Wright (1932) felt that in a population with a large effective size there would be virtually no chance for a peak shift. Hence, the population would evolve toward the nearest local peak, and then selection would maintain the population on that local peak, even if a vastly superior adaptive alternative existed. We have already seen an example of this with the j -Hb locus in human populations living in the malarial regions of Africa; most such populations have evolved toward the A/S polymorphic peak even though the fixation for the C peak seems to be a far superior adaptive outcome in terms of average fitness (Figure 11.8), percentage of the population protected against malaria, degree of protection against malaria, and eliminating deleterious side effects such as hemolytic anemia from the population. Because selection could keep populations on inferior adaptive peaks, Wright regarded natural selection as a potential impediment to adaptation when it dominated over genetic drift.
On the other hand, an isolated population of small variance effective size is also not optimal for adaptive evolution. Genetic drift is powerful in such a population, but the primary manifestation of the power of drift is the rapid loss of genetic variation. Without genetic variation, there is no evolution of any sort. What is needed is to simultaneously have demes of small variance effective size but with access to much genetic variation. As shown in Chapters 6 and 9, population subdivision can both induce small variance effective sizes at the local deme level yet maintain higher levels of genetic diversity and additive genetic variance at the global level than an equally sized panmictic species (see equations 6.44 and 9.28). What is needed is the right amount of gene flow between the local demes: too much, and the population becomes effectively panmictic and selection prevents genetic drift from allowing populations to explore the adaptive surface; too little, and the local demes generally have little genetic variation and hence low adaptive potential. Under an island model of population subdivision, Wright (1932) argued that an Nm term (the product of the local variance effective size and the migration rate) of about 1 provided the right balance. This conclusion has been supported by Barton and Rouhani (1993), who showed that peak shifts are most likely if Nm is slightly below 1.
To illustrate peak shifts in a subdivided population, a computer simulation was run using the same fitness parameters as that given in Figure 12.3 and with the local variance effective size being 100, as in Figure 12.4, but now with an island model of gene flow with m = 0.01. Hence, Nm = 1 in these simulations. As before, we assume that the population had adapted to a previous environment with an equilibrium p = 0.3. More of the local demes will tend to go to the p = 0 peak than the p = 1 peak upon the environmental change, so we will assume that the average frequency of the A allele in the global population is 0.1. Note that 0.1 is well below the threshold of p = 3 that separates the domain of the lower peak from that of the higher. Hence, gene flow in these simulations acts as a directional force to bring local populations to the lower fitness peak. Figure 12.5 shows the evolution of eight local demes, each starting with p = 0.3 but receiving input at rate m = 0.01 every generation from the global gene pool with p = 0.1. As can be seen, two of the eight local demes still evolved toward the higher peak (although gene flow now prevents fixation of A). Obviously,
Was this article helpful?