|
|
|---|---|
Some Problems With Studying Discrete Traits
A quick note concerning some of the problems with studying discrete, or non-metric traits in the cranium and dentition include issues of intra- and inter-observer error, phenotypic expression, quasi-continuity, age and sex dimorphism, asymmetry, inter-trait correlation, causation factors, heritability, and statistical methods (Saunders, 1989). Intra-observer error describes the inaccuracies within the individual researcher when measuring the presence of discrete traits within a sample. Inter-observer error accounts for differences in scoring between observers. In this study problems of observer error were minimized because all of the traits were scored by Dr. Lukacs. The individual’s age can influence expression as well. Dental genotypes are obviously not expressed at birth, they are progressive through the lifespan and some teeth are more affected by this fact than others, the third molar does not erupt until age18 for instance. Also, living teeth are dynamic systems in which enamel is subject to environmental stress during development and teeth are worn during life. Therefore, traits which are found on occlusal surfaces or those which are characterized by extra enamel may be lost through attrition. Traits can also be lost through AMTL, because of periodontal disease or caries and some traits may aggravate these processes. The individuals in this study were all living young adults of similar age, between 15 and 19. Missing teeth were scored as separate from absence of trait. The sex of the individual can also influence expression in part because males in general, have more robust teeth and there may be an association between tooth size and expression, or penetrance of traits (Turner and Scott, 1997). The significance tests for sex dimorphism are presented here and show that the frequencies for significant levels of dimorphism have a mean of 15.63 % (standard deviation of 5.24%). The probability of getting these results randomly is 0.01, so there is a significant level of sex dimorphism and both sexes were included in this study. The pattern of side distribution for discrete traits is generally considered to be random, a phenomenon referred to as fluctuating asymmetry. However, there may be some effects from directional asymmetry which disturb the randomness of the distribution. There were very few significant differences caused by asymmetry, the actual p-values are reported above. The low level of observed asymmetry allowed the use of the teeth from the left side only, cutting the sample size significantly. Inter-trait correlation is related to the concept of pleiotropism, a single genetic locus which confers multiple and otherwise sometimes unrelated characteristics (Turner and Scott, 1997 and Saunders, 1989). These affected attributes may be alternate expressions of the same trait, traits associated by commonalties in development, or traits associated with common environmental factors. The effects of pleiotropism should be apparent in the Pearson inter-trait and inter-tooth correlations as the sample size is so large. The problem of causation arises when differences occur between populations in the frequency of discrete traits. Are the differences attributable to biological distance or has environment played a significant role? Assuming a high level of heritability vs. environmental factors in the given trait’s expression, are the differences in frequency due to migration, diffusion, or assimilation (Saunders, 1989)? The answer to this question generally lies in convergent lines of evidence, using metric traits, any other genetic evidence available, archaeological, historic and linguistic evidence to corroborate the conclusion. Problems of heritability are especially relevant because measuring discrete traits is based on phenotypic expression and there is the problem of understanding just how much appearances reflect their genetic base. This question is also reflected in problems of statistical methods. If environment is a significant factor in the expression of a discrete trait, the distribution may not be normal. This concern is reflected in the use of chi-square distributions and tests rather than student’s t or standard normal inference procedures. Also, the study is predicated on the assumption that the distance between the expression of these traits is equivalent to the distance between populations (Saunders, 1989). The validity of these assumptions is questionable and resolution is crucial to any study of discrete traits.
