Reader Comments

Post a new comment on this article

Data now released with a corrigendum

Posted by JonathanSilvertown on 03 Oct 2011 at 15:21 GMT

We have now archived the data used in this paper at Data Dryad where it is available under creative commons:


Some minor numerical errors were discovered in preparing the data for release. They do not affect the results or conclusions of the original paper. Briefly:

1. The last sentence of the first paragraph of Results has some wrong numbers. This is the bit looking at increase in midbanded in historical-only and in non-dune sites. What got into the paper is the versions of these from the analysis of midbanded as % total, and for consistency with everything else it should be midbanded as % banded. The right numbers are, for historical only, alpha = 0.0106, t = 7.694, P<0.001, and for non-dune, alpha= 0.0044, t=3.089, P=0.002. Still pretty strong evidence that the effect is still there in these restricted analyses, even though it’s a bit weaker (particularly for non-dune) than the paper said.

2. In the penultimate paragraph of the results, the correlation between change in midbanded and temp diffs is even more negative than we reported. The correlation shown as r=-0.108 should be r = -0.273, (d.f. and P are OK).

No competing interests declared.