Correction to: Nature https://doi.org/10.1038/nature04738Published online 15 June 2006
In this Letter, there is an error in the analysis of data that affects some of the conclusions. The three main conclusions of the Letter were as follows. First, that the Heliconius heurippa wing pattern arose by hybridization between H. cydno cordula and H. m. melpomene, demonstrated by hybridization experiments. Second, that H. heurippa is a distinct species reproductively isolated from both H. cydno cordula and H. m. melpomene, demonstrated by patterns of genetic differentiation at multilocus microsatellite genotypes and by mate choice experiments with live butterflies. Third, that mate choice on the basis of colour patterns explained a notable proportion of H. heurippa mating preference, such that hybridization led directly to both pre- and post-mating reproductive isolation in this species. It is this final conclusion that is affected by the error.
In the original Letter, the results of the wing model experiments are reported in Fig. 3a as the ‘Probability relative to controls’ (see Fig. 1a of this Amendment), whereas in fact the data plotted would be more correctly labelled as ‘Proportion of interactions with the heterospecific model’. Furthermore, the likelihood function was not corrected for the proportion of conspecific interactions, as previously reported1 (and as reported in the Supplementary Information of the original Letter), but was instead similar to that later used by Merrill et al.2. The null hypothesis (that is, no preference) would be 1 in the former, but 0.5 in the latter, which therefore changes the interpretation of the results. When we reanalyse the data with likelihoods based on the proportion of interactions with the heterospecific model (that is, in which 0 = a complete preference for the H. heurippa model, and 1 = a complete preference for the alternative model), the results map directly onto those shown in Fig. 3 of our original Letter (see left panels of Fig. 1b of this Amendment). However, there is no consistent statistical evidence from the data for a colour-based mate preference when the correction is applied (see right panels of Fig. 1b of this Amendment).
The implication of this re-analysis is that there is not good evidence for the key claim of the Letter that H. heurippa had a preference for the combined red and yellow hybrid wing phenotype. Furthermore, the inference that the colour pattern alone directly causes reproductive isolation is not supported. However, colour patterns could still have an important role in the reproductive isolation of H. heurippa, perhaps in combination with behavioural displays or chemical clues, which cannot be mimicked with the type of mate choice experiments performed here. The evidence for a hybrid origin of the H. heurippa colour pattern and for partial reproductive isolation of H. heurippa from H. cydno and H. melpomene (from the no-choice and tetrad experiments) remains valid. In particular, the introgression of colour alleles from H. melpomene into the H. cydno/H. timareta/H. heurippa clade is now well established and supported by extensive genetic evidence1,3. Our Letter was an early demonstration of this process, supported by hybridization experiments.
We thank Alex Hausmann and Richard Merrill for drawing attention to this issue. The original Letter has not been corrected.
The Heliconius Genome Consortium. Butterfly genome reveals promiscuous exchange of mimicry adaptations among species. Nature 487, 94–98 (2012).
Merrill, R. M., Van Schooten, B., Scott, J. A. & Jiggins, C. D. Pervasive genetic associations between traits causing reproductive isolation in Heliconius butterflies. Proc. Biol. Sci. 278, 511–518 (2011).
Salazar, C. et al. Genetic evidence for hybrid trait speciation in Heliconius butterflies. PLoS Genet. 6, e1000930 (2010).
About this article
Cite this article
Mavárez, J., Salazar, C.A., Bermingham, E. et al. Author Correction: Speciation by hybridization in Heliconius butterflies.
Nature (2021). https://doi.org/10.1038/s41586-021-03330-8