Relaxation of purifying selection
Genetic drift can lead to relaxation of selection by lowering selection efficacy. To evaluate the level of selection relaxation, we used the descriptive model in RELAX (Wertheim et al., 2015) to infer a relaxation parameter k for 7,835 orthologous genes between the two spoonbill species and an outgroup, the crested ibis Nipponia nippon (S. Li et al., 2014). A relaxation parameter k> 1 indicates that the gene was under intensified selection, and < 1 implies relaxed selection. We counted the number of genes with a k < 1 for each spoonbill species. First, we used the program Inparanoid v. 4.1(Remm, Storm, & Sonnhammer, 2001) to identify homologous genes between the crested ibis and the black-faced spoonbill. Then, we usedgffread v. 0.11.4 (https://github.com/gpertea/gffread) to extract the coding region sequences of each homologous gene from each black-faced and royal spoonbill individual. We used MACSEv0.9b1(Ranwez, Harispe, Delsuc, & Douzery, 2011) to align the homologous sequences from all three species. We removed sequences that contained frameshifts and/or stop codons flagged by MACSE from the subsequent analysis. Then we used RAxMAL8 v. 8.2.12(Stamatakis, 2014) to draw the maximum likelihood gene tree of all 5,657 well-aligned homologous genes with the GTRGAMMA model. Gene trees of all orthologous genes from all three species in focus were used for the RELAXanalysis. The likelihood ratio test in RELAX was used to estimatek for each homologous gene by comparing the model which fixedk =1 and the model which allowed k to be estimated. The MRCA branch and each branch of each spoonbill species were alternatively set as the test branch in the RELAX analysis.