So-called junk DNA is not really junk. Functions have been found for
noncoding DNA which was previously thought to be junk, and we cannot be
sure that the rest of the junk DNA is not functional as well.
It has long been known that some noncoding DNA has important
functions. (This was known even before the phrase "junk DNA" was
coined.) However, there is good evidence that much DNA has no
function:
Sections of DNA can be cut out or replaced with randomized sequences
with no apparent effect on the organism (Nóbrega et al. 2004).
Some sections of DNA are corrupted copies of functional coding DNA,
but mutations in them, such as stop codons early in the sequence,
show that they cannot have retained the same function as the coding
copy.
The fugu fish has a genome that is about one third as large as its
close relatives.
Mutations in functional regions of DNA show evidence of selection --
nonsilent changes occur less often that one would expect by
chance. In other sections of DNA, there is no evidence that any
changes are selected against.
Nóbrega, Marcelo A., Yiwen Zhu, Ingrid Plajzer-Frick, Veena
Afzal and Edward M. Rubin, 2004. Megabase deletions of gene deserts
result in viable mice. Nature 431: 988-993.