Page 1 of 2
Out of all the things that live and die on Earth, a small fraction becomes entombed in mud or sand and fossilizes, forming over time a grand tableau of life’s history. We measure the richness of this history in terms of biodiversity, the number of different kinds of creatures present at a given time. Though life’s story stretches back billions of years, only its somewhat recent chapters are at all clear, and even then mostly in the ocean, where fossils are more easily formed and preserved. Starting with the advent of abundant marine fossils about 500 million years ago, biodiversity’s development over time can be seen as a rising curve; more kinds (genera) of creatures exist now than existed 500 million years ago. But closer scrutiny reveals wiggles along this steady, gradual rise.
Though they may appear minor, some of these short-term fluctuations were the most dramatic events in life’s history, precipitous drops in biodiversity known as mass extinctions. Scientists have identified nearly 20 throughout the fossil record. Some 250 million years ago, one of the worst, the end-Permian event, killed off about 95 percent of life on Earth. Such transitions can be so radical and rapid that the rocks are fundamentally altered, inaugurating an entirely new geological period. The signs of ancient cataclysms are visible on most exposed cliff faces, where the oblivion of whole classes of creatures is often compressed into a centimeters-thin layer of rock.
Other declines occurred gradually, perhaps due to a decreased emergence of new species rather than increases in extinctions. Fortunately, all known dips in biodiversity seem to be followed by periods of rapid diversification called radiations. It was the end-Permian event that allowed the dinosaurs to develop and flourish. Of course, yet another mass extinction ended their reign. In their place, birds and then mammals ascended. And now humans have emerged. But even with all our intelligence and technology, we still don’t really understand what causes extinctions or radiations.
One of the big mysteries associated with these phenomena is also a key question for life’s future: Do they occur with any regularity? If we discovered a cycle to these events, it might suggest what’s driving the changes on Earth, and what linkage, if any, they have to events elsewhere in the universe.
I was always interested in this, but only as a spectator. Most of my scientific career had been in cosmology, working on computational models for the formation of structure after the big bang, dreaming of vast sheets and filaments of dark matter upon which galaxies congealed like dust settling on soap bubbles. But a few years ago I was drawn into a novel field that really doesn’t yet have a name, a blending of astrophysics and paleontology. Each year more evidence shows that past and future cosmic events, such as the deaths of distant stars or the orbits of comets, profoundly shape life on Earth. With this in mind, I began watching the literature for anything related to extinctions or the long-term history of the fossil record.
Soon I was losing sleep over something I’d read in a 2005 issue of Nature. Robert Rohde and his mentor, Richard Muller of UC Berkeley, had reported a fascinating cycle in biodiversity within a major compendium of fossil data sets, a regular 62-million-year rise and fall in the count of all kinds of creatures. They had explored several mechanisms to explain it and found them lacking, but both they and their editors deemed the signal so significant that it was published.
There’s a long history of people seeing cycles in the fossil record, but none of the work has been particularly rigorous, and rock strata don’t come with date labels. Paleontologists have rightly been skeptical of the claims. But the work of Rohde and Muller was different. They had capitalized on a careful study in 2004 that revised the dates for all those ancient environmental transitions seen in layers of rock. They also used a better quantitative method than past researchers, something called Fourier analysis.
Given reasonable limitations, almost any mathematical function—a line you’d graph with Xs and Ys—can be represented as a sum of sines and cosines, which have the familiar squiggly shape of oscilloscopes in old sci-fi movies. These “sinusoids” have different wavelengths, long for the long-term trends and short for sudden changes. Something called the “power spectrum” measures how much energy is in the different wavelengths. By looking at the varying contributions of sinusoids of different lengths to the power spectrum, you can reliably make estimations about the trends within a data set. Music is a good example. If you examined the power spectrum of the sound waves of different instruments playing middle C, each would look unique—which is why you could tell the instruments apart. But each would have a big peak at middle C—which is how you could tell they were all playing that note. So power-spectral analysis is a very effective way to search for regularities; one has to wonder why it wasn’t used sooner on the fossil record.
Using the revised timescales and Fourier analysis, Rohde and Muller looked for a periodic signal in the history of biodiversity. They began by subtracting out biodiversity’s long-term growth—a vital step if you want to find any short-term signal (the wiggles) superimposed upon the rising curve. They were looking for evidence of a 26-millionyear cycle that had been hinted at in the 1980s; the strong peak in their power spectrum indicating a 62-million-year cycle was a surprise. Using the same data, Bruce Lieberman and I checked their results. We estimated the 62-million-year peak had a 1 in 100 probability of arising through random chance. Then, collaborating with paleobiologist Richard Bambach, we found evidence of the same cycle in three more data sets.
Page 1 of 2