pleiotropy
Orr HA 2000 Adaptation and the cost of complexity. Evolution 54:13-20.
- fitness is determined by n independent (orthogonal) characters
- at each of our n characters, fitness falls off as a Gaussian function of distance, zi, from the optimum
- Fisher's model thus allows for a kind of universal pleiotropy
- we consider a sexual haploid in which evolution is due to new unique mutations
- the expectation of the product Pa Π Δzfix in eq. (1) equals the product of the individual expectations
- when using mutations of fixed size r, the speed of adaptation slows with greater organismal complexity, n
- more complex organisms pay a cost of about n−1/2 in terms of probability of fixation
- more complex organisms pay a cost of about n−1/2 in terms of the gain in fitness that results when a substitution does occur
- more complex organisms pay a third cost
- the probability that a random mutation will be favorable decreases with n
- random mutations of a given size have a smaller chance of being favorable in complex organisms
- the probability of fixation for mutations of a given size declines with organismal complexity
- when favorable mutations do get substituted, a smaller gain in fitness accrues to complex organisms
- there may well be other, and perhaps more than compensatory, ecological advantages to complexity
- when using mutations of the same size, complex organisms adapt more slowly than simple ones
- we are not concerned with the cost of producing a more complex organism
- in Rechenberg (1984), mutations are built from the "character up"
- mutations change the value of each character by a random amount
- these changes are independent and identically distributed across characters
- the square of the magnitude of mutational effects is chi-square distributed
- such a distribution appears unbiological as mutational effects are maximized away from zero
- small mutations are rare or nonexistent, intermediate-sized ones common, and large ones rare or nonexistent
- this nonmonotonic distribution qualitatively differs from the leptokurtic ones usually assumed in evolutionary biology
- equation (A6) seem inconsistent with the nearly neutral theory
- because fitness effects scale with mutation size, exponential or gamma distributions (with small values of the shape parameter) of fitness effects cannot be naturally obtained from equation (A6)
- in Kimura (1983), Orr and Coyne (1992), and Orr (1998), mutations are made from the "top down"
- we begin with some biologically plausible distribution of mutational magnitudes (e.g., exponential)
- any particular mutation is forced to have a random direction
- each mutation represents a random displacement in n-dimensional space