Clocks Versus Rocks

The researchers’ main criticism is that O’Leary’s team took the age of the oldest fossil from various placental groups to be the age of the group itself. This is unlikely to be true: even older fossils probably exist but have not been found yet.

“Imagine you have a mother with a 10-year-old child,” explained Mario dos Reis from University College London, the lead author on the new paper. “You know the mother can’t be younger than ten, but to say that the mother is ten would be a mistake.”

Dos Reis’ team did its own “molecular clock” study, comparing the genomes of 36 mammals to determine when they diverged from each other, and calibrating these splits using the ages of known fossils. The team also accounted for how often fossils of different groups were found, to estimate the gap in time between a group’s oldest fossils and its true ancestors.

The researchers concluded that the placental mammals arose between 72 million and 108 million years ago. “Most of these early placentals were probably similar and likely didn’t resemble modern groups, like primates or bats or carnivores,” said dos Reis. “After the dinosaurs died out, they diversified explosively.”

Mark Springer from the University of California, Riverside, supports the new critique; last year, he and his colleagues published a response to O’Leary’s paper that made similar points.

But O’Leary said that the new study provides no new data and changes very little. “It’s another clock paper and tells us pretty much what clock analyses have been telling us,” she said.  

She acknowledged that gaps must exist between a group’s oldest fossils and its last common ancestor, but her team did not want to make assumptions about the size of that gap. Dos Reis’s technique for estimating that gap is “just a prediction, like a weather prediction,” she said. “It’s another way of tweaking the model on a computer. They put a lot of stock in statistical assumptions about the fossil record.”

“I’d say the real test belongs in the field,” O’Leary added. That is, someone has to go and find a placental fossil that clearly originated from before the K/T boundary.

“The debate is getting a little tiresome, in my opinion,” Olaf Bininda-Emonds from the University of Oldenburg told The Scientist in an e-mail. “Dos Reis et al. are perfectly correct: fossils underestimate the true divergence time of a clade,” he said. “But this doesn't mean that clock-based [models] can't overestimate that time.” Such models have become increasingly sophisticated, but they still rely on large sets of assumptions.

“There’s nothing really wrong with either set of analyses,” Bininda-Emonds continued. “Both are robust. The real problem is that the methods are fundamentally different and make fundamentally different assumptions, so that there's little point in comparing the apples with the oranges.”

M. dos Reis et al., “Neither phylogenomic nor palaeontological data support a Palaeogene origin of placental mammals,” Biology Letters, doi:10.1098/rsbl.2013.1003, 2014.