This week’s BEACON Researchers at Work post is by MSU graduate student Carlos Anderson.
That the structure and laws of our universe enabled the origin of life is an incredible coincidence. Without the gravity that aggregates matter into galaxies, stars, and planets, or without the light that radiates from the sun and heats the Earth, or without the chemical bonds between atoms that store and release energy, life, as we know it, would have probably never arisen. But would life, as we don’t know it, have?
I thought about this question in the summer of 2003, after reading a curious novel by Michael Crichton called Prey. In it, scientists create self-reproducing artificial nanobots, which evolve out of control and prey on their creators (most of them die, of course). But the concept of an artificial life fascinated me, and I soon began reading on evolutionary computation, starting with a wonderful book by Melanie Mitchell. She describes a class of algorithms that use the principles of evolution by natural selection—inheritance, variation, and differential reproduction—to search for solutions to various kinds of problems, such as the design of efficient engines or the sequence reconstruction of the human genome. These ‘genetic algorithms’ demonstrate that evolution is not unique to biological life but to any system having certain basic properties.
With this new interest in evolution I began taking classes in biology, and although I had recently obtained a B.S. degree in Computer Science, I decided to pursue an M.S. degree in Biology. This first exposure to graduate school helped me transition from by background in computer science to biology, and it allowed me to develop a specific interest within evolution that I could study further as a dissertation.
This interest was speciation, the process by which new species arise. There are many definitions of ‘species,’ some based on morphology, others on ecology, and still others on genetic differences. But the one most widely accepted is the biological species concept, in which populations are considered to be different species if they are reproductively isolated; that is, if they cannot produce fertile or viable offspring, either because they are unable to mate or because their hybrid offspring are sterile or inviable. Reproductive isolation is thought to evolve most readily when populations become geographically isolated and each subpopulation adapts independently to its local environment. Two big questions in speciation are (1) can speciation proceed even if environments are similar to each other, and (2) can speciation proceed when migration between populations occurs? I wanted to study speciation in an artificial life system, where my findings could be generalized to life as we don’t know it.
Searching for Ph.D. programs in 2006, I found Michigan State University, where a team of scientists and students have been developing Avida. Avida is an artificial life software designed to study questions in evolution and ecology. In Avida, digital organisms consist of a sequence of instructions (or ‘genome’) that encodes their ability to replicate and perform computational functions. The precise sequence of instructions that allow organisms to perform functions evolve through natural selection and genetic drift, two evolutionary processes that occur in biological organisms. With Avida, one can observe millions of generations of evolution in a short period of time, perform many replications, easily manipulate genomes, and accurately record measurements like fitness and events like mutations.
One of my studies addresses whether speciation can occur when environments are similar between populations. One hypothesis is that speciation can happen by compensatory adaptation, in which a deleterious mutation (a mutation that has a negative effect on the organism) rises in frequency in a population and is subsequently compensated by secondary mutations. Imagine that two populations become divided and each undergoes the process of compensatory adaptation. If the populations were now to come into contact, their hybrids would inherit a combination of deleterious and compensatory mutations, which, because they evolved independently, may not be compatible with each other and possibly cause inviability or sterility. I found this hybrid incompatibility in Avida, and it wasn’t simply because hybrids inherited deleterious mutations, but also because compensatory mutations between populations were incompatible. These findings show that compensatory adaptation is one way in which speciation can occur when the external environment does not change.
Another of my studies tests the effect of migration between diverging populations on the probability of speciation. I evolved populations that had to adapt to a new environment while migration between them occurred. Although the environments were new, I had treatments in which the two environments were different from each other and in which they were the same. I found that when the environments are different, migration does not prevent speciation from starting—even at 10% migration. However, when the environments are the same, even 1% migration prevents speciation. It appears that when the environments are the same, the population that adapts to it first and lets an individual migrate to the other population, effectively gives away its solution. This causes both populations to adapt similarly, preventing reproductive isolation between them.
Interestingly, these findings agree with theoretical expectations. Such unsurprising results are actually good because they show that Avida behaves like biology, and therefore demonstrates that evolution does not require the intricacies of biological life. Although Avida was not designed to study the origin of life, it does make one ponder whether artificial life could evolve de novo, and thus show that life can emerge not from specific universal structures or laws but from general ones.
For more information on Carlos’ work, you can contact him at carlosja at msu dot edu.