To test whether the changes in percent identity across the promoter profile shown in Figure 2a were significant we performed the following analysis.  Because the average nucleotide identity acorss all positions in all promoters was 34% we chose one 25 bp window (from position 464 to 488) in which the conservation was 34% to be our baseline for comparison. We grouped the nucleotide frequencies within this window together and compared them to the nucleotide frequencies in all 25 bp windows across the distribution using both a parametric (t-test) and a non-parametric (Mann-Whitney) statistical test (shown below). Yellow indicates 25 bp windows that were not significantly different from the baseline, blue indicates windows that were significantly greater than the baseline (P<.05) and pink indicates 25 bp windows that were significantly lower than the baseline (P<.05).  The results indicate that the peak in conservation between 100 and 250 bp upstream of the ATG is significant. Interestingly the first  15 windows directly upstream of the ATG also show significantly increased conservation over the baseline. We think this is due to a subset of translationaly regulated genes that show very high conservation around the ATG (see the manuscript).  There is also a significant dip in the conservation around 50 bp upstream of the ATG which we believe represents some physical constraints on the placement of regulatory motifs relative to the start site.