What effect does recombination have when there is linkage disequilibrium?

Suppose now that there is linkage disequilibrium, and it is due to coadaptation between loci. Let us suppose that A1B1 and A2B2 are coadapted pairs of alleles, like the tail genes and color genes in Papilio memnon (opposite), and are present in greater than random frequencies.

The amount of linkage disequilibrium will be a balance between the increase due to selection and the decrease due to recombination. In this case, recombination is disadvantageous. The lower the recombination rate, the stronger the association between the coadapted alleles that can be built up, and the fitter on average the individuals will be. There is selection to reduce the recombination rate between the loci.

So far, it seems that recombination either makes no difference or is disadvantageous:

If there is linkage equilibrium, the genotype frequencies are unaltered by recombination; if beneficial associations between genes have built up, recombination tends to destroy them.