Ribosomal frameshifting, a translational error, catastrophically alters the amino acid composition of the nascent protein by shifting the reading frame from the intended contiguous trinucleotide reading. Frameshift events waste energy and resources, and peptide products have unpredictable cytotoxic effects. The 'Ambush Hypothesis' (Seligmann and Pollock 2004, DNA Cell Biol 23:701-5) suggests there is a selective pressure favouring the evolution of out-of-frame ('hidden') stop codons. Although this hypothesis has gained empirical support through whole-genome studies, it is presently unknown whether it can be applied at a single-gene scale. Herein, we report such an investigation using the gene, polyketide synthase (PKS), among species of fungi. Contrary to expectation, genes presented with significantly lower number of hidden stop codons than expected in a selection-neutral model (p < 0.0005), suggesting both non-adherence to the ambush hypothesis as well as suppression of hidden stop codon evolution. It is known that there are multiple adaptive considerations determining codon selection during evolution, and that the information-holding potential of the genetic code is finite. We hypothesize that the reason for low hidden stops in PKS genes is due to competing 'codon biases' that are prioritized over the selective pressure favouring the emergence of hidden stops. Future studies of the ambush hypothesis in the context of other drivers of codon bias may allow this hypothesis to be molded into a comprehensive genetic theory that can be integrated within the broader genetic theory of codon bias and applied to the genetic code at any scale of analysis.