As I just alluded to in the previous planting, exit polls have not reported vote breakdowns for indigenous (Native American) voters, and this makes South Dakota hard to call. Silver notes that there is some evidence that Obama has performed well with that demographic group.
I would add that Oklahoma, another state with a large indigenous population, was one of the early states where Obama most over-performed his late polling estimate (from a limited number of polls). See the graph.
Although Obama lost Oklahoma by a big margin, he outperformed his polls substantially.
We may be well past the point at which this contest can produce surprises, but maybe Obama will surprise in South Dakota, a state with 10% indigenous (and very little polling).