates from 20 to 45 to a projected 100 as your criteria sharpen.
  6. What “simple” vs “complex” multicellularity actually means, and why the contrast between the two is the most important pattern in the data.
  7. What an “honest scientific disagreement” looks like at the deep root of the animal tree — and how this companion treats such disagreements throughout.

The chapter is in seven parts.

PartTitleApprox. length
1Where you are on the tree of life~1100 words
2The bacterial origins — older than the air you breathe~2400 words
3The archaeal surprise — biology that wasn’t in the textbooks two years ago~1200 words
4The eukaryotic explosion — animals, plants, algae, fungi, slime moulds~1900 words
5Why the count is contested, and what “simple” and “complex” mean here~1900 words
6A live disagreement, and how this companion handles disagreements~1300 words
7The picture — what the bush actually shows~1300 words

Part 1 — Where you are on the tree of life

The documentary opens with three lines drawn from a trunk. BACTERIA. ARCHAEA. EUKARYOTES. That’s a diagram of every known kind of cellular life on Earth, sorted into three categories. Multicellularity — many cells living together as one organism — has been invented on each of those three branches, independently, more than once.

Before any of that can mean anything, you need to know what those three branches actually are, why they are separate, and which one you are on.

The unit being classified is the cell

A cell is the smallest thing that is alive. A bacterium is one cell. You are tens of trillions of cells. A blade of grass is somewhere in between.

When biologists talk about “kinds of life,” the thing they are actually sorting is kinds of cell. Two organisms with similar-looking cells are usually closer relatives than two organisms with very different cells, even if the bigger organisms look nothing alike. A human and a mushroom are made of cells that are extremely similar at the level of internal machinery. A human and a bacterium are not.

This is the first move in the story: forget the organism, look at the cell. Multicellularity is a thing organisms do with cells, and the question of how it happened is, in the end, a question about what cells can do together that they could not do alone.

Three deep categories

Until the 1970s, biology recognised two kinds of cell: prokaryotes (small, no internal compartments, no nucleus) and eukaryotes (larger, with a nucleus and other internal compartments). Bacteria were prokaryotes; animals, plants, fungi, and protists were eukaryotes. The line between the two was the deepest division in life.

In 1977, a microbiologist named Carl Woese published a result that broke the two-kingdom picture in half. He had been sequencing a particular molecule — ribosomal RNA, a piece of the cell’s protein-making machinery that changes very slowly over evolutionary time, which makes it a good clock for measuring deep ancestry — across many different organisms. When he looked at his results, the things that had been called “prokaryotes” did not cluster together. They split into two groups, genetically as far apart from each other as either group was from us [Woese & Fox 1977; Woese et al. 1990].

The reshuffled picture has three deep categories, called domains:

  • Bacteria. Single-celled. No nucleus. Typically one to five micrometres across (a micrometre is a thousandth of a millimetre — small enough that several hundred fit across the diameter of a human hair). Found everywhere on Earth where life is found at all: in soil, in the sea, in ice, in clouds, in your gut, on every surface in your home. Most are harmless; some are useful; a small minority cause disease. They have been on Earth for at least three and a half billion years — more than a billion years before the first eukaryotic cell appeared — and by raw count of cells they outnumber every other form of life put together by many orders of magnitude.

  • Archaea. Also single-celled. Also no nucleus. Also one to five micrometres across. Under a microscope, they look like bacteria — that’s why Woese’s 1977 result was a surprise. They are not. Their molecular machinery, especially the membranes they wrap themselves in and the way they read DNA into protein, is closer to ours than it is to a bacterium’s. Famous for thriving in extreme environments (hot springs at near-boiling, lakes saturated with salt, deep-ocean hydrothermal vents) but also abundant in ordinary places — your gut has some, the open ocean has a lot.

  • Eukaryotes. Cells with a nucleus. The word means true nucleus in Greek. Bigger — typically ten to a hundred micrometres. Internally elaborate: the DNA is sequestered in a membrane-bound compartment (that’s the nucleus), there are other compartments (mitochondria, which do the cell’s energy chemistry; the endoplasmic reticulum, which folds proteins; many more), and there is a cytoskeleton (a scaffolding of protein fibres) that lets the cell change shape and move things around inside itself. Every animal, plant, fungus, and alga is built from eukaryotic cells. So are many single-celled organisms (yeasts, Paramecium, the malaria parasite Plasmodium, amoebae, slime moulds). You are a eukaryote. So is every mushroom in the forest and every blade of grass.

The three-domain picture is the deepest cut you can make through life on Earth. Everything else is sorting within one of these three categories.

The eukaryote category happens once

This is going to matter several times across the documentary. Bacteria and Archaea are deeply old branches of life. Eukaryotes, on the other hand, are not a third deep branch in the same sense. The eukaryotic cell appeared once, around two billion years ago, by what is now generally accepted to be a partnership: an archaeon took up a bacterium and, instead of digesting it, kept it as an internal partner. The bacterium became the mitochondrion, the energy organelle present in essentially every eukaryotic cell today [Martin & Müller 1998; Williams et al. 2013]. Current syntheses lean toward the archaeon being the host throughout the process, with the bacterium internalised; the exact mechanism is still in play [Imachi et al. 2020 on Asgard archaea; Baum & Baum 2014 on the inside-out model].

Every eukaryote alive — you, every animal, every plant, every fungus, every alga, every protist — descends from that one event. Which means: you are more closely related to a mushroom than you are to any bacterium. You are more closely related to a tree than you are to any archaeon. The closest relatives of animals are not bacteria or archaea; they are single-celled eukaryotes (we’ll meet them in Chapter 3).

This is the topology of the tree. Bacteria, deep and broad. Archaea, deep and broad. Eukaryotes, a single twig coming off of (more or less) one of the archaeal branches, then radiating outward.

Multicellularity has been invented on all three.

What “lineage” means

A lineage is a continuous line of descent. Your lineage goes you → your parents → grandparents → great-grandparents → and so on back, through generations of human ancestors, then pre-human apes, mammals, fish, worms, single-celled eukaryotes, an archaeon, and eventually to something we cannot see in the fossil record, called LUCA — the Last Universal Common Ancestor, which lived around four billion years ago. LUCA is the most recent ancestor that all life on Earth shares.

Two organisms share a lineage back to the point where their family trees diverge. You and a chimpanzee share a lineage back to about seven million years ago, where it splits. You and a mushroom share a lineage back to perhaps a billion years ago, where the ancestors of animals and the ancestors of fungi went their separate ways. You and an oak tree share a lineage back roughly 1.5 billion years. You and a bacterium share a lineage back to LUCA itself.

This matters because the question of multicellularity is a question about lineages. When we ask “did this lineage invent multicellularity?” we are asking whether, somewhere along that branch of the tree, the cells started living together permanently. When we ask “did two lineages invent multicellularity independently?” we are asking whether they got there separately — without their common ancestor having already been multicellular.

What “independent origin” actually means

Two lineages share a common ancestor — they have to, because all life on Earth shares LUCA. The question is what that common ancestor was like.

If the common ancestor of two multicellular lineages was already multicellular, those two lineages share a single origin of multicellularity — they both inherited it. Lions and tigers, for instance, share a multicellular common ancestor; their multicellularity is one origin, not two.

If the common ancestor was single-celled, and each lineage went multicellular separately on its own branch, that’s two independent origins. The two lineages each had to figure it out from scratch — often using completely different molecular machinery, and often hundreds of millions of years apart.

Here is the surprise the chapter is built around. When life invented multicellularity in different lineages, it almost always reached for different parts. The protein that holds animal cells together (a thing called a cadherin — for now, just a protein that sticks animal cells to their neighbours) is not the same kind of molecule as the thing that holds plant cells together (a wall built from cellulose and pectin, two structural sugars). The thing that holds brown algae together (alginate, yet another sugar polymer) is something else again. The thing that holds bacterial cells together in a biofilm (a meshwork of secreted DNA, proteins, and sugars laid down outside the cell) is something else still [Abedin & King 2008; Michel et al. 2010; details in content/03-molecular-toolkit/cell-adhesion.md].

We will look at adhesion molecules properly in Chapter 3. For now, the headline is: when we say multicellularity “has been invented in all three domains,” we are not saying the trick was invented once and inherited by everyone who now does it. We are saying it was re-invented, separately, with different molecular tools, in many different lineages — at least twenty times, possibly more than a hundred, depending on how strictly you define your terms [Grosberg & Strathmann 2007; Lamża 2023].

Why this rules out a single ladder

Once the topology of the tree is in view, one thing becomes immediately impossible: a single sequence of multicellularity that begins in bacteria, passes through some intermediate stages, and ends at animals. Independent inventions on three separate domains, with different molecular parts in each, cannot also be a sequence in which the later cases descend from the earlier ones. Independent and sequential are not compatible claims about the same set of lineages.

The popular ladder you may have encountered — bacteria → biofilms → slime moulds → Volvox → choanoflagellates → animals — runs into exactly this problem. Each of those lineages we’ll meet by name and in mechanism over Parts 2, 3, and 4 of this chapter, and you’ll see in each case that the lineage is alive right now, at the tip of its own branch, not on the way to anywhere. Part 7 returns to the picture once the dots are on the page and finishes the demolition there, lineage by lineage.

For now, the move you need is the geometric one: the right picture is not a ladder, even a complicated one. It is a tree, with light scattered across all its branches, with each glowing point of light marking a separate invention of multicellularity. That’s the “bush” the chapter is named after. (→ See the bush — an interactive map of every origin this chapter is about to introduce.)

In the next part, we’ll start lighting up the dots — beginning with the bacterial cases, which are the oldest, and where the story starts in a place older than the oxygen in the air you’re breathing.

→ Continue with Part 2 — The bacterial origins — older than the air you breathe. (Not yet written.)

What this part draws on:

  • The three-domain framework: content/01-polyphyly/independent-origins.md; [Woese & Fox 1977]; [Woese et al. 1990].
  • The eukaryotic-cell origin: [Martin & Müller 1998]; [Williams et al. 2013]; expanded in content/00-framework/major-evolutionary-transitions.md.
  • The polyphyly thesis (multicellularity as a recurring, independent invention): content/01-polyphyly/independent-origins.md; [Grosberg & Strathmann 2007]; [Lamża 2023].
  • Adhesion-molecule diversity across lineages (foreshadowed; expanded in Chapter 3): content/03-molecular-toolkit/cell-adhesion.md; [Abedin & King 2008]; [Michel et al. 2010].

Part 2 — The bacterial origins — older than the air you breathe

The bacterial branch of the tree lights up with at least four glowing dots. None of them is anyone’s ancestor; each is a separate lineage that, somewhere along its own branch, found a way to make many cells act as one organism. None of them uses the same molecular machinery as any of the others.

We’ll look at the four in order: cyanobacterial filaments (the oldest and the strangest), Streptomyces (the bacterial lineage that gave us most of our antibiotics), myxobacteria (a bacterial parallel to slime moulds), and biofilms — which need a careful word, because they are not multicellularity in the same sense and the popular story tends to confuse the two.

Cyanobacteria: a chain of cells with one job, and one specialist who can’t go back

Cyanobacteria are the bacteria that invented photosynthesis with oxygen as a byproduct. Geologically speaking, they are the reason there is breathable air on Earth at all. Many cyanobacteria are single cells. But a major group of them, called Nostocales (Anabaena and Nostoc are the two you’ll see referenced most often), grow as long chains of cells that don’t fully separate after dividing. They are clonal multicellular organisms in the most straightforward sense: one cell divides, doesn’t release its daughter, and the result is a filament dozens or hundreds of cells long.

Here is the mechanism that makes this interesting. Cyanobacteria photosynthesize — they take in light, water, and carbon dioxide, and produce sugars and oxygen as a waste product. They also need nitrogen — not the gas in the air (which is chemically inert and not usable as-is) but nitrogen in a reactive form like ammonia, which they can build into proteins and DNA. Many Nostocales can do this conversion themselves; they pull N₂ gas out of the air and turn it into ammonia using an enzyme called nitrogenase [Flores & Herrero 2010].

The catch is that nitrogenase is destroyed by oxygen — the same oxygen the cyanobacterium itself produces during photosynthesis. The cell has two essential chemistries, and they are incompatible in the same place at the same time.

There are two possible solutions. One is to do them at different times — photosynthesize during the day, fix nitrogen at night — and some unicellular cyanobacteria do exactly that. The other is to do them in different places — and that is the solution Nostocales found. When the surroundings run low on fixed nitrogen, about one cell in every ten to fifteen along the filament transforms. It thickens its outer wall, shuts down its oxygen-producing photosynthesis machinery, switches on nitrogenase, and starts pulling N₂ out of the air. This specialised cell is called a heterocyst. The thick wall keeps oxygen out; the rest of the filament’s photosynthesizing cells keep running their chemistry, and they feed the heterocyst sugars. The heterocyst, in turn, feeds them fixed nitrogen through direct cell-to-cell channels that link adjacent cells in the chain [Flores & Herrero 2010; Mariscal et al. 2007]. The cells are now stuck to each other in a way that means neither type can survive alone.

The key word is transforms. A heterocyst is not a slightly different cell; it is a permanently committed one. It can no longer divide. It cannot go back. In some species, the cell physically rearranges chunks of its DNA during this commitment, deleting genes it no longer needs — an irreversible genome edit that locks the cell into its new role [Rossetti et al. 2010].

The spacing of heterocysts along the filament — roughly one per ten to fifteen photosynthetic cells — is held steady by a signalling system that has its own elegance. A differentiating heterocyst releases a small molecule called PatS that travels short distances along the filament and prevents nearby cells from following the same path; a related molecule, HetN, keeps mature heterocysts vetoing their neighbours [Flores & Herrero 2010]. The mathematics is the same kind of local-inhibition-plus-self-promotion that produces evenly-spaced stripes on a zebrafish — or that lets a zebrafish regenerate its stripe pattern after injury without needing a stored template. Different molecules, different lineage, the same kind of pattern-making.

The timing of all this is the chapter’s most surprising single number. The gene that controls heterocyst commitment, called hetR, can be dated by comparing how much it has changed across different cyanobacterial species and using known fossil dates as anchors. The most careful recent work places its appearance at about 2.6 to 2.7 billion years ago [Boden et al. 2025]. The genes that build the cell-to-cell channels in the filament are similarly old. The first fossils of one of the cyanobacterial specialised cell types — akinetes, dormant spore-like cells — are at least 2.0 billion years old [Tomitani et al. 2006]. All of this is before the Great Oxidation Event, which is the moment around 2.45 to 2.32 billion years ago when oxygen first accumulated to detectable levels in Earth’s atmosphere [Lyons et al. 2014]. The pre-GOE timing of cyanobacterial multicellularity itself comes from the molecular-clock work [Schirrmeister et al. 2013, 2015; Boden et al. 2025]. The trick — being a multicellular organism with a permanently specialised cell type — was already in place before there was breathable air for anything else to evolve into.

That is the punchline of the title. The oldest case of multicellularity we know of is older than the oxygen in the air you’re breathing — because the lineage that made oxygen possible is also one of the lineages that invented multicellularity, in a form that includes heritable, irreversible specialisation between cells.

Streptomyces: a bacterium that grows like a tiny fungus

The next case looks, under a microscope, almost nothing like a bacterium. Streptomyces species — common soil bacteria — grow as networks of branching threads called mycelia, with the threads infiltrating their growth medium hunting for nutrients. When the colony has consumed what it can reach below the surface, a second type of thread, an aerial hypha, grows upward against gravity, pushing through the water film at the surface using little secreted surfactant peptides. The tip of each aerial hypha then partitions, in a synchronized snap, into a chain of dormant spore cells, which can drift on a breeze to a new patch of soil [Flärdh & Buttner 2009].

The trigger for this whole sequence is local nutrient depletion. As the substrate mycelium runs the colony’s resources down, a set of regulatory genes called the bld (for “bald”, because mutants in these genes grow only the substrate threads and never the fuzzy aerial layer) switches the developmental program from “keep growing outward” to “build aerial hyphae and sporulate.” Some of the bld genes encode an uptake system for short oligopeptides — small fragments of protein that diffuse through the colony and act as developmental cues — and in some Streptomyces species the switch is also set off by a small secreted signal molecule called A-factor, a population-density signal whose accumulation triggers the differentiation cascade [Flärdh & Buttner 2009]. The same kind of spend resources building a dispersal stage when the patch is exhausted logic that drives the myxobacterial fruiting body, except executed clonally inside one growing colony rather than by aggregation.

Mechanistically this is multicellular in every sense we care about: cells of common ancestry, physically attached, with at least three distinguishable cell fates (substrate threads, aerial threads, spores). The transition between fates is fuelled by something especially direct: when the colony begins its upward growth, large stretches of the substrate mycelium die in a coordinated way — a kind of regulated cellular self-destruction — and the surviving aerial growth uses their remains as building material [Flärdh & Buttner 2009; Claessen et al. 2014]. The aerial structure is, very literally, made out of the death of the foundation that grew first.

Streptomyces shows up in popular biology mostly because, as it turns out, this lineage produces about two-thirds of the natural-product antibiotics in clinical use — streptomycin (which gave the genus its name), tetracycline, erythromycin, the precursors of vancomycin [Bérdy 2005]. The medical relevance has somewhat overshadowed the developmental interest. The developmental interest is the part that matters here: another, completely separate invention of multicellularity, by a lineage on the other side of bacterial phylogeny from cyanobacteria, using completely different molecular tools, with a body plan that happens to look more like a fungus than like the cyanobacterial chain.

Myxobacteria: a bacterial slime mould

The third case is the one that, at the cellular level, looks the most like the popular story of multicellularity. Myxobacteria (the canonical species is Myxococcus xanthus) live in soil as predatory swarms — sheets of single cells that slide along surfaces together, secreting digestive enzymes that kill other microbes, and then absorbing the products of the kill collectively [Kaiser 2003; Muñoz-Dorado et al. 2016].

When the swarm runs out of prey, something remarkable happens. The cells start communicating using diffusible chemical signals, converge on focal points, and pile up into mounds. Each mound — made of roughly a hundred thousand cells — constructs a small raised structure called a fruiting body, about a millimetre tall. Inside the fruiting body, the cells choose between three fates. About ten percent become myxospores, tough resistant dispersal cells. A smaller fraction become “peripheral rods” that remain around the outside of the structure. The remaining roughly eighty percent deliberately die, releasing their contents — which feed the survival and sporulation of their neighbours [Muñoz-Dorado et al. 2016].

This is aggregative multicellularity, and it’s a different kind of thing from what cyanobacteria and Streptomyces do. The defining feature is not who the cells happen to be related to — in nature, a Myxococcus swarm is often substantially clonal because cells in a patch are usually descendants of local division — but what triggers the developmental program. In cyanobacteria and Streptomyces, the program is keyed to continued clonal growth from a founder cell; the multicellular structure is built out from one starting cell. In myxobacteria, the program is keyed to aggregation — independent cells finding each other, signalling, and committing together. The same trigger logic appears in the eukaryotic slime moulds (we’ll meet Dictyostelium in Chapter 2): aggregation under starvation, three cell fates, massive regulated cell death in service of the colony. Same trick, completely different lineage, separated by something like four billion years of evolution. The clonal/aggregative distinction will do a lot of work in Chapter 4 — it predicts which lineages can permanently lock multicellularity in and which can’t.

Biofilms: not the same kind of thing

You may have heard the version of this story that begins with biofilms — slimy surface communities of bacteria, often of mixed species, embedded in a matrix of secreted polymers. Biofilms are real, important, and ancient (biofilm-like fossils go back at least 3.5 billion years). They are sometimes called the first multicellularity, or the precursor to multicellularity, or the bacterial version of multicellularity. None of those framings is quite right.

Biofilms differ from the three cases above in the ways that matter for our question. They are typically polymicrobial — many different species mixed together — not the cooperative product of cells of common ancestry. They generally do not have heritable cell-type differentiation in the strong sense we just saw in heterocysts or in Streptomyces aerial hyphae; what differentiation they do show is reversible and environmentally triggered. And the cells are not held together by direct cell-to-cell connections, but by being trapped in the same secreted matrix [Claessen et al. 2014].

How sharp that distinction is depends on your definition. Under the broader topology-first frameworks we’ll meet in Part 5, single-species biofilms do count as a very simple form of multicellularity in their own right. Under the stricter sense this chapter is using — where we want heritable, lineage-level developmental commitment of the kind heterocysts, Streptomyces aerial hyphae, and myxobacterial fruiting bodies all show — biofilms sit alongside the multicellular cases rather than among them. Either way, biofilms are a different phenomenon from the bacterial multicellular lineages we just walked through, and the popular conflation of the two is one of the specific failure modes the documentary refuses.

What the bacterial section gets you

Four glowing dots on the bacterial branch, then. At least one of them — the cyanobacterial filaments — predates the atmosphere you currently need to breathe, with terminal cell-type differentiation that was thought, until quite recently, to be a eukaryotic invention. One of them — Streptomyces — is the lineage that gives us most of our antibiotics, and is, geometrically, more similar to a fungus than to a typical bacterium. One of them — myxobacteria — solves the same aggregative-multicellularity problem that the eukaryotic slime moulds solve, with completely different molecules and from completely the wrong direction in the tree. And one common phenomenon that is sometimes called multicellularity — biofilms — is actually a different category of thing, sitting alongside the multicellular cases rather than overlapping with them.

A note on the count: these four are the best-attested bacterial cases, not an exhaustive list. There are other candidates with weaker or partial claims — filamentous sulfur bacteria (Beggiatoa, Thiothrix) that grow as motile multicellular threads in marine sediments; Magnetoglobus, an obligately multicellular magnetotactic bacterium that swims as a coordinated sphere of 17±4 cells; cable bacteria that conduct electrons along centimetre-scale filaments — and almost certainly more in lineages that haven’t been sampled yet [Claessen et al. 2014; corpus file content/02-model-systems/bacterial-multicellularity.md §D]. When Part 7 totals the bacterial count as “at least four,” the at least is doing real work.

The next part picks up the third domain — the archaea — where almost everything you might have read before 2023 is out of date.

→ Continue with Part 3 — The archaeal surprise — biology that wasn’t in the textbooks two years ago. (Not yet written.)

What this part draws on:

Part 3 — The archaeal surprise — biology that wasn’t in the textbooks two years ago

For most of the time biologists have been arguing about which lineages count as multicellular, the third domain — the archaea — was the one nobody could agree about. Some candidate cases had been flagged in the older literature: transient cell aggregates of Methanosarcina in low-energy environments; clumps of Sulfolobus responding to ultraviolet stress. Under a microscope, these looked superficially multicellular. On closer examination, they turned out to be stress responses — single cells huddling together temporarily when something went wrong — not the kind of developmental program where one cell type permanently becomes another and the organism builds a stable body plan generation after generation. As recently as 2022, you could read a serious survey of multicellular life and find archaea listed as the one major branch of the tree where multicellularity had not been credibly documented at all.

That picture changed in two papers, both published in the last three years.

Tang 2023: a salt-loving archaeon with a Streptomyces-style life cycle

In 2023, a group led by Shengjie Tang described an organism they proposed to call Actinoarchaeum halophilum — a haloarchaeon (a salt-loving archaeon, isolated from saturated-salt habitats; you can think of these as the cells you’d find living in concentrated brine where almost nothing else can grow) — that runs a multicellular life cycle astonishingly similar to what we just walked through in Streptomyces [Tang et al. 2023]. The cells grow as branching mycelia. Under specific conditions, they switch to producing chains of differentiated spore-like cells. The developmental commitment is terminal — once a cell enters the spore-forming pathway, it does not revert.

This case is mechanistically more striking than just “another lineage that figured out mycelia.” The Tang group identified, in this archaeon’s genome, two pieces of molecular machinery that matter here. One was a member of the Cdc48 family of ATPases — a class of cellular machines whose job is to unfold and remodel proteins, and that is present in all three domains of life (your cells use Cdc48 too). The other was a putative transporter for short oligopeptides — the same small protein-fragment signals the Streptomyces bld pathway we met in Part 2 also depends on. When they expressed the archaeal version of this transporter in a Streptomyces coelicolor mutant that was missing its own version (the bldKA–bldKE genes), the archaeal protein was able to functionally substitute for the missing bacterial one. The archaeal version of the protein was, in other words, similar enough to its bacterial counterpart at the level of biochemistry to rescue a bacterial developmental defect.

That does not mean the two lineages inherited their multicellular machinery from a common multicellular ancestor — they almost certainly did not, given that Streptomyces sits inside the bacterial domain and Actinoarchaeum sits inside the archaeal domain, with more than three billion years of independent evolution between them. The much more interesting reading is that when each lineage built a multicellular life cycle, separately, it reached for proteins from a deeply old toolkit — old enough that the archaeal and bacterial versions of one of those proteins are still close enough to be interchangeable. This is the first time we’ll see this pattern in the explainer; in Chapter 3 we’ll see it everywhere.

Rados 2025: an archaeon that builds tissue when you squeeze it

The second case came two years later, in a paper from a different group [Rados et al. 2025]. They were studying another haloarchaeon — a different lineage from Tang’s, though whether the two are distantly related within the haloarchaea or are genuinely separate origins of multicellularity within archaea is itself an open question (the kind of question Part 5 will return to: one origin or two often depends on how strictly you define what an “origin” has to look like). The experimental setup was simple to describe. Take cells of this haloarchaeon, place them on a soft substrate, and apply controlled pressure from above — squeeze them in one direction. Under that compression, the cells did something nobody had documented in archaea before.

They stopped behaving like a population of single cells. Individual cells stopped dividing into separate daughters and instead became multinucleate — single cell bodies containing several copies of the genome — and then partitioned into multiple cells within a shared outer wall, in a process that did not depend on tubulin (the protein most eukaryotic cells, including yours, use to organise division). The resulting structure was tissue-like: two distinct cell types, geometrically arranged, with the cells at the periphery showing different surface chemistry and different shape from the cells in the centre — and packing into a specific 3D geometry called the scutoid — a polyhedral cell shape first described in 2018 from animal epithelia [Gómez-Gálvez et al. 2018], where it was at the time treated as a hallmark of animal tissue-level packing. The whole thing was clonal — all descended from one original cell — and it formed in response to a single environmental cue: mechanical force.

There are two ways to read this result, and both are interesting. The narrow read is that Actinoarchaeum and the Rados haloarchaeon are two cases of archaeal multicellularity, and the archaeal branch of the bush now has at least one and possibly two glowing dots. The broader read is that some archaeal lineages are capable of organising into multicellular tissue when the right physical signal arrives. The toolkit, in other words, is present — even in a domain that the textbooks treated for decades as unicellular by nature.

Why this matters for the chapter

Two things.

First, the existence of even one defensible archaeal case completes the picture. Multicellularity has been invented on all three domains of life, not on two-out-of-three with a question mark over the third. The bush you’ll see in Part 7 is now genuinely three-branched, not “bacteria, eukaryotes, and a maybe in archaea.”

Second, the recency of these results says something about the state of the field that is worth absorbing. Until very recently — last year, in the case of Rados — the most basic question about archaeal life cycles (“does anyone in this domain even do multicellularity?”) did not have a clean answer. Most popular accounts of the tree of life that you can find on the internet predate the resolution. Most introductory biology textbooks still in classrooms predate it. If at any point in this chapter you find yourself thinking I’m sure I read something different about this in school, that may well be why. The picture is still being filled in.

The next part is the longest in the chapter, and the most familiar — the eukaryotic explosion. Your home branch, where the count of independent origins climbs from twenty to forty-five to a projected hundred as the criteria for what counts as an “origin” sharpen.

→ Continue with Part 4 — The eukaryotic explosion — animals, plants, algae, fungi, slime moulds. (Not yet written.)

What this part draws on:

  • Archaeal multicellularity, scope and prior candidate cases: content/02-model-systems/bacterial-multicellularity.md scope note; content/01-polyphyly/independent-origins.md archaeal rows.
  • The Tang 2023 Actinoarchaeum halophilum case (Cdc48-family ATPase implicated; oligopeptide transporter functionally complements a Streptomyces coelicolor bldKA–bldKE mutant): [Tang et al. 2023] Nature Communications.
  • The Rados 2025 compression-induced tissue case (multinucleate stage → tubulin-independent cellularisation → two-cell-type scutoid arrangement): [Rados et al. 2025] Science.
  • Scutoid geometry, originally described in animal epithelia: [Gómez-Gálvez et al. 2018] Nature Communications. Cited here because the Rados et al. 2025 compression result documents the same geometry in an archaeal cellularisation event, which is what makes the comparison case load-bearing.

Part 4 — The eukaryotic explosion — animals, plants, algae, fungi, slime moulds

Your home branch of the tree is the busy one. On the eukaryotic side, multicellularity has been invented so many times — and the definition of “invented” gets so much harder to pin down — that the count itself escalates depending on what question you’re asking. Sixteen, if you count loosely and lump similar cases. Twenty-five, with a stricter inventory. Forty-five, under the most careful recent survey of eukaryote-only cases [Lamża 2023]. A projected hundred, if the still-unsampled single-celled eukaryote lineages turn out to harbour what we expect they harbour. None of these numbers is wrong; each is a different answer to a slightly different question. Part 5 is where we unpack the question.

For now, the work of this part is to put the dots on the page, one major lineage at a time. The bacterial section gave you the mechanism for every case it covered; this section is going to give you breadth instead. There are too many eukaryotic origins to mechanise them all in one part. Later chapters return to specific lineages in more depth.

Animals: one origin, around 700 to 800 million years ago

You. Sponges. Comb jellies. Jellyfish, worms, insects, snails, fish, birds, mammals — every animal that has ever existed traces back to a single multicellular common ancestor that lived somewhere between roughly 700 and 800 million years ago [dos Reis et al. 2015]. The first soft-bodied animal-style fossils show up in the rock record only by about 575 million years ago [Erwin et al. 2011], leaving a long gap between when animals must have existed (per molecular clock) and when they appear clearly as bodies in the rock record. This molecular-clock-versus-fossil gap — sometimes called the “long fuse” — is the kind of unresolved fight that pushes paleontology forward, and we’ll touch it again in Chapter 6.

One origin. Not the oldest. Not the most complex by every criterion. One lineage among many.

There is an active scientific argument about which of the very deepest animal lineages branched off first — sponges or comb jellies — which matters because the answer changes the kind of cells we think the first animal had. That argument is Part 6 of this chapter, and the documentary’s method for handling it (steelmanning rather than picking a side) is also Part 6. We’ll come back to it.

If you are wondering why this section is shorter than the one on Volvox or the one on the fungi: that is the answer. In the picture this chapter is drawing, the animal lineage is one origin among many, and giving it disproportionate space here would be a small betrayal of the picture.

Land plants: one origin, around 470 million years ago

Everything green growing on a continent — every moss, fern, grass, tree, flower, vegetable in your kitchen — traces back to a multicellular common ancestor that emerged onto land roughly 470 million years ago. Land plants are descendants of a freshwater group of green algae called the charophytes, which themselves were already multicellular in a simpler way. So the “origin of land plants” is more precisely the origin of terrestrial multicellular green plants — moving ashore rather than going multicellular in the strict sense; their ancestral lineage had already done the multicellular work in water.

One origin. Recent — younger than the animal lineage by perhaps two or three hundred million years.

Red algae: one origin, possibly older than any other multicellular eukaryote

The red algae — the seaweeds that give nori, agar, and carrageenan — trace back to a multicellular common ancestor at least 1.05 billion years ago. That number is the date of the Bangiomorpha pubescens fossils, which are the oldest taxonomically resolved multicellular eukaryote remains in the rock record [Butterfield 2000]. Recent work pushes the floor for unambiguous multicellular eukaryotes back further still: filamentous fossils from the Chuanlinggou Formation in China have been dated to about 1.63 billion years ago [Miao et al. 2024], though their lineage affinity (red algae? something else?) is uncertain.

That puts the red-algal origin somewhere between five and nine hundred million years before the animal one. One origin in the conservative count; with several additional internal origins of complex multicellularity within the red-algal lineage that some authors count separately. The deep antiquity matters for the chapter’s argument: animals are demonstrably not the oldest case of eukaryotic multicellularity. They are not even close.

Brown algae: an independent origin on a separate branch

The brown algae — the lineage that includes kelps and the various seaweeds you’d see on a temperate rocky coast — sit on a totally separate branch of the eukaryotic tree from animals, plants, and red algae. (Their nearest single-celled relatives include diatoms, the microscopic algae that build glass shells.) They evolved multicellularity independently somewhere between roughly 450 and 200 million years ago [Silberfeld et al. 2010; Choi et al. 2024], and went on to develop some of the most morphologically complex multicellular bodies known outside of animals, plants, and fungi — giant kelps reaching tens of metres in length, with internal transport tissues and rooting structures that are functionally analogous to but ancestrally independent of the equivalents in land plants [Cock et al. 2010].

One detail worth flagging now and returning to in Chapter 3: a chunk of the genetic machinery that makes brown-algal multicellularity possible — the synthesis of alginate, the gel-like polymer in their cell walls — appears to have been acquired by horizontal gene transfer from bacteria [Michel et al. 2010]. Horizontal gene transfer is the (very common, in microbes) process by which genes move sideways between unrelated lineages, rather than only being passed down from parent to offspring. The brown-algal case complicates the simple story we’ve been telling about adhesion molecules being independent in each lineage. Sometimes a key piece of the toolkit isn’t even strictly the lineage’s own — it’s borrowed from elsewhere on the tree.

Green algae: at least two origins inside one small family

Green algae are not the same thing as land plants. They include freshwater and marine lineages that never moved ashore. Multicellularity has arisen in green algae more than once. The best-studied case is the volvocine lineage, which contains the famous Volvox — a microscopic green ball, often visible to the naked eye, with thousands of cells embedded in a transparent sphere swimming together with coordinated flagellar beats. Recent phylogenetic work [Hanschen et al. 2018; Ma et al. 2023] shows that multicellularity arose at least twice within the volvocines alone — so a single small family of green algae accounts for two of the origins on the count.

Volvox is also one of the youngest cases of multicellularity with cell-type differentiation that we know of. The volvocine lineage with germ–soma separation (some cells are reproductive; most are sterile, permanently somatic) is something like 200 million years old [Hanschen et al. 2016] — far younger than the animal origin. This matters: the chapter’s refusal of the ladder gets quietly reinforced every time the timeline doesn’t line up the way the popular story expects. Volvox is younger than the dinosaurs.

The mechanism of Volvox’s germ–soma split is one of the cleanest cases biology has of a single gene being redeployed from its ancestral single-celled function into a new multicellular role. The master regulator gene regA, which keeps somatic cells permanently committed to their non-reproductive fate, descends from a gene that, in single-celled green-algal ancestors, was part of the circadian (day-night) regulation machinery [Nedelcu & Michod 2006]. Chapter 3 will give this case the full treatment; for the chapter you’re in, just notice that the toolkit-reuse pattern we already saw in archaea (Cdc48, oligopeptide transport) shows up here too.

Slime moulds: at least six independent aggregative origins, in unrelated lineages

Aggregative multicellularity in eukaryotes — single-celled amoebae that come together temporarily under starvation to build a dispersal structure, exactly like the myxobacteria of Part 2 — has been independently invented at least six times across the eukaryotic tree. The famous case is the dictyostelid slime moulds (Dictyostelium discoideum, the lab favourite), in the supergroup Amoebozoa. But the same trick has been invented separately by:

  • Copromyxa, also within the Amoebozoa, but on a different sublineage [Brown et al. 2012];
  • the acrasid slime moulds, in a completely different eukaryotic supergroup called Heterolobosea — a group of soil and freshwater amoebae and flagellates [Brown et al. 2012];
  • Sorogena, an aggregating ciliate inside the alveolates (the supergroup that also contains Paramecium and the malaria parasite Plasmodium) [Brown et al. 2012];
  • Guttulinopsis, in the cercozoans — yet another supergroup, this one made up of mostly microscopic amoeboid and flagellate organisms [Brown et al. 2012];
  • Sorodiplophrys, in the stramenopiles (the same supergroup as brown algae and diatoms) [Brown et al. 2012];
  • Fonticula alba, in the Holomycota — the broader group that contains fungi [Brown et al. 2009].

Each of these is a separate origin of the gather under starvation, build a fruiting body, sacrifice most cells to disperse a few trick. None of them inherited it from any of the others. They all independently figured out that, when food runs out, aggregating to lift a few survivors above the depleted patch is a strategy worth investing in. The convergence is striking enough on its own — but it is even more striking when you remember that the bacterial myxobacteria are doing the same thing, on the other side of the tree, with completely different molecular tools. The myxobacterial fruiting body and the dictyostelid fruiting body are visually and functionally near-identical solutions to the same selection problem, on lineages separated by something like four billion years of evolution.

Fungi: 8 to 11 origins of complex multicellularity, all within one fungal subgroup

This is the chapter’s most spectacular case of within-clade reinvention. The fungi we recognise — mushrooms, truffles, morels, lichens, the moulds in your bread — are mostly in a single fungal subgroup called the Dikarya. Within the Dikarya alone, complex multicellularity has been invented somewhere between eight and eleven times, depending on how you cut the phylogeny [Nagy et al. 2018, 2020].

Each of those independent origins is a separate evolutionary path to fruiting-body multicellularity. Mushrooms in one Dikarya lineage. Cup fungi and morels in another. The lichen-forming groups in others again. Same kingdom, same general fungal toolkit available in every Dikarya genome — and yet eight to eleven distinct paths to a complex multicellular body, all from a starting point of single-celled or hyphal-but-not-developmentally-complex fungal ancestors.

If you had to choose one fact from this part of the chapter to keep in your head when the rest fades, this would be a strong candidate. Eight to eleven independent inventions, in one fungal kingdom, of a complex multicellular body. The fungi did not learn the trick once and inherit it; they learned it eight to eleven times, in lineages that had a common multicellular ancestor available to copy from and chose not to copy it.

Sixteen, twenty-five, forty-five, a hundred

Now you can see why the count escalates.

A traditional survey — lumping similar cases, being conservative about which protist lineages to admit — gives you something in the sixteen to twenty-five range [Grosberg & Strathmann 2007; Knoll 2011; Niklas et al. 2013b]. That’s the number you usually see in textbooks and popular accounts: multicellularity arose about 25 times.

A recent survey using explicit criteria — six topological types that a multicellular case has to fit into, each lineage individually assessed against those criteria — gives you forty-five in eukaryotes alone [Lamża 2023]. The escalation between 25 and 45 is not new fieldwork; it is the same tree, counted under a more careful and more explicit definition of what an origin is.

And if you take that framework and project it across the still-unsampled diversity of single-celled eukaryotes — there are entire microbial-eukaryote groups in which we have not yet looked carefully — the projected estimate climbs to around a hundred [Lamża 2023].

The numbers are not contradictory. They are answers to different questions, asked of the same tree, with different definitions of origin in hand. Part 5 is about exactly that — why one observer counts sixteen and another counts forty-five from the same data, and why the answer turns out to be more interesting than a definitional squabble.

→ Continue with Part 5 — Why the count is contested, and what “simple” and “complex” mean here. (Not yet written.)

What this part draws on:

Part 5 — Why the count is contested, and what “simple” and “complex” mean here

Part 4 ended on a count that escalated — 25 to 45 to a projected 100. None of those numbers was wrong; they are answers to different questions. Before we go any further, you should know what the questions actually are, and what biology is choosing between when it picks one count over another. The decision is not about who is right; it is about what you want to measure.

Two frameworks are doing most of the work in the recent literature. They sound like they disagree. Once you see what each one is asking, you will see they are not in conflict at all.

The first framework: count by how cells stay together (Lamża 2023)

The first framework comes from a 2023 review by Łukasz Lamża [Lamża 2023]. Lamża’s premise is that “multicellular” is not one phenomenon but a family of related ones, and you can sort the family by topology — by the geometric and mechanical way the cells of an organism are held together. He proposes six types:

  1. Septated multinucleated thalli. A single fungal-like body containing multiple nuclei in shared cytoplasm, partitioned by cross-walls that are not always complete cell separations. Some early-branching fungi and the fungus-like water moulds (oomycetes) fit here.
  2. Plasmotomy-based reproductive structures. Organisms that spend most of their lives as multinucleate masses (plasmodia) and partition into reproductive cells only for spore formation. Some plant-root parasites (Phytomyxea, e.g. Plasmodiophora).
  3. Endogenous (internal) budding. Some marine parasites (Paramyxida) reproduce by making daughter cells inside the mother cell. The “multicellular” stage is the mother-with-internal-buds.
  4. Clonal non-separation after cell division — the most familiar kind. A cell divides, and the daughters don’t fully separate; instead they remain stuck together. Every cell in the resulting organism descends from one founder. Animals, land plants, red algae, brown algae, fungi (the mushrooming kinds), Volvox — all of these are Type 4. (The bacterial filaments from Part 2 — cyanobacteria, Streptomyces — also fit the mechanism, although Lamża’s count is eukaryote-only.)
  5. Pseudoplasmodial aggregative. Single-celled organisms come together temporarily under starvation and build a fruiting body. Dictyostelium and the other aggregative slime moulds from Part 4 — Sorogena, Guttulinopsis, Sorodiplophrys, acrasids. This is the eukaryotic version of what the bacterial myxobacteria do.
  6. Meroplasmodial aggregative. A variant of aggregation in which the cells, once they come together, form a connected network of shared cytoplasm rather than a tightly packed body. Found in a couple of small protist groups (Variosea, Filoreta).

(Lamża also floats a possible seventh type — multi-species superorganisms in which one lineage’s life cycle is obligately tied to another’s, like certain protist–bacterial symbioses — but leaves it on the table without committing to it.)

The point of the framework is that every multicellular case fits into one of these six topological boxes. Once you have the boxes, you can count. Lamża’s eukaryote-only inventory totals 45 independent origins across the six types — about half of them Type 4, the rest spread across the others — and he estimates that another fifty-or-so origins are likely to exist in protist lineages we haven’t sampled adequately yet, pushing the projected count toward a hundred.

The second framework: count only when there’s sustained, integrated complexity (Knoll 2011)

The second framework is the older one, from a 2011 review by the paleobiologist Andrew Knoll [Knoll 2011]. Knoll’s premise is different. He does not deny that there are many forms of multicellularity, but he wants to draw a line between organisms that are simple multicellular and organisms that are complex multicellular — and he picks a specific, functional criterion to draw the line.

His criterion is bulk transport. In a simple multicellular organism, every cell is in direct contact with the outside environment at least some of the time, so each cell can take in nutrients and dispose of waste by diffusion across its own membrane. In a complex multicellular organism, some cells are buried deep inside the body, surrounded by other cells, and cannot reach the outside world directly — they need other cells to pipe nutrients in and waste out for them. That requires plumbing: blood vessels in animals, xylem and phloem in plants, the equivalent structures in kelps and in larger red algae.

Bulk transport, in turn, requires sustained cooperation, division of labour, durable cell-type differentiation, and (usually) irreversibility. So “complex multicellularity” in Knoll’s sense is a package. By Knoll’s 2011 count, the package had been assembled fewer than ten times in the history of life: animals, land plants, the larger red algae, the kelp-grade brown algae, and (notably) some of the fungi.

The fungi are where the count gets dramatic. Knoll’s 2011 review treated the fungi as essentially one or two complex origins; subsequent higher-resolution work by Nagy and colleagues, looking inside the fungal subgroup Dikarya, showed that complex multicellularity has arisen separately in eight to eleven lineages within Dikarya alone [Nagy et al. 2018, 2020] — origins Knoll’s original tally did not yet absorb. Adding the Nagy within-Dikarya expansion to Knoll’s outside-fungi cases gets you a current total of something like 13 to 18 origins of complex multicellularity in the entire history of life — small, but not tiny.

How the two frameworks fit together

Now the crosswalk. Lamża has 45 origins (of multicellularity of any kind). Knoll has 13–18 (of complex multicellularity). These are not contradictory counts of the same thing — they are counts of different things.

The cleanest way to see the relationship is this: every one of Knoll’s complex-multicellular cases — animals, plants, kelps, red algae, complex fungi — is a Lamża Type 4 (non-separation after division). Every single one. The complex multicellular body, in the literature so far, has only been built by lineages that stayed clonal — that didn’t disperse, didn’t aggregate, didn’t bud internally, didn’t form temporary plasmodia, didn’t do any of the other five topological strategies. The complex form has only ever emerged from one mechanism, in different lineages, repeatedly. (This crosswalk is a project synthesis, not Lamża’s or Knoll’s claim — they were operating in different conceptual layers; the observation that they overlap perfectly within Type 4 is the project’s editorial contribution.)

So the two frameworks read together say:

  • Many simple origins. Multicellularity in the broadest sense is evolutionarily easy. Forty-five separate eukaryotic lineages have invented some form of it, with many more probably awaiting sampling. The trick of “many cells acting as one organism” is cheap; life keeps finding it.
  • Few complex origins, only in one of the six mechanisms. Within those forty-five-plus origins, only a small subset (maybe a dozen and a half) crossed Knoll’s threshold into bulk transport, multiple cell types, and the kind of body that needs internal plumbing. And every member of that subset arrived at complex multicellularity by the same mechanism — clonal non-separation after division.

Two things to clarify before moving on. The first: where do the 16-to-25 numbers from Part 4 fit in this two-framework picture? Those are the older lumping surveys — pre-Lamża, mostly pre-Nagy — counting some cases that Knoll would exclude as not-complex and missing many of the eukaryotic origins Lamża later catalogued. They are not a third framework; they are the historical compromise number that survives in textbooks because nothing has displaced it as the one-line headline. The second: both Lamża 2023 and Knoll 2011 are eukaryote-centric — Lamża’s count is eukaryote-only by construction, and Knoll’s complex-MC list contains no bacterial or archaeal cases. The bacterial cases from Part 2 (cyanobacteria, Streptomyces, myxobacteria, plus the candidates that share their mechanisms) and the archaeal cases from Part 3 (the Tang and Rados haloarchaea) sit alongside the frameworks discussed here — at least four bacterial origins and at least one archaeal origin, depending on how strictly the same definitional choices are applied. They have not dropped out of the count; they are simply tallied separately, because the frameworks that did most of the recent definitional work were drawn up for eukaryotes.

That contrast — many simple origins, few complex origins, the complex ones routed through a single mechanism — is the chapter’s most important pattern. It is the question the rest of the documentary is built to answer. Why is the simple form so frequent? Why is the complex form so rare? Why does the complex form only show up when the lineage is clonal? Chapter 4 picks this up in detail, with an answer that involves something called a ratchet.

A guardrail before we move on

A reasonable reader might leave this part with a slightly uncomfortable feeling. The chapter has spent four parts arguing against a ladder — a single sequence of lineages climbing toward animals — and then closed Part 4 by saying the count escalates from 25 to 45 to a hundred. Did we just smuggle a ladder back in?

No, and the distinction is worth naming. The ladder the chapter refuses is a claim about ancestry: that one lineage descended from another, in a chain leading up to you. That claim is false, and the topology of the tree makes it false. The count escalation in Parts 4 and 5 is a claim about definitions: that depending on what you call an “origin,” the number you arrive at gets bigger or smaller. Different question. The escalation does not climb toward us; it spreads outward, sideways, across the tree. Twenty-five becomes a hundred not because cleverer lineages keep being added at the top of a chain, but because slower and more thorough cataloguing keeps finding new origins on the sides of the same bush.

The bush has more dots on it than you’d think, in more places than you’d think, with most of them simple and a few of them complex. That is the picture this chapter is trying to put in your head. Part 7 returns to it.

The next part is the chapter’s first explicit “we are not going to pick a side” beat — a live disagreement about which animal lineage branched off first, and the named method this companion uses for handling such disagreements.

→ Continue with Part 6 — A live disagreement, and how this companion handles disagreements. (Not yet written.)

What this part draws on:

  • The Lamża–Knoll crosswalk, the simple/complex contrast, and the typology of multicellularity: content/00-framework/defining-multicellularity.md.
  • Lamża’s six-types framework, the 45-and-projected-100 eukaryote-only count: [Lamża 2023].
  • Knoll’s bulk-transport criterion and the canonical complex-multicellular clades: [Knoll 2011].
  • The 8–11 independent origins of complex multicellularity within Dikarya: [Nagy et al. 2018, 2020].
  • The conservative ~16–25 simple-MC range, for contrast with Lamża’s count: [Grosberg & Strathmann 2007]; [Knoll 2011]; [Niklas et al. 2013b].
  • The observation that “every Knoll-complex case is a Lamża Type 4” is a project-synthesis crosswalk, not Lamża’s or Knoll’s claim: content/00-framework/defining-multicellularity.md §“Lamża–Knoll crosswalk”.

Part 6 — A live disagreement, and how this companion handles disagreements

Up to now, almost every claim in this chapter has been settled enough that the field would broadly agree with it. The polyphyly thesis itself is settled. The Tang and Rados archaeal cases are recent but uncontested. The 8–11 fungal origins, the multiple volvocine origins, the Bangiomorpha date floor — all of it sits inside the range of what a biologist writing in 2026 would call established.

This part is different. The question we are about to look at is not settled, has moved actively in the last three years, and may resolve one way or the other in the next few. We use it for two purposes at once: as a piece of the bush you should know about, and as the place where the companion introduces the method it will use every time it meets a question of this kind.

The question

Within the animal lineage — Lamża Type 4, single origin, somewhere around 700–800 million years ago — there is a smaller tree branching outward. Sponges off one side. Comb jellies off another. Then the rest of the animals, the worms and snails and fish and us, coming off a trunk that the sponges and comb jellies are not on.

The argument is over which of those very first branches came off first.

Sponges are the simplest living animals. They have no neurons. They have no muscles. They have no nervous system and no true tissues in the strict sense (organised sheets of cells with co-ordinated specialised function). They are made of a handful of specialised cell types arranged loosely around a system of internal water channels, and they feed by pulling water through those channels and filtering food particles out of it. Some of the oldest plausible animal fossils we have are sponge-like in form.

Comb jellies — the scientific name is ctenophores — are transparent gelatinous animals you would find drifting in coastal plankton. They have neurons. They have muscles. They have organised tissues. They move by beating long rows of cilia (tiny hair-like extensions sticking out of cells in their body wall) in coordinated waves that ripple down the body. To the eye they look more elaborate than sponges.

The disagreement comes to this. The very first animal — the single ancestor that all of the rest of the animals, comb jellies and sponges and you included, descend from — was that organism more like a sponge (simple, no neurons, no muscles), or more like a comb jelly (already with neurons, already with muscles)?

If the answer is sponge-like, neurons and muscles were invented somewhere along the long trunk that leads to comb jellies, jellyfish, worms, and you — a single later origin, downstream of the sponge split.

If the answer is comb-jelly-like, neurons and muscles were already in the very first animal, and the sponges, sitting where they do on the tree, lost them. Their simplicity isn’t original; it’s a secondary loss.

These are very different stories about how animal nervous systems got started.

Where the evidence sits

Two kinds of evidence are in play.

The first is phylogenomics — the standard, well-established method of comparing the DNA sequences of many genes across many species and asking, statistically, what tree of relationships best explains the patterns of similarity and difference. Different research groups using slightly different gene sets, slightly different statistical models, and slightly different methods for correcting for known biases have gone back and forth across the last twenty years. Recent careful work using improved methods for identifying which genes are truly the same gene across distant species has tended to support sponges branching first [McCarthy et al. 2022].

The second is synteny — a different kind of evidence altogether. Synteny is the order in which genes sit along a chromosome. If two different species have the same handful of genes lined up next to each other in the same order along a chromosome, that arrangement is almost certainly inherited from a common ancestor rather than independently arrived at by chance (there are too many possible orderings for matching coincidences to add up). A 2023 paper [Schultz et al. 2023] looked across the deepest animal lineages at these large-scale chromosomal arrangements and argued that comb jellies share fewer such arrangements with the rest of animals than sponges do — which the authors read as evidence that comb jellies branched off first, before the arrangements had finished stabilising.

In 2025, a follow-up [Copley 2025] argued that the statistical model behind the 2023 result made unrealistic assumptions about how often chromosomal arrangements should break apart by chance, and that with a more careful baseline the synteny evidence weakens. Separately, work on the molecules that build comb-jelly neurons [Burkhardt & Sprecher 2021] has shown features they share with the neurons of more-complex animals — which is consistent with either tree, but slightly easier to fit under a comb-jellies-late reading than a comb-jellies-early one.

So: sponges-first has strong recent gene-comparison support. Ctenophores-first has the synteny evidence (now under statistical challenge) plus some neuron-molecular evidence that nudges its way. There are smart, careful, well-resourced groups on both sides. The field is moving.

The method

This is the first we are not going to pick a side moment in the explainer. It will not be the last. The companion treats live disagreements by a method that has a name in the project’s editorial standards — steelmanning — and a simple shape:

  • When serious researchers disagree, we show you each position at its strongest — the strongest version of the sponges-first argument, the strongest version of the ctenophores-first argument — and we tell you where the live edge is.
  • We do not pretend the field has settled questions it hasn’t settled.
  • And we use one tiebreaker for whether to engage seriously with a newer idea. Newer ideas that build on the older empirical work — taking what is known and adding to it — earn engagement, even before they are consensus. Newer ideas that abandon older empirical content without explanation earn less.

The asymmetry is a tiebreaker for whether to take a view seriously, not for whether it is right. Both the Schultz 2023 synteny argument and the Copley 2025 critique pass the bar — both are built on top of, not in dismissal of, the prior literature. Both get serious engagement here. Neither gets a tick mark from us.

We are not going to pick a side is not the same as we don’t know what to say. It is the explicit choice not to manufacture certainty the science doesn’t have. The companion will use this move several times — in Chapter 2 about whether the language of decision and memory genuinely extends to cells without nervous systems, in Chapter 5 about whether cancer is best understood as a return to an ancient cellular program or as a failure of bioelectric coordination, and in Chapter 6 about the biggest open questions in the field — and each time it lands here, on this method.

You are being asked, as a reader, to be more comfortable with disagreement than most popular accounts of biology will have invited you to be. The science is more honest than the popular account. So is this companion.

The next part is the chapter’s closing image — the bush itself, with the popular ladder taken down lineage by lineage.

→ Continue with Part 7 — The picture — what the bush actually shows.

What this part draws on:

  • The Porifera-vs-Ctenophora debate, both readings and the 2025 critique: BIG-PICTURE.md §“What is actively contested” #4; [Schultz et al. 2023]; [Copley 2025]; [McCarthy et al. 2022]; [Burkhardt & Sprecher 2021].
  • The steelmanning method and the asymmetry rule for newer ideas that build on older work: README.md §editorial standard 8.
  • The documentary parallel: chapters/01-the-bush.md Beats 1.9 and 1.9b.

Part 7 — The picture — what the bush actually shows

Pull back. Take in the whole tree. (→ Open the bush — every dot below appears there with its mechanism, dates, and citations.)

Three deep branches: bacteria, archaea, eukaryotes. On each branch, glowing dots — every dot a separate lineage that, somewhere along its own line of descent, invented multicellularity from a single-celled starting point. None of them inherited it from any of the others.

The count, by section:

  • Bacteria — at least four origins. Cyanobacterial filaments, Streptomyces, myxobacteria, and (under permissive definitions of where multicellularity begins) single-species biofilms. The “at least” is doing real work; other candidate cases exist, and many bacterial lineages have not been carefully sampled.
  • Archaea — at least one origin (Tang’s Actinoarchaeum halophilum), and a strong second candidate (the haloarchaeon that forms tissue under compression, Rados 2025). Both confirmed in the last three years. Until 2023, this branch had a question mark on it.
  • Eukaryotes — 16–25 in the older lumping surveys; 45 under the careful explicit modern inventory (Lamża 2023); ~100 projected if the still-unsampled microbial-eukaryote lineages turn out to harbour what we expect. Forty-five is the closest thing the field currently has to a single careful answer; the other numbers are answers to slightly different questions about the same tree.

And on top of the count, the contrast. Most of those origins are simple multicellular lineages — cells stuck together, doing one or two coordinated things, without bulk internal transport or many permanent cell types. The simple form is everywhere; life keeps finding it. Only a small subset — somewhere in the neighbourhood of fifteen origins across all of life, by current count, with most of those inside the fungal kingdom — crossed the threshold into complex multicellularity, with internal plumbing, distinct cell types in different places, and the kind of body that needs cells specialised to feed other cells. And every one of those complex cases used the same mechanism: clonal non-separation after division. Cells stayed stuck together after dividing, rather than dispersing or aggregating.

Many simple origins. Few complex ones. The complex ones all routed through one mechanism. That contrast is the central pattern of the data. Why? That question is the engine of the rest of the documentary.

Where you are on it

You are one dot. Animals — one origin, around 700–800 million years ago — and that origin is one. Not the earliest (red algae have you beaten by at least a quarter of a billion years; cyanobacterial filaments by a factor of three). Not the latest (Volvox’s germ-soma split is a couple of hundred million years younger than you are). Not the most complex by every criterion (some of the kelp and fungal cases are extraordinary in their own ways). Not the simplest. One of many.

The ladder, taken down a rung at a time

You have probably encountered, somewhere, the popular version of this story. It goes roughly like this: first there were bacteria; then there were biofilms; then slime moulds; then simple things like Volvox; then choanoflagellates; then animals. As if each lineage on the list were a more sophisticated version of the previous one, with each step a rung climbing toward us.

You now have the parts of this chapter to take that ladder down rung by rung.

Bacterial biofilms are not a simple version of animal tissue. They are a separate phenomenon in their own right — typically mixed-species, held together by trapped polymers rather than direct cell-to-cell connections, with reversible rather than heritable differentiation (Part 2, end). The biofilms we see today share no recent common ancestor with animal tissue — the bacterial and animal lineages have been evolving separately for some four billion years. Biofilms are not an earlier version of you; they are a contemporary solution by a contemporary lineage to a problem the bacterial lineage has been solving in its own way the whole time.

Slime moulds are not an intermediate stage between single cells and animals. They are an aggregative trick — single-celled amoebae that come together temporarily under starvation to build a fruiting body, then go back to being single cells — invented at least six independent times on the eukaryotic side (Part 4), and at least one more time in bacteria (the myxobacteria, Part 2). The dictyostelid lineage that biologists like to film has been doing the aggregative thing for hundreds of millions of years and is doing it right now. It is not on its way to becoming anything else.

Volvox is not a stepping stone the animal lineage came through. Volvox is a green alga. The green-algal lineage split off from the rest of the eukaryotic tree before either you or any other animal existed. The famous Volvox germ-soma differentiation — the most-cited single example of multicellular cell-type specialisation in the popular literature — happened around 200 million years ago, long after the first animals already existed. The standard documentary use of Volvox as a model for early multicellularity is a story told backwards in time: a young, recently-differentiated lineage being used to imagine what an old, ancestrally-differentiated lineage might once have looked like. Volvox is younger than the dinosaurs.

Choanoflagellates — single-celled organisms with a single whip-like flagellum surrounded by a collar of microvilli (tiny finger-like extensions of the cell membrane) — are the closest extant single-celled relatives of animals. We will spend real time with them in Chapter 3, because they matter. The thing to say here is that they are cousins, not ancestors. The choanoflagellate lineage and the animal lineage diverged from a common single-celled ancestor about 800 million years ago, and have been evolving along separate paths since. The choanoflagellate of today is not a snapshot of what the common ancestor of animals looked like; it is what 800 million years of independent evolution downstream of that ancestor have produced in the cousin lineage.

Every single rung on the popular ladder is, on closer inspection, a contemporary lineage at the tip of its own branch — alive right now, doing its own thing, not on the way anywhere else. The lineages are not steps. They are siblings.

The line

This is the documentary’s most precise sentence about its own thesis, and the explainer has now done enough work to earn the right to repeat it:

The ladder is a story about us, told backwards.

The ladder gets built by starting at the animals (because they are the lineage telling the story), then walking backwards through the lineages that happen to look progressively simpler to a creature like us. Each step is chosen because it superficially resembles a less-complicated version of the next step up — even though the lineages chosen have no descent relationship to each other, were evolving independently for hundreds of millions or billions of years before the present, and are still evolving independently now. The ladder is selected, not derived. It is the shape we get when we sort other people’s lineages by how much they resemble simpler versions of ourselves. The bush is the picture actually written in the data.

What’s next

Now the chapter’s central question can be the chapter’s exit question. If multicellularity has been invented twenty, forty-five, possibly a hundred times — independently, in lineages with no common multicellular ancestor — then the lineages that pulled it off must each have had to solve the same set of problems separately. Adhesion. Coordination. Cooperation under the temptation to cheat. The risk of any one cell going its own way at the expense of the others.

What do all of these inventions actually have in common? What did each of them have to solve?

That is Chapter 2.

→ Continue with Chapter 2 — The Problem. (Not yet written.)

What this part draws on:

  • The polyphyly thesis as the chapter’s closing image: content/01-polyphyly/independent-origins.md; BIG-PICTURE.md §“What we are confident about” #1.
  • The lineage-by-lineage demolition: cumulative across Parts 2, 3, 4, and 5 of this chapter; content/02-model-systems/*.md.
  • Choanoflagellates as cousins not ancestors (returned to in Chapter 3): content/02-model-systems/choanoflagellates.md; the ~800 Mya divergence time anchored in [dos Reis et al. 2015] and the choanoflagellates content file.
  • The closing thesis sentence is borrowed verbatim from chapters/01-the-bush.md Beat 1.10.

End of Chapter 1 (draft state).

When complete, this chapter should be readable in one sitting (~10,000 words across seven parts) by someone with no prior biology, and should leave them with a picture they can hold without consulting any other document. The length is on the high end deliberately: Chapter 1 has both what multicellularity is and here are 20+ independent cases to deliver, and downstream chapters can be tighter because they assume the picture this chapter draws. The next chapter — 02-the-problem.md — picks up from Part 7.