Fig. 5

Size dependent intron size evolution. A) Distributions of intron sizes in D. rerio discretised by their inferred size in the joint vertebrate ancestor (JVA). Colours indicate the size in the JVA as shown at the top. Introns that are short in the JVA have bimodal distributions in D. rerio with peaks at 76 and ca. 2000 bp. B-D) Heatmap representations of the joint distributions of intron sizes in the JVA and D. rerio (B), T. rubripes (C) and M. musculus (D). Solid blue lines indicate mode lines (distribution peaks); dashed lines indicate the diagonal. The marginal distributions are shown in the margins. The joint distributions were scaled by column and blurred to facilitate the extraction of the mode lines. E) Mode lines of the change in intron size from the JVA to extant vertebrates. Lines are equivalent to those extracted in B-C, but rotated to emphasise change in size. Colours indicate taxons as in (F). F) Proportion of introns minimised (between 76 and 100 bp long) as a function of size in the JVA. Background histogram gives the inferred distribution of intron sizes in the JVA (right axis). Thick solid lines give the proportions of introns minimised in the clade specific common ancestors. Note that only the teleost ancestor (red) is significantly different from the extant species indicating extensive heterogeneity in the sets of minimised introns in teleosts. G) Proportion of minimised introns (75-100 bp) in vertebrates plotted against genome size. Colours indicate taxons. H) Intron sizes in O. latipes plotted against D. rerio. The division into short and long sets of introns used to calculate correlation coefficents for (I) are indicated by the shaded regions. I) Introns in D. rerio lying between 76 bp and the 99th percentile were divided into two sets of equal size ranges (Fig. 5H) and the correlation in sizes between these introns and their orthologues in other species calculated. The plot shows the ratios of correlation coefficients for introns long and short in D. rerio plotted against the Kimura two-factor distances between the respective species pairs. Long introns are generally more correlated than short ones and this effect is stronger in more evolutionary diverged pairs of species. All intron sizes are log2 transformed