Thomas G. Aubier1,2, Michael Kopp3, Isaac J. Linn2, Oscar Puebla4,5, Marina Rafajlović6,7 and Maria R. Servedio2 1Centre de Recherche sur la Biodiversité et l'Environnement (CRBE), Université de Toulouse, CNRS, IRD, Toulouse INP, Université Toulouse 3 – Paul Sabatier (UT3), Toulouse, France 2Department of Biology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599, USA 3Aix Marseille Université, CNRS, Centrale Marseille, I2M, UMR 7373, 13331 Marseille Cedex 3, France 4Leibniz Centre for Tropical Marine Research, 28359 Bremen, Germany 5Institute for Chemistry and Biology of the Marine Environment (ICBM), 26111 Oldenburg, Germany 6Department of Marine Sciences, University of Gothenburg, 405 30 Gothenburg, Sweden 7Linnaeus Centre for Marine Evolutionary Biology, University of Gothenburg, 453 96 Strömstad, Sweden Correspondence: thomas.aubiernormalesup.org

Speciation can be mediated by a variety of reproductive barriers, and the interaction among different barriers has often been shown to enhance overall reproductive isolation, a process referred to as “coupling.” Here, we analyze a population genetics model to study the establishment of linkage disequilibrium (LD) among loci involved in multiple premating barriers, an aspect that has received little theoretical attention to date. We consider a simple genetic framework underlying two distinct premating barriers, each encoded by a preference locus and its associated mating trait locus. We show that their interaction can lead to a decrease in overall reproductive isolation relative to a situation with a single barrier, a process we call “negative coupling.” More specifically, in our model, negative coupling results either from sexual selection that reduces divergence at all loci, or from reduced LD that occurs because the presence of many females with “mismatched” preferences causes the mating success of recombinant males to become high. Interestingly, the latter effect may even cause LD among preference loci to become negative when recombination rates among loci are low. We conclude that coincident reproductive barriers may not necessarily reinforce each other, and that the underlying loci may not necessarily develop a positive association.