Museomics help resolving the phylogeny of snowfinches (Aves, Passeridae, Montifringilla and allies)

Historical specimens from museum collections provide a valuable source of material also from remote areas or regions of conflict that are not easily accessible to scientists today. With this study, we are providing a taxon-complete phylogeny of snowfinches using historical DNA from whole skins of an endemic species from Afghanistan, the Afghan snowfinch, Pyrgilauda theresae. To resolve the strong conflict between previous phylogenetic hypotheses, we generated novel mitogenome sequences for selected taxa and genome-wide SNP data using from ddRAD sequencing for all extant snowfinch species endemic to the Qinghai-Tibet Plateau (QTP) and for an extended intraspecific sampling of the sole Central and Western Palearctic snowfinch species (Montifringilla nivalis).

Methods
We extracted DNA from 40 samples of snowfinches from all eight species of the three genera (Montifringilla, Onychostruthus, and Pyrgilauda) and one further sample of the rock sparrow, Petronia petronia. All samples were either frozen blood or tissue samples preserved in ethanol or preserving buffer except two toe pad samples taken from two historical specimens of the Afghan snowfinch, P. theresae. We generated whole mitochondrial genomes for this species and representatives from the two other snowfinch genera using specific protocols for museum material. To complete a previous single-marker data set, we amplified a 1079-bp-long cytochrome-b fragment for 19 samples using the primer combination of O-L14851/ O-H16065 primers.

For inference of a genome-wide SNP data set were used for double-digest restriction site associated DNA sequencing (ddRAD seq). DdRAD seq was performed at the Deep Sequencing Facility in the Center for Molecular and Cellular Bioengineering (CMCB) Dresden. We used Qubit (Thermo Fisher Scientific, Waltham, MA, USA), dsDNA High-Sensitivity (HS) and Broad-Range (BR) assays for DNA concentration measurement following the manufacturer’s protocol. According to our Qubit measurements, we selected 38 samples with sufficient DNA concentrations for ddRAD seq. For sample preparation, 50 ng gDNA were double-digested with SbfI and MspI (NEB) for 120 minutes at 37°C followed by heat inactivation at 65°C for 20 min. SbfI specific library barcodes carrying Truseq-i5 Illumina adapters with cohesive ends were ligated to the cohesive ends of the SbfI restriction sites of the digested DNA fragments. The same was done for the MspI site with a MspI-specific truncated universal TruSeq-i7 adapter. Samples with different P5 Barcodes were pooled and purified using XP beads (Beckman Coulter, Krefeld, Germany) at a ratio of 1:1 to remove non-ligated adaptors. Libraries were equimolarly pooled before sequencing them in single-end mode on an Illumina NextSeq 500 system to a read length of 75 bp and a depth of at least 1 million reads per sample.

We mapped our cleaned ddRAD seq read data to three different reference genomes of 1.) Pyrgilauda ruficollis (NCBI acc. no: GCF_017590135.1), 2.) Onychostruthus tazcanowskii (NCBI acc. no: GCA_017590055.1) and 3.) to a house sparrow (Passer domesticus) genome which was assembled to chromosome-level. We used ipyrad v.0.9.42 for data assembly and read mapping to the three different reference genomes. We applied a clustering threshold of 85% and a minimum sequencing depth for clustering ≥6X. We applied default parameter settings of the reference-based ipyrad pipeline with a maximum of 8 indels, 0.5 heterozygous sites, 20% SNPs per locus, and a minimum of four samples per locus. After these first filtering steps, the ipyrad pipeline (Eaton and Overcast, 2020) produced three independent VCF output files from mapping against three different reference genomes. From each of the VCF files we generated final SNP data sets allowing for 0%, 10%, 20% and 30% missing data that were used as input data for phylogenetic analysis. For variant calling we used vcftools 1.1.5 and bcftools v1.8 with a quality value ≥30 applied to separate autosomal from Z-chromosomal data sets [which was possible only for the data set inferred from alignment to the house sparrow reference genome that was annotated to the chromosome level].