Selection based on the ability to find a mate can take two forms. Organisms can compete among themselves for access to a mate. In animals this is usually competition among males, since females are normally the limiting resource. A trait which improves a male's ability to compete will be favored, though this can be opposed by other forms of selection.
If the sexual selection is very strong, it can overcome strong natural selection. Some researchers think that the extinct Irish Elk evolved its big antlers for sexual competition and compromised its ability to survive.
A more complicated situation arises where one gender chooses which member of the other to mate with. In mammals and birds this is usually female choice of males.
It would be straightforward if females chose the healthiest and best adapted males, since that would tend to lead to healthy, well-adapted offspring. However, we observe that many females prefer traits such as peacock tails which are probably detrimental to the health and survival of the possessor. Why? At least two theories have been proposed.
Under this theory, the trait (say, peacock tail) started out mild and advantageous. Females evolved a preference for the trait because it was advantageous, and they would have better quality sons if they mated with a male possessing it.
However, if the female preference was for a longer tail and not just a useful-length one, males could evolve to have longer tails specifically to attract females. Females would evolve to prefer long-tail mates not only because the trait was advantagous in itself, but because their sons would be desirable to females and thus give them more grandchildren.
This positive feedback process could push the tail far past the point where it stopped being an advantage in itself. In the end, the only advantage of the long tail is that it's attractive, and the only reason to mate with a long-tailed male is to have attractive sons, but that's enough to keep the system in place.
In theory a population could destroy itself via run-away sexual selection, reducing its viability so much that it would be wiped out by some fluctuation of the environment.
This theory argues that the showy male trait acts as a proof of the male's general fitness. Only a healthy peacock would be able to have a huge tail, and only an otherwise superior peacock could hope to survive the liability of the tail. Thus, females are selecting for overall good genes for their offspring, as demonstrated by the male's ability to waste resources.
Why doesn't the male evolve a cheaper form of proof? The difficulty is that if the proof is cheap, even an inferior, unhealthy male can display it, so females have no genetic incentive to believe it. (Throughout this discussion, words like prove and believe should be in quotes. Sexual selection can work even in organisms which are almost surely not thinking about it.)
One difficulty with this theory is that the female may be selecting for overall good genes, but they have to be good enough to overcome the bad trait of the enormous tail. She risks having sons who will kill themselves trying to drag that tail around, despite their otherwise good genes. At some point she would be better off picking a slightly less superior, but short-tailed, male.
A variation of this theory is the parasite-load theory of Hamilton and Zuk. They argue that it is hard for a sickly, parasite- ridden male to maintain colorful plumage, so fancy plumage is a good handicap sign-it is hard to fake and advertises freedom from parasites. Evidence for this theory: fancy plumage is more common in highly parasitized species of birds, and in an experimental study, males with fancier plumage had offspring with fewer parasites, even when the offspring were moved to foster nests.
Most species that have two sexes have nearly equal numbers of each. Why? It would seem that if, for example, most male sea lions never reproduce, it would make sense to have fewer males. While our X/Y system looks as though it has to produce 50/50, it is actually easy for modifier genes to change the sex ratio. Yet it remains close to 50/50.
Haldane argued that half the genes in the next generation came from males, and half from females. The best way to make sure your genes win out, then, is to put them in the rarer sex where there will be less competition. This leads to pressure to have whichever sex is rarer, until it is not rarer anymore. This theory fits several observations:
Some genes can influence the Mendelian ratio in their own behalf. A classic example is the t haplotype of mice. t/+ heterozygote males make more t sperm than + sperm. Apparently the t locus damages the non-t chromosome during spermatogenesis.
t chromosomes in wild mouse populations always carry some lethals closely linked to the t locus (and generally in an inversion which prevents recombination). If this was not the case, t would increase so fast we'd probably never see a case where it wasn't fixed.
t is an example of selfish DNA: a gene which increases not because it is good for the organism, but because it is good for itself. The behavior of t is called meiotic drive or segregation distortion.
If a Y has meiotic drive, it will be passed to more than half of a male's offspring, so it will increase: but he will have more than 50% sons, which is not to his advantage. In this case it can be shown that the selection at the sperm level will defeat selection at the individual level. The population is on a death spiral to being all males.
However, if a mutation arises anywhere in the genome that can stop the Y from being driven, it will spread (since its possessors will produce daughters, who are in great demand) and this may save the population.
[If I have time, I will talk about selfish X chromosomes here.]
Transposons are another form of selfish DNA. These are genetic elements that can move and reproduce within the genome. A transposon that makes many copies of itself is more likely to survive (they are vulnerable to mutation and deletion) than one that makes few copies. So transposons which replicate wildly will win the within-cell competition. But they may be very bad for the organism. Again, modifier genes that can stop the selfish gene will be selected for. We observe that in most organisms with transposons, they don't move often except under stressful or unusual circumstances.
Barbara McClintock suggested that the movement under stressful circumstances is actually adaptive-it may produce a favorable mutation and rescue the organism from its troubles. An alternative explanation is that a stressed cell simply can't afford the resources to suppress its transposons. These theories are fairly hard to test.