Ancient DNA

DNA has been recovered and sequenced from fossils. Neanderthal and Denisovan genomes are now available at high coverage, and they show that these populations interbred with modern humans and that their DNA is present in living people.

This is a distinct line of evidence from anatomy or the fossil sequence: it reads genetic material directly from extinct populations.

What has been sequenced#

Svante Pääbo received the 2022 Nobel Prize in Physiology or Medicine for this work.

The Denisovans are the strongest single case. They were identified from a fingerbone fragment and two molars, and the genome established them as a distinct population before anyone could describe their skeleton. A prediction about the anatomy of a group nobody had seen was later confirmed when a mandible from Xiahe, Tibet, was matched by protein analysis.

What the genomes show#

The distribution of Neanderthal DNA is not uniform. It clusters outside Africa, which matches a model in which modern humans left Africa and encountered Neanderthals already living in Eurasia. Sub-Saharan African populations carry much less, and the small amount they do carry is attributed to later back-migration.

Why the results are trusted#

Ancient DNA is fragmented, chemically damaged, and easily contaminated by the DNA of the people handling it. The field had a serious early credibility problem, including claims of dinosaur DNA and insect DNA from amber that could not be replicated.

The controls that were developed in response:

• Damage signatures. Genuinely old DNA carries a characteristic pattern of cytosine-to-thymine substitutions at fragment ends, produced by deamination over time. Modern contaminant DNA does not. The pattern is used to authenticate sequences.
• Fragment length. Ancient DNA is broken into short pieces, typically under 100 base pairs, with a length distribution that follows a predictable decay.
• Clean-room protocols. Extraction is done in dedicated facilities with positive air pressure, and all handlers are sequenced so their DNA can be excluded.
• Independent replication. Key results are reproduced in separate laboratories on separate extractions.

Common objections#

"DNA cannot survive that long"

A reasonable objection, and it was the field's central problem. DNA degrades through hydrolysis and oxidation, and early claims of very old DNA were wrong.

The measured half-life gives the range. Work on moa bones estimated a DNA half-life of about 521 years for a 242-base-pair fragment under typical conditions, implying that useful sequence should be unrecoverable beyond roughly 1.5 million years, and much less in warm environments.

That prediction has held. The oldest genomes come from cold, stable sites — permafrost, deep caves — and the oldest hominin DNA is around 430,000 years, well inside the limit. Claims of dinosaur DNA are rejected by the same reasoning, by the same researchers.

"These are just contamination from modern human handlers"

The correct question to ask, and it invalidated several early studies.

Contamination is detectable rather than merely denied. Modern DNA lacks the deamination damage pattern and the short fragment length distribution, so contaminant reads can be identified and excluded rather than assumed absent.

The decisive point is that the sequences are not modern human. Neanderthal genomes differ from every living person's at consistent positions, and the Denisovan genome represented a population nobody knew existed — contamination cannot produce a sequence that matches no known source.

"Interbreeding proves Neanderthals were just humans, which supports one created kind"

The strongest young-earth reading, and it has been argued by Answers in Genesis and Todd Wood. It is partly correct: Neanderthals were close enough to interbreed and produce fertile offspring, and treating them as non-human is not supported.

Interbreeding does not establish a single created kind. Many pairs of species interbreed — lions and tigers, polar and brown bears, coyotes and wolves — and are not therefore one population.

The genomes also show what the model needs to explain. The lineages separate around 600,000 years ago, which exceeds a young-earth chronology by two orders of magnitude, and the separation is calculated from mutation counts rather than assumed.

The pattern of introgression is also directional and regional in a way a single recent population does not predict.

"Molecular clock dates assume constant mutation rates"

A real assumption, and rates do vary between lineages and across the genome.

Modern rates are measured directly rather than inferred. Sequencing parent-offspring trios counts new mutations per generation, which is where the human rate of roughly 1.2 × 10⁻⁸ per base pair per generation comes from.

Ancient genomes also provide an internal check. A sample of known radiocarbon age carries fewer accumulated mutations than a modern one, and the deficit corresponds to the elapsed time — which calibrates the clock against an independent dating method rather than against itself.

What the evidence shows#

Genetic material has been recovered from hominin fossils up to about 430,000 years old, using authentication methods that distinguish ancient DNA from modern contamination by its chemical damage and fragmentation.

Neanderthals and Denisovans were distinct populations that interbred with modern humans. Their DNA is present in living people at measurable and regionally structured frequencies, and specific inherited variants have known functions.

The Denisovans were identified from DNA alone, before any diagnostic skeletal material was recognised — a case where the genetic method made a claim about a population's existence that was subsequently confirmed by other evidence.

Related material is in Hominin Fossils and Chromosome 2.