Venomous snakes often display extensive variation in venom composition both between and within species. However, the mechanisms underlying the distribution of different toxins and venom types among populations and taxa remain insufficiently known. Rattlesnakes (Crotalus, Sistrurus) display extreme inter- and intraspecific variation in venom composition, centered particularly on the presence or absence of presynaptically neurotoxic phospholipases A2 such as Mojave toxin (MTX). Interspecific hybridization has been invoked as a mechanism to explain the distribution of these toxins across rattlesnakes, with the implicit assumption that they are adaptively advantageous. Here, we test the potential of adaptive hybridization as a mechanism for venom evolution by assessing the distribution of genes encoding the acidic and basic subunits of Mojave toxin across a hybrid zone between MTX-positive Crotalus scutulatus and MTX-negative C. viridis in southwestern New Mexico, USA. Analyses of morphology, mitochondrial and single copy-nuclear genes document extensive admixture within a narrow hybrid zone. The genes encoding the two MTX subunits are strictly linked, and found in most hybrids and backcrossed individuals, but not in C. viridis away from the hybrid zone. Presence of the genes is invariably associated with presence of the corresponding toxin in the venom. We conclude that introgression of highly lethal neurotoxins through hybridization is not necessarily favored by natural selection in rattlesnakes, and that even extensive hybridization may not lead to introgression of these genes into another species.

Association and genetic mapping studies aimed at linking genotype to phenotype are powerful tools that require large numbers of samples, complicating their use in long-lived species with low fecundity. Shed skins of snakes and other reptiles contain DNA; are a safe and ethical way of non-invasively sampling large numbers of individuals; and provide a simple mechanism by which to involve the public in scientific research. Here we test whether the DNA in dried shed skins mailed to us from citizen scientists is suitable for reduced representation sequencing approaches, specifically genotyping-by-sequencing (GBS). We find that shed skin samples provide DNA of sufficient quality and quantity for GBS, although libraries from shed skin resulted in fewer sequenced reads than libraries from snap-frozen muscle, and contained slightly fewer variants (70,685 SNPs versus 97,724). This issue is a direct result of lower read counts of the shed skin samples, and can be rectified quite simply with deeper sequencing. Skin-derived libraries also have a very slight (but significantly different) profile of transitions and transversions, suggesting that DNA damage occurs but is minimal. We conclude that shed skin-derived DNA is a good source of genomic DNA for a variety of genetic studies, and use it to identify sex-linked scaffolds in the corn snake genome.