Do animal viruses need to adapt to humans before causing an outbreak?

Not necessarily. A 2026 study analyzing multiple zoonotic outbreaks found that most animal viruses did not show detectable genome-wide selection changes before spilling into humans. The traditional model holds that viruses must gradually adapt through directional selection to become human-compatible, but the genomic evidence from H1N1pdm09, Ebola, Marburg, mpox, and SARS-CoV-2 does not support this for most recent outbreaks.

What is the RELAX framework and how does it detect viral adaptation?

RELAX is a phylogenetic framework that detects changes in selection intensity across viral genomes. It uses the dN/dS ratio, comparing protein-changing mutations to silent ones, and tracks a parameter called K. A K value above 1 indicates intensified selection; below 1 suggests relaxed selection. The framework classifies evolutionary branches into natural host background, pre-outbreak stem branches, and human outbreak branches to identify shifts in selection regimes.

What did the study find about SARS-CoV-2 origins?

The analysis of 31 SARS-CoV-2-like sarbecovirus genomes across 15 non-recombinant regions showed no significant selection intensification or relaxation on the pre-outbreak stem branch compared to bat host background. This result held across different methods for defining recombination breakpoints. It suggests that if the SARS-CoV-2 precursor circulated in an intermediate host, that phase was too brief to leave a genome-wide selection signal. The alternative hypothesis of lab passage or artificial selection also finds no support in the selection data.

Did any natural outbreak show pre-outbreak adaptation?

Yes — SARS-CoV (2002-2004) was the one natural outbreak where the researchers detected a selection signal. When combining the stem branch with civet cat and early human isolates, they found selection relaxation relative to bat background. This suggests SARS-CoV spent enough time circulating in intermediate hosts (likely civet cats) to leave a detectable genomic footprint. This exception actually strengthens the methodology by showing it can detect adaptation when it occurs over a sufficiently long period.

What does the 1977 H1N1 case reveal about viral origins?

The 1977 H1N1 re-emergence is an anomaly. The virus accumulated only 48 substitutions across its entire coding genome over roughly 27 years, and showed elevated dN/dS with significant selection relaxation (K ≈ 0.71). This pattern matches what you would expect from a virus that was cold-stored and then lab-passaged, not one evolving naturally in animals or humans. It provides genomic evidence supporting the long-held suspicion that the 1977 H1N1 strain was associated with laboratory activities.

What does this mean for pandemic surveillance?

The findings suggest that surveillance should not only focus on viruses showing obvious signs of adapting to humans, since many zoonotic viruses may already be near the cross-species transmission threshold. The limiting factor in spillover is often ecological exposure, not viral evolution. The study also provides a framework for distinguishing natural spillover from intermediate host adaptation and lab-related events based on their genomic signatures. The better question for pandemic prevention may be 'what ecological and demographic conditions make spillover more likely?' rather than 'is this virus adapting to humans?'

What are the limitations of phylogenetic selection analysis?

The method has several limitations. Complex recombination, repeated host-switching, and sampling biases can affect branch classification. Short adaptation windows with only a few key mutations may not produce enough substitutions to trigger a genome-wide signal. The analysis depends on homologous nucleotide site comparisons, so it cannot fully capture adaptations from insertions, deletions, or structural changes. The authors recommend using the framework alongside epidemiological, ecological, functional, and field sampling evidence rather than as a standalone tool.

Does this mean viruses never acquire adaptive mutations before crossing to humans?

No. The study explicitly does not claim that viruses never acquire adaptive mutations before spillover. A handful of key functional mutations that matter for transmission may not accumulate enough substitutions to register as a genome-wide selection regime shift. The RELAX framework requires enough time and enough substitutions to detect changes. The conclusion is about the absence of broad, sustained selection regime shifts, not about the absence of any functional adaptation.

Viruses Don't Always Adapt Before Spilling Over: New Study Rewrites How Pandemics Begin

The standard story about animal viruses jumping to humans usually goes something like this. A virus hangs out in bats or pigs, picks up mutations over time, and eventually becomes human-ready. Then it spills over, and we get a pandemic.

Nice story. But the genomic data doesnt always cooperate.

A 2026 study from UC San Diego, Temple University, and a few other labs dug into the evolutionary fingerprints of the most consequential viruses from the last twenty years. They came back with findings that upend a lot of the conventional wisdom.

The Two Models

So heres the core question: do animal viruses actually need to adapt to humans before triggering an outbreak?

One camp says yes. A virus passes through intermediate hosts or brief human-to-human chains, collects adaptive mutations, gets better at infecting us, and leaves evolutionary breadcrumbs in its genome.

The other camp says not really. A lot of animal viruses already have the capacity to infect people. They just need the right circumstances. Cross-species transmission in this view is more about screening than training. The viral pool already contains variants sitting near the transmission threshold. An outbreak happens when ecology, exposure frequency, and contact networks create the right sieve.

Which one is right?

A Different Way to Read Viral History

Instead of hunting for the one magic mutation, the team used a phylogenetic framework to track selection intensity across whole genomes. The tool is called RELAX and it runs on dN/dS, which compares the rate of protein-changing mutations to silent ones.

If a virus is under heavy pressure to adapt, the ratio of protein-changing mutations goes up. If selection relaxes, that ratio drops. The researchers split the evolutionary tree into three kinds of branches: natural host background, pre-outbreak stem leading to humans, and early human outbreak. Then they checked whether the stem branch looked different from the background.

They tracked a parameter called K. Above 1 means selection is cranking up. Below 1 means its easing off.

Six Viruses, Six Answers

H1N1pdm09. The 2009 swine flu came from a reassortment event. The stem branch represented about 9 to 12 years of evolution, but it looked totally normal compared to swine flu background. No elevated selection, no relaxation. The virus that caused a pandemic was evolving just like any ordinary pig virus.

Ebola 2013-2016. Biggest outbreak ever recorded. The stem branch showed zero selection shift relative to the bat reservoir. Not before crossing. After hitting human transmission, selection changes did appear – but not before.

Marburg 2004-2005. Same result. Bats and the pre-outbreak branch were statistically identical on selection intensity.

Mpox 2022. This was an interesting test because clade IIb had been circulating unnoticed in Nigeria for years. If long hidden transmission involved adaptation, youd expect a signal. It wasnt there. The stem branch looked normal. But once the virus started sustained human spread, selection intensified. The method clearly distinguished before from after.

SARS-CoV 2002-2004. This was the one natural case where they found a signal. Combining the stem branch with civet cat and early human samples showed selection relaxation relative to bats. SARS-CoV apparently spent enough time in intermediate hosts to leave a genomic footprint. Which actually validates the approach – when adaptation lasts long enough, the method catches it.

SARS-CoV-2. The heavy hitter. Using 31 related sarbecovirus genomes across 15 non-recombinant regions, the stem branch showed no significant selection change compared to bat background. Different methods of defining recombination breakpoints gave the same answer.

If the SARS-CoV-2 precursor circulated in an intermediate host before hitting humans, that phase was too short to leave a genome-wide signal. The alternative – lab passage or artificial selection – also gets no support from these data. When they combined the stem branch with early human sequences from the first three months, selection changes did show up. The point is, the pre-outbreak branch itself looked like any other bat sarbecovirus.

1977 H1N1: The Freezer Anomaly

The 1977 H1N1 re-emergence has always been weird. The strain was practically identical to 1950s viruses. Hard to explain without invoking cold storage and lab work.

The new analysis strengthens that case. The stem branch accumulated only 48 substitutions across the whole coding genome over 27 years. Thats glacially slow. It also showed elevated dN/dS and significant relaxation (K around 0.71). This is what youd expect from a virus that was frozen and then passaged in a lab, not one evolving naturally in animals or people.

The Positive Controls

A fair question: what if the method just misses everything?

The researchers ran a bunch of control systems to check. Lab-passaged viruses, attenuated vaccine strains, cold-adapted flu, ferret-passaged H5N1. Every single one showed detectable selection shifts. Measles and mumps vaccine strains showed clear relaxation compared to wild viruses. The “ice lake” H1N1 samples had signatures closer to lab passage than natural preservation.

These controls are important. They show the method works. And they build a reference library where natural spillover, intermediate host adaptation, and lab passage all leave different fingerprints.

What This Actually Means

Lets be precise about what this study says and what it doesnt.

It doesnt say viruses never adapt before crossing to humans. A couple of key mutations that matter for transmission may not add up to enough substitutions to trigger a genome-wide signal. RELAX needs enough time and enough changes to work.

What it does say: across most recent zoonotic outbreaks, we cant find broad, sustained selection changes in the pre-outbreak stem branch. The viruses behind H1N1pdm09, West Africa Ebola, Marburg, mpox, and SARS-CoV-2 were not undergoing prolonged directional evolution targeting humans before they emerged.

So the mental model shifts. Outbreaks arent the final chapter of a long adaptation narrative. Theyre screening events. The reservoir population already has variants with different levels of human compatibility. Most exposures go nowhere. The ones that succeed do so because of ecological opportunity – the right contact, the right density, the right timing – not because the virus was practicing.

What This Means for Surveillance

This changes how we watch for the next one.

We cant just track viruses showing obvious signs of human adaptation. Some bat sarbecoviruses may already be closer to the transmission threshold than we think. The bottleneck isnt always viral evolution. Its exposure.

The study also gives us a way to distinguish outbreak scenarios by their genomic signatures. Natural spillover, intermediate host adaptation, and lab-related events all leave different marks. Systematic animal surveillance plus early outbreak sequencing can make this a standard diagnostic tool.

And maybe the biggest shift: asking “is this virus adapting to humans” is the wrong question if the answer is usually “no, and it doesnt need to.” The better question is “what conditions make spillover happen?”

The Limitations Worth Knowing

The study doesnt pretend to be perfect. Complex recombination, frequent host-switching, and sampling gaps can mess up branch classification. They couldnt even analyze MERS-CoV because the recombination history was too messy.

Short adaptation windows are a blind spot. If the important changes happened fast with just a couple of mutations, RELAX might miss them. And since the method compares homologous sites, it cant fully capture adaptations from insertions, deletions, or structural changes.

These arent fatal. Theyre honest constraints. The authors say the framework works best when combined with epidemiology, ecology, lab experiments, and field sampling.

The Takeaway

This study flips the script on viral emergence. The old picture was viruses slowly training to become human pathogens. The new one, backed by genome-wide selection data from multiple major outbreaks, suggests many zoonotic viruses already sit near the cross-species threshold in their natural hosts. Its not adaptation that filters them. Its opportunity.

Thats an unsettling thought. It means the next pandemic virus could already be out there, doing fine in bats or birds, not bothering to prepare for us. Just waiting for the right door to swing open.

The real challenge is not predicting whether a virus will adapt. Its predicting when we hand it the chance.

The Two Models

A Different Way to Read Viral History

Six Viruses, Six Answers

1977 H1N1: The Freezer Anomaly

The Positive Controls

What This Actually Means

What This Means for Surveillance

The Limitations Worth Knowing

The Takeaway

You May Also Like

How Dissolution Testing Ensures Your Dog’s Praziquantel Chewable Tablets Actually Work

Veterinary Oxytetracycline: A Comprehensive Guide for Pig Clinical Applications

FDA’s Approval of Faunamor: A Breakthrough for Ornamental Fish Health