Description and validation of a method for simultaneous estimation of effective population size and mutation rate from human population data.

Abstract

A method is presented for utilizing population data on electrophoretic variants of proteins to estimate simultaneously the effective sizes (Ne values) of the populations in question and the rate of mutation resulting in electromorphs at the loci whose products were surveyed. The method is applied to data from 12 relatively unacculturated Amerindian tribes for whom census data and independent estimates of the number of different electrophoretic variants at 27 loci are available. Because oftribal demographic structure, Ne should be less than the current number of reproductive-aged adults. In fact, it is substantially greater for 7 tribes, most likely due to intertribal migration and a recent decrease in tribal size. Estimates of locus mutation rates for the 27 loci vary by more than a factor of 20, with an average of 1.1 x 10-5 per locus per generation. This latter estimate is in satisfactory agreement with the results of other indirect approaches to the estimation of mutation rates in these tribes but about two times higher than the results of direct estimates based on these same loci in studies on civilized populations. This discrepancy could be due to the above-hypothesized migration and to decreases in tribal size.The effective size of a population (Ne) and the mutation rate at genetic loci (v) are two basic parameters in understanding the genetic structure of a population. We present a maximum likelihood method whereby the results of surveys of discrete and endogamous Amerindian populations for genetic variation at loci encoding proteins are used to estimate simultaneously these two parameters. The results of this method are then contrasted with independent estimates of the same parameters in these populations.NOTATIONS AND THE THEORY Consider a situation where J evolutionarily isolated populations have been assayed for genetic variation at I loci. Let vi denote the mutation rate at the ith locus (constant over all taxa) and Nj, the effective population size (Ne) of thejth taxa.We want to estimate I + J parameters (v1, .. , vi; N1, ... . NJ). When each population is at steady state under mutationdrift balance, Ewens (1) has shown that the statistic kij, the observed number of alleles in a sample of nij genes at the ith locus from thejth population (i = 1, 2, . . . , I;j = 1, 2, ... J), is a sufficient statistic for the above parameters. In addition, the probability of obtaining kij alleles in a sample of nij genes is given by ii(kinl +j)k# ) [1] where I 1.(kij) is the coefficient of ok in rIH'd, (Qij + m), termed a Stirlings number of the first kind, Oij = 4viNj, and F( ) is the Gamma function (1, 2).When a set of independently segregating loci is used in the survey, the evolutionary independence of these populations yields a joint likelihood function I J

Genetic variation of the mitochondrial DNA genome in American Indians is at mutation‐drift equilibrium

“…Our analytical test finds no such departure, when the number of distinct mtDNA-types seen in this sample is contrasted with its expectation. Note that a recent estimate of effective size in this population demonstrated no profound effect of its recent expansion on genetic variation in nuclear genes (Chakraborty and Neel, 1989).…”

The prospects for polymorphisms shared between species

“…Ns of 1000 does not require actual selection coeffi- Non-passerine birds Barrowclough & Shields, 1984 87 Passerine birds Barrowclough & Shields, 1984 30-200 Mammals Lande, 1979 40-300 Lower vertebrates Lande, 1979 Begon, 1977 400 Olivefruitfly Pollak, 1983 368 Bufo marinus Easteal, 1985 4286 Amerindian tribes Chakraborty & Neel, 1989 *The studies by Avise eta!. (1988) estimate female effective population size (these species were also chosen on the basis of their large census sizes and high rates of historical gene flow).…”

Rates of nucleotide substitution, mutation at a locus, and the "beanbag" gene number in man

“…We are mostly concerned here with mutations that affect a single nucleotide (point mutations) in which one nucleotide (base) is replaced by another. The average rates of mutation at a classical gene locus are estimated to be ~10 Ϫ5 per generation (Mukai and Cockerham 1977;Chakraborty and Neel 1989). They are much higher than the estimates for base substitution rates for two reasons.…”