Reproductive interference is widespread, despite the theoretical expectation that it should be eliminated by reproductive character displacement (RCD). A possible explanation is that females of sympatric species are too similar phenotypically for males to distinguish between them, resulting in a type of evolutionary dilemma or catch-22 in which reproductive interference persists because male mate recognition (MR) cannot evolve until female phenotypes diverge further, and vice versa. Here we illustrate and test this hypothesis with data on rubyspot damselflies (Hetaerina spp.). First, reproductive isolation owing to male MR breaks down with increasing interspecific similarity in female phenotypes. Second, comparing allopatric and sympatric populations yielded no evidence for RCD, suggesting that parallel divergence in female coloration and male MR in allopatry determines the level of reproductive isolation on secondary contact. Whenever reproductive isolation depends on male MR and females of sympatric species are phenotypically similar, the evolutionary catch-22 hypothesis offers an explanation for the persistence of reproductive interference.

In nonterritorial damselflies, females often come in multiple color morphs, perhaps because females with rare colors experience reduced sexual harassment, and thus have a frequency‐dependent fitness advantage, compared to females of the most common color morph, but such polymorphisms are rare in territorial species. We consider three hypotheses to explain the rarity of female color polymorphisms in territorial species: (a) misdirected male aggression, (b) poor male mate recognition, and (c) low mating harassment rates. The first hypothesis has some empirical support, and can account for the absence of andromorphs (i.e., females that resemble males), but does not explain the absence of multiple heteromorphs. We tested the second hypothesis by presenting females of two novel color morphs (green‐ or red‐banded abdomens) to territorial male Hetaerina capitalis. Females of both novel color morphs elicited fewer sexual responses than control females, and the red morph occasionally elicited aggressive responses. These results indicate that novel female color morphs would experience reduced mating harassment in this species, contradicting the hypothesis that male mate recognition is too poorly developed to reduce harassment of novel female morphs. By process of elimination, the third hypothesis, that harassment rates are too low in territorial species to provide rare female morphs a fitness advantage, is favored, but remains untested. Our findings also suggest that the common practice of color‐marking odonates for behavioral research is likely to interfere with mate choice, as has long been known to be the case in birds.