For those that don't know much about either the significance of the science or the technology involved with generating the data, this might be useful. One big gray area in our understanding of evolution is how quickly genomes are changing, where they change, and the types of changes that are occurring. Yes, a genome is usually made up from DNA (RNA viruses being the major exception), and encoded in the DNA are genes, many of which get translated into proteins that do much of the "work" in an organism. However, depending on the organism, much of the DNA does not code for genes. The human genome for example is ~3,100,000,000 nucleotides (DNA's building blocks) long. Of that, ~1.5 percent codes for protein. Of the rest, the vast majority are ancient, dead, "selfish" chunks of DNA such as retroviruses (RNA viruses that convert to DNA and integrate into a genome. HIV is an example of one of these guys) and transposons (a major class of which are just like retroviruses but lack the genes for cell-to-cell transfer). Periodically in the evolution of many multicellular organisms (e.g. plants and animals), there are explosions or blooms of these types of elements that suddenly take off and integrate around a genome. This is one type of mutation (or genome evolution), and there are many others. Single nucleotides can change (e.g. C->T, as discussed in the paper), individual genes can get duplicated through a process known as unequal crossing-over or nonhomologous recombination, and the entire genome can be duplicated (known as polyploidy and is a dominant feature in flowering plant genome evolution.)
Our current understanding of how dynamic a genome is, the types of changes that occur, and the factors that limit these changes is very limited. Much of this is because getting a genome of an organism can be expensive and laborious, depending on the size of the genome (RNA virus 15,000 nt, DNA virus: 150,000 nt, bacteria: 5,000,000 nt, yeast: 20,000,000 nt, multicellular organisms: 100,000,000-10,000,000,000). Since our understanding of how genomes evolve depend on getting genomes sequenced that are appropriately related to one another (e.g. populations of organisms versus diversity of organisms), we can only get answers for those genomes we currently have (current ~8000 for all viruses, bacteria, archaea, and eukaryotes). Fortunately, there is currently a major technological revolution happening in biology: generating DNA sequences fast and cheap. For example, the first human genome was approx a 10 year project and cost ~$1,000,000,000. Now, the record for a human genome takes less than a week and costs ~$15,000.
This project is a major milestone as the authors sequenced 6 plant genomes (a mustard known as Arabidopsis thaliana) that are related to one another by 30 generations. Because of the close evolutionary relationships of these organisms, the authors can characterize the types of genomic change happening over very short time periods.
The emerging picture is that genomes, the fundamental genetic blueprint for a lineage of organisms, are much more dynamic than we had previously thought.