Janzen’s seasonality hypothesis predicts that organisms inhabiting environments with limited climatic variability will evolve a reduced thermal tolerance breadth compared with organisms experiencing greater climatic variability. In turn, narrow tolerance breadth may select against dispersal across strong temperature gradients, such as those found across elevation. This can result in narrow elevational ranges and generate a pattern of isolation-by-environment, or neutral genetic differentiation correlated with environmental variables that is independent of geographic distance. We tested for signatures of isolation-by-environment across elevation using genome-wide SNP data from five species of Andean dung beetles (subfamily Scarabaeinae) with well-characterized, narrow thermal physiologies and narrow elevational distributions. Contrary to our expectations, we found no evidence of population genetic structure associated with elevation and little signal of isolation-by-environment. Further, elevational ranges for four of five species appear to be at equilibrium and show no evidence of demographic constraints at range limits. Taken together, these results suggest physiological constraints on dispersal may primarily operate outside of a stable realized niche, and point to a lower bound on the spatial scale of local adaptation.

We sampled populations of our five focal species (Deltochilum speciosissimum, Deltochilum tesselatum, Dichotomius podalirius, Dichotomius satanas, Eurysternus affin. flocossus) along two elevational and two horizontal transects on the eastern slope of the Andes Mountains in Napo Province, Ecuador. Our two elevational transects spanned 730 to 1175 m and 1500 to 1950 m in largely undisturbed humid forest, with four sampling localities spaced as close to 125 m apart as possible given local constraints of soil and topography. Our two horizontal transects were each approximately 2 km long, with one transect having four points and the other having two points. Both horizontal transects were located in replicate corridors of montane forest at 2150 m asl separated by a natural barrier, the Cosanga River. No species were shared between the two elevational transects. We extracted genomic DNA from all samples using Qiagen DNeasy kits and the manufacturer’s recommended protocol for insects. To prepare libraries for reduced representation high throughput sequencing, we used a double restriction enzyme digest approach adapted from Gompert et al., 2014. After initially demultiplexing our libraries using our unique barcodes and a Python script implemented in the first step of ipyrad (Eaton, 2014), we assembled sequencing reads and called and filtered variants for each species independently using the dDocent pipeline (v. 2.7.8), a set of bash wrapper scripts optimized for population genomics of nonmodel organisms requiring de novo assemblies (Puritz, Hollenbeck, & Gold, 2014). We then used VCFtools, vcffilter (1.0.0) (https://github.com/vcflib/vcflib), and the dDocent script dDocent_filters to filter this file to high quality SNPs alone. 

Uploaded data can be used in conjunction with scripts in the publication's GitHub repository (https://github.com/elinck/scarab_migration) to recreate all figures and results. 

data.zip: zipped archive of all intermediate (processed) data files, including locality information, PCA results, ABC results, BEDASSLE output, and data frames for plotting.

vcfs.zip: zipped archive of .vcf files for each species (with MAF filter)

abc_vcfs.zip: zipped archive of .vcf files for each species for ABC (without MAF filter)