al., 1984; Luckinbill and Clare, 1985) and other cases where they have not been (Partridge and Fowler, 1992). Still other studies find evidence both for and against antagonistic pleiotropy, as well as for and against mutation accumulation (Service et al., 1988; Leroi et al., 1994; Promislow et al., 1996; Tatar et al., 1996). The question is, then, how these results reflect on the validity of the two mechanisms of demographic evolution.

Many of the aforementioned studies have been plagued by artifacts. Indeed, it may be that almost all experiments of this kind are afflicted to at least some degree by problems like inbreeding depression, inadequate replication, genotype-by-environment interaction, and so on (Rose and Service, 1985). Nonetheless, it is sometimes possible for further experimentation to sort out many of these problems (e.g., Rose, 1984b; Service and Rose, 1985; Leroi et al., 1994; Chippindale et al., 1994).

Another point to make is that the experimental evidence can't plausibly be viewed as an either-or competition between mutation and pleiotropy models. There is some empirical support for both, some of it coming from the same experiments (e.g., Service et al., 1988; Tatar et al., 1996). Therefore, the only balanced point of view is that both of these genetic mechanisms are somewhat supported. They may both be involved in the evolution of demography, but neither is necessarily involved. Either is, in any case, sufficient for the evolution of demography to be dominated by the force of natural selection.

Gompertz Models: Phenomenological Theory

From age 15 years or so, the rate of mortality accelerates exponentially in long-industrialized human populations, and this has been a commonplace of both demography and gerontology (e.g., Comfort, 1964). The Gompertz equation

attempts to summarize this pattern using what amounts to a linear regression of the logarithm of the mortality rate (µ) on age, x, where a gives the slope of that regression and α is the initial rate of mortality. It has been argued that this ad hoc demographic model is key to understanding aging (e.g., Finch et al., 1990; Finch, 1990), and it certainly has much to recommend it, particularly in comparison with alternative indices, like mean or maximum life span, when the data are collected under dubious conditions.

My colleagues at the University of California, Irvine, and I have recently explored the extent to which the Gompertz model can be used to predict accurately the demography of experimental Drosophila populations (Mueller et al., 1995; Nusbaum et al., 1996). We did so because we do not regard human data as a particularly good way to test any general population-level theory; there is too much complexity introduced by culture and history. We also have reservations concerning the uncritical use of field data. We began with considerable doubts

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement