Serpwidgets

This is because a normal is not carrying any copies of the Salmon mutant. If it were, the snake would not be a normal. The only way a normal "sibling to salmon" would have salmon offspring is if its mate were carrying the mutant gene.

When you cross a salmon to a normal and get normals and salmons, it all makes sense when you look at what is happening underneath the hood... the normal is S⁺·S⁺ (homozygous normal) and the salmon is S⁺·S^S (heterozygous normal and salmon) and the Punnett square for this cross reflects the results you expect:

<table border=1><tr align=right><td> </td><td>S⁺</td></tr>
<tr><td>S⁺</td><td>S⁺·S⁺</td></tr><tr><td>S^S</td><td>S^S·S⁺</td></tr></table>

The normal (on top) is giving the wild-type allele to all offspring, the salmon (on the left) is heterozygous and thus giving the wild-type to some, the salmon mutant to others. The resulting genotypes in the litter are normal (homozygous or genotype S⁺·S⁺) and the salmons are hets (S⁺·S^S)

Obviously a normal will not throw the salmon mutant to any of its offspring, because it does not have a salmon mutant to give. Nobody is saying that a normal is het for salmon. But a salmon is het for salmon and wild-type at the salmon locus, the same as a "normal het for albino" is a normal het for albino and wild-type at the albino locus. What is being called a "super salmon" is simply homozygous for the salmon mutant at the salmon locus.