New tools filter noise from evolution data

While rates of evolution have appeared to accelerate over short time periods, new analysis suggests that statistical noise is affecting the data patterns. A professor at the University of Tennessee, Knoxville, and his colleague have developed new tools to help researchers filter the data.

“Our work is an important step in showing how substantially error can affect rate estimates,” said Professor Brian O’Meara in the Department of Ecology and Evolutionary Biology.

He worked with Professor Jeremy Beaulieu, a former postdoctoral researcher at UT who is now an associate professor at the University of Arkansas, on the research published September 13 in PLOS Computational Biology.

“I have long been interested in odd patterns of diversification rates, especially the observation that recently originated groups of organisms have fast rates,” O’Meara said. “We generally expect the past to look like the present, but this pattern suggests that rates of everything are increasing towards the present.”

The rates at which species form, body size changes, and even rates of extinction increase over short time scales. For example, relatively young perching birds appear to evolve faster than birds as a whole. “Watching something for 10 years results in a faster rate than watching it over 50 years,” he said.

“We thought it was due to a bias in what people study,” he explained. “To use an analogy from one of our papers, people study sports cars and ignore minivans, so they only look at the fast or otherwise compelling examples and do not sample the slow or boring ones, creating a bias.”

Instead, O’Meara and Beaulieu show the pattern can be explained by statistical noise or related factors in the equation used to calculate the rate of change. They propose the term “tip fog” to describe the variances resulting from different mechanisms.

“They could be short-term evolutionary changes: a change in bird beak size as only those with large beaks can crush seeds available during a drought, for example,” O’Meara said. “Or they could be things like uncertainty in measurements: How long is a stretchy squid tentacle? Another possibility is short-term ecological changes: a warm summer leading to a taller plant than plants from a cooler summer 50 years ago.”

The equation and software they developed assume one kind of error. “It’s likely a pretty good first approximation, but there could be other kinds of error that make interpretations of reconstructed rates still uncertain in unexpected ways,” he said. “I would love for our solution to fully fix the problem, allowing for unlimited examination of residual rates, but I do not think we are quite there yet.”

More accurate estimates can lead to better answers to the many questions related to rate change, such as whether an extinction rate is rising due to human impact or whether changing antibiotics leads to faster population growth of bacteria.