Biologists have reaffirmed a century-old "law" that said evolution cannot reverse itself.
The law, which was put forward by 19th century Belgian paleontologist Louis Dollo, argued that once natural selection for a complex functional trait is relaxed, mutations that degrade the genes needed for the trait accumulate, and the sequence of mutations is unlikely to be exactly reversed.
That means a species trait, once lost to the sands of time through evolution, can never be regained.
But, over the past two decades, many biologists have challenged Dollo's Law, often by using statistical tools to reconstruct trait evolution and ancestry by looking only at existing species and their present-day traits.
Boris Igic, assistant professor of biological sciences, and Emma Goldberg, a post-doctoral student in Igic's University of Illinois at Chicago (UIC) laboratory, became suspicious of those methods, which they also had used in evolutionary studies of plant fertilization.
They found the methods flawed, prompting them to examine the challenges to Dollo's Law.
"We used computer simulations to demonstrate that this methodology consistently gives incorrect results when the loss of a trait truly is irreversible," said Igic.
"Fifteen years of studies have relied on these procedures to show that Dollo's Law is frequently violated. But, they used what we found was faulty methodology," he added.
Igic and Goldberg identified two problems with the challenges: a logical error in how the common ancestor of a group of organisms is treated mathematically, and a disregard for the likelihood that complex characters affect the chances either of speciation - the origin of new species - or of extinction.
The biologists suggest a better way of testing Dollo's Law.
They used mathematical tools to refute two Dollo challenges which claimed certain animals regained traits such as winged flight and sexual reproduction, and say many other claims of regained complex traits should be re-evaluated.
According to the researchers, Dollo was correct in saying that transitions between two character states are unidirectional.