How emerging adaptive variants interact is an important factor in the evolution of wild populations, but the opportunity to empirically study this interaction is rare. We recently documented the emergence of an adaptive phenotype "curly-wing" in Hawaiian populations of field crickets (Teleogryllus oceanicus). Curly-wing inhibits males' ability to sing, protecting them from eavesdropping parasitoid flies (Ormia ochracea). Surprisingly, curly-wing co-occurs with similarly protective silent "flatwing" phenotypes in multiple populations, in which neither phenotype has spread to fixation. These two phenotypes are frequently coexpressed, but since either sufficiently reduces song amplitude to evade the fly, their coexpression confers no additional fitness benefit. Numerous "off-target" phenotypic changes are known to accompany flatwing, and we find that curly-wing, too, negatively impacts male courtship ability and affects mass and survival of females under lab conditions. We show through crosses and genomic and mRNA sequencing that curly-wing expression is associated with variation on a single autosome. In parallel analyses of flatwing, our results reinforce previous findings of X-linked single-locus inheritance. By combining insights into the genetic architecture of these alternative phenotypes with simulations and field observations, we show that the co-occurrence of these two adaptations impedes either from fixing, despite extreme fitness benefits, due to fitness epistasis. This co-occurrence of similar adaptive forms in the same populations might be more common than is generally considered and could be an important force inhibiting adaptive evolution in wild populations of sexually reproducing organisms.