Results
We found that six out of the nine candidate SNPs included in the OpenArray could not inform on sexual differences among individuals. Four of them (F1C, F2C, F4C and F6C) showed a fixed allele expression without variability between samples. Another SNP (F3C) did not amplify in any sample. The last SNP (F70A) did amplify for the “A” and “G” alleles but not for the heterozygote genotypes “A/G”, indicating that the SNP was located within the PAR region from the Z chromosome. We therefore discarded these six SNPs and focused our attention to the remaining two W-markers SNPs (F5C and F7C) and the Z-marker F9A.
For F9A, 323 samples (34.8%) expressed the allele “A”, 360 (38.79%) expressed the allele “G” whereas 236 samples (25.43%) expressed both “A” and “G”. Nine samples (0.97%) could not be genotyped because they did not amplify in any of the 928 samples. Scatter plots for allelic discrimination confirmed the existence of three clusters for F9A (Fig. S1a; see also Table S1), corresponding to the homozygote genotype for allele 1 (“G/G” with high values of VIC and low values of FAM), the homozygote genotype for allele 2 (“A/A” with high values of FAM and low values of VIC) and finally the heterozygote genotype (“A/G” with high values of VIC and FAM). Since F9A was found in the Z-chromosome, heterozygote samples could only correspond to ZZ male individuals, while either male or female individuals could express a homozygote genotype.
The results for the F5C and F7C confirmed that they were W-linked and located outside the PAR region (Table 1). Fluorescence for F7C and F5C clustered into two groups separating samples that amplified (“A” with high values of FAM and VIC) from those which did not amplify (“NOAMP” with low values of FAM and VIC) (Fig. S1b and Fig. S1c respectively). In F7C, 466 samples (50.21%) amplified for the variant “A” and in F5C, 443 samples (47.73%). No other allele variant was expressed in these two SNPs. The fact that only one copy was present indicated hemizygosity; thus, only female individuals carrying the W chromosome amplified for F5C and F7C. Being both W-linked SNPs complementary, one is enough to identify females instead of two, which would reduce the number of SNPs for genotyping. F7C showed less specificity to amplify other regions of the genome, suggesting the use of F7C primers was more reliable for sexing (see Section S3, Table S4, Table S5 and Fig. S4 in Supplementary Information for more details regarding this justification).
Therefore, we assigned an individual as female if amplified for the variant “A” or “G” in the Z-linked SNP F9A and “A” for the W-linked SNP F7C. We assigned an individual as male if amplified for the variant “A/A”, “A/G” or “G/G” in the SNP F9A and did not amplify for the SNP F7C (Fig. 1 and Fig. 2). To validate the discriminating power of our method, we used 18 males and 18 females from 18 breeding pairs. Within couples, morphological measures can be reliably used to sex the bigger- and the smaller-sized individuals as males and females respectively (see Material and Methods). We found that our SNP approach correctly classified 100% of females and males (Table S2).
Using our method, we identified 456 individuals of the genotyped samples as females and 457 as males, including both adults and fledglings. This represents the 99.5% of all genotyped samples successfully sexed. The few cases in which individuals could not be sexed (N=5) showed a heterozygote genotype call for F9A but amplified for F5C (purple triangles in Fig. 1, observation in group b and c in Fig. 2). Accordingly, these individuals (with the sample codes CMF1598, CMF2121, CMF2223, CMF0858 and CMF1808) should be composed by two Z chromosomes and one W chromosome.