Nature 466, 478–481 (2010)

In Figure 1a and b of this Letter, the sign in the regression equations should have been a plus symbol, not a minus symbol. This does not affect the reported results or conclusions of the paper.