The Chemoton and the Dream of Rebuilding Lost Evolutionary Stages

Box 2 introduces the chemoton, a theoretical protocell model proposed by Tibor Gánti.

The chemoton has three coupled subsystems:

A metabolic engine, an autocatalytic chemical cycle.

A self-replicating template macromolecule.

A bilayer membrane.

The metabolic cycle consumes nutrient X and produces waste Y. It also produces building blocks for the template and membrane. A byproduct of template replication helps membrane growth, coupling the subsystems stoichiometrically.

In the figure, the chemoton is drawn as a circular system with nutrient entering, waste exiting, internal metabolic intermediates cycling, template polymers replicating, and membrane units growing. The point is that the whole system can grow in synchrony and may divide spontaneously through interactions among growth, osmotic forces, and membrane surface tension.

Why the chemoton matters

The chemoton is not presented as a confirmed historical ancestor. It is a model for thinking clearly about the first major transitions.

It captures the article’s themes in miniature.

Complexity. A simple cycle is probably unrealistic. Real protocells would need networks. Complexity might increase through chemical symbiosis, network extensions, and template families produced by mutation, duplication, and divergence.

Division of labour. The three subsystems do complementary jobs. The membrane is good at boundary-making, not metabolism. Templates are good at digital information, not enclosing the cell. The metabolic network powers the whole unit. Each subsystem is bad at being everything, good at being itself.

Competition of replicators. Even if the whole chemoton is coupled, selfish mutants can arise among digital information carriers. The stochastic corrector principle from Box 1 may prevent deterioration.

Heredity. The membrane and metabolic cycle carry mostly analog information and can support only limited heredity. The template subsystem can carry digital information and therefore unlimited heredity.

The chemoton is essentially a tiny philosophical machine: it asks what a minimal living system must include.

Constructive evolution

Near the end, the authors propose “constructive evolution”: recreating vanished intermediate stages experimentally.

Examples include:

De novo synthesis of a living chemical system such as the chemoton.

Construction of a truly self-replicating RNA in vitro.

Generation of ribozymes through amplification and selection.

Experimental work on RNA molecules relevant to early coding and translation.

Artificial symbioses that clarify how once-separate organisms become integrated.

Recreation of extinct or ancestral forms from living genomes, such as work on fossil fern species from extant polyploids.

The spirit here is delightfully practical. If the ancient intermediates are gone, build plausible versions in the lab and see what they can do.

The conclusion: biology as information history

The article ends by framing biology around information. Developmental biology studies how genomic information becomes adult structure. Evolutionary biology studies how that information came to exist.

The authors’ excuse, as they put it, for discussing genes, cells, sex, societies, and language in one article is that all concern storage and transmission of information.

The review is not a final answer. It is an agenda. But its wager is powerful: the transitions are formally similar enough that understanding one may illuminate the others.

Genes became chromosomes. Bacteria became organelles. Cells became bodies. Insects became colonies. Signals became language.

Life’s story is not a staircase. It is a sequence of mergers, lock-ins, rebellions, treaties, and new alphabets. 🧬📜

Division of Labour, Or Why Evolution Loves Specialists

Once parts are held together, another possibility opens: specialization.

The article calls this division of labour, borrowing the concept from Adam Smith and classical economic thought. Biological systems can become more efficient when different parts perform different tasks.

But specialization is risky. A specialist depends on others. That is why division of labour usually appears only when conflict is controlled and cooperation is stable.

Volvox and the invention of cellular jobs

The article’s key example is the Volvocales, especially Volvox.

Many members of this algal group have only one cell type. Each cell handles both vegetative and reproductive functions. But in Pleodorina, some cells begin with vegetative functions and later become reproductive gonidia. In Volvox, the division is sharper: germ cells are immotile and sit inside the spheroid, while somatic cells bear cilia and cannot divide.

This is a miniature evolutionary drama: cells that once did everything become specialists.

The authors point out an important constraint. In these organisms, motile cells cannot divide, and dividing cells cannot move, because the same organelles are used as basal bodies for movement or centrioles for mitosis. That physical tradeoff makes specialization beneficial.

Division of labour across the transitions

The article lists several cases:

Multifunctional, low-efficiency enzymes can duplicate and diverge into more specific, efficient enzymes.

In the RNA world, RNA served both as genetic material and catalyst. Today DNA stores genetic information, while proteins do most catalytic work.

In prokaryotes, the cell is one compartment. In eukaryotes, nucleus and cytoplasm are separated, and organelles perform specialized tasks.

Sexual populations often evolve from isogamy, where gametes are similar, to anisogamy, where sperm and eggs specialize.

Hermaphrodites can be replaced by separate sexes when reproductive specialization pays.

The pattern is unmistakable: once the parts are bound together, evolution starts handing out job descriptions.

Synergy matters

The authors argue that cooperation requires non-additive or synergistic fitness interactions. Two cooperating individuals must be able to achieve something that isolated individuals cannot.

Their image is simple and memorable: two people each with one oar can row a boat, while one person with one oar goes in circles.

But synergy alone is not enough. Relatedness matters too. If cooperation creates benefits but cheaters can capture those benefits, the system may collapse. Major transitions need both payoff and policing.

The big lesson

Division of labour is one of the main ways complexity grows. But it does not float down from the clouds. It evolves when parts become sufficiently aligned that specialization is profitable rather than suicidal.

A multicellular body is not just many cells. It is many cells with constrained conflict, coordinated development, and specialized function.

From Naked Genes to Chromosomes, The First Great Packaging Problem

Box 1 tackles one of the deepest origin-of-life problems: how could early genes cooperate before chromosomes existed?

The article begins with Eigen’s paradox.

Early replication was probably error-prone. If genomes were too long, mutations would destroy them. That gives an upper limit, called the error threshold, on how much information a primitive genome could contain. Early genomes may not have been much longer than modern transfer RNA.

But here is the trap: a single tiny gene cannot encode a whole organism-like system. You need multiple different genes. Yet if those genes are unlinked and replicate independently, they compete. The fastest-replicating gene wins, and the cooperative system collapses.

So long chromosomes are unstable because mutation wrecks them. Collections of short genes are unstable because internal competition wrecks them.

That is Eigen’s paradox, and it is a nasty little evolutionary mousetrap. 🪤

The stochastic corrector model

The authors describe a solution called the stochastic corrector model.

Imagine compartments containing two kinds of genes. One type has an average replication advantage inside compartments. But compartments grow best when they contain balanced numbers of both kinds.

Inside a compartment, selfish replication pushes the composition away from balance. But random replication and random assortment during division occasionally regenerate compartments with the optimal mix. Those better-balanced compartments grow faster and leave more offspring compartments.

So selection at the compartment level can maintain cooperation despite competition inside compartments.

The figure in Box 1 shows empty and filled circles representing two gene types. Some compartments, marked with asterisks, regain the optimal gene composition. This is the “corrector” part: stochastic randomness keeps generating variation that selection can rescue.

Why chromosomes help

Chromosomes solve the same problem more directly. If complementary genes are physically linked, one cannot replicate without the other. Linkage prevents one gene from outrunning its partner.

The article notes that simulated chromosomes can spread even when they suffer a within-cell replicative disadvantage. Why? Because linked genes avoid the risk of being separated into a low-fitness compartment missing a crucial partner.

A chromosome is therefore not just a string of genes. It is a peace treaty written in chemistry.

Figure 1 revisited: symbiosis into linkage

Figure 1b illustrates this logic visually. It begins with independent replicators A, B, and C. Then they interact in a hypercycle, then become enclosed inside a compartment, then become physically linked. This is a progression from ecological cooperation to inherited unity.

That matters because the article’s whole story is about units of selection being rebuilt. Evolution begins with entities competing and cooperating loosely, then sometimes binds them into a new individual.

The chromosome is one of the earliest and most profound examples: genes stop being lone replicators and become members of a shared hereditary vehicle.

The Strange Career of Formerly Independent Things

One of the article’s most important ideas is that major transitions often convert independent replicators into dependent parts.

Before the transition, the units can reproduce on their own. Afterward, they can replicate only as components of a larger whole.

This is one of evolution’s great mergers and acquisitions. 🧫

Genes become chromosomes. Bacteria become mitochondria and chloroplasts. Single cells become parts of animals, plants, and fungi. Individual insects become workers in colonies. Individual humans become participants in language-based societies.

The problem: lower-level selfishness

The authors insist that this transformation is not easy. Natural selection acting at the lower level can sabotage the higher-level unit.

Examples:

A gene may cheat Mendelian inheritance through meiotic drive or transposable elements.

An asexual female may have a short-term advantage over sexual reproduction because she does not pay the cost of producing males.

A somatic plant cell could, in principle, improve its own genetic transmission by becoming a flower bud even if this harms the plant.

Worker bees may lay male eggs rather than exclusively help the queen reproduce.

These examples show that “integration” is always vulnerable. A body, colony, genome, or society is a political arrangement among replicators. The parliament can be stormed from within.

Why higher-level units do not collapse immediately

The authors argue that major transitions cannot be explained by their eventual long-term benefits. Eukaryotic chromosomes later allowed larger genomes, but that does not explain why eukaryotic chromosome segregation evolved in the first place. Sex later helped eukaryotes diversify, but it could not have originated because of benefits millions of generations in the future.

Instead, the transitions must be explained by immediate selective advantages to replicators.

This is where the gene-centered perspective enters. Szathmáry and Maynard Smith lean on the tradition of George Williams and Richard Dawkins: selection must be explained in terms of benefits to replicators now, not future glory.

The small-founder trick

A key stabilizing principle is that higher-level organisms often pass through a bottleneck with one or very few genetic founders.

A multicellular animal develops from a single fertilized egg. That means its cells are genetically almost identical. Most eukaryotes inherit organelles from one parent only, making organelles within an individual closely related. Early protocells, the authors suggest, may have worked similarly.

This is powerful because high relatedness reduces internal conflict. If all the cells in a body share the same genes, a cell’s evolutionary interests are largely aligned with the body’s success. Not perfectly, as cancer reminds us, but enough for bodies to function.

When does a group become an organism?

The article discusses the idea of the “superorganism.” A group qualifies when it has functional organization like an organism and when selection can act at the group level.

For group selection to work well, several conditions help:

The number of groups should be large.

Migration between groups should be low.

Each group should have no more than one parental group.

These conditions create differences between groups but similarity within groups. That lets selection act on whole groups rather than being drowned by competition among their parts.

Two forces that lock transitions in place

The article names two processes that help maintain higher-level entities once they evolve.

Contingent irreversibility. A formerly independent entity may lose the ability to live alone. Mitochondria cannot go back to free-living bacterial life because many of their genes have moved to the nucleus. Worker bees cannot simply found independent bee civilizations. Cancer cells may escape body control, but they do not become successful protists.

The irreversibility is “contingent” because the reasons are historically accidental. Evolution closes doors not by design, but by piling furniture in front of them.

Central control. If a selfish mutation arises in one gene, suppressor mutations elsewhere in the genome can evolve to restrain it. Leigh’s “parliament of genes” is not democracy by ballot. It is more like every other locus having an incentive to stop the rogue actor.

The message: major transitions require mechanisms that suppress internal rebellion. Without them, the larger unit dissolves back into squabbling parts.

Evolution Does Not Promise Complexity, So Why Did Complexity Happen?

Evolution has no built-in ladder. There is no law saying bacteria must become amoebas, amoebas must become animals, or primates must become poets. Szathmáry and Maynard Smith begin from that bracing point: there is neither a theoretical necessity nor a clean empirical rule that all lineages increase in complexity over time.

And yet, here we are.

Eukaryotic cells are more internally elaborate than prokaryotes. Animals and plants are more complex than single-celled protists. Human societies transmit information in ways no bacterium ever dreamed of, assuming bacteria dream in plasmids.

The authors’ central proposal is that complexity increased in some lineages because evolution passed through a small number of “major transitions.” Each transition changed not merely what organisms looked like, but how biological information was stored, replicated, transmitted, and organized.

The major transitions

The article’s Table 1 lists the great evolutionary handoffs:

1. Replicating molecules became populations of molecules inside compartments.
2. Unlinked replicators became chromosomes.
3. RNA, once both gene and enzyme, gave way to DNA plus protein, via the genetic code.
4. Prokaryotes became eukaryotes.
5. Asexual clones became sexual populations.
6. Protists became animals, plants, and fungi through cell differentiation.
7. Solitary individuals became colonies with non-reproductive castes.
8. Primate societies became human societies through language.

At each step, previously independent units became locked into a larger evolutionary unit. Free-living bacteria became organelles. Individual cells became parts of multicellular bodies. Individual insects became components of colonies. Words and gestures became grammar-bearing language.

Complexity, but how do we measure it?

The article is cautious about complexity. There is no universally accepted biological complexity-meter, no little dashboard reading “complexity: 87%.”

The authors discuss two rough measures.

First, genome size and coding DNA. Table 2 compares organisms such as E. coli, yeast, nematodes, fruit flies, newts, humans, lungfish, and flowering plants. The general pattern is that eukaryotes have larger coding genomes than prokaryotes, and animals and plants often have more genetic material than protists. But genome size is a slippery clue. Lungfish and some plants have enormous genomes without being obviously “more complex” than humans.

Second, behavioral and morphological richness. A bacterium does many impressive things, but it does not phagocytose prey with a cytoskeleton, compose music, or build a bee colony. Cell types, behaviors, and developmental possibilities may better capture the intuitive sense of complexity.

The article’s deeper point is not simply that complexity increased. It asks: by what mechanisms could the amount and organization of information increase?

The three engines of added information

Figure 1 gives three major routes:

Duplication and divergence. A gene is copied. One copy keeps the old job, while the other is free to mutate into a new role. This is the classic “photocopy, then improvise” engine of genetic innovation.

Symbiosis. Separate replicators or organisms join into a cooperative unit. The figure moves from independent replicators, to a hypercycle, to enclosure in a compartment, to physical linkage. This is the visual seed of mitochondria, chloroplasts, and other once-independent entities becoming parts of a larger whole.

Epigenesis. Genes do not merely exist as sequences. They can be switched on or off in heritable states. Figure 1 shows genes A, B, and C with activity states passed through cell division. This foreshadows the evolution of differentiated cell types in multicellular organisms.

The series thesis

The rest of the article keeps circling a single question with different masks:

How can evolution make a new individual out of old individuals?

That question applies to genes on chromosomes, organelles inside cells, cells inside bodies, insects inside colonies, and minds inside language communities. The answer is not sentimental cooperation. It is a rugged evolutionary bargain: cooperation can evolve when conflicts are suppressed, relatedness is high, division of labour pays, and information transmission becomes more powerful.

The Evolution of Heredity, From Chemical Echoes to DNA and Grammar

The article’s most ambitious section follows heredity itself through a series of upgrades.

Heredity means like begets like. But there are different kinds of “like,” and different systems for transmitting information.

The authors distinguish between limited heredity, where only a few states can be transmitted, and unlimited heredity, where an indefinitely large number of messages can be transmitted.

This distinction links genes, epigenetic marks, and language in one shimmering information-thread. 🧵

Stage 1: Simple autocatalytic systems

Autocatalysis means a molecule helps produce more molecules of the same kind. This is essential for growth, but not enough for true heredity. Heredity requires that if the original molecule changes, the system reproduces the changed type.

Some autocatalytic networks may have shown limited heredity, but only among a small number of molecular states.

Stage 2: Polynucleotide-like molecules and unlimited heredity

The origin of polynucleotide-like molecules was a decisive shift because they could encode open-ended digital information.

But the article emphasizes that this transition is hard. Problems include enantiomeric cross-inhibition, where mirror-image building blocks interfere with chain formation, and failure of template and copy to separate because they bind too strongly.

Short oligonucleotides may have been intermediate. Their shorter length allows spontaneous separation, but their growth dynamics can be parabolic rather than exponential. That produces “survival of everybody” rather than sharp survival of the fittest.

For full Darwinian competition, replicators need something closer to exponential growth.

Stage 3: The genetic code before translation

The article suggests that the genetic code may have begun before full translation. The key idea is that amino acids became attached to specific oligonucleotide handles.

The authors favor a scenario in which amino acids acted as coenzymes for ribozymes. Each amino acid had a trinucleotide “handle” allowing it to bind by base pairing. This could let different ribozymes recruit the same amino acid, gradually building the logic later used in the genetic code.

This avoids the “all at once” problem. Translation does not need to appear fully formed, wearing a tuxedo and carrying a ribosome.

Stage 4: Encoded protein synthesis

The origin of translation and encoded protein synthesis is treated briefly in the article, with details deferred to the authors’ larger book. But in the series of transitions, this is enormous: proteins become the main catalytic workforce, while nucleic acids specialize in information storage.

Stage 5: DNA replaces RNA

The article argues that DNA may have replaced RNA because DNA is chemically more stable. Thymine is more stable than uracil, and deoxyribose more stable than ribose.

The authors challenge a common argument that RNA lacks repair systems. In principle, damage repair could be chemically feasible in double-stranded RNA. Stability itself may have been the main advantage.

Stage 6: Epigenetic heredity

The authors then turn to heritable regulatory states. In prokaryotes and simple eukaryotes, methylation patterns can be transmitted through cell division. That means inheritance can depend not only on DNA sequence but also on gene-activity states.

This is central for development. Multicellular organisms need cells with the same genome to behave differently. A neuron and a liver cell are not different because they have different genes, but because they maintain different gene-expression states.

Figure 1c illustrates this idea with genes A, B, and C carrying heritable activity states, marked by asterisks.

Stage 7: Multicellular heredity

Animals, plants, and fungi evolved epigenetic inheritance systems with enough richness to support many differentiated cell types. The authors note that this happened three times, suggesting the transition may not have been extraordinarily difficult once the relevant epigenetic machinery existed.

Stage 8 and 9: Protolanguage and true language

The final heredity transition is cultural.

Proto-language in Homo erectus may have allowed limited communication without grammar. Human language, by contrast, has grammar and unlimited semantic representation. With finite vocabulary and rules, humans can generate indefinitely many meanings.

The authors compare this directly to the genetic code: finite components, infinite combinatorial potential.

They accept Chomsky’s argument that grammar is uniquely human and specific to language, but they criticize reluctance to think evolutionarily about grammar. They argue that intermediate forms are possible. A partial grammar can still be useful, just as a light-sensitive patch can be useful before a full eye evolves.

The article even discusses hereditary variation in linguistic competence, including a family with inherited difficulty automatically generating plurals and past tense. This suggests that grammar can be biologically dissected, much as development can.

The grand move here is bold: heredity includes genes, epigenetic states, and culture. Evolutionary transitions are information revolutions.