The spatial distribution of synonymous substitutions in enterobacteria
l genes is investigated. It is shown that synonymous substitutions are
significantly clustered in such a way that a synonymous substitution
in one codon elevates the rate of synonymous substitution in an adjace
nt codon by about 10%. The level of clustering does not appear to be r
elated to the level of gene expression, and it is restricted to a rang
e of two or three codons. There are at least three possible explanatio
ns: (1) sequence-directed mutagenesis, (2) recombination, and (3) sele
ction.