Skip to main content
Dryad logo

Anthropogenic and natural barriers affect genetic connectivity in an Alpine butterfly

Citation

Trense, Daronja et al. (2020), Anthropogenic and natural barriers affect genetic connectivity in an Alpine butterfly, Dryad, Dataset, https://doi.org/10.5061/dryad.c866t1g3m

Abstract

Dispersal is a key biological process serving several functions including connectivity among populations. Habitat fragmentation caused by natural or anthropogenic structures may hamper dispersal, thereby disrupting genetic connectivity. Investigating factors affecting dispersal and gene flow is important in the current era of anthropogenic global change, as dispersal comprises a vital part of a species’ resilience to environmental change. Using fine-scale landscape genomics, we investigate gene flow and genetic structure of the Sooty Copper butterfly (Lycaena tityrus) in the Alpine Ötz valley system in Austria. We show surprisingly high levels of gene flow in L. tityrus across the region. Nevertheless, ravines, forests and roads had effects on genetic structure, although rivers did not. The latter is surprising as roads and rivers have a similar width and run largely in parallel in our study area, pointing towards a higher impact of anthropogenic compared with natural linear structures. Additionally, we detected eleven loci potentially under thermal selection, including ones related to membranes, metabolism, and immune function. This study demonstrates the usefulness of molecular approaches in obtaining estimates of dispersal and population processes in the wild. Our results suggest that, despite high gene flow in the Alpine valley system investigated, L. tityrus nevertheless seems to be vulnerable to anthropogenically-driven habitat fragmentation. With anthropogenic rather than natural linear structures affecting gene flow, this may have important consequences for the persistence of species such as the butterfly studied here in altered landscapes.

Methods

SNPs were called using a de novo pipeline without a reference genome. We used the process_radtags program within stacks version 2.3 (Catchen, Hohenlohe, Bassham, Amores, & Cresko, 2013; Catchen, Amores, Hohenlohe, Cresko, & Postlethwait, 2011) to demultiplex the sequence data. This separated individuals by their unique barcode, removed adapter sequences, trimmed the reads to 120 bp length, and discarded low-quality reads (Phred score < 20). We then ran the stacks denovo_map pipeline using one individual per outgroup and 26 individuals from the Ötz valley system to generate a catalogue of loci. Each of the 186 individuals was then matched against this catalogue and SNPs were subsequently called. We filtered the data to retain SNPs which were present in more than 50% of the 186 individuals, and then retained the first SNP in each ddRAD locus. Further filtering with the program vcftools version 0.1.11 (Danecek et al., 2011) was done to retain loci that were at Hardy-Weinberg equilibrium, had a minor allele count of > 2, missing data of < 5%, and mean depth of 20-45. The final dataset contained 12,806 SNPs for analyses of relatedness, genetic structure, and outlier loci.