Arabidopsis thaliana (pictured) is the dry erase board of plant biology; which is to say it's a good model organism. We know a lot about it, not because it's particularly useful or interesting to humans, but because it's very easy to study. It has a short life cycle (six weeks), it's allegedly easy to grow (although I, personally, suck at it), you can grow a lot of it in a very limited amount of space, it has a small, fully-sequenced genome and it self-fertilizes. A paper published in last week's Nature explores how Arabidopsis' capacity for self-fertilization evolved.
Most plants take steps to avoid self-fertilization, (or 'selfing') because inbreeding in general is undesirable in most cases. In most members the mustard family, which includes A. thaliana, selfing is prevented by a relatively small segment of DNA called the S-locus, which contains two genes called SCR, which is expressed in pollen, and SRK, which functions in stigma (the organ which accepts pollen). You can think of them like a lock and key system in reverse: when the particular variety ('allele') of SCR protein on the pollen doesn't match the SRK protein on the stigma, the lock is opened and fertilization can take place. When the SCR and SRK proteins match (i.e., when they are from the same plant), the lock shuts and fertilization is prevented.
It has been known for a while that A. thaliana can self because this system is broken: the 'key' can never fit in the 'lock', so fertilization is always allowed. The system is badly broken, in fact: the S-locus in any given Arabidopsis population contains several mutations, any one of which by itself would cause the whole system to be non-functional. This is to be expected: since the S-locus is non-functional in A. thaliana, mutations therein are not selected against and tend to accumulate. This makes it difficult to determine which mutation originally caused self-compatibility.
Tsuchimatsu et al. did several very clever things in order to figure this out. They conducted crosses between members of various A. thaliana populations and a close relative (Arabidopsis halleri) in which the S-locus is intact and functional. They found that when A. halleri deposits pollen on the stigma of A. thaliana, the pollen is sometimes rejected. This (along with a number of control experiments which I'm glossing over) shows that SRK (which is the 'lock' part of the S-locus, remember) still works in at least some A. thaliana populations: the mutation which originally caused self-compatibility must have happened in SCR.
The authors examined the S-locus in detail in a number of A. thaliana populations from across a large part of its range (which is huge: it covers all of Europe, most of Asia and a good chunk of northern Africa). They found that there is great variation across populations, but one mutation in particular is almost ubiquitous: a large inversion in the SCR sequence. An inversion is a rather unusual mutation: a whole string of sequence is basically lifted out of the genome and re-inserted backwards. To test whether this mutation actually caused self-compatibility, the authors took the direct route: they turned the inverted sequence back around, resulting in a totally functional S-locus.
This is one of those rare – and very cool – situations where we can actually turn back the clock on evolution, undo the change and get a good look at the ancestral state. It's very exciting to be a botanist at a time when things like this are possible.
T. Tsuchimatsu, K. Suwabe, R. Shimizu-Inatsugi et al. (2010). Evolution of self-compatibility in Arabidopsis by a mutation in the male specificity gene. Nature. 464. pp. 1342-46,