Thursday, February 16, 2006

Why do fish need bicycles?

The evolution of sexual reproduction is one of the great mysteries of evolutionary biology. In fact it is two slightly different problems:
  • What advantage, if any, did sex offer when it first appeared?

  • Why does sex persist in modern organisms? That is, what stops them from becoming asexual again?
These questions, although related, might actually have slightly different answers. It may seem strange to ask such questions at all, but the reason is that there are many costs associated with reproducing sexually. I'll give two examples. First, sexually transmitted diseases are widespread in sexually reproducing populations, which makes sex risky. Second, there's the so-called "two-fold cost of sex". As feminists have been telling us for a while, males are pretty useless. Well, this is seems to be true in evolutionary terms as well. A mutant human female able to reproduce asexually and give birth to more females like her, would give rise to a population with twice the reproductive rate per capita of the normal human population, and these mutants would probably become dominant within a few centuries. (Actually, this is extremely unlikely to happen in our case because, due to a genetic quirk of mammals called genomic imprinting, asexual reproduction is very difficult to evolve in humans. However, asexuality can and has re-evolved many times in other animals, such as reptiles, fish and insects.)

Evolutionary biologists have been grappling with these questions for over a century and many hypotheses (over 20 by a recent count) have been proposed to explain the origin and maintenance of sexual reproduction. However, there is still a lot of debate, partly because many of the hypotheses are not mutually exclusive and are difficult to test. Many of the hypotheses that are currently favored have in common a central idea originally proposed by August Weismann over a century ago. This is that the benefits of sex are not direct (in the sense that the offspring of sexually reproducing individuals have a higher mean fitness than those of asexually reproducing ones) but indirect such that the offspring of sexually reproducing individuals have a higher variance in fitness than that of asexually reproducing ones. In other words, according to Weismann, sex makes natural selection more efficient, thus allowing sexual populations to adapt better to their environments. This can be achieved in many ways (hence the different hypotheses), such as eliminating deleterious (bad) mutations or allowing the spread of beneficial (good) mutations.

One of the main hypotheses from the Weismann "family" is the mutational deterministic hypothesis (MDH), developed by Alex Kondrashov and others. MDH postulates that sexual reproduction confers an advantage by helping natural selection remove bad mutations from the population. The MDH is very attractive because, in order for sexual populations to overcome the two-fold cost of sex, only two things must be true, and these can, in principle, be tested using data from real organisms.
  1. The rate of production of bad mutations must be relatively high, such that each individual acquires on average one or more bad germline mutations not inherited from their parents. This has been observed in some species, but not all. For example, humans have an even higher deleterious mutation rate than the one required by the MDH. The jury is still out over whether this assumption is generally valid in the real world -- there's a lot more work to be done there.

  2. The bad mutations must interact in a special way, called negative epistasis, such that adding more and more bad mutations makes you disproportionately sicker and sicker. For example, imagine that a single bad mutation lowers your fitness by 5% on average. If bad mutations don't interact, adding successive mutations should lead to a progressive decline in 5% steps. Negative epistasis would occur if, for example, the second mutation decreased fitness by 10%, the third by 15%, and so on. The evidence for this second assumption is also equivocal, partly because it is even more difficult to measure than the deleterious mutation rate.
In the next post I'll introduce the other concept needed to understand our paper: robustness.