Solely from the genetic code itself, we know the importance of the first two positions in a codon in coding for an amino acid. A mutation in the third position, on the other hand, does not usually have an effect on the resultant amino acid; it usually a synonymous mutation or a mutation that results in an amino acid with very similar physic-chemical properties to that of the original one.
The lesser importance of the base at the third position allows for greater variability and results in the observation of increased GC content, as shown in Figure 1A, below.
Figure 1B highlights the correlation between GC content in coding versus intergenic regions of bacterial genomes.
Due to selection pressure, the first two positions in a codon are relatively AT-rich. The third position, therefore, seems to have a higher GC-content, so as to maintain some sort of equilibrium which resembles that of the intergenic regions. I find this fascinating!
Note that there seems to be no mention of which (and how many) genomes were studied. Let me know if I missed something!
This blog post is based on the following paper, from which the figure has also been taken:
Brocchieri, L., 2013. The GC content of bacterial genomes. Journal of Phylogenetics & Evolutionary Biology, pp.1-3.