I. Genetics And BehaviorP. L. Broadhurst
BIBLIOGRAPHY
II. Demography and Population GeneticsJean Sutter
BIBLIOGRAPHY
III. Race and GeneticsJ. N. Spuhler
BIBLIOGRAPHY
Behavior genetics is a relatively new cross-disciplinary specialization between genetics and Psychology. It is so new that it hardly knows what to call itself. The term behavior genetics is gaining currency in the United States; but in some quarters there, and certainly elsewhere, the term psycho-genetics is favored. Logically, the best name would be genetical psychology, since the emphasis is on the use of the techniques of genetics in the analysis of behavior rather than vice versa; but the in evitable ambiguity of that term is apparent. Psy chologists generally use the terms genetic or genetical in two senses: in the first and older sense of developmental, or ontogenetic; and in the second, more recent usage relating to the analysis of inheritance. The psychologist G. Stanley Hall coined the term genetic before the turn of the century to denote developmental studies (witness the Journal of Genetic Psychology), and Alfred Binet even used the term psychogenetic in this sense. But with the rapid rise of the discipline now known as genetics after the rediscovery of the Mendelian laws in 1900, William Bateson, one of the founders of this new science, pre-empted the term genetic in naming it, thereby investing genetic with the double meaning that causes the current confusion. Psychological genetics, with its obvious abbreviation, psychogenetics, is probably the best escape from the dilemma.
Importance of genetics in behavior. The importance of psychogenetics lies in the fundamental nature of the biological processes in our understanding of human social behavior. The social sciences, and psychology in particular, have long concentrated on environmental determinants of behavior and neglected hereditary ones. But it is clear that in many psychological functions a substantial portion of the observed variation, roughly of the order of 50 per cent for many traits, can be ascribed to hereditary causation. To ignore this hereditary contribution is to impede both action and thought in this area.
This manifold contribution to behavioral variation is not a static affair. Heredity and environment interact, and behavior is the product, rather than the sum, of their respective contributions. The number of sources of variability in both he redity and environment is large, and the consequent number of such possible products even larger. Nevertheless, these outcomes are not incalculable, and experimental and other analyses of their limits are of immense potential interest to the behavioral scientist. The chief theoretical interest lies in the analysis of the evolution of behavior; and the chief practical significance, so far as can be envisaged at present, lies in the possibilities psychogenetics has for the optimization of genetic potential by manipulation of the environmental expression of it.
Major current approaches. The major approaches to the study of psychogenetics can be characterized as the direct, or experimental, and the indirect, or observational. The former derive principally from the genetical parent of this hybrid discipline and involve the manipulation of the heredity of experimental subjects, usually by restricting the choice of mates in some specially defined way. Since such techniques are not possible with human subjects a second major approach exists, the indirect or observational, with its techniques derived largely from psychology and sociology. The two approaches are largely complementary in the case of natural genetic experiments in human populations, such as twinning or cousin marriages. Thus, the distinction between the two is based on the practicability of controlling in some way the essentially immutable genetic endowmentin a word, the genotypeof the individuals subject to investigation. With typical experimental animals (rats, mice, etc.) and other organisms used by the geneticist, such as the fruit fly and many microorganisms, the genotype can often be specified in advance and populations constructed by the hybridization of suitable strains to meet this specification with a high degree of accuracy. Not so with humans, where the genotype must remain as given, and indeed where its details can rarely be specified with any degree of accuracy except for certain physical characteristics, such as blood groups. Observational, demographic, and similar techniques are therefore all that are available here. The human field has another disadvantage in rigorous psychogenetic work: the impossibility of radically manipulating the environmentfor example, by rearing humans in experimental environ ments from birth in the way that can easily be done with animals in the laboratory. Since in psychogenetics, as in all branches of genetics, one deals with a phenotypein this case, behavior and since the phenotype is the end product of the action, or better still, interaction of genotype and environment, human psychogenetics is fraught with double difficulty. Analytical techniques to be mentioned later can assist in resolving some of these difficulties.
Definition. To define psychogenetics as the study of the inheritance of behavior is to adopt a misleadingly narrow definition of the area of study, and one which is unduly restrictive in its emphasis on the hereditarian point of view. Just as the parent discipline of genetics is the analysis not only of the similarities between individuals but also of the differences between them, so psychogenetics seeks to understand the basis of individual differences in behavior. Any psychogenetic analysis must therefore be concerned with the environmental determinants of behavior (conventionally implicated in the genesis of differences) in addition to the hereditary ones (the classic source of resemblances). But manifestly this dichotomy does not always operate, so that for this reason alone the analysis of environmental effects must go hand in hand with the search for genetic causation. This is true even if the intention is merely to exclude the influence of the one the better to study the other; but the approach advocated here is to study the two in tandem, as it were, and to determine the extent to which the one interacts with the other. Psychogenetics is best viewed as that specialization which concerns itself with the interaction of heredity and environment, insofar as they affect behavior. To attempt greater precision is to become involved in subtle semantic problems about the meanings of terms.
At first sight many would tend to restrict environmental effects to those operating after the birth of the organism, but to do so would be to exclude prenatal environmental effects that have been shown to be influential in later behavior. On the other hand, to broaden the concept of environment to include all influences after fertilization the point in time at which the genotype is fixed permits consideration of the reciprocal influence of parts of the genotype upon each other. Can environment include the rest of the genotype, other than that part which is more or less directly concerned with the phenotype under consideration? This point assumes some importance since there are characteristics, not behavioralat least, none that are behavioral have so far been reported whose expression depends on the nature of the other genes present in the organism. In the absence of some of them, or rather certain alleles of the gene pairs, the value phenotypically observed would be different from what it would be if they were present. That is, different components of the genotype, in interplay with one another, modify phenotypic expression of the characteristic they in fluence. Can such indirect action, which recalls that of a chemical catalyst, best be considered as environmental or innate? It would be preferable to many to regard this mechanism as a genetic effect rather than an environmental one in the usually accepted sense. Hence, the definition of the area of study as one involving the interaction of heredity and environment, while apparently adding complexity, in fact serves to reduce confusion.
It must be conceded that this view has not as yet gained general acceptance. In some of the work reviewed in the necessarily brief survey of the major findings in this area, attempts have been made to retain a rather rigid dichotomy between heredity and environmentnature versus nurture in fact, an either/or proposition that the facts do not warrant. The excesses of both sides in the controversies of the 1920sfor example, the famous debate between Watson and McDougall over the relative importance of learned (environmental) and instinctive (genetic) determinants of behavior show the fallacies that extreme protagonists on either side can entertain if the importance of the interaction effect is ignored.
Gene action. The nature of gene action as such is essentially conducive to interaction with the environment, since the behavioral phenotype we observe is the end product of a long chain of action, principally biochemical, originating in the chromosome within the individual cell. A chromosome has a complex structure, involving DNA (deoxyribonucleic acid) and the connections of DNA with various proteins, and may be influenced in turn by another nucleic acid, RNA (ribonucleic acid), also within the cell but external to the nucleus. There are complex structures and sequences of processes, anatomical, physiological, and hormonal, which underlie normal development and differentiation of structure and function in the growth, development, and maturation of the organism. Much of this influence is determined genetically in the sense that the genotype of the organism, fixed at conception, determines how it proceeds under normal environmental circumstances. But it would be a mistake to regard any such sequence as rigid or immutable, as we shall see.
The state of affairs that arises when a number of genetically determined biochemical abnormalities affect behavior is illustrative of the argument. Many of these biochemical deficiencies or inborn errors of metabolism in humans are the outcome of a chain of causation starting with genie structures, some of them having known chromosomal locations. Their effects on the total personalitythat is, the sum total of behavorial variation that makes the individual uniquecan range from the trivial to the intense. The facility with which people can taste a solution of phenylthiocarbamide (PTC), a synthetic substance not found in nature, varies in a relatively simple genetical way: people are either tasters or nontasters in certain rather well-defined proportions, with a pattern of inheritance determined probably by one gene of major effect. But being taste blind or not is a relatively unimportant piece of behavior, since one is never likely to encounter it outside a genetical experiment. (It should perhaps be added that there is some evidence that the ability to taste PTC may be linked with other characteristics of some importance, such as susceptibility to thyroid disease.) Nevertheless, this example is insignificant compared with the psychological effect of the absence of a biochemical link in patients suffering from phenylketonuria. They are unable to metabolize phenylalanine to tyrosine in the liver, with the result that the phenylalanine accumulates and the patient suffers multiple defects, among which is usually gross intellectual defect, with an IQ typically on the order of 30. This gross biochemical failure is mediated by a single recessive gene that may be passed on in a family unnoticed in heterozygoussingle doseform but becomes painfully apparent in the unfortunate individual who happens to receive a double dose and consequently is homozygous for the defect.
Alternatively, a normal dominant gene may mutate to the recessive form and so give rise to the trouble. While mutation is a relatively rare event individually, the number of genes in each individualprobably on the order of ten thousand and the number of individuals in a population make it statistically a factor to be reckoned with. One of the best documented cases of a deleterious mutation of this kind giving rise to a major defect relates to the hemophilia transmitted, with certain important political consequences, to some of the descendants of Queen Victoria of England. The dependence of the last tsarina of Russia on the monk Rasputin was said to be based in part on the beneficial therapeutic effect of his hypnotic techniques on the uncontrollable bleeding of the Tsarevitch Alexis. Victoria was almost certainly heterozygous for hemophilia and, in view of the absence of any previous record of the defect in the Hanoverian dynasty, it seems likely that the origin of the trouble was a mutation in one of the germ cells in a testicle of Victorias father, the duke of Kent, before Victoria was conceived in August 1818.
But however it comes about, a defect such as phenylketonuria can be crippling. Fortunately, its presence can be diagnosed in very early life by a simple urine test for phenyl derivatives. The dependence of the expression of the genetic defect on the environmental circumstances is such that its effect can be mitigated by feeding the afflicted infant with a specially composed diet low in the phenylalanine with which the patients biochemical make-up cannot cope. Here again, therefore, one sees the interaction of genotype and environment in this case the type of food eaten. Many of the human biochemical defects that have been brought to light in recent years are rather simply determined genetically, in contrast with the prevailing beliefs about the bases of many behavioral characteristics including intelligence, personality, and most psychotic and neurotic disorders. This is also true of several chromosomal aberrations that have been much studied recently and that are now known to be implicated in various conditions of profound behavioral importance. Prominent among these is Downs syndrome (mongolism) with, again, effects including impairment of cognitive power. [SeeIntelligence and Intelligence Testing; Mental Disorders, articles onBiological AspectsandGenetic Aspects.]
Sex as a genetic characteristic. The sex difference is perhaps the most striking genetically determined difference in behavior and the one that is most often ignored in this connection. Primary sex is completely determined genetically at the moment of fertilization of the ovum; in mammals sex depends on whether the spermatozoon effecting fertilization bears an X or a Y chromosome to combine with the X chromosome inevitably contributed by the ovum. The resulting gamete then has the form of an XX (female) or an XY (male) individual. This difference penetrates every cell of every tissue of the resulting individual and in turn is responsible for the observable gross differences in morphology. These, in turn, subserve differences of physiological function, metabolism, and endocrine function which profoundly influence not only those aspects of behavior relating to sexual behav ior and reproductive function in the two sexes but many other aspects as well. But behavior is also influenced by social and cultural pressures, so that the resulting sex differences in behavior as observed by the psychologist are especially good examples of a phenotype that must be the and product of both genetic and environmental forces. There is a large literature on sex differences in human behavior and a sizable one on such differ ences in animal behavior, but there has been little attempt to assess this pervasive variation in terms of the relative contribution of genetic and environmental determinants. This is partly because of the technical difficulties of the problem, in the sense that all subjects must be of either one sex or the othercrossing males with females will always result in the same groups as those one started with, either males or femalesthere being, in general, no genetically intermediate sex against which to evaluate either and identical twins being inevitably of like sex. It is also partly because the potential of genetic analyses that do not involve direct experi mentation has not been realized. This is especially so since the causal routes whereby genetic determinants of sex influence many of the behavioral phenotypes observed are often better understood than in other cases where the genetic determinants underlying individual differences manifest in a population are not so clear-cut. [SeeIndividualDifferences, article onSex Differences.]
Sex linkage. There is one exception to the general lack of interest in the biometrical analysis of sex differences having behavioral connotations: sex-linked conditions. That is to say, it is demonstrated or postulated that the gene or genes responsible for the behavioroften a defect, as in the case of color blindness, which has a significantly greater incidence in males than in femalesare linked with the sex difference by virtue of their location on the sex chromosome determining genetic sex. Thus it is that sex can be thought of as a chromosomal difference of regular occurrence, as opposed to aberrations of the sort which give rise to pathological conditions, such as Downs syndrome. Indeed there are also various anomalies of genetic sex that give rise to problems of sexual identity, in which the psychological and overt be havioral consequences can be of major importance for the individual. While the evidence in such cases of environmental modification of the causative genetic conditions is less dramatic than in phenylketonuria, interaction undoubtedly exists, since these chromosomal defects of sex differentiation can in some cases be alleviated by appropriate surgical and hormonal treatment. [SeeSexual BEHavior, article onSexual Deviation: Psychological Aspects; andVision, article onColor Vision and Color Blindness.]
Human psychogenetics. It is abundantly clear that most of the phenotypes the behavioral scientist is interested in are multidetermined, both environmentally and genetically. The previous examples, however, are the exception rather than the rule, and their prominence bears witness that our understanding of genetics and behavior is as yet so little advanced that the simpler modes of genetic expression have been the first to be explored. In genetics itself, the striking differences in seed configuration used by Mendel in his classic crosses of sweet peas are determined by major genes with full dominance acting simply. But such clear-cut expression, especially of dominance, is unusual in human psychogenetics, and more complex statistical techniques are necessary to evaluate multiple genetic and environmental effects acting to produce the observed phenotype.
Whatever the analysis applied to the data gathered in other fields, in human psychogenetics the method employed cannot be the straightforward Mendelian one of crossbreeding which, in various elaborations, remains the basic tool of the geneticist today. Neither can it be the method of selection artificial, as opposed to naturalthat is other wise known as selective breeding. Indeed, none of the experimental techniques that can be applied to any other organism, whatever the phenotype being measured, is applicable to man, since experimental mating is effectively ruled out as a permissible technique in current cultures. It may be remarked in passing that such has not always been the case. The experiment of the Mogul emperor, Akbar, who reared children in isolation to determine their natural religion (and merely produced mutes) and the eugenics program of J. H. Noyes at the Oneida Community in New York State in the nineteenth century are cases in point. The apparent inbreeding of brother with sister among the rulers of ancient Egypt in the eighteenth dynasty (sixteenth to fourteenth century B.C.), which is often quoted as an example of the absence in humans of the deleterious effects of inbreeding (inbreeding depression), may not be all it seems. It is likely that the definition of sister and brother in this context did not necessarily have the same biological relevance that it has today but was rather a cultural role that could be defined, at least in this case, at will.
Twin study. In the absence of the possibility of an experimental approach, contemporary re search in human psychogenetics must rely on natural genetic experiments. Of these, the one most widely used and most industriously studied is the phenomenon of human twinning. Credit for the recognition of the value of observations on twins can be given to the nineteenth-century English scientist entist Francis Galton, who pioneered many fields of inquiry. He may be justly regarded as the father of psychogenetics for the practical methods he introduced into this field, such as the method of twin study, as well as for his influence which extended, although indirectly, even to the American experimenters in psychogenetics during the early decades of the present century.
Twin births are relatively rare in humans and vary in frequency with the ethnic group. However, the extent to which such ethnic groups differ among themselves behaviorally as a result of the undoubted genetic differences, of which incidence of multiple births is but one example, is controversial. As is well known, there are two types of twins: the monozygotic or so-called identical twins, derived from a single fertilized ovum that has split into two at an early stage in development, and the dizygotic or so-called fraternal twins, developed from two separate ova fertilized by different spermatozoa. These two physical types are not always easy to differentiate, although this difficulty is relatively miner in twin study. Nonetheless, they have led to two kinds of investigation. The first relates to differences in monozygotic twins who have identical hereditary make-up but who have been reared apart and thus subjected to different environmental influences during childhood; and the second relates to the comparison of the two types of twins usually restricted to like-sex pairs, since fraternal twins can differ in sex. The latter method supposes all differences between monozygotic pairs to be due to environmental origin, whereas the (greater) difference between dizygotic pairs is of environmental plus genetic origin. Thus, the relative contribution of the two sources of variation can be evaluated.
Findings obtained from either method have not been especially clear-cut, both because of intractable problems regarding the relative weight to be placed upon differences in the environment in which the twins have been reared and because of the sampling difficulties, which are likely to be formidable in any twin study. Nevertheless, interesting inferences can be drawn from twin study. The investigation of separated monozygotic twins has shown that while even with their identical heredity they can differ quite widely, there exists a significant resemblance in basic aspects of personality including intelligence, introversion, and neurotic tendencies, and that these resemblances can persist despite widely different environments in which the members of a pair are reared. Such findings emphasize the need to consider the contribution of genotype and environment in an inter active senseclearly some genotypes represented in the personality of monozygotic twin pairs are sensitive to environmentally induced variation, whereas others are resistant to it.
Comparisons between monozygotic and dizygotic twins reared together suggest that monozygotic twins more closely resemble each other in many aspects of personality, especially those defining psychological factors such as neuroticism and introversion-extroversion. The increase in the differences between the two types of twins when factor measures are usedas opposed to simple test scoressuggests that a more basic biological stratum is tapped by factor techniques, since the genetic determination seems greater than where individual tests are employed. Here again, the de gree to which any phenotype is shown to be hereditary in origin is valid only for the environment in which it developed and is measured; different environments may well yield different results. The problems of environmental control in human samples are so intractable that some students of the subject have questioned whether the effort and undoubted skill devoted to twin study have been well invested, in view of the inherent and persisting equivocality of the outcome.
Multivariate methods. Methods of twin study, introduced largely to improve upon the earlier methods of familial correlation (parents with off spring, sib with sib, etc.), have been combined with them. Familial correlation methods them selves have not been dealt with here, since within-family environments are bound to be even greater contaminants in determining the observed behavior than environments in twin study methods. Never theless, used on a large scale and in conjunction with twin study and with control subjects selected at random from a population, multivariate methods show promise for defining the limits of environmental and genotypic interaction. So far, the solutions to the problems of biometrical analysis posed by this type of investigation have been only partial, and the sheer weight of effort involved in locating and testing the requisite numbers of subjects standing in the required relationships has deterred all but a few pioneers. Despite the undoubtedly useful part such investigations have played in defining the problems involved, the absence of the possibility of experimental breeding has proved a drawback in the provision of socially useful data.
Animal psychogenetics. Recourse has often been had to nonhuman subjects. The additional problem thereby incurred of the relevance of comparative data to human behavior is probably balanced by the double refinements of the control of both the heredity and the environment of the experimental subjects. Two major methods of genetics have been employed, both intended to produce subjects of predetermined genotype: the crossbreeding of strains of animals of known genotype; and phenotypic selection, the mating of like with like to increase a given characteristic in a population.
Selection. Behavioral phenotypes of interest have been studied by the above methods, often using laboratory rodents. For example, attributes such as intelligence, activity, speed of conditioning, and emotionality have been selectively bred in rats.
Selection for emotional behavior in the rat will serve as an example of the techniques used and the results achieved. Rats, in common with many other mammalian species, defecate when afraid. A technique of measuring individual differences in emotional arousal is based on this propensity. The animal under test is exposed to mildly stressful noise and light stimulation in an open field or arena. The number of fecal pellets deposited serves as an index of disturbance, and in this way the extremes among a large group of rats can be characterized as high or low in emotional responsiveness. Continued selection from within the high and low groups will in time produce two distinct strains. Control of environmental variables is achieved by a rigid standardization of the conditions under which the animals are reared before being subjected to the test as adults. Careful checks on maternal effects, both prenatal and postnatal, reveal these effects to be minimal.
Such an experiment does little beyond establishing the importance of the genetic effect on the given strains in the given environment. While there are techniques for assessing the relative importance of the genetic and environmental contributions to the variation observed under selection, they are better suited to the analysis of the outcome of experiments using the alternative major genetical method, that of crossbreeding of inbred strains.
Crossbreeding. Strains used in crossbreeding experiments have usually been inbred for a phenotypic character of interest, although not usually a behavioral one. However, this does not preclude the use of these inbred strains for behavioral studies, since linkage relationships among genes ensure that selection for factors multidetermined genetically often involves multiple changes in characteristics other than those selected for, and behavior is no exception to this rule. Moreover, the existence of such inbred strains constitutes perhaps the most important single advantage of animals as subjects, since it enables simplifying assumptions regarding the homozygosity or genetic uniformity of such strains to be made in analysis of the outcome of crosses. Members of inbred strains are theoretically as alike as monozygotic twin pairs, so that genetic relationshipswhich in human populations can be investigated only after widespread efforts to find themcan be multiplied at will in laboratory animals.
This approach allows a more sensitive analysis of the determinants, both environmental and genetic, of the behavioral phenotype under observation. In addition, the nature of the genetic forces can be further differentiated into considerations of the average dominance effects of the genes in volved, the extent to which they tend to increase or decrease the metrical expression of the behavioral phenotype, and the extent to which the different strains involved possess such increasers or de creases. Finally, rough estimates of the number of these genes can be given. But the analysis depends upon meeting requirements regarding the scaling of the metric upon which the behavior is measured and is essentially a statistical one. That is, only average effects of cumulative action of the relatively large number of genes postulated as in volved can be detected. Gone are the elegantly simple statistics derived from the classical Men-delian analyses of genes of major effect, often displaying dominance, like those encountered incertain human inborn errors of metabolism. There is little evidence of the existence of comparable genes of major effect mediating behavior in laboratory animals, although some have been studied in in sects, especially the fruit fly.
A typical investigation of a behavioral phenotype might take the form of identifying two inbred strains known to differ in a behavioral trait, measuring individuals from these strains, and then systematically crossing them and measuring all offspring. When this was done for the runway performance of mice, an attribute related to their temperamental wildness, the results, analyzed by the techniques of biometrical genetics, showed that the behavior was controlled by at least three groups of genes (a probable underestimate). The contributions of these groups were additive to each other and independent of the environment when measured on a logarithmic scale but interacted with each other and with the environment on a linear scale. These genes showed a significant average dominance effect, and there was a preponderance of dominant genes in the direction of greater wildness. The heritability ratio of the contributions of nature and nurture was around seven to three.
The use of inbred lines may be restricted to first filial crosses if a number of such crosses are made from several different lines. This increases precision of analysis in addition to allowing a proportionate decrease in the amount of laboratory work. One investigation examined the exploratory behavior of six different strains of rats in an open field of the kind used for the selection mentioned above. On a linear scale there were no untoward environmental effects, including specifically prenatal maternal ones. The heritability ratio was high, around nine to one; and while there was a significant average dominance component among the genes determining exploration, there was no preponderance of dominants or recessively acting genes among increasers or decreasers. The relative standing in this respect of the parental strains could be established with some precision.
Limitations. While the methods described above have allowed the emergence of results that ultimately may assist our understanding of the mechanisms of behavioral inheritance, it cannot be said that much substantial progress has yet been made. Until experiments explore the effect of a range of different genotypes interacting with a range of environments of psychological interest and consequence, little more can be expected. Manipulating heredity in a single standard environment or manipulating the environment of a single standard genotype can only provide conclusions so limited to both the genotypes and conditions employed that they have little usefulness in a wider context. When better experiments are performed, as seems likely in the next few decades, then problems of some sociological importance and interest will arise in the application of these experiments to the tasks of maximizing genetic potential and perfecting environmental control for the purpose of so doing. A new eugenics may well develop, but grappling with the problems of its impact on contemporary society had best be left to future generations.
P. L. Broadhurst
[Directly related are the entriesEugenics; Evolution; Mental Disorders, article onGenetic Aspects. Other relevant material may be found inIndividual Differences, article onSex Differences; Instinct; Intelligence and Intelligence Testing; Mental Ertardation; Psychology, article onConstitutional Psychology.]
Broadhurst, P. L. 1960 Experiments in Psychogenetics: Applications of Biometrical Genetics to the Inheritance of Behavior. Pages 1-102 in Hans J. Eysenck (editor), Experiments in Personality. Volume 1: Psychogenetics and Psychopharmacology. London: Routledge. Selection and crossbreeding methods applied to laboratory rats.
Catteix, RaymondB.; Stice, GlenF.; and Kristy, Nor TonF. 1957 A First Approximation to Nature-Nurture Ratios for Eleven Primary Personality Factors in Objective Tests. Journal of Abnormal and Social Psychology 54:143159. Pioneer multivariate analysis combining twin study and familial correlations.
Fuller, JohnL.; and Thompson, W. Robert 1960 Be havior Genetics. New York: Wiley. A comprehen sive review of the field.
Mather, Kenneth1949 Biometrical Genetics: The Study of Continuous Variation. New York: Dover. The classic work on the analysis of quantitative char acteristics.
Shields, James1962 Monozygotic Twins Brought Up Apart and Brought Up Together: An Investigation Into the Genetic and Environmental Causes of Variation in Personality. Oxford Univ. Press.
The best available definition of population genetics is doubtless that of Malcot: It is the totality of mathematical models that can be constructed to represent the evolution of the structure of a population classified according to the distribution of its Mendelian genes (1955, p. 240). This definition, by a probabilist mathematician, gives a correct idea of the constructed and abstract side of this branch of genetics; it also makes intelligible the rapid development of population genetics since the advent of Mendelism.
In its formal aspect this branch of genetics might even seem to be a science that is almost played out. Indeed, it is not unthinkable that mathematicians have exhausted all the structural possibilities for building models, both within the context of general genetics and within that of the hypothesesmore or less complex and abstractthat enable us to characterize the state of a population.
Two major categories of models can be distinguished: determinist models are those in which variations in population composition over time are rigorously determined by (a) a known initial state of the population; (b) a known number of forces or pressures operating, in the course of generations, in an unambiguously defined fashion (Male-cot 1955, p. 240). These pressures involve mutation, selection, and preferential marriages (by consanguinity, for instance). Determinist models, based on ratios that have been exactly ascertained from preceding phenomena, can be expressed only in terms of populations that are infinite in the mathematical sense. In fact, it is only in this type of population that statistical regularities can emerge (Malecot 1955). In these models the composition of each generation is perfectly defined by the composition of the preceding generation.
Stochastic models, in contrast to determinist ones, involve only finite populations, in which the gametes that, beginning with the first generation, are actually going to give birth to the new generation represent only a finite number among all possible gametes. The result is that among these active, or useful, gametes (Malecot 1959), male or female, the actual frequency of a gene will differ from the probability that each gamete had of carrying it at the outset.
The effect of chance will play a prime role, and the frequencies of the genes will be able to drift from one generation to the other. The effects of random drift and of genetic drift become, under these conditions, the focal points for research.
The body of research completed on these assumptions does indeed form a coherent whole, but these results, in spite of their brilliance, are marked by a very noticeable formalism. In reality, the models, although of great importance at the conceptual level, are often too far removed from the facts. In the study of man, particularly, the problems posed are often too complex for the solutions taken directly from the models to describe concrete reality.
Not all these models, however, are the result of purely abstract speculation; construction of some of them has been facilitated by experimental data. To illustrate this definition of population genetics and the problems that it raises, this article will limit itself to explaining one determinist model, both because it is one of the oldest and simplest to under stand and because it is one of those most often verified by observation.
A determinist model. Let us take the case of a particular human population: the inhabitants of an island cut off from outside contacts. It is obvious that great variability exists among the genes carried by the different inhabitants of this island. The genotypes differ materially from one another; in other words, there is a certain polymorphism in the populationpolymorphism that we can define in genetic terms with the help of a simple example.
Let us take the case of autosome (not connected with sex) gene a, transmitting itself in a mono-hybrid diallely. In relation to it individuals can be classified in three categories: homozygotes whose two alleles are a (a/a); heterozygotes, carriers of a and its allele a (a/a); and the homozygotes who are noncarriers of a (a/a). At any given moment or during any given generation, these three categories of individuals exist within the population in certain proportions relative to each other.
Now, according to Mendels second law (the law of segregation), the population born out of a cross between an individual who is homozygote for a (a/a) and an individual who is homozygote for a (a/a) will include individuals a/a, a/a, and a/a in the following proportions: one-fourth a/a, one-half a/a, and one-fourth a/a. In this popu lation the alleles a and a have the same frequency, one-half, and each sex produces half a and half a. If these individuals are mated randomly, a simple algebraic calculation quickly demonstrates that individuals of the generation following will be quan titatively distributed in the same fashion: one-fourth a/a, one-half a/a, and one-fourth a/a. It will be the same for succeeding generations.
It can therefore be stated that the genetic structure of such a population does not vary from one generation to the other. If we designate by p the initial proportion of a/a individuals and by q that of a/a individuals, we get p + q = 1, or the totality of the population. Applying this system of symbols to the preceding facts, it can be easily shown that the proportion of individuals of all three categories in the first generation born from a/a and a/a equals p2, 2pq, q2. In the second and third generation the frequency of individuals will always be similar: p2, 2pq, q2.
Until this point, we have remained at the individual level. If we proceed to that of the gametes carrying a or a and to that of genes a and a, we observe that their frequencies intermingle. In the type of population discussed above, the formula p2, 2pq, q2 still applies perfectly, therefore, to the gametes and genes. This model, which can be regarded as a formalization of the Hardy-Weinberg law, has other properties, but our study of it will stop here. (For a discussion of the study of isolated populations, see Sutter & Tabah 1951.)
Model construction and demographic reality. The Hardy-Weinberg law has been verified by numerous studies, involving both vegetable and animal species. The findings in the field of human blood groups have also been studied for a long time from a viewpoint derived implicitly from this law, especially in connection with their geographic distribution. Under the system of reproduction by sexes, a generation renews itself as a result of the encounter of the sexual cells (gametes) produced by individuals of both sexes belonging to the living generation. In the human species it can be said that this encounter takes place at random. One can imagine the advantage that formal population genetics can take of this circumstance, which can be compared to drawing marked balls by lot from two different urns. Model construction, already favored by these circumstances, is favored even further if the characteristics of the population utilized are artificially defined with the help of a certain number of hypotheses, of which the following is a summary description:
(1) Fertility is identical for all couples; there is no differential fertility.
(2) The population is closed; it cannot, there fore, be the locus of migrations (whether immigration or emigration).
(3) Marriages take place at random; there is no assortative mating.
(4) There are no systematic preferential marriages (for instance, because of consanguinity).
(5) Possible mutations are not taken into consideration.
(6) The size of the population is clearly denned. On the basis of these working hypotheses, the whole of which constitutes panmixia, it was possible, not long after the rediscovery of Mendels laws, to construct the first mathematical models. Thus, population genetics took its first steps forward, one of which was undoubtedly the Hardy-Weinberg law.
Mere inspection of the preceding hypotheses will enable the reader to judge how, taken one by one, they conflict with reality. In fact, no human population can be panmictic in the way the models are.
The following evidence can be cited in favor of this conclusion:
(1) Fertility is never the same with all couples. In fact, differential fertility is the rule in human populations. There is always a far from negligible sterility rate of about 18 per cent among the large populations of Western civilization. On the other hand, the part played by large families in keeping up the numbers of these populations is extremely important; we can therefore generalize by emphasizing that for one or another reason individ uals carrying a certain assortment of genes reproduce themselves more or less than the average number of couples. That is what makes for the fact that in each population there is always a certain degree of selection. Hypothesis (1) above, essential to the construction of models, is therefore very far removed from reality.
(2) Closed populations are extremely rare. Even among the most primitive peoples there is always a minimum of emigration or immigration. The only cases where one could hope to see this condition fulfilled at the present time would be those of island populations that have remained extremely primitive.
(3) With assortative mating we touch on a point that is still obscure; but even if these phe nomena remain poorly understood, it can nevertheless be said that they appear to be crucial in determining the genetic composition of populations. This choice can be positive: the carriers of a given characteristic marry among each other more often than chance would warrant. The fact was demonstrated in England by Pearson and Lee (1903): very tall individuals have a tendency to marry each other, and so do very short ones. Willoughby (1933) has reported on this question with respect to a great number of somatic characteristics other than heightfor example, coloring of hair, eyes and skin, intelligence quotient, and so forth. Inversely, negative choice makes individuals with the same characteristics avoid marrying one another. This mechanism is much less well known than the above. The example of persons of violent nature (Dahlberg 1943) and of red-headed individuals has been cited many times, although it has not been possible to establish valid statistics to support it.
(4) The case of preferential marriages is not at all negligible. There are still numerous areas where marriages between relatives (consanguineous marriages) occur much more frequently than they would as the result of simple random encounters. In addition, recent studies on the structures of kinship have shown that numerous populations that do not do so today used to practice preferential marriagemost often in a matrilinear sense. These social phenomena have a wide repercussion on the genetic structure of populations and are capable of modifying them considerably from one generation to the other.
(5) Although we do not know exactly what the real rates of mutation are, it can be admitted that their frequency is not negligible. If one or several genes mutate at a given moment in one or several individuals, the nature of the gene or genes is in this way modified; its stability in the population undergoes a disturbance that can considerably transform the composition of that population.
(6) The size of the population and its limits have to be taken into account. We have seen that this is one of the essential characteristics important in differentiating two large categories of models.
The above examination brings us into contact with the realities of population: fertility, fecundity, nuptiality, mortality, migration, and size are the elements that are the concern of demography and are studied not only by this science but also very often as part of administrative routine. Leaving aside the influence of size, which by definition is of prime importance in the technique of the models, there remain five factors to be examined from the demographic point of view. Mutation can be ruled out of consideration, because, although its importance is great, it is felt only after the passage of a certain number of generations. It can therefore be admitted that it is not of immediate interest.
We can also set aside choice of a mate, because the importance of this factor in practice is still unknown. Accordingly, there remain three factors of prime importance: fertility, migration, and preferential marriage. Over the last decade the progressive disappearance of consanguineous marriage has been noted everywhere but in Asia. In many civilized countries marriage between cousins has practically disappeared. It can be stated, therefore, that this factor has in recent years become considerably less important.
Migrations remain very important on the genetic level, but, unfortunately, precise demographic data about them are rare, and most of the data are of doubtful validity. For instance, it is hard to judge how their influence on a population of Western culture could be estimated.
The only remaining factor, fertility (which to geneticists seems essential), has fortunately been studied in satisfactory fashion by demographers. To show the importance of differential fertility in human populations, let us recall a well-known cal culation made by Karl Pearson in connection with Denmark. In 1830, 50 per cent of the children in that country were born of 25 per cent of the parents. If that fertility had been maintained at the same rate, 73 per cent of the second-generation Danes and 97 per cent of the third generation would have been descended from the first 25 per cent. Similarly, before World War I, Charles B. Davenport calculated, on the basis of differential fertility, that 1,000 Harvard graduates would have only 50 descendants after two centuries, while 1,000 Rumanian emigrants living in Boston would have become 100,000.
Human reproduction involves both fecundity (capacity for reproduction) and fertility (actual reproductive performance). These can be estimated for males, females, and married couples treated as a reproductive unit. Let us rapidly review the measurements that demography provides for geneticists in this domain.
Crude birth rate. The number of living births in a calendar year per thousand of the average population in the same year is known as the crude birth rate. The rate does not seem a very useful one for geneticists: there are too many different groups of childbearing age; marriage rates are too variable from one population to another; birth control is not uniformly diffused, and so forth.
General fertility rate. The ratio of the number of live births in one year to the average number of women capable of bearing children (usually defined as those women aged 15 to 49) is known as the general fertility rate. Its genetic usefulness is no greater than that of the preceding figure. Moreover, experience shows that this figure is not very different from the crude birth rate.
Age-specific fertility rates. Fertility rates according to the age reached by the mother during the year under consideration are known as age-specific fertility rates. Demographic experience shows that great differences are observed here, depending on whether or not the populations are Malthusianin other words, whether they practice birth control or not. In the case of a population where the fertility is natural, knowledge of the mothers age is sufficient. In cases where the population is Malthusian, the figure becomes interesting when it is calculated both by age and by age group of the mothers at time of marriage, thus combining the mothers age at the birth of her child and her age at marriage. This is generally known as the age-specific marital fertility rate. If we are dealing with a Malthusian population, it is preferable, in choosing the sample to be studied, to take into consideration the age at marriage rather than the age at the childs birth. Thus, while the age at birth is sufficient for natural populations, these techniques cannot be applied indiscriminately to all populations.
Family histories. Fertility rates can also be calculated on the basis of family histories, which can be reconstructed from such sources as parish registries (Fleury & Henry 1965) or, in some countries, from systematic family registrations (for instance, the Japanese koseki or honseki). The method for computing the fertility rate for, say, the 25-29-year-old age group from this kind of data is first to determine the number of legitimate births in the group. It is then necessary to make a rigorous count of the number of years lived in wedlock between their 25th and 30th birthdays by all the women in the group; this quantity is known as the groups total woman-years. The number of births is then divided by the number of woman-years to obtain the groups fertility rate. This method is very useful in the study of historical problems in genetics, since it is often the only one that can be applied to the available data.
Let us leavefer tility rates in order to examine rates of reproduction. Here we return to more purely genetic considerations, since we are looking for the mechanism whereby one generation is replaced by the one that follows it. Starting with a series of fertility rates by age groups, a gross reproduction rate can be calculated that gives the average number of female progeny that would be born to an age cohort of women, all of whom live through their entire reproductive period and continue to give birth at the rates prevalent when they themselves were born. The gross reproduction rate obtaining for a population at any one time can be derived by combining the rates for the different age cohorts.
A gross reproduction rate for a real generation can also be determined by calculating the average number of live female children ever born to women of fifty or over. As explained above, this rate is higher for non-Malthusian than for Malthusian populations and can be refined by taking into consideration the length of marriage.
We have seen that in order to be correct, it is necessary for the description of fertility in Malthusian populations to be closely related to the date of marriage. Actually, when a family reaches the size that the parents prefer, fertility tends to approach zero. The preferred size is evidently related to length of marriage in such a manner that fertility is more closely linked with length of marriage than with age at marriage. In recent years great progress has been made in the demographic analysis of fertility, based on this kind of data. This should en ablegeneticists to be more circumspect in their choice of sections of the population to be studied.
Americans talk of cohort analysis, the French of analysis by promotion (a term meaning year or class, as we might speak of the class of 1955). A cohort, or promotion, includes all women born within a 12-month period; to estimate fertility or mortality, it is supposed that these women are all born at the same moment on the first of January of that year. Thus, women born between January 1, 1900, and January 1, 1901, are considered to be exactly 15 years old on January 1, 1915; exactly 47 years old on January 1, 1947; and so forth.
The research done along these lines has issued in the construction of tables that are extremely useful in estimating fertility in a human population. As we have seen, it is more useful to draw up cohorts based on age at marriage than on age at birth. A fertility table set up in this way gives for each cohort the cumulative birth rate, by order of birth and single age of mother, for every woman surviving at each age, from 15 to 49. The progress that population genetics could make in knowing real genie frequencies can be imagined, if it could concentrate its research on any particular cohort and its descendants.
This rapid examination of the facts that demography can now provide in connection with fertility clearly reveals the variables that population genetics can use to make its models coincide with reality. The models retain their validity for genetics because they are still derived from basic genetic concepts; their application to actual problems, however, should be based on the kind of data mentioned above. We have voluntarily limited ourselves to the problem of fertility, since it is the most important factor in genetics research.
The close relationship between demography and population genetics that now appears can be illustrated by the field of research into blood groups. Although researchers concede that blood groups are independent of both age and sex, they do not explore the full consequences of this, since their measures are applied to samples of the population that are representative only in a demographic sense. We must deplore the fact that this method has spread to the other branches of genetics, since it is open to criticism not only from the demographic but from the genetic point of view. By proceeding in this way, a most important factor is overlookedthat of genie frequencies.
Let us admit that the choice of a blood group to be studied is of little impor tance when the characteristic is widely distributed throughout the populationfor instance, if each individual is the carrier of a gene taken into account in the system being studied (e.g., a system made up of groups A, B, and O). But this is no longer the case if the gene is carried only by a few individualsin other words, if its frequency attains 0.1 per cent or less. In this case (and cases like this are common in human genetics) the structure of the sample examined begins to take on prime importance.
A brief example must serve to illustrate this cardinal point. We have seen that in the case of rare recessive genes the importance of consan guineous marriages is considerable. The scarcer that carriers of recessive genes become in the pop ulation as a whole, the greater the proportion of such carriers produced by consanguineous marriages. Thus if as many as 25 per cent of all individuals in a population are carriers of recessive genes, and if one per cent of all marriages in that population are marriages between first cousins, then this one per cent of consanguineous unions will produce 1.12 times as many carriers of recessive genes as will be produced by all the unions of persons not so related. But if recessive genes are carried by only one per cent of the total population, then the same proportion of marriages between first cousins will produce 2.13 times as many carriers as will be produced by all other marriages. This production ratio increases to 4.9 if the total frequency of carriers is .01 per cent, to 20.2 if it is .005 per cent, and to 226 if it is .0001 per cent. Under these conditions, one can see the importance of the sampling method used to estimate the frequency within a population, not only of the individuals who are carriers but of the gametes and genes themselves.
Genealogical method. It should be emphasized that genetic studies based on genealogies remain the least controversial. Studying a population where the degrees of relationship connecting individuals are known presents an obvious interest. Knowing one or several characteristics of certain parents, we can follow what becomes of these in the descendants. Their evolution can also be considered from the point of view of such properties of genes as dominance, recessiveness, expressivity, and penetrance. But above all, we can follow the evolution of these characteristics in the population over time and thus observe the effects of differential fertility. Until now the genealogical method was applicable only to a numerically sparse population, but progress in electronic methods of data processing permits us to anticipate its application to much larger populations (Sutter & Tabah 1956).
Dynamic studies. In very large modern populations it would appear that internal analysis of cohorts and their descendants will bring in the future a large measure of certainty to research in population genetics. In any case, it is a sure way to a dynamic genetics based on demographic reality. For instance, it has been recommended that blood groups should be studied according to age groups; but if we proceed to do so without regard for demographic factors, we cannot make our observations dynamic. Thus, a study that limits itself to, let us say, the fifty- to sixty-year-old age group will have to deal with a universe that includes certain genetically dead elements, such as unmarried and sterile persons, which have no meaning from the dynamic point of view. But if a study is made of this same fifty- to sixty-year-old age group and then of the twenty- to thirty-year-old age group, and if in the older group only those individuals are considered who have descendants in the younger group, the dynamic potential of the data is maximized. It is quite possible to subject demographic cohorts to this sort of interpretation, because in many countries demographic statistics supply series of individuals classified according to the mothers age at their birth.
This discussion would not be complete if we did not stress another aspect of the genetic importance of certain demographic factors, revealed by modern techniques, which have truly created a demographic biology. Particularly worthy of note are the mothers age, order of birth, spacing between births, and size of family.
The mothers age is a great influence on fecundity. A certain number of couples become in capable of having a second child after the birth of the first child; a third child after the second; a fourth after the third; and so forth. This sterility increases with the length of a marriage and especially after the age of 35. It is very important to realize this when, for instance, natural selection and its effects are being studied.
The mothers age also strongly influences the frequency of twin births (monozygotic or dizygotic), spontaneous abortions, stillborn or abnormal births, and so on. Many examples can also be given of the influence of the order of birth, the interval between births, and the size of the family to illustrate their effect on such things as fertility, mortality, morbidity, and malformations.
It has been demonstrated above how seriously demographic factors must be taken into consideration when we wish to study the influence of the genetic structure of populations. We will leave aside the possible environmental influences, such as social class and marital status, since they have previously been codified by Osborn (1956/1957) and Larsson (19561957), among others. At the practical level, however, the continuing efforts to utilize vital statistics for genetic purposes should be pointed out. In this connection, the research of H. B. Newcombe and his colleagues (1965), who are attempting to organize Canadian national statistics for use in genetics, cannot be too highly praised. The United Nations itself posed the problem on the world level at a seminar organized in Geneva in 1960. The question of the relation between demography and genetics is therefore being posed in an acute form.
These problems also impinge in an important way on more general philosophical issues, as has been demonstrated by Haldane (1932), Fisher (1930), and Wright (1951). It must be recognized, however, that their form of Neo-Darwinism, although it is based on Mendelian genetics, too often neglects demographic considerations. In the future these seminal developments should be renewed in full confrontation with demographic reality.
Jean Sutter
[Directly related are the entriesCohort Analysis; Fertility; Fertility Control. Other relevant ma terial may be found inNuptiality; Race; SocialBehavior, Animal, article onThe Regulation of Animal Populations.]
Barclay, George W. 1958 Techniques of Population Analysis. New York: Wiley.
Dahlberg, Gunnar(1943)1948 Mathematical Methods for Population Genetics. New York and London: Inter-science. First published in German.
Dunn, Leslie C. (editor) 1951 Genetics in the Twentieth Century: Essays on the Progress of Genetics During Its First Fifty Years. New York: Macmillan.
Fisher, R. A. (1930) 1958 The Genetical Theory of Natural Selection. 2d ed., rev. New York: Dover.
See original here:
genetics facts, information, pictures | Encyclopedia.com ...
