|MadSci Network: Evolution|
I am not sure where you are getting your information, but in order for me to answer your question we need to spend a lot of time defining exactly what we mean by terms such as "evolution", "artificial", "natural" and "mutation". Evolution has many different definitions. One definition is, "A change in the frequency of an allele in a poputlation over time." This definition works for diploid populations which mate to produce offspring—if we can easily measure the allele frequency—and usually it implies that the population is relatively stable in size, with one allele outcompeting the other in the same environment. This definition is most often used in very short time scales within one species or subspecies. Another definition of evolution is the change in a genome over time. This definition would be more applicable to larger timescales involving many speciation events, or to very rapidly evolving organisms such as viruses.
Mutations also come in many types. There are point mutations involving a single base change in a single gene, such as a change in the hemoglobin gene which results in the sickle-cell anemia beta globin allele. There are large deletions which eliminate many genes from a genome in one step. There are transpositions which insert transposons into genes to eliminate the function of the gene. There are transpositions which put a new promoter in front of a gene to alter the gene's regulation. There are chromosomal fusions and translocations which can potentially create, in one generation, a new species unable to cross breed with the parental species.
Life on earth has been evolving for roughly 4 billion years, and most scientists can spend less than 20 years following populations of fruit flies or some other animal in a lab. The study of viruses has the advantage that they can evolve up to ten million times faster than eukaryotes, but the disadvantage that we often can't look at them and see an obvious phenotypic change. Molecular biology techniques, which allow us to study the changes in genes directly, have only been available for roughly 30 years, but they are rapidly becoming more affordable. In 1980 it cost roughly $1 per DNA base to sequence DNA, and now it costs less than $.10 per base in many labs.
It is certainly true that most mutations are deleterious and result in a gene or organism that is less fit than the "wild type" version. So most mutations observed in a lab or in the wild would be expected to be lost, or to revert back to wild type over time if the environment remains constant, whether "natural" or "artificial". An alternative method of studying evolution is to look at environmental changes, such that the "wild type" is not expected to be "optimally fit". For example we can change the temperature at which we grow a bacteria or bacteriophage and rapidly select mutants which outcompete the wild type at the new temperature. Instead of taking a particular mutation and searching for the environment that would favor it, we are changing the environment and looking for any mutation which increases survival in that environment. If this meets your criteria for "survival conditions loosely set by the scientist", then the answer is YES, such mutations have been observed to be stable.
We also now have DNA sequences from immunodeficiency viruses, influenza viruses, and many other viruses which mutate rapidly, have absolutely enormous population sizes, have very short generation times, and therefore evolve millions of times faster than eukaryotes. We can chart the evolution of the human influenza virus type A hemaglutinin and neuraminidase genes over time from 1918 to 2005 and see that it changes at a relatively constant rate. Some amino acids in these proteins are conserved over time because they are necessary for protein function (negative selection). Other amino acids change very rapidly because they do not affect protein function and their variability helps the virus evade the host immune system (positive selection).
Try the links in the MadSci Library for more information on Evolution.