A phylogeny of Antirrhinum reveals parallel evolution of alpine morphology
Durán-Castillo, Mario et al. (2021), A phylogeny of Antirrhinum reveals parallel evolution of alpine morphology, Dryad, Dataset, https://doi.org/10.5061/dryad.xgxd254gr
• Parallel evolution of similar morphologies in closely related lineages provides insight into the repeatability and predictability of evolution. In the genus Antirrhinum (snapdragons), as in other plants, a suite of morphological characters are associated with adaptation to alpine environments.
• We test for parallel trait evolution in Antirrhinum by investigating phylogenetic relationships using Restriction-site associated DNA (RAD) sequencing. We then associate phenotypic information to our phylogeny to reconstruct patterns of morphological evolution and relate this to evidence for hybridization between emergent lineages.
• Phylogenetic analyses show that the alpine character syndrome is present in multiple groups, suggesting that Antirrhinum has repeatedly colonised alpine habitats. Dispersal to novel environments happened in the presence of intraspecific and interspecific gene flow.
• We find support for a model of parallel evolution in Antirrhinum. Hybridisation in natural populations, and a complex genetic architecture underlying the alpine morphology syndrome, support an important role of natural selection in maintaining species divergence in the face of gene flow.
Raw reads were demultiplexed using the process_radtags script from the Stacks software (Catchen et al., 2013). Trimmomatic 0.36 (Bolger et al., 2014) was used to remove adaptor sequences, clip sequences with a phred score of ≤20 and remove any read shorter than 30 bp. Filtered reads were mapped to the A. majus genome (Li et al., 2019) using Bowtie2 (Langmead & Salzberg, 2012) and duplicate sequences removed using Picard tools (Broad Institute, 2018). SNPs were called using samtools 1.6 and the multiallelic caller implemented in bcftools 1.4 (Li, 2011), retaining invariant sites. This dataset was then filtered by mapping quality (≥40), depth (≥3x) and missing data, both per taxon (removing individuals with >70% missing data) and per site (removing sites present in less than 50% of individuals). The final data included 16,061,293 sites from 86 samples corresponding to 24 taxa.
To root the phylogenetic trees we used available whole genome sequence data from Misopates orontium (A. Whibley and E. Coen, unpublished). Misopates has been shown to belong to the Antirrhinum clade, and diverged from the genus Antirrhinum in the last 10-15 million years (Ogutcen and Vamosi, 2016). Variant calling was done as above retaining only the loci present in the alignment of Antirrhinum samples.
Natural Environment Research Council, Award: NE/R010609/1
Natural Environment Research Council, Award: NE/L011336/1
Biotechnology and Biological Sciences Research Council, Award: BB/D552089/1