Results
The proportion of reads mapping to each of the autosomes was highly consistent between samples (Fig. 1). Further, autosomal read-dosage appeared to be positively correlated with scaffold length, as expected if restriction motifs are randomly distributed. We tested this correlation by performing a linear regression in RStudio (proportion of reads ~ scaffold length), which resulted in a slope coefficient of 3.833e-10 (adjusted R2 = 0.7, p < 2e-16). Unlike the autosomes, values for the proportion of reads mapping to the X chromosome formed two clusters, putatively representing females (with higher read-dosage values) and males (with lower read-dosage values). However, the mean proportion of reads mapping to the X chromosome (length = 171,031,299 bp) for the putatively female samples (0.0308) was substantially lower than the expectation (0.0656) based on the relationship between the proportion of reads mapped and scaffold length inferred from the autosomes. This difference may represent an inherent bias against sex-linked loci in the DArT pre-sequencing workflow or a depletion in the restriction motif on the X chromosome relative to the autosomes.
The read-dosage sex-assignment program (sexassign ) allowed us to successfully assign all individuals in the dataset as either male (heterogametic, XY; X read-dosage = ~0.5X) or female (homogametic, XX; X read-dosage = ~1X, Fig. 2, Table 1). Of the 60 individuals sequenced, 33 were determined to be female and 27 to be male, consistent with the typical sex ratio in rodent populations under normal conditions (Labov et al., 1986; Rosenfeld et al., 2003). Genetic sex determination had a ~94% concurrence rate with field determined sex, a typical human error margin considering the lack of obvious sexual dimorphism within the species and the difficulty of accurately sexing rodents in the field, particularly during non-reproductive periods (Hoffmann et al., 2010; Jacques et al., 2015).