The Genetic Bottleneck and Adam and Eve: Can Two Be Enough?

Q: Does Luke’s genealogy require Adam to be a real person?

Luke traces Jesus’ line backwards and ends with the son of Adam, the son of God (Luke 3:38). He does not change his manner of speaking partway through, as if the recent names were history and the early ones were poetry. Every name in that list is treated the same way. A genealogy that runs from a real Jesus through a real David to a symbolic Adam would be a strange document indeed—and Luke, who opens his Gospel by promising an orderly and researched account, gives no hint that he thought he was doing anything of the kind.

There were two people to start with—one man, one woman. And every human being who has ever lived descended from them.

For most of history, nobody found the idea strange. Then the science of genetics arrived, and a new objection appeared. It sounded final. Our DNA, we were told, carries too much variety to have come from just two people. The human family must have begun with thousands.

If that’s true, it’s a major problem for the reliability of Scripture. Apostle Paul builds his whole explanation of sin and salvation on one man’s single act. Take away the one man and the argument collapses.

So the question deserves a careful answer. And the answer, it turns out, is far more interesting than either side perhaps expects.

The claim that started the argument

In 2017, Dennis Venema, a biologist then writing for BioLogos, published a book called Adam and the Genome with theologian Scot McKnight. Venema’s section made a bold claim: the human population never dropped below about 10,000 individuals at any point in the last several centuries. A starting couple was, in his words, about as believable as a flat earth.

He gave four lines of evidence. Here they are in plain language.

THE EVIDENCE	WHAT IT MEANS IN SIMPLE TERMS
Linkage disequilibrium	DNA is inherited in chunks. Over generations, those chunks get shuffled and broken into smaller pieces. By measuring how broken up the chunks are today, scientists estimate how many people were doing the shuffling in the past. Fewer people means less shuffling.
Human leukocyte antigen (HLA) gene variety	HLA genes help your immune system recognise invaders. Humans carry a huge number of different versions of these genes. Two people can only carry four versions of any one gene between them (two each). So where did all the rest come from?
Alu insertions	Alu elements are short pieces of DNA that copy themselves and jump into new positions in the genome. The pattern of where they sit, and how common each one is, is used to estimate past population size.
Incomplete lineage sorting	At some points in the genome, a human sequence looks more like a gorilla sequence than a chimpanzee sequence. Standard evolutionary models explain this only if the shared ancestral population was large.

That looks like a wall. Four independent methods, one answer. And for a while, almost everyone on both sides accepted it.

One quiet question that changed the conversation

Then Richard Buggs asked a question.

Buggs, an evolutionary geneticist, is a working scientist in exactly this field, and he wasn’t defending a young earth or a six-day creation. He simply asked something nobody had asked:

THE QUESTION NOBODY HAD ASKED

Has anyone actually tested the hypothesis that humanity began with a single couple?

Not: does the data fit a large ancestral population? It does.

But: has a short, sharp bottleneck of two been modelled at a specific date and then ruled out?

The answer was no. The methods had been built to estimate population trends over long periods. They’d never been asked to detect a brief squeeze down to two people and back up again.

Buggs’s own conclusion was even-handed, and we should report it exactly as he gave it. It cuts both ways:

Against a recent couple. The evidence that humanity did not pass through a two-person bottleneck in recent years is genuinely strong.
But not further back. Beyond that, the methods lose their resolution. A brief bottleneck of two cannot be excluded there.

That second point matters enormously. “We haven’t detected it” and “we’ve proved it didn’t happen” are two very different statements. Only one of them was being made, and it wasn’t the one being repeated in public.

What Dennis Venema conceded

Here’s the part of the story most articles leave out.

Venema acknowledged he hadn’t himself tested the single-couple hypothesis. He also acknowledged one of his four pillars, the HLA argument, had been weakened by later research.

That argument had come from work by Francisco Ayala in 1994, which claimed certain immune-system gene versions were so ancient and so numerous that tens of thousands of individuals must have been breeding continuously to carry them. Later work showed the apparent age of these gene versions had been overestimated. Recombination—the reshuffling of DNA between chromosome pairs—can make a gene look far older than it is. When that was corrected for, the number of genuinely ancient versions turned out to be far smaller than Ayala had counted.

Venema didn’t abandon his position. He still insists a recent bottleneck of two is ruled out. What he gave up was the certainty. The flat-earth comparison quietly disappeared from the conversation.

WHY THIS EXCHANGE IS WORTH STUDYING

Two scientists disagreed in public, on the evidence, without either one shouting. One asked a fair question; the other admitted the question was fair. The claim that had been presented as settled science turned out to have an untested assumption sitting underneath it.

That is not a scandal. That is how honest science works. But it is a good reason to be slow to surrender a doctrine the moment somebody says the word “consensus”.

A population estimate isn’t a headcount

If you remember one thing from this article, make it this one.

When geneticists say the ancestral human population was about 10,000, they’re not reporting a census. Nobody counted. They’re reporting a number called effective population size, usually written Ne.

Effective population size is the size of an imaginary, idealised population that would produce the pattern of genetic variety we actually see today. That imaginary population is assumed to mate completely at random, stay the same size, experience no natural selection, and breed in neat separate generations. No real population has ever behaved like that.

So Ne isn’t an observation. It’s the output of a model. Feed the model different assumptions and it gives us different numbers. In living species today, effective population size often differs from the real headcount by a factor of ten or more.

Here are the assumptions doing the work:

THE ASSUMPTION BUILT INTO THE MODEL	WHAT HAPPENS IF IT ISN’T TRUE
The DNA studied is neutral—natural selection is not acting on it	If selection preserves variety at a gene (which is exactly what immune-system genes are built to do), the estimated population size is pushed far too high
Everyone mated at random across the whole population	Real people live in villages, clans and regions. Structure in the population distorts the estimate
Mutations happened at a steady, known rate	Mutation rates are measured in different ways that give different answers. Change the rate and every date changes with it
Recombination maps measured today applied equally in the deep past	The chunk-shuffling clock is calibrated on modern data and then run backwards for hundreds of thousands of years
The first couple carried only one version of each gene between them	This is the big one. If a created couple began with up to four different versions of each gene, most of today’s variety needs no long history at all
No sudden, brief population events occurred	A short squeeze followed by rapid growth is precisely the event these methods are least able to see

Notice the fifth row. It’s not a discovery. It’s a rule that was decided before the calculation began: every difference we see today must have been produced by accidental mutation building up from a single starting version.

That rule isn’t neutral. It quietly assumes the answer to the very question being asked. If a designer created a first couple already carrying genetic variety, the rule is simply false, and every date built on it needs a rejig.

That’s where the design perspective actually contributes something. It doesn’t require rejecting the data. It removes one assumption—an assumption that was never tested, only presumed.

THE “MITOCHONDRIAL EVE” AND “Y-CHROMOSOME ADAM” CONFUSION

We often see headlines about these two. They’re almost always misunderstood, in both directions.

Mitochondrial Eve is the woman from whom all living people inherit their mitochondrial DNA, which passes only from mother to child.

Y-chromosome Adam is the man from whom all living men inherit their Y chromosome, which passes only from father to son.

They don’t need to have been alive at the same time, and standard estimates place them at different dates.

Their existence doesn’t mean only two people were alive. Many other people could have lived alongside them whose particular lines simply died out.

Equally, their existence does not disprove Adam and Eve. These are calculations about two specific pieces of DNA, not a headcount of humanity.

Be careful with this evidence. It doesn’t do the work either side often wants it to do.

Testing the other possibility: Gauger and Hössjer

Ann Gauger, a molecular biologist, worked with mathematician Ola Hössjer and geneticist Colin Reeves to do what nobody had done: actually build the model and run it.

They removed the hidden assumption. Instead of requiring all variety to grow from one starting version, they allowed a founding couple to carry up to four different versions of each gene from the beginning. Then they ran forward-time simulations to see what the population would look like today.

The result: a single-couple origin is mathematically compatible with the genetic variety we observe. The date of that couple depends heavily on how much created variety we allow at the start.

However, let’s be honest about what this does and doesn’t prove.

WHAT THE GAUGER AND HÖSSJER WORK DOES ESTABLISH	WHAT IT DOES NOT ESTABLISH
That a single-couple origin hasn’t been mathematically excluded	That a single-couple origin has been positively demonstrated from the data
That the exclusion claim depended on an assumption, not a measurement	That created genetic variety has been independently observed
That the question is open, and open questions deserve testing	That any particular date for Adam has been fixed by genetics

The wider point is one that Douglas Axe and Michael Behe have pressed for years in a different area of biology: the question isn’t whether the mathematics can be made to work, but whether the mechanism being assumed is actually up to the job it’s being given. Assumptions about what mutation can achieve, unexamined, quietly shape the results that come out the other end.

Joshua Swamidass and the genealogical Adam

The most inventive response came from S Joshua Swamidass, a computational biologist at Washington University. In The Genealogical Adam and Eve (2019), he refused to fight on genetic ground at all.

His key distinction is simple once we see it, and genuinely important.

TWO KINDS OF ANCESTRY—DON’T CONFUSE THEM

Genetic ancestry means an ancestor who gave us some of our DNA.

Genealogical ancestry means an ancestor on our family tree, whether or not any DNA came down to us.

Our ancestors double every generation we go back: 2 parents, 4 grandparents, 8 great-grandparents. Go back a few hundred years and the number is in the millions. But we only carry a limited amount of DNA. Beyond about ten generations, most of our genealogical ancestors gave us no measurable DNA at all.

So a real ancestor of ours may be genetically invisible. Not hidden—genuinely, mathematically invisible.

The mathematics here isn’t controversial. Work by Joseph Chang, and later by Douglas Rohde, Steve Olson and Joseph Chang, showed the most recent person who is a genealogical ancestor of everyone alive today lived surprisingly recently—possibly only a few thousand years ago. Go a little further back and we reach a point where every person then living is either an ancestor of everybody alive today or of nobody.

Swamidass draws the conclusion: Adam and Eve could have been created directly, a few thousand years ago, in the Ancient Near East, and be the genealogical ancestors of every human alive by the first century. Population genetics would have nothing to say about it, because they’d leave no genetic signature to find.

It’s a clever argument. On its own terms, it’s also correct.

Where the genealogical model runs into trouble

The difficulty isn’t scientific. It’s what the model requires us to believe about the people outside the garden.

On this view, Adam isn’t the first human being. Other biologically human people were alive at the same time, outside Eden, and Adam’s descendants later married into them. Swamidass himself doesn’t decide whether those people bore the image of God. He passes the question to theologians.

So let’s take the question.

WHAT THE MODEL PRESERVES	WHAT THE MODEL COSTS
A real, specially created Adam and a real Eve	Adam is no longer the first man, and Eve is no longer, at the time she is named, the mother of all living (Genesis 3:20)
A real fall, at a real moment in history	At the moment of the fall, human beings exist who aren’t descended from Adam at all
Descent from Adam for everyone by the time of Christ	Paul’s claim that God made every nation from one man (Acts 17:26) becomes true by later intermarriage rather than by creation
Complete immunity from genetic objections	Image-bearing has no clear beginning and no clear boundary—it spreads gradually through a mixed population

The confessional problem, stated plainly

The Westminster Confession says our first parents, being the root of all mankind, had the guilt of their sin imputed and their corrupted nature conveyed to all their posterity descending from them by ordinary generation.

Read that phrase slowly. Both the guilt and the corruption travel along a single road: ordinary descent from Adam.

The Heidelberg Catechism says the same in its own words: our corrupt nature comes to us from the fall and disobedience of our first parents. The Canons of Dort trace the corruption of all people back to Adam in the same way.

Now apply the genealogical model. For a period after the fall, there are human beings who’re not Adam’s descendants. Two options follow, and neither is comfortable:

If they bear God’s image, then there are image-bearing human beings outside the covenant God made with Adam, and the confession’s account of how guilt reaches every person doesn’t reach them.
If they’re not image-bearers, their children and grandchildren acquire it gradually through intermarriage. Notice how we’re then left with a humanity in which being made in God’s image is a matter of degree.

The strongest objection to our position. And our answer

Here’s the best case for the other side:

Federal headship, the objection runs, is a covenantal arrangement, not a biological one. Adam represents us because God appointed him to, not because of DNA. So why could God not appoint Adam as covenant head over all image-bearers, whether or not they descend from him? That would solve the whole problem at once.

The idea does sound coherent. But the older writers never separated the two, and they had a reason.

Francis Turretin insisted Adam’s headship was both natural and moral. He’s our covenant head because he is the root of the race. The covenant is made with him as the fountain of all humanity, not with a man picked out from among many. Cut headship loose from descent, and God’s choice of Adam becomes formally arbitrary—he might just as well have chosen someone else. That’s exactly the conclusion the older writers were determined to avoid, because an arbitrary appointment makes imputed guilt look unjust in a way that natural headship doesn’t.

Why Romans 5 cannot bend

We now come to the heart of the debate. Everything above is a matter of models and probabilities. This isn’t.

The word “one” is not decoration—it is the argument

Read Romans 5:12–19 and count. One man. One trespass. One act of righteousness. One man’s obedience. Paul doesn’t mention Adam in passing. Singularity is the load-bearing beam of the entire passage.

The comparison is the whole point

Therefore, as one trespass led to condemnation for all men, so one act of righteousness leads to justification and life for all men. For as by the one man’s disobedience the many were made sinners, so by the one man’s obedience the many will be made righteous (Romans 5:18–19).

The structure is as … so also. Adam is the pattern; Christ is the fulfilment. Whatever Adam is, Christ answers to.

That’s why the passage cannot be softened. If Adam is a symbol standing for humanity in general, then by Paul’s own logic Christ becomes a symbol standing for humanity in general. Empty the type of historical reality and the antitype follows it out the door.

Herman Ridderbos, John Murray, CEB Cranfield and Douglas Moo differ from one another on plenty, but on this they all agree: the comparison only carries weight if both men are real and both acts are real.

Paul blocks the escape route himself

Some readers try to take Romans 5:12 to mean simply that everyone dies because everyone personally sins. Paul closes that door two verses later.

For sin indeed was in the world before the law was given, but sin is not counted where there is no law. Yet death reigned from Adam to Moses, even over those whose sinning was not like the transgression of Adam (Romans 5:13–14).

Death reigned over people whose own sins couldn’t account for it. Something other than their personal record was already counting against them. That’s the doctrine of imputed guilt, stated by Paul before any confession put it into a paragraph.

And Paul says it twice, in two different letters

For as by a man came death, by a man has come also the resurrection of the dead. For as in Adam all die, so also in Christ shall all be made alive (1 Corinthians 15:21–22).

Thus it is written, “The first man Adam became a living being”; the last Adam became a life-giving spirit (1 Corinthians 15:45).

Two letters, two audiences, two occasions—one architecture. Paul isn’t leaning on a passing illustration. He’s building on a foundation.

WHAT ROMANS 5 SETTLES, AND WHAT IT LEAVES OPEN

It settles: that Adam was a particular man; that his one act had legal consequences for everyone he represented; that the whole logic of justification rests on the match between his act and Christ’s.

It leaves open: when Adam lived; how old the earth is; how much biological continuity exists between Adam’s body and other creatures.

This matters. The boundary is narrower than most readers fear—but inside that boundary, there’s no room to negotiate.

What’s not at stake here

It is worth saying clearly, because many readers arrive at this subject expecting a different argument.

This is not an argument about the age of the earth.
It’s not an argument about whether the days of Genesis 1 are 24 hours long.
It’s not even, strictly, an argument about whether human bodies share ancestry with other creatures.

Vern Poythress, who trained as a mathematician before turning to theology, has argued for years the deepest problem in these debates is rarely the data itself. It’s the unexamined philosophy that decides in advance what kinds of explanation are allowed to count. That’s exactly what we’ve found here. The data didn’t exclude a first couple. An assumption did.

William Lane Craig has taken a different route in In Quest of the Historical Adam (2021), placing Adam far back in time—around 750,000 years ago—in exactly the window Buggs identified as still open. That preserves descent from one man for everybody. But he pairs it with a reading of Genesis 1–11 as mytho-history and a rejection of imputed guilt. He solves the scientific difficulty by giving away the theological point. That’s a trade we dare not make.

So: can two be enough?

Genetically, the honest answer is that nobody has shown two couldn’t be enough. The claim that they could not rested on an assumption nobody had tested—and when it finally was tested, it didn’t hold up as firmly as advertised.

Theologically, the answer is that they must be. Not because two is a comfortable number, but because Paul’s gospel is built on a comparison between one man’s disobedience and one man’s obedience, and a comparison needs two real terms.

The church has never required a date for Adam. It has never required a position on the age of the earth. What it has required is this: a real Adam, a real covenant, a real fall, and real guilt reaching every human being who has ever lived—so a real Saviour can reach them too.

If Adam’s disobedience didn’t truly reach us, then tragically, neither does Christ’s obedience. That’s the cable everything hangs from. It’s not a cable we can afford to cut, and thankfully, science hasn’t asked us to.

Tough Questions, Honest Answers

Where did Cain get his wife?

He married a sister or a niece. Genesis 5:4 tells us Adam had other sons and daughters, and Genesis names only those people who matter to the story it is telling. This is not an embarrassed guess—it is simply what the text says once you stop assuming that everyone alive got a mention. The marriage laws in Leviticus 18 that forbid marrying close relatives came much later, at Sinai, and were given for reasons that did not yet apply in the first generation.

Humans and chimpanzees share about 98% of their DNA. Doesn’t that settle the question?

That figure is less solid than it sounds. It comes from comparing regions that align well and setting aside the regions that do not, and different methods of counting give noticeably different percentages. But even if the number were exact, similarity by itself does not tell you its cause. Shared design and shared descent both predict similarity, so the figure cannot decide between them. What matters is which explanation handles the differences—and the differences, particularly in gene regulation and in genes with no counterpart in any other species, are where the real argument lies.

Were Neanderthals and Denisovans human beings made in God’s image?

We should be careful here, because we’re reasoning beyond what Scripture directly addresses. If Adam is placed early enough, these groups fall within his descendants, and there is no difficulty—they are simply human populations who lived and died. What we can say is that image-bearing is not something a fossil can measure. It is a status God confers, not a trait we can dig up. Anyone descended from Adam bears it, and anyone who does not is not our concern to judge.

Does Luke’s genealogy require Adam to be a real person?

Luke traces Jesus’ line backwards and ends with the son of Adam, the son of God (Luke 3:38). He does not change his manner of speaking partway through, as if the recent names were history and the early ones were poetry. Every name in that list is treated the same way. A genealogy that runs from a real Jesus through a real David to a symbolic Adam would be a strange document indeed—and Luke, who opens his Gospel by promising an orderly and researched account, gives no hint that he thought he was doing anything of the kind.

If Adam’s children married each other, wouldn’t that have been genetically disastrous?

The danger in close-relative marriage comes from inherited copying errors. When both parents carry the same faulty version of a gene, their child may receive two copies and suffer for it. But those faults accumulate over many generations. A freshly created couple would carry none, and their immediate children would carry very few. The risk grows with time, which fits neatly with the fact that God gave the laws against close-relative marriage centuries later, not in the first generation.

Is original sin passed down through the body, or is it purely a legal matter?

Both, and the distinction matters more than it first appears. Guilt is imputed—it is credited to our account because Adam acted as our representative. Corruption is conveyed—we actually receive a nature bent away from God. The Westminster Confession holds the two together and ties both to descent from Adam. This is precisely why models that keep Adam as a legal representative while loosening the line of descent do not solve the problem: they save the imputation and lose the conveyance.

Does this debate settle how old the earth is?

No, and it’s worth saying so plainly. Nothing in this article depends on the age of the earth, and Christians who hold different views on that question can agree on everything argued here. The population-genetics debate is about how many people, not about how many years. Keeping the two questions apart is not a compromise—it is simply accuracy, and it lets us put our weight where the real issue lies.