Been meaning to contribute my thoughts to this subject for some time, but wasn't sure there was an appropriate thread on which to post. After reading everybody's thoughts on the matter, I think every comment on this thread has relevance to this unique phenomena, and I thought I'd chime in as well. My comments will be limited to those changes that take place within dough systems (internal factors) versus those from external influences (such as air temperature or moisture content during baking, etc.).
Browning in dough-related products comes either from direct caramelisation of residual sugars or from non-enzymatic Maillard reactions produced under thermal degradation, as Omid has rightly pointed out. Even more astutely, scott123 writes leoparding is, "[a]t it's simplest . . . just uneven browning. Longer fermentation times are residual sugar generators and sugar accelerates browning. The more sugar you have, the faster browning occurs, the more likelihood of leoparding."
To add further to the above, longer fermentation times, although an indirect agent of leoparding, are not the direct cause, and not because they generate residual sugars, per se, as yeast can easily remove 95% of glucides in a dough system, which are among the most prevalent reducing sugars for pyrolytic-based browning (caramelisation), along with maltose (which is also substantially removed from doughs during elongated fermentations). This removal of the most prevalent sugars in a dough leads to a nice pale canvas on which the leopard spots can appear.
The uneven browning scott123 refers to has to do with selected increased browning, due to abundant amino acids left in a dough system; hence, the selected unevenness of the browning. Longer fermentations increase Maillard reactions because they lead to a larger window for wheat's native proteases to release abundant amino acids. A similar effect could be achieved through prolonged autolysis (10 - 16 hours) and without the need for longer fermentations, though there may be more residual reducing sugars that contribute to the overall browning of the dough if the inoculant is not increased. This is not to say reducing sugars do not play a role, as the sugars resultant from longer fermentations and/or of doughs with a high amount of prefermented flour lead to the characteristic dark spots as well. In fact, many of these sugars (like arabinose, ribose or xylose) generate darker spots than do glucose or maltose when combined with amino acids, although they occur in much smaller amounts.
The second major component has to do with the dehydration of a dough ball's exterior, alluded to by f.montoya. bakeshack writes, "A cold fermented dough baked in a very hot WFO will form blistering much easier although they tend to be larger blisters compared to the micro blistering you get from RT fermentation," going on to say it "is very similar to the surface blistering you get from a retarded Tartine loaf . . . During baking, this allows the CO2 gasses to escape up to the surface of the crust and form blisters." Though, he incorrectly states the reason these blisters form ("gluten degradation"), TXCraig1 hits the nail on the head: "I think the second factor is that the dough needs to be developed enough that there is sufficient strength to maintain the bubble past the surface of the dough."
This "sufficient strength" is the result of long fermentations especially when the dough proofs in its final shape, thus allowing for increased surface exposure and a higher degree of moisture loss to the outside environment (especially pronounced under refrigerated conditions). The reason the Tartine (or any long-proved loaves) have blisters is the same reason they have significantly thicker crusts than similar loaves with shorter proofing times: lower water-activity on the surface of the dough at the time of baking. Outwardly migrating moisture is hence trapped by the dough's having developed a thicker skin, with the characteristic blisters forming, protruding like goosebumps that, in a very hot environment, will selectively brown more than an even, lower surface.
In short, less reducing sugars (general dough caramelisation), more amino acids (Maillard browning) and a thicker dough skin (greater rate of blisters formed) are the primary reasons for leoparding from a dough's perspective.