2.2 Quality control of genetic data
While checking chromatograms we found several cases of double peaks on the sequences of cox1 , cytb and CR (Supplementary Material 3). Bird blood contains relatively few mitochondria, and it is therefore likely to amplify nuclear copies of mitochondrial markers, or “numts” (Sorenson & Quinn 1998). Such nuclear copies may diverge from the original mitochondrial genes since they are non-coding, which result in double-peaks on the chromatograms. To check this scenario and avoid such copies, we digested nuclear DNA with the ExonucleaseV (ExoV; NEB-M0345S) and sequenced again the mitochondrial markers for all individuals showing double-peaks for cox1 , plus 5 individuals showing no double-peaks randomly chosen, using a protocol modified from (Jayaprakash et al. 2015, see Supplementary Material 3). Before running analyses, we checked that coding sequences contained no stop codons or indels. Some analyses required phased data (e.g. *BEAST analysis), so the gametic phase of nuclear markers was determined probabilistically using PHASE 2.1 (Stephens et al. 2001) implemented in DNAsp v.5.10.01 (Librado & Rozas 2009). Additional Genbank sequences (Supplementary Material 1) were aligned to our sequences using MAFFT v 7.187 (Katoh et al. 2002).