I have read of this topic, in reference with genetic interaction with epistasis: The segregation proportions 9:3:3:1 typical of a dihybrid changes to 9:6:1. If the dominant condition (A_bb and aaB_) at either locus, but not both, produces the same phenotype, the F2 ratio becomes 9 : 6 : 1. There is a good example in the colour coat of the Duroc breed of pigs, which can be sandy, red and white, here the epistatic genes are involved in producing various amounts of a pigment, the dominant genotypes of each locus may be considered to produce one unit of pigment independently. Thus, genotypes A_bb and aaB_ produce only one unit of pigment each and therefore have the same phenotype, which is sandy. The genotype aabb produces no pigment, therefore they are white, but in the genotype A_B_ have a cumulative effect and two units of pigment are produced, and the coat is red.
We said that ther is epistasis because the number of phenotypes is less than 4 in this case 3. and also the typical segregation 9:3:3:1 changes to 9:6:1.