Nei's Modified Genetic Identity and Distance Measures and Their Sampling Variances

Abstract

-The conditions under which Nei's (1972) genetic identity measure (I) yields results which are discordant with changes recorded in the gene identities at single loci are defined. We noticed that upon reassessment of allele frequencies, the value of I can in some cases change in the opposite direction of changes recorded in single locus gene identities. This anomaly may affect phylogenetic reconstructions especially when closely related populations and/or rare alleles are involved. We illustrate this problem using two examples, one based on real electrophoretic data from two Macaca species, the other based on hypothetical allele frequencies. We propose to use instead of Nei's I, an alternative measure, which we call Nei's modified genetic identity (I). This measure is based on the arithmetic mean of single locus gene identities. Nei's modified distance (D) is derived analogically to Nei's D. We present the sampling variances of these modified estimates. [Nei's distance; allele frequencies; philogeny; electrophoresis.] The most widely used genetic distance estimate is Nei's (1972) D. Its popularity stems from its simplicity and facility of application (Hedrick, 1983; Kimura, 1983). Nei's D has also been shown to have an approximate linear relationship with time of divergence (Nei, 1987). However, during a reassessment of data on the gene diversity at the amylase and hemoglobin loci in two, species of macaques (Tomiuk, unpubl.), we noticed that it is possible to record a decrease in the single locus gene identity at one or several loci, and concomitantly record an increase in the overall gene identity across all loci between the two species. This means that despite of a larger divergence at single loci, the total genetic identity may in some cases assume higher values, thus indicating undue genetic similarity. In this note we assess the conditions under which this type of anomaly may occur. Furthermore, we propose modified versions of Nei's genetic identity and Nei's genetic distance, as suggested but not recommended by Nei in his 1972 paper, and * Dedicated to Prof. K. Wohrmann on his sixtieth birthday. 3To whom all correspondence should be addressed. presented again in Hillis (1984). In this note we also calculate the sampling variances of the modified formulae.