Natural selection in human populations : the measurement of ongoing genetic evolution in contemporary societies / edited Carl Jay Bajema.

Date:
[1971]
Catalogue details

Licence: Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0)

Credit: Natural selection in human populations : the measurement of ongoing genetic evolution in contemporary societies / edited Carl Jay Bajema. Source: Wellcome Collection.

    77/420 (page 65)
    Selection in Natural Populations 65 Weight classes with 100% observed mortaility were ignored. These classes would bias the result because the scale of mortality does not go beyond 1 even though the mean disability of the infant (in units of, e.g., lethal equivalents) might still in¬ crease. The equation for males is m = 0.118X2 _ 0.855 X+ 1.519, (2) while that for females is m - 0.134 - 1.030 X + 1.921. (3) From differentiation of these equations, the optimal birth weight for males is about 3.62 kg. and that for females is about 3.84 kg. Other regression models give slightly different estimates of the optima. The parabolic regressions of the kind used by Kam & Penrose (1951) are m = -0.890 + 6.22 Y - 6.32 (4) for males and m = -0.745 У2 + 5.81 Y - 6.38 (5) for females, where Y is the natural logarithm of the ratio of survivors to non- survivors. The optimal weights derived from these equations are 3.49 kg. for males and 3.90 kg. for females. Additional aspects of the regression curves will be dis¬ cussed in a later section. The unselected distribution of birth weights is not normal, but rather is left- skewed and somewhat leptokurtic, as shown by graphs against the normal distribu¬ tion expected for each sex with the given mean and variance. Much of this deviation from normality (on an arithmetic scale) is due to an excess of infants with very low birth weights. Because most of these infants die, the distribution of the survivors is more nearly normal. Graphs of the expected and observed distributions of the nor¬ mal survivors of each sex, however, also show some excess at the lowest birth weights, although the rest of the distribution is close to normality. There is some bias towards low weight among the foetal deaths because many of them die before birth, and the time from death to their expected time of delivery would have per¬ mitted additional growth if the foetus were alive. This bias cannot be estimated and has been ignored. Its effect is to make the estimated selection intensities somewhat lower than the real ones. The time of birth is in any event to some degree arbitrary from the viewpoint of the foetus: even though the size of the foetus may influence the time of its birth, this is certainly not the only important cause. It is possible to estimate the total number of selective deaths. The expected death rate for those individuals with the optimal birth weight, by equations (2) and (3), is about 0.009 for males and about 0.005 for females. The expected mortality at the optimal birth weight by equations (4) and (5) is about 0.011 for males and about 0.007 for females. The statistics derived from the latter values will be given in