Endogenous Retroviruses

An endogenous retrovirus is a viral infection that became heritable. A retrovirus copies its genome into a host chromosome; if it does this in a sperm or egg precursor rather than an ordinary body cell, the insertion is passed to every descendant of that individual.

Vertebrate genomes are full of them. Roughly 8% of the human genome consists of retroviral remnants, and comparable fractions appear in mice, cats, sheep, and birds.

The evidence is not that these sequences exist. It is where they sit. Insertion position is close to random across billions of base pairs, so two species carrying the same viral family at the same coordinate is a record of a single infection in a shared ancestor — or an extraordinary coincidence, repeated tens of thousands of times.

Endogenization observed in progress#

The process is not inferred from ancient sequences alone. It is happening now, in a living species.

Koala retrovirus (KoRV) is midway through invading the koala germ line. Tarlinton, Meers and Young (2006) showed that northern Australian koalas carry the virus in every cell, including germ cells, while some southern island populations carry none. Museum specimens allow the spread to be tracked across the last century and a half.

That is the mechanism being watched in real time: a virus that is exogenous in one population and inherited in another, with the geographic boundary still visible.

Sheep provide the complementary picture from the other end. The endogenous Jaagsiekte sheep retroviruses sit in the genome as fixed insertions, some of them older than the domestic sheep lineage, alongside a still-circulating exogenous relative that causes lung tumours. The two forms can be sequenced side by side.

Why position carries the argument#

Retroviral integration shows mild preferences — some families favour transcriptionally active regions — but nothing approaching site specificity. Across a genome of billions of positions, independent infections in separate lineages land in different places.

This makes every insertion a test. Common descent predicts that shared insertions should form a nested hierarchy: the deepest insertions shared by the widest groups, recent ones confined to close relatives, and no insertion distributed in a pattern that crosses branches.

The prediction is falsifiable in an unusually clean way. A single ancient insertion found in, say, cats and cows but not in the lineages between them would not fit any tree.

Insertions used to settle a disputed relationship#

The strongest demonstration is a case where the insertions were consulted before the answer was known.

Whale ancestry was contested through the 1990s: morphologists linked whales to an extinct group of hoofed carnivores, while early molecular work pointed toward hippopotamuses, a result many anatomists rejected.

Nikaido, Rooney and Okada (1999) surveyed retroposon insertions across even-toed ungulates and found insertions shared by whales and hippos and by no other group, along with a nested set of older insertions ordering the rest of the family. Their conclusion — hippos are the closest living relatives of whales — was subsequently supported by fossil discoveries and by whole-genome sequencing.

The same approach has resolved relationships among felids, among primates, and among birds. In each case, insertion data recovered the tree that anatomy and unrelated gene sequences had produced independently.

Shared damage, not just shared presence#

A retroviral insertion arrives flanked by two identical long terminal repeats. Over time these sometimes recombine and excise the interior, leaving a lone "solo LTR" at the original site.

When two species share a solo LTR at matching coordinates, that records two events at the same location: one insertion, then one deletion, both in a shared ancestor. Independent origin would have to reproduce both.

Related non-retroviral cases point the same way. Horie et al. (2010) found bornavirus sequences — from a virus family that does not integrate at all — embedded in mammalian genomes at matching positions across species, evidence of ancient insertion events preserved by inheritance.

Common objections#

"Many ERVs have functions, so they were designed rather than inherited"

The premise is correct and well documented. The clearest example is syncytin, a retroviral envelope protein co-opted for placental development and essential to it.

Co-option is expected. A sequence sitting in a genome for millions of years occasionally lands somewhere useful and is retained.

The distribution is what the design reading cannot accommodate. Syncytin was captured from different, unrelated retroviruses independently in primates, rodents, carnivores, and marsupials (Dupressoir et al., 2012). Separate lineages solving placentation by capturing separate viruses is a record of separate historical accidents.

And function does not explain nested placement. Useful components would be distributed by need. Tens of thousands of mostly-defective viral remnants distributed in a pattern that reproduces the anatomical tree would not.

"The same virus infected both lineages independently"

The same virus family infecting multiple lineages is ordinary — that is how KoRV is spreading through koalas now.

The argument rests on coordinates, not on presence. Independent infections put insertions in different places. What is observed across primates is the same family at the same base pair, on the order of 200,000 times.

Genuine independent insertions into the same region do occur and are documented, which is why they are reported as noteworthy exceptions rather than as the general pattern.

"These sequences aren't viral at all — they're original genomic elements that became viruses"

A position advanced by some creationist authors, and it inverts the direction of transfer: ERVs are held to be designed elements that later escaped to become pathogenic viruses.

It runs into the koala case, where the direction is directly observed — an exogenous, infectious virus entering the germ line of a species that previously lacked it, with uninfected populations still available for comparison.

It also has to explain why the "original" elements carry the structural signature of retroviral integration: flanking long terminal repeats, target-site duplications of host sequence created by the integrase enzyme, and degraded copies of the gag, pol, and env genes in their normal viral order.

"This assumes common descent in order to demonstrate it"

The observation does not. Determining that a sequence occupies the same coordinate in two aligned genomes requires no assumption about how it arrived.

Common descent supplies a prediction about the pattern those coordinates should form. That prediction could have failed at any of hundreds of thousands of sites, and the whale–hippo case shows it can be run before the answer is known.

What the evidence shows#

Retroviral endogenization is directly observed in koalas, with infected and uninfected populations of the same species available for comparison.

Insertion position is close to random, so matching positions in separate species are not expected from independent infection. Shared insertions across mammals, birds, and primates form a nested hierarchy matching trees built from anatomy and from unrelated sequence data, and shared solo LTRs add a second event at the same site.

The method has predicted a contested relationship — whales and hippopotamuses — that later fossil and genomic work confirmed.

Applied specifically to human ancestry, the same evidence is set out in Endogenous Retroviruses in humans. Related molecular evidence appears in Pseudogenes and Common Descent.