A Multivariate Genetic Analysis of Ridge Count Data From the Offspring of Monozygotic Twins

Abstract

Inheritance patterns of digital ridge counts have been analyzed using multivariate statistical methods and data from the offspring of half-sib twin kinships. Prior studies found the univariate measure total ridge count to be highly heritable and the counts on individual fingers to be somewhat less heritable, and exploratory factor analytic studies indicated that at least two, and possibly three, independent genetic influences are responsible for this ten variable multivariate trait.Two statistical methods have been employed to elucidate the factors controlling ridge count development on all ten fingers. An exploratory method developed by Bock and Vandenberg [4] has been applied to the among and between mean square matrices from a multivariate nested analysis of variance on thirty balanced male twin kinships. A principal component analysis on the resulting matrix of pure genetic effects has revealed two substantial genetic factors. One strongly influences the counts on all ten fingers, with the largest loadings on the three central fingers of each hand, while the other has an impact on the thumbs and fifth fingers. For both factors the loadings on homologous fingers are nearly equal. This exploratory procedure is wasteful of the data that is available in half-sib twin kinships, however.Confirmatory factor analyses, employing the LISREL IV program, have been conducted on all available ridge count data from the offspring of forty-eight unbalanced male twin kinships and fiftynine unbalanced female twin kinships. Nested analyses of variance performed on sex-adjusted data yielded five 10 × 10 variance-covariance matrices containing 275 unique statistics for the estimation of genetic and environmental parameters and the testing of hypotheses.A series of ten genetic and environmental hypothetical models for ridge count development, each more complex than the previous one, have been tested. They include a simple environmental model, an additive genetic and environmental model proposed by Holt [16], a full additive genetic model including five separate finger factors, two laterality factors and a general genetic factor, and seven models augmenting this full additive genetic model with factors for maternal epistatic and general environmental effects. The most complete model, which includes eight additive (one general, two laterality, and five finger) as well as maternal, epistatic, and general environmental factors cannot be rejected at a .05 level of significance. This model accounts for 99% of the variance that cannot be accounted for by a simple environmental model, and 95% of the variance unaccounted for by Holt's model. It suggests that while a strong genetic factor influences the ridge counts on all ten fingers, there are other factors affecting the counts on the homologous fingers separately as well as different factors affecting the counts on the left and right hands. In addition to these additive effects, influences due to the maternal environment common to all pregnancies of the mother, and those due to the unique environment of each pregnancy of the mother, and those due to the interaction of genes at separate loci have also been detected.Results of the Bock and Vandenberg analysis are concordant with those obtained by the LISREL program. While the former only requires the availability of standard statistical packages, it is wasteful of data from the half-sib families. The latter, on the other hand, while it requires the use of a specific program, LISREL or its equivalent, uses all half-sibship data and allows one to test genetic and environmental hypotheses as well as conduct exploratory factor analyses.