Virtual Life-Forms Mutate, Shedding Light on Evolution

By John Roach
for National Geographic News
May 7, 2003
They don't sting or bite. They don't cause diarrhea or headaches. They don't even exist in a tangible form. But "digital organisms"— special programs that reproduce, mutate, and adapt —can thrive inside computers, and they are teaching scientists several lifetimes worth of information about evolution.

These artificial "bugs" show that complex functions that are the digital equivalent to something like human eyesight can evolve from the simplest of functions via a long and winding road of gradual mutation, according to a team consisting of a biologist, a computer scientist, a philosopher, and a physicist.

"From comparative studies of living organisms, as well as the fossil record, it's long been clear that new functions don't just suddenly appear out of nowhere, but rather organisms have evolved new structures and functions out of old ones, by tweaking bits and pieces here and there," said Richard Lenski, a biologist at Michigan State University in East Lansing.

Lenski and his colleagues Charles Ofria and Robert Pennock at Michigan State University and Christoph Adami at the California Institute of Technology in Pasadena report on their research with the digital organisms in the May 8 issue of Nature.

Thomas Ray, a biologist at the University of Oklahoma in Norman, who designed digital organisms in the early 1990s, said the study "doesn't surprise evolutionary biologists, but rather is a completely clear and detailed analysis of what we believe."

Digital Evolution

Ever since Charles Darwin theorized about evolution in the 19th century, biologists have believed that complex features such as eyesight must have arisen through many steps and that some of the intermediate steps must have served different functions from what is observed today.

But while functions change, biologists have assumed that each step along the path to a complex new function is a step up—an improvement—in terms of producing an organism that is better adapted to its environment.

The problem is that evolution in the natural world is a slow process, making it difficult to watch the process play itself out and thus test these assumptions, said Lenski.

To speed things up, he and his colleagues placed digital organisms in a computer environment that was programmed to allow the organisms to replicate, mutate, and compete.

"We created a world in which these things are possible and the right organisms can take advantage of it," said Adami, a physicist, who designed the simulated environment to allow the evolutionary process to work as it does in the natural world.

And since the process is saved on a computer, the scientists can analyze the sequence of events that leads to the ability to perform a complex new function.

"What this system allows us to do is to look in incredible detail at the whole process," said Lenski. "Most of what we saw pretty much confirms what biologists have reported from other lines of evidence, but there was an interesting twist."

The twist is that some of the intermediate steps rather than always being steps up, or even sideways, were steps down. That is, some of the key mutations were harmful in the short term but survived the forces of natural selection and ultimately played a crucial role in the genetic development of a newly evolved complex function.

Squash, Bugs, and Computers

Lenski's laboratory is crammed with petri dishes filled with thousands of generations of the bacteria Escherichia coli which he uses to study evolution in real—or wet—organisms. The bacteria replicate, mutate, and compete relatively quickly, allowing Lenski to watch and manipulate the process of evolution.

During a friendly game of squash with a colleague in the physics department, Lenski learned that Adami was to speak at Michigan State University about his experiments with digital organisms.

Lenski says he was skeptical, but went to the talk and found that Adami was using a different language to describe "similar dynamics to what we were finding with bacteria." The scientists decided to collaborate.

The researchers use the computer program designed by Adami, which is called Avida. The program is basically an artificial petri dish in which organisms reproduce and if they evolve the right skills, can perform mathematical calculations to obtain rewards.

The reward is more computer time that the digital organisms use to copy themselves. To mimic real life, Avida is programmed to randomly add mutations to the copies, thus spurring natural selection and evolution.

"As an evolutionary biologist who does experiments rather than looking at ancient fossils, I like to joke 'what has evolution done for us lately?,'" said Lenski. "I like to be able to watch the process. Microorganisms are one way of doing that. With digital ones we can measure all aspects of a complex system as it mutates and evolves."

The researchers found that if the digital organisms lived in a computer environment that only rewarded them for performing a complex mathematical task—akin to solving a logic puzzle—they never could evolve the ability to do it.

But when the scientists repeated the experiment in a computer environment that also rewarded the digital organisms for solving several simpler puzzles, the organisms eventually evolved the ability to solve the most complicated problem they were given.

"In order to get to the point of doing the most complex operations, the experiments showed it was necessary that they had to solve easier problems first," said Lenski.

The digital organisms solved the most complicated problem by borrowing and modifying bits and pieces of the "genetic code" that their ancestors had used to solve the simpler tasks, just as predicted by Darwin.

The surprise, says Lenski, is that the evolutionary process is not a ladder in which the fittest organisms are descended from the fittest organisms in earlier generations. Instead, some mutations are harmful in the short term but can set up subsequent changes that are quite beneficial.

"What we are able to do is show how all components of the evolutionary process, the random and non-random, get together to form a highly complex gene which could not have evolved by random drift," said Adami.

This research is funded by a grant from the U.S. National Science Foundation.

© 1996-2008 National Geographic Society. All rights reserved.