http://en.wikipedia.org/wiki/Melanin << Production of melanin is stimulated by DNA damage induced by UVB-radiation, and it leads to a delayed development of a tan. This melanogenesis-based tan takes more time to develop, but it is long lasting. >>
i.e., DNA-damage, when caused by sunlight, is self-limiting (negative feedback loop), though maybe any form of DNA damage will cause long-term tans too... otherwise, this mechanism is only responsive in the specific regions of DNA that UVB is likely to damage.
I can't imagine that the damaged DNA is coding for the melanin directly (that would be too crazy), but rather the detection/correction mechanisms must activate transcription factors for melanin-related genes, or something like that, in a permanent way. I wonder what happens when the original damaged cells die, and are replaced by copies of themselves. Does the tan still persist? If so, I imagine it would be due to epigenetic (Lamarckian) inheritance, rather than the mutated DNA.
I think of genetics as a cooperative version of a game known as
Core Wars, with all its strange loopiness. But Douglas Hofstadter noticed this long before I did.
Agar N, Young AR (April 2005). "Melanogenesis: a photoprotective response to DNA damage?". Mutation research 571 (1-2): 121-32. doi:10.1016/j.mrfmmm.2004.11.016. PMID 15748643.