Scientists’ most basic assumptions about the fossil record may be going extinct, thanks to the work of a large research team at UCSB’s National Center for Ecological Analysis and Synthesis.

The researchers assembled information about the fossil record on a computer database called the Paleobiology Database, and results suggest that the number of species on earth has not been growing as rapidly as previously thought – and possibly, it has not grown at all. John Alroy, a postdoctoral fellow at NCEAS and one of the project’s founders, was one of 25 authors of a paper published May 22 in The Proceedings of the National Academy of Sciences.

Paleontologists – researchers who specialize in reading the fossil record – have long taken for granted a model of diversity known as the Sepkoski Compendium. This model says that the number of species on the planet has long been increasing.

“Sepkoski’s work showed this really enormous spike [in diversity] that goes back about 250 million years,” Alroy said. “You get this huge, huge, huge climb that’s like a straight line going up.”

The new data from Alroy and his team indicate that the Sepkoski Compendium might be wrong because its conclusions are based on oversimplified data. Jackson Sepkoski of the University of Chicago, who conceived the model, made a database containing the names of different organisms and the dates that they first and last appeared in the fossil record. Most databases since, say NCEAS researchers, have been more of the same.

“The typical global database almost looks like a phonebook,” said Arnie Miller, a project member and geology professor at the University of Cincinnati. “You can actually go through them and see a list of names, along with the time of first known appearance and last known appearance.”

NCEAS researchers claim that a list of this type is not enough data to construct an accurate graph of species diversity over time.

For a variety of reasons, the fossil record is incomplete. Some geological eras laid down more rock and therefore more fossils. Some fossils are found only in relatively inaccessible areas of the world, such as deep oceans or large deserts. Some animals have been the subjects of more interest than others.

In the end there are many reasons for the large gaps in the fossil record. Even if the earth retained a perfect record of life, it is probable that more recent species would be overrepresented because their fossils are more exposed and easier to recover. It is not surprising that classic diversity graphs show a constant increase over time in the number of known species.

This is where the NCEAS’ Paleobiology Database differs from classic diversity curves: it takes into account these gaps in the fossil record.

Alroy compared mapping species diversity to taking a census. “People in certain neighborhoods tend to be missed. People in certain socioeconomic strata tend to be missed, along with certain age groups, etc,” he said.

Alroy said that the way the database counts species is similar to the way the census systematically adjusts its numbers to include underrepresented groups. “They make a lot of local estimates based on detailed relationships between what kind of a person you are and how likely it is that you’re going to be captured in a census,” he said.

Of course, the census can make an educated guess about the total number of people who are out there to be counted. Paleobiologists, on the other hand, have no idea how many organisms may have constituted a species or how many species may really have been on the earth at a given time. So NCEAS researchers have used a slightly different method called interpolation.

Interpolation does not attempt to determine the population of a certain type of organism. Instead it attempts to identify the frequency of organisms in proportion to one another; for every organism of species A, there will be two of species B, four of species C and so on.

To do this, the researchers needed to make all their data equally reliable. Since they could not make the poor data better, they did this by placing a lower value on the best data.

“[This] of course makes people upset because people have a sort of built in dislike of doing things where you make the data equally bad everywhere,” Alroy said. “They like to do it the other way around where you try to pretend what it would be like if the data were perfect. That’s everyone’s natural inclination, but that turns out to be a bad idea for statistical reasons. You get a much more reliable estimate if you did it the way we did it with interpolation.”

The net result of constructing a diversity graph in this way is a number of species that does not spike but remains relatively constant over time.

“Our results are more compatible with what’s called a density dependence model where the number of new species you can make is a function of how many are already out there,” Alroy said. “If the world were crowded with species, it’d be hard to make more of them and if the world is empty and doesn’t have a lot of species, it’s easy to make more species.”