rogers@ARSUN.ANTHRO.UTAH.EDU (Alan R. Rogers) (10/18/90)
>How do you measure fitness? What are the units? I think of fitness as the conditional expectation of reproductive success, given genotype (or, depending on context, phenotype). Being an expectation, it is a parameter, not a variable that you can measure directly. But this answer only raises another question: what is reproductive success? Some models define it as the expected number of one's children, but that may not be a useful definition if children vary in quality. Others have used the expected number of grandchildren, but this ignores variation in the quality of grandchildren. One way out is to define a "tau-generation fitness", w(tau,x), which is expected the number descendants in tau generations expected by an individual with characteristic (genotype or whatever) x. Then the long-term fitness of characteristic x relative to characteristic y is the limit as tau --> infinity of w(tau,x)/w(tau,y). Believe it or not, you can actually make progress with that definition, and with a suitable model you can even estimate it. I took this approach in a paper that is now in press at Ethology and Sociobiology. Fisher did more or less the same thing in 1930. >My second objection is that if fitness is the time geometric mean, then >highly successful animals like the dinosaur and (yes) humans AND ALMOST >ALL SPECIES have zero fitness, since most species die off eventually! True. Eventually, the sun will explode and all life on earth will perish. In the long run, therefore, everyone has zero fitness. But it's best to ignore this eventuality if you want to figure out how evolution will procede in the meantime. Alan Rogers INTERNET: rogers@anthro.utah.edu USMAIL : Dept. of Anthropology, Univ. of Utah, S.L.C., UT 84112 PHONE : (801) 581-5529