Graphs of Codon Adaptation Index (CAI) vs. normalized substitution rates. Data for the sensu stricto yeasts (S. cerevisiae, S. bayanus, S. mikatae, S. paradoxus) (right) and for N. crassa-C. globosum (left) are both shown. Although the r distribution for N. crassa-C. globosum (Fig 1c) is wider than a Normal distribution, this appears to be due to a large number of genes under codon usage selection, rather than regional mutational biases. Genes under codon usage bias tend to have low z-scores and high CAI values (red ellipse), while the remaining genes have low CAI values and r values centered around zero (yellow ellipse). N. crassa has a much larger number of genes in the cluster likely to be under codon usage selection (949 genes with CAI≥0.70, red ellipse) than does S. cerevisiae (121 genes with CAI≥0.40), suggesting that more genes are under codon usage selection in N. crassa.