Scientists debunk key evolutionary theory as "statistical noise"

There is an apparent paradox in evolutionary theory. The rate of evolution seems to run like a sprinter over a few million years yet crawl over tens of millions, a pattern so regular that textbooks use it to link microevolution with macroevolution. 

For decades, biologists treated that apparent increase in evolutionary rate as biological truth, arguing that young groups diversify faster than old ones, shed species more quickly, and tweak body plans at record pace.

Evolutionary rate paradox

Early molecular studies showed that mutation rates estimated from contemporary samples were 5-20 times higher than those calibrated with fossils drawn from deep time, jolting confidence in the evolutionary “clock.”

Paleontologists spotted the same mismatch in the rock record: the evolutionary rate in newly arisen dinosaur lineages seemed faster than in lineages that had persisted for tens of millions of years.

Some theorists proposed that ecological opportunity drives early bursts, while competitive saturation dulls the pace later on.

Others hinted that developmental constraints lock traits into “optimal” configurations and slow further change, an idea that made intuitive sense but resisted firm testing.

Statistical noise warps the picture

A research team led by Brian C. O’Meara at the University of Tennessee and Jeremy M. Beaulieu at the University of Arkansas has recently investigated this paradox of evolutionary theory.

O’Meara and Beaulieu built a simulation that treats every evolutionary rate estimate as a fraction – change in trait divided by elapsed time – and then distributed measurement error evenly across every numerator.

Because the denominator (time) varies wildly while the error does not, the ratio traces a hyperbola that plummets with age – exactly the pattern that inspired all those stories of adaptation.

“By employing a novel statistical approach, we found that this time‑independent noise … makes it seem like evolutionary rates increase over shorter time frames when, in fact, they do not,” explained O’Meara. 

When the team randomized real datasets, shuffling trait differences while leaving times intact, the fake data reproduced more than 85 percent of the scatter found in genuine evolutionary rate studies. This level of mimicry stunned even longtime rate‑modelers.

Running the numbers

The analysis covered molecular substitutions in birds and primates, body‑size shifts in mammals and lizards, speciation surges in flowering plants, and extinction tallies drawn from conservation lists. 

Across five independent datasets totaling almost 9,000 evolutionary rate estimates, the fitted curves showed the same steep drop in the first 10 million years followed by a flattening tail, just what the noise model predicts.

In one mammal dataset, the median coefficient of variation for repeated measurements at the same age was 0.59, meaning the standard error was more than half the mean value.

Put differently, five percent of those “measurements” would have placed the Eiffel Tower shorter than a giraffe or taller than twice its real height – an error band that is wide enough to drown any subtle biological signal.

Viral evolution and the cosmos

The statistical noise idea echoes work on viral evolution, where apparent mutation‑rate “decay” with time has been traced to mutational saturation and sampling bias rather than to fundamental shifts in replication chemistry. 

Astronomers face a parallel dilemma: the “Hubble tension” arises because recent supernova surveys and ancient cosmic‑microwave‑background data give different expansion rates.

Some physicists now attribute the Hubble discrepancy, at least partly, to mismatched error structures.

Both cases remind scientists that plotting a ratio (rate) against one of its components (time) courts mathematical traps unless errors are modeled explicitly.

The new study argues that evolutionary biology has been caught in just such a trap, mistaking geometry for genetics.

Next steps in evolutionary rate research

Comparative methods that search a phylogeny for bursts of trait change may flag any young clade as “rapidly evolving” even when its true rate of change matches the background, simply because less time has accrued to dilute measurement error.

Diversification analyses that exclude small groups with zero speciation events risk overestimating average rates near the present, further skewing the perceived trend.

O’Meara suggests that researchers of evolutionary theory adopt fixed‑window approaches, examining change per 100,000 years or per 1 million years, much as dendrochronologists study tree rings, thereby holding the denominator constant.

He also urges journals to require explicit error terms for trait means, branch‑length estimates, and dating calibrations – allowing models to partition signal from noise instead of forcing error to zero by fiat.

Why evolutionary theory matters

Far from closing the doors of research into evolution rates, the work clears the field to tackle fresh questions: which traits truly speed up or slow down over time, and under what ecological or genetic circumstances?

With the hyperbolic mirage out of the way, biologists can focus on signals that remain after accounting for error, signals that may illuminate how life diversifies, adapts, and sometimes stalls.

Tools developed for sequencing error in virology or for tip‑dating in paleogenomics could be repurposed to model measurement uncertainty in comparative datasets from diverse branches of the tree of life. 

“Biology is rich in mysteries: actually answering one lets us move on to the next,” wrote the authors in an unpublished summary of their work. The statement highlights their view that solving the rate paradox frees, rather than constrains, evolutionary thought.

The study is published in PLOS Computational Biology.

—–

Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates. 

Check us out on EarthSnap, a free app brought to you by Eric Ralls and Earth.com.

—–