the frozen plasticity theory
1. This theory is based ( I think) on one wrong key assumption that all the offspring need to possess that useful trait which allowed parents to have more offspring than any other pair, for evolution to work. Evolution can still proceed even if none of the immediate descendants possesses the trait, because in the process of evolution what is changing is a species. Important for evolution is therefore the frequency of a trait in a population. A species become taller when there is more tall individuals in the population. Although genotype for some trait of an individual is combination of different alleles this combination is not unique in population. Specific combination, which results in advantageous trait is not vanishing but has the same probability to from as any other combination. Actually if some trait allows to have more offspring the alleles forming that trait are copied more than other alleles and therefore the chance that they will meet again in this combination is increased ( in positive feed-back loop- higher frequency mean higher chance of meeting and that results in even higher frequency and so on). Also the higher variability allows for more combinations to be tested, some of them could be as good as parental, or even better and some could be not as good but still better than the population average. All these possibilities will lead to a higher fitness.
It is not so important to know how many genes control a trait, all what is needed to be known is variability of the trait in a population. For example when the difference between the highest and lowest male is let’s say 20 cm and height is controlled by (let’s suppose) 10 genes (two alleles for each gene and co-dominance), there are 59049 possible phenotypes. If every genotype had to have its own phenotype, two most similar genotypes would differ by 0.003 mm (for evolution absolutely invisible).Or where there are only two colors (e.g. white and blue) controlled by 10 genes, half of genotypes stand for one color and half for another.
Hallmark of epistasis is that many genotypes stand just for a few phenotypes.
Worth of considering is also a supergene theory. It suggests that traits controlled by alleles of many genes are selected to move closer on a chromosome, what results in a lower probability of the cross-over. This way the trait is inherited as one locus.
2. It is not true that: “Population genetics models usually suggest that each allele can be characterized by a constant, a value that describes the average relative fitness of the carriers of a particular allele“. Not alleles but (logically) phenotypes are associated with different fitness. For example in a case of sickle cell disease there are three genotypes and each one stands for different phenotype. Homozygote with two mutated alleles have abnormal red blood cells that can lead to death (low fitness), homozygote with two normal alleles have functional red blood cells. Heterozygote with one mutated and one normal allele produces a few sickled red blood cells (abnormal) not enough to cause symptoms, but enough to give resistance to malaria. So both homozygotes have a lower fitness than the heterozygotes. Two alleles (normal and mutated) are in an equilibrium due two opposing selection forces. If one of this forces disappear (for example as a result of environmental change mosquitoes would get extinct and malaria would vanish with them) then the frequency of the mutated alele will decrease. No matter how low the frequency will be, its fitness will not increase. So no “frequency dependent selection“ will act againts this decrease.
3.Also a process of freezing seems to be problematic. First of all, not species but induvidual traits could freeze (for example when a color of a fur is frozen, lenght of claws could still be unfrozen). When a trait is so fragile that only one combination can code for it, as the frozen plasticity theory suggests, than any new mutation (any new variability) should destroy it. Therefore strong purifing selection will be acting to sweep away any variability and the trait will never get frozen.