In spite of this deleterious cost, introns have spread in certain genomes; notably, in mammals and flowering plants. How do we explain the spread of introns? Is it consistent with the null hypothesis of random genetic drift?
According to Lynch the answer could be, yes. Here's what he says in his book The origins of genome architecture.
For newly arisen introns having no functional significance for the products of their host genes, the primary force opposing their ability to spread throughout a population is their excess mutation rate to defective allele(s), and because this force is expected to be quite weak, selection will be ineffective in preventing intron colonization in populations experiencing substantial levels of random genetic drift.The selection coefficient for intron deletion has to be above a certain threshold in order to prevent introns from spreading. This threshold depends on the population size: in large populations the deleterious effect of introns is sufficient to ensure that they will be kept to a minimum, or eliminated entirely.
For species with small populations there will be a cutoff where the selection coefficient cannot overcome the effect of random genetic drift and intron insertion is effectively neutral.
Lynch calculates the cost of the extra target nucleotides as a function of the mutation rate (μ) and explains why the cutoff is 2Ngμ = 0.04 (Ng is the effective number of genes ~ 2Ne). You can estimate 2Ngμ by counting the nucleotide diversity at silent sites in protein-encoding genes (πs). Thus, a plot of number of introns vs πs [2Ngμ] is a test of the hypothesis.
Here's the figure from Lynch's book.
The data indicates that species with small values of πs the spread of introns cannot be prevented even though introns may be deleterious. The cutoff is about 0.04 as predicted.
This does not prove that intron proliferation in some species is due to random genetic drift but it does show that the hypothesis cannot be ruled out. There's no need to invoke adaptive explanations for the initial spread of introns in vertebrate and plants genomes.
Lynch, M. (2010) Rate, molecular spectrrum, and consequences of human mutation. Proc. Natl. Acad. Sci. USA 107:961-968. [doi: 10.1073/pnas.0912629107]