The effect of population structure on genetic drift was studied using three computer models to simulate the populations of interest. For each model the effective breeding population size was 100, and the initial gene frequency for a locus with two alleles (A, a) was 0.5. These populations were considered to be made up of sexually reproducing diploid organisms. The population size was held constant from generation to generation, and generations were nonoverlapping. It was assumed that selection and mutation were absent. In the panmictic model, mating was at random within the population. In the other two models the populations were subdivided into 10 subpopulations of equal size, each with an initial gene frequency of 0.5. Mating between subpopulations was structured according to an area migration scheme based on the proximity of subpopulations to one another. The two levels of migration were 40% and 5%.
Gametes were simulated by random numbers generated by computational methods. In each model 400 replicates were made by using different sequences of random numbers. Each replicate was run for 200 generations or until one allele was fixed. At every 25th generation the computer printouts included the total gene frequency for the population for each model and the gene frequency in each subpopulation or the subdivided models.
The results showed a significantly reduced cumulative rate of fixation starting at the 100th generation in the 5% model as compared with the 40% and panmictic models. The latter two were almost identical in fixation rate. The variances among subpopulations for any one generation were computed for the two subdivided models and compared with variances of the panmictic model taken over 10 generations around the generation of interest. The variances of the 5% model were significantly higher than either of the other models for generations 25 through 175 but the variances of the 40% model were significantly different from the panmictic model only through the 100th generation. This indicates that for either subdivided model the variance within any one generation is equal to or greater than that found in the panmictic model in 10 generations. The distributions of gene frequencies over time did not differ significantly among models nor did they deviate from predicted theoretical distributions.
It was tentatively concluded that population structure can alter the effects of random drift on gene frequencies as well as the genetic variability maintained in a population. Further studies employing selection and mutation coefficients are suggested.
Level of Degree
UNM Biology Department
First Committee Member (Chair)
William Wayne Johnson
Second Committee Member
Abraham P. Hillman
Third Committee Member
Loren David Potter
Berger, Beverly Jane. "Computer Models of Genetic Drift in Small Populations with Three Different Structures." (1967). https://digitalrepository.unm.edu/biol_etds/367