A rare model limits the distribution of its more common mimic: a twist on frequency-dependent Batesian mimicry.

Batesian mimics are predicted to lose their fitness advantage not only in the absence of an unpalatable model, but also when the mimic becomes relatively abundant. The phenotypic hybrid zone between mimetic and nonmimetic admiral butterflies, comprising the polytypic Limenitis arthemis species complex, offers an ideal opportunity to test these predictions because the position of the hybrid zone is hypothesized to be controlled by the geographic range of Battus philenor, the chemically defended model. We used 29 years of observational field data from a continental-scale butterfly monitoring program, the 4th of July Butterfly Counts, to show that (1) the advantage of mimicry does not extend beyond the range of the model, (2) in contrast to expectations, the mimicry complex is maintained even where the model is rare and (3) the sharp phenotypic transition between mimetic and nonmimetic admiral populations occurs over a very narrow spatial scale corresponding to the limit of the model's range. These results suggest that, even at very low densities, there is selection for Batesian mimicry and it maintains the geographic position of this hybrid zone. Our findings highlight the value of large-scale, long-term citizen science monitoring programs for answering basic ecological and evolutionary questions.