Some of what these notes call lecture 4 was presented in lecture 3 and vice versa. Funny things happen when one is actually lecturing....
X-linked traits can be analyzed like autosomal ones. There are five genotypes instead of three: females have two homozygotes and a heterozygote, males have two hemizygotes (only one gene present, either a or A ).
Recessive genes are exposed in males, so deleterious recessives are eliminated faster, and advantageous recessives are fixed faster. The effects of heterozygote advantage and disadvantage are weakened because there are fewer heterozygotes.
A gene which is favored when rare will have dynamics like heterozygote advantage; a gene which is favored when common will have dynamics like heterozygote disadvantage. It is actually hard to tell whether you have heterozygote advantage or frequency-dependent selection unless you can measure selection on individuals, since their overall results are the same.
Multiple loci can be analyzed by the same reasoning as single loci if you are willing to make some assumptions about how they interact. Two common assumptions are additive fitness (the fitness increase is the same for each additional favorable allele) and multiplicative fitness (the fitness is the product of the fitness for each gene on its own). Real situations are often more complex than these.
In a two-locus case with no linkage, we expect Hardy-Weinberg at gamete formation for each locus independently. The only interaction will come in the fitness terms: is AAbb better, worse, or the same as aaBB or AaBb?
However, if there is genetic linkage and we do not start out with all possible combinations, we will see different results. At the limit of no recombination, if we start with only aB and Ab gametes, we will never form aabb or AABB zygotes; the population cannot reach those parts of fitness space, even if they might be good. In essence the two loci behave like one locus with two alleles (aB and Ab).
With limited recombination we will slowly break up the combinations over time, so initially the system will look like a 1-locus system but eventually it will look like a 2-locus one.
Some workers think that it is advantageous for loci which need to work well together to be tightly linked, so that working combinations are not broken up. These groups of cooperating loci are called co-adapted gene complexes. The genes involved in butterfly mimicry, for example, are tightly linked, as are the human genes involved in the cellular immune response.
Early researchers were worried about the cost imposed on a population by selection, which they called genetic load. For example, the sickle-cell polymorphism imposes a cost due to the deaths of homozygotes (one kind via malaria, the other via sickle-cell disease).
Prior to molecular tests such as electrophoresis, it was thought that most organisms were homozygotes for most of their genes. This turns out not to be the case. If we assume that all that heterozygosity is due to balancing selection, can the population really survive?
Additional load is provided by the rare recessives found in all populations; for example, it has been suggested that each human carries, on average, 1-4 recessive genes which would be lethal in homozygous form. (This estimate is made by studying the health of children produced by mating among close relatives.)
Part of the answer to this question is that all populations have excess reproductive capability and can survive well even when far below maximum fitness. But it is still possible to show that no organism could maintain variability in tens of thousands of genes via balancing selection-so many homozygotes would need to die that no one would be left. This observation was one of the foundations of the Neutral Theory, which Peter will discuss in a few weeks.
It's also important to remember that maximum fitness is an arbitrary concept. If a new mutation suddenly arises that is more fit than anything currently existing, that doesn't mean that the current population (which is far below the new maximum fitness) is sickly or endangered, just that it will tend to lose out in competition with the new form.