07. The Great Merger

About 2.2 billion years ago, in an ocean lit by a fainter sun, under a sky still adjusting to a new supply of oxygen, something merged that is still running today, inside a beating human heart. The era is the Paleoproterozoic, the opening chapter of Earth's Proterozoic history — a stretch of time roughly two billion years after LUCA, life's last universal common ancestor, first appeared. Nothing with eyes existed yet. Nothing had a nerve. What happened in that ocean was smaller than one cell noticing another, and it turned out to be larger than almost anything that came after it.

No fossil records the moment itself; cells that small and that old almost never leave a shape behind. Two rival reconstructions compete for what the scene actually looked like. In one, a larger host cell engulfed a smaller bacterium whole, through a process called phagocytosis — cell-eating — and then, instead of digesting its catch, kept it alive inside its own body, permanently. In the other, called the hydrogen hypothesis, there was no ambush at all: an archaeon (one of life's three great domains, alongside bacteria and the more complex cells that would follow) that could only survive by consuming hydrogen gas settled into permanent partnership beside a bacterium that released hydrogen as a waste product of its own breathing, until the two became so mutually dependent that separation stopped being an option. William F. Martin and Miklós Müller proposed the hydrogen version in 1998, arguing that energetic interdependence, not predation, was the glue from the very start. It still competes today with the classical predation story and with a third idea — that the partnership began as protection from oxygen poisoning — rather than having settled into one accepted account.

Whichever scene is closer to true, the general idea has a name and a modern author. Endosymbiosis means, literally, living together, one cell inside another, permanently. The serial endosymbiotic theory — proposing that mitochondria (the compartments that generate a cell's usable energy), plastids (the compartments that carry out photosynthesis in plant and algal cells), and, more speculatively, the whip-like flagellum all began as free-living bacteria taken up and retained by a host cell — was assembled into scientific respectability by the biologist Lynn Margulis, publishing in 1967 under her name at the time, Lynn Sagan. She built on ideas about symbiosis inside cells first raised in the early twentieth century by Mereschkowski and Wallin. Her paper carried no new experiment — it was a synthesis, a case built from existing evidence pointed in a new direction — and more than a dozen journals rejected it before the Journal of Theoretical Biology finally printed it.

Time has sorted her claims into two very different piles. The mitochondrial and plastid parts of the theory are now mainstream science, accepted across cell biology, evidenced by every eukaryotic cell (a cell with a nucleus and internal compartments, the design shared by everything from yeast to whales) that still carries mitochondria, and by every photosynthetic eukaryote that still carries a plastid descended from a captured cyanobacterium. The flagellum-from-spirochete part — the idea that the whip-like tail some cells use to swim also began as a captured bacterium, in this case a corkscrew-shaped spirochete — never gathered comparable support, because no one has found the genomic evidence for it that turned up so clearly for mitochondria. A theory, it turns out, can be half right and still remake a field.

Margulis dated none of this; her argument was about mechanism, not timing. The dates arrived decades later, from a study led by Paschalia Kapli, Davide Pisani and Philip C. J. Donoghue, of the Bristol Palaeobiology Group, published in Nature with an online release in December 2025 and print in 2026. Their method, a molecular clock, treats mutations as a kind of metronome: genes accumulate changes at roughly steady rates, so comparing how different two copies of a gene have become can estimate how long ago they split from a shared original. The team built a hundred and thirty-five time-resolved family trees for individual genes — ninety-five rooted in archaea, forty rooted in bacteria — layered across a master tree of sixty-two marker genes, and calibrated the whole structure against eighteen independent fossils.

What emerged reorders the story. The host lineage, which the paper labels nFECA — the nuclear first eukaryotic common ancestor, in its own terminology — split away from the rest of the Asgard archaea, its closest surviving relatives, somewhere between 3.05 and 2.79 billion years ago. Critically, this host had already built itself an internal skeleton (a cytoskeleton), internal membrane compartments (an endomembrane system), and the cellular engulfing machinery of phagocytosis, before it ever acquired a mitochondrion. Only later did the mitochondrial ancestor — mFECA, the mitochondrial first eukaryotic common ancestor — split from its own closest relatives, a large bacterial family called the Alphaproteobacteria, sometime between 2.37 and 2.13 billion years ago. Specific gene duplications tied to importing proteins into the mitochondrion, in genes named TIM14 and TIM44 — essentially molecular doorkeepers that let proteins cross the mitochondrion's inner membrane — push the working estimate for the actual onset of the merger to around 2.2 billion years ago. Only after the merger was fully absorbed into a single working cell did the last eukaryotic common ancestor, LECA, radiate into the lineages that would become every plant, animal, fungus and protist alive today, between 1.80 and 1.67 billion years ago.

Living relatives of both halves of the merger still exist. Asgard archaea — a group that includes the aptly named Lokiarchaeota and Heimdallarchaeia — are the closest surviving kin of the ancient host cell. The Alphaproteobacteria remain the closest surviving kin of the ancient mitochondrial symbiont. The order this study reconstructs, a complex host first and a captured bacterium second, directly challenges older mitochondrion-early scenarios that pictured a simple archaeon transformed into complexity only after picking up its symbiont; the 2026 paper explicitly argues against that older picture, though the precise timeline, and the deeper root of the archaeal-bacterial tree itself, remain live questions, discussed in an accompanying commentary by Hua in the same journal family.

The Energy Behind Complexity

None of this timing explains why the merger mattered so much. For that, the clearest argument comes from Nick Lane and William F. Martin, of University College London and Heinrich Heine University Düsseldorf, in a 2010 paper in Nature. Their case is bioenergetic, built from a mismatch of geometry. A cell makes its usable energy currency, a molecule called ATP, across membrane surfaces, but the cost of building and running a larger genome scales with the volume of the whole cell. Surface area and volume do not grow at the same rate — area lags behind as an object gets bigger — so a bacterium scaled up to the size of a typical complex cell would starve its own genes of power long before it got there. By multiplying the amount of energy-generating membrane packed inside a single cell, without also multiplying that cell's outer surface, mitochondrial endosymbiosis broke through that ceiling. Lane and Martin calculate the resulting gain at roughly two hundred thousand-fold in the number of genes a cell could afford to keep switched on.

This is not settled science. Michael Lynch and Georgi K. Marinov called the entire energy-per-gene framing ecologically and evolutionarily meaningless, arguing that once cell volume is properly accounted for, mitochondria do not actually expand a cell's bioenergetic capacity at all. Lane and Martin answered in 2016, in a PNAS paper whose title turned the criticism back around, framing mitochondrial complexity as a kind of evolutionary deficit spending; Lynch and Marinov published a counter-reply the same year. The dispute remains open and active — a reminder that a mechanism can be repeated confidently in popular science while professional disagreement about it continues underneath.

The clearest surviving proof that any of this happened at all is not an argument but an object: mitochondrial DNA. It is small, circular, and shaped like a bacterial chromosome rather than like the long chromosomes coiled inside a cell's nucleus, because it is, in every meaningful sense, a shrunken bacterial chromosome that never finished being absorbed. In humans it runs to 16,569 base pairs (the paired chemical letters, A-T and G-C, that spell out DNA's instructions) and carries just 37 genes: 13 for proteins used in respiration, 22 for the small RNA adaptors called tRNAs, and 2 for the RNA scaffolding of the ribosome. Over the two billion years since the merger, most of the original bacterial genome moved out of the mitochondrion and into the cell's own nucleus; what is left behind is itself a kind of clock, its smallness a rough gauge of how much time has passed since the takeover began.

Even the question of exactly which bacteria mitochondria are closest to remains disputed, three ways at once. Rickettsiales — the bacterial family that includes the microbes responsible for typhus and spotted fever — has long been the classical candidate. A 2022 reanalysis by Sergio A. Muñoz-Gómez and colleagues, comparing 108 mitochondrial proteins with modeling designed to correct for compositional differences between distantly related genomes, instead places mitochondria as sister to the Alphaproteobacteria as a whole, not to Rickettsiales specifically. Joran Martijn and colleagues, in 2018, argued for something stranger still: that the true sister lineage split off even earlier, and is not represented by any alphaproteobacterium currently sampled and sequenced at all — a relative that has, so far, evaded discovery. All three positions remain in circulation, and the disagreement turns on how each team samples its bacteria and models the statistical noise of two billion years.

What is certain is simpler than any of the three positions. The merger happened; energy became abundant inside a single cell for the first time; and every large, complex cell alive today descends from that one partnership, however it began. Nothing that followed — no thought, no sensation, no signal passed from one cell to another on purpose — was yet possible in that Paleoproterozoic ocean. But the energy to build such things was, for the first time, sitting inside a cell with room to spend it. The molecules that would one day carry feeling had already begun appearing in cells simpler than this one, and it is to them, still blind, still alone, that the story turns next.

Overturned if: a robust phylogenomic dataset reverses the sequence the 2026 Donoghue, Kapli and Pisani gene-duplication study reports — showing the mitochondrial ancestor's arrival preceding, rather than following, the evolution of the host lineage's cytoskeleton, endomembrane system and phagocytotic machinery.

Sources

• 1967 · Lynn Sagan, "On the Origin of Mitosing Cells" — Journal of Theoretical Biology, via Embryo Project Encyclopedia; 50-years retrospective — Molecular Biology of the Cell
• 2026 · Donoghue, Kapli, Pisani et al., "Dated gene duplications elucidate the evolutionary assembly of eukaryotes" — Nature 650, 129–140; commentary — Hua, Nature Genetics 58, 13 (2026)
• 2010 · Lane & Martin, "The energetics of genome complexity" — Nature 467, 929–934; rebuttal — Lynch & Marinov, PNAS 113(6):E668–E669 (2016)
• 1998 · Martin & Müller, "The hydrogen hypothesis for the first eukaryote" — Nature 392, 37–41
• 2022 · Muñoz-Gómez et al., "Site-and-branch-heterogeneous analyses of an expanded dataset favour mitochondria as sister to known Alphaproteobacteria" — Nature Ecology & Evolution 6, 253–262; mtDNA structure — Anderson et al., PMC5267354
• 2018 · Martijn et al., "Deep mitochondrial origin outside the sampled alphaproteobacteria" — Nature 557, 101–105