Environmental Variability Selects for Large Families only in Special Circumstances:
Another Objection to Differential K Theory

Appeared: Personality and Individual Differences, Vol. 19 (December 1995), No. 6, pp. 903-918.

Abstract

Rushton has argued that less predictable climates should select for r traits. However, because selection calls for maximizing the geometric mean rather than the arithmetic mean, and having small families could contribute to survival in adverse times, unpredictability should select for K characteristics. An alternative in which excess births provide an option on larger families is considered and rejected. Biological evidence from the Great Tit and other birds is used for illustration. This effect may explain why humans have such low conception rates. Several additional pieces of evidence for Miller's (1994, this journal) differential paternal investment theory are offered. Evidence for seasonality in prehistoric death rates is offered. Comparisons of fixation coefficients for Y chromosome genes versus mitochondrial ones suggest that polygyny has been more common in Africa than in Europe since prehistoric times. Additional evidence for racial differences in motivations for marriage and child bearing is offered. Higher black expenditures on clothing is explained. In cold climates the greater importance of detecting a mate's deception is argued to select for higher intelligence.

Contents

The Geometric Mean versus the Arithmetic Mean 2
More Offspring versus More Investment Per Offspring 3
An Example from Nature
The Great Tit 3
Numerous Births as an Option Permitting Large Families in Good times 4
Disease 6
Relevance to Other Species 6
Other Applications 7
Caucasoids as a Mixture of Mongoloids and Negroids 9
New Evidence 9
New Research on Intelligence 11
Conclusions 13
References 13

Rushton (1985, 1987, 1988, 1995) has argued that Negroids (i.e. Negroes) were more r selected than Caucasians and Mongoloids. This idea has produced considerable scientific (Flynn 1989; Leslie, 1990; Lynn, 1989; Roberts & Gabor, 1990; Silverman, 1990, Mealey, 1990) and popular controversy (Gross, 1990; Pearson, 1991, Chap. 5), which Rushton (1989a, 1990, 1991) has responded to.
In an earlier paper (Miller, 1993), to which Rushton & Ankney (1993) replied, the author presented several objections to Rushton's differential K theory. The primary objections were that Africa was no more variable than other continents, and questions as to whether the African rainfall variability really would select for the traits Rushton reported Africans to have. Miller (1994b) later presented an alternative theory (differential paternal investment) to explain most of the effects Rushton described.

There is not space here to do more than briefly outline Rushton's original theory. For details the reader should see his recent book (1995), or earlier articles in this journal (1985, 1988). In essence Rushton argued that where the environment was most unpredictable selection would be for those characteristics that contributed to rapid population growth in a very favorable environment (r is the rate of growth of the population in such an environment). A short period between births, and a large number of offspring per birth (in humans the twinning rate) are examples of such traits. In conditions where the population was close to the environment's carrying capacity (which is K in the theory), selection was for traits which contributed to competition with conspecifics including high paternal investment, intelligence, long intervals between birth, the ability to cooperate, etc. The argument was impressive in the number of traits that could be explained. The essence of r/ K theory in biology is that the organism has only a certain quantity of resources and the organism has a trade off between a large number of offspring, each of whom receives little resources (or parental investment), or fewer offspring with more investment in each. The same tradeoff occurs in humans as occurs between species of animals or plants. It will be argued here that the tradeoff is determined by how many offspring the organism has to divide its resources among (family size in humans), and that a more variable environment actually selects for a higher parental investment per capita, and for smaller families.

The biologist's r/K selection is a generalization that in certain conditions species or sub-species tend to evolve certain traits. One should not be surprised to find that the generalizations do not apply to certain species (say humans) or to certain varieties within the species, or if the traits are observed in a certain species that the evolutionary conditions that actually led to them are indeed the conditions discussed in r/K theory. Saying that certain humans (say Negroids) displays traits that could have been produced by r selection does not necessarily imply that the traits in humans were produced by the selective forces that often lead to r traits in animals.

One issued raised by Rushton and Ankney (1993) should be discussed. They argued I had confused unpredictability and variability, clearly preferring unpredictability. I must confess I fail to see the distinction between the two terms, regarding them as essentially synonyms. My usage follows that of financial economics where variability is a quantitative measure of the width of a probability distribution (more formally variance) and hence implies unpredictability. Optimally an organism adjusts its behavior to predictable changes in the environment using whatever information it can find in the environment. The variability I am discussing is the variability left after this optimal adjustment, which appears to be the unpredictability Rushton is discussing.

The only advantage to the usage "unpredictable" versus "predictable", is that intelligence may assist in recognizing patterns and selecting the optimal behavior. Thus predictability may select for intelligence, while unpredictability does not. A classic example is seasonal fluctuations, where a more intelligent human may be better able to discern the seasonal pattern and adjust behavior to it. Thus it could be argued that predictable variations select for intelligence, Rushton may be doing so. Certainly, Miller (1991) has argued that the need to plan for storage of food to survive the winter has selected for intelligence, and for the ability to defer gratification.

It may be noted that unpredictability may select for intelligence. This can happen if the less intelligent use simple rules of thumb such as planning for last year's conditions, or for average conditions, while the more intelligent devise strategies that involve being prepared for rare events, or for rare opportunities. For instance, the more intelligent may know that stream flow in the desert riverbeds is unpredictable, and avoid camping there because of the risk of a flash flood. The less intelligent merely see that it is currently dry, that it is usually dry, and that the riverbed is flat ground near to water, and camp there. If this behavior causes the less intelligent to occasionally drown, selection will be for intelligence. Many other examples could be provided where unpredictability selects for intelligence, and against stupidity.

When discussing the value of an option, variability is the traditional term (measured technically by the variance). This makes it more suitable for discussing behavior that amounts logically to the purchase of options.

In the course of my original critique, it was pointed out that conditions suitable for fast reproduction (abundant food and absence of competition) would indeed select for r characteristics. However, the very nature of a fluctuating environment is that the good times are followed by bad times. In bad times, selection favors K characteristics. The conclusion was that one could not be certain what the net effect was. An argument has now been discovered for specifying what the net effect is. It will be argued that a fluctuating environment (interpreted as one that fluctuates in its carrying capacity) will select against at least one very important r characteristic, large family size, and for one important K characteristic, high parental investment.

The Geometric Mean versus the Arithmetic Mean

The genotype that comes to dominate a population will be the one with the highest growth rate over a long period of time. Let us simplify by imagining all generations to be of equal length. Also the complications arising from sexual reproduction will be ignored. (The reader may imagine that individuals of each genotype mate only with other individuals of the same type). Let gi be the number of descendants each individual leaves in the next generation, with i as the number of the generation. After a large number of generations, n, the number of individuals of a genotype alive is g1g2g3 . . . gn or Pgi. Thus, the genotype with the highest Pgi will eventually become the most frequent. The genotype that wins out is the genotype that achieves the highest "average" growth rate. However, the "average" growth rate is a special average, the geometric average, not the arithmetic average. The geometric average is defined as (Pgi)1/n, where the average is taken over n generations. Notice this differs from the arithmetic average taught in schools which is (Sgi)/n.

As an aside it should be noted that the belief in "survival of the fittest" is incorrect if fittest is interpreted as being the species or genotype with the largest expected number of offspring in the next generation, since the expected value is an arithmetic average

Why all the attention to the different types of averages? The reason is that unless rates of growth over all generations are equal, the geometric average will be less than the arithmetic average. This fact is well known to financial economists (such as the author) who are concerned with the growth of a sum of money over a number of years.

A useful approximation (Markowitz, 1959; Gillespie, 1977) is:

geometric mean = arithmetic mean - variance/2

where both means refer to the mean number of offspring per generation. Markowitz (1959, pp. 121-125) discusses this approximation, showing that from -.3 to .3 the approximation is good. Of course, some extreme cases where lines almost go extinct could fall outside of this range.

The key question in determining whether a particular reproductive strategy will be more strongly selected for by environmental predictability is how the different strategies affect the generation to generation variability in growth rate.

In particular, a strategy that has a very adverse effect on the size of the next generation during environmentally unfavorable times loses out to a more conservative strategy that retains a breeding population even during adverse times.

The reader may wish to think about this in the context of the source of the environmental variability Rushton apparently had in mind. These are the periodic droughts that hit the Sahel in Africa, the part of Africa where the evolutionary adaptation of Negroids apparently occurred. The question is how a particular strategy affects population growth over a generation that includes a major drought. Since in a drought, the population growth rate is negative, the question is better expressed as how do different strategies affect the fraction of the population that survives a major drought?

The extreme case of low population growth is -100%. This, of course, implies extinction. In the long run any strategy that involves an appreciable chance of extinction will lose out in competition, even if it results in a large number of surviving offspring in average conditions, or even the vast majority of conditions.

Lewontin & Cohen (1969) have made the point that a random series representing reproductive success can have a high, or even infinite expected value, but yet the population can have a very high probability of extinction. They consider the case where most years are good, but a few are bad. In their example, a species might have in 9 out of 10 years a growth to 1.1 times the initial population, and in the remaining year a growth to .3 times the initial population. This causes the expectation of population size to grow at 2 percent per year, while the geometric mean would be only .841, making extinction very likely. The argument is even stronger when it is realized that carrying capacity places an upper limit on population size. Once this upper limit is reached, further increase is impossible, while a series of poor years can lead to extinction. The evolutionary success of the species may be more sensitive to how it does in the occasional poor years than in the more common good years. The large family strategy, which implies dividing the parental investment among many offspring, succeeds in good periods, but fails in the occasional poor periods.

More Offspring versus More Investment Per Offspring

The most important distinction between r and K selected organisms is the number of offspring, and the resulting difference in per capita investment. By definition, r selected organisms produce more offspring than K selected organisms. The energy that does not go into producing offspring goes into increasing per capita investment. The practical trade off in humans might be between having four offspring among which are divided the parent's food and time, or having five offspring, with each receiving less food and time.

In good times, high birth rates will indeed result in more offspring surviving to the next generation. However, in adverse periods, such as drought, a large family's survival is poor.

In the early stages of the drought, the per capita allocation might easily drop below that needed to sustain weight. All of the children would then fall into the badly malnourished category. When the drought worsened, or even continued longer than expected, the badly nourished would then perish, ending the line.

This argument suggests that environmental fluctuations should reduce r selection, not increase it.

Let us look at it from the viewpoint of differing degrees of K selection. Remember, K characteristics enable winning competition for resources when the population density is high relative to resources. In the original explications (MacArthur & Wilson, 1967; Pianka, 1970), this occurred when population growth had elevated the population density. However, exactly the same situation of a high population relative to resources occurs when environmental fluctuations make resources small in relation to the population. There is then strong selection for traits that promote survival during adversity.

An Example from Nature: The Great Tit

One naturally asks if there are any well studied natural examples resembling humans, which have been studied for long enough to know how they are affected by a fluctuating environment.

The Great Tit (Parus major) is a bird whose offspring are fed by the parents after hatching (Boyce & Perrins, 1987). There is very extensive data on the Great Tit. Its reproductive success has been followed in a large population in Wytham Wood near Oxford, England, from 1963 to 1983 for 4,489 clutches, making it one of the best studied bird populations. The number of eggs per nest, and the number of offspring from each clutch surviving to breed in the next year, have been recorded. The bird lays an average of 8.53 eggs per nest (the clutch size). When averaged over all years, the maximum number of surviving offspring would be achieved by laying 12 eggs. The problem, (which has been observed in other birds) is that this calculated apparent optimum far exceeds the actual number of eggs laid. Why? Evidence permits rejecting an explanation involving skewness in fitness as a function of clutch size, and rejecting the possibility that rearing a large family reduces the parents' survival sufficiently to make smaller families optimal.

Has the theory of survival of the fittest been disproved? In a sense it has, since evolution has apparently led to birds pursuing a strategy which fails to maximize their fitness (with fitness defined as the expected number of surviving offspring).

However, when the data was analyzed to determine which clutch size would give the greatest geometric mean fitness over the study period, it was found to be 8 for a single data point, and 9 for the maximum of a smooth curve fitted through the data. This corresponded very closely to the actual average of 8.53 eggs. Why was this?

The offspring of large clutches fared especially poorly in years of poor food supply (when the parents had to divide the food they brought to the nest among more offspring). The standard deviation of fitness (calculated across years) was found to increase with clutch size. In bad years, the largest clutches did appreciably worse than the smaller clutches. Clutches of 3 and 4 had standard deviations of fitness (number of surviving offspring) of .2 or less. Even in bad years such small families could be adequately fed. At the other extreme, for clutches of 12 (the clutch size that produced the largest absolute number of surviving offspring) the standard deviation was .8. In good years, there was apparently food to support such large families. However, in the bad years, dividing the food among so many greatly reduced survival.

Not surprisingly, the optimal number of eggs varied from year to year, with the number being less in the years with a poor food supply (which for the tit meant a late spring). The tits had no way of knowing when laying eggs whether the food supply would be good or poor. Instead, they had evolved to lay a little fewer eggs than would be optimal if every year was an average year (which would call for laying 12 eggs). This increased the probability of their offspring being adequately fed during the occasional poor years. The effect of the environmental variability was to reduce the optimal family size.

The phenomenon of actual clutch size in birds being below the calculated optimum has been reported to be common. Most birds care for their young, and it would seem logical that in good times most offspring could be adequately provisioned, resulting in a small variance in year to year reproductive success. In poor years attempts to rear large families result in spreading food among too many mouths, and only a few surviving,

I suggest the effect demonstrated in the Great Tit also applies to humans, who also divide the available food and time among several young. Insuring survival during adverse times affects the optimal number of young to bear more than insuring fast population growth during good times. In this case, the effect of increased variability would be to decrease the family size and increase the per capita parental investment, not to decrease it.

The feature that corresponds to clutch size in humans is litter size, and in particular twinning rates. Human normal litter size is one, with occasional twins raising the average slightly above one. Rushton makes much of the fact that twinning rates vary among the races of the world, being highest among Negroids (especially the Yoruba), lower in Caucasoids, and lowest in Mongoloids.

Another variable he discussed was birth rates, and the elements of gestation lengths, age at start of reproduction, and sexual drives that he believed caused racial differences in birth rates. The factors contributing to large families he showed were highest in Negroids, lowest in Mongoloids, and intermediate in Caucasoids. Many of the variables he discussed related to parental investment, such as altruism, and willingness to obey rules. He argued these were greatest in Mongoloids, and least in Negroids, with Caucasoids intermediate. However, the parental investment per capita is inversely related to family size. Drives to give the first or the second child the best possible opportunities lead to small families. These drives would be selected for in the environments that selected for small families, which as shown are likely to be the least predictable environments, not the most predictable ones. The conclusion is that on pure logical grounds the effects of environmental variability are likely to be the opposite of those Rushton argued for. Not only does environmental variability not select for the characteristics Rushton thought it did, it would appear to select for the opposite.

Numerous Births as an Option Permitting Large Families in Good times

Rushton's coauthor, Ankney (personal communication), has provided an alternative bird model. He points out that the Great Tit's eggs all hatch at the same time. In other species eggs hatch at different times, and some are bigger than others and able to out compete the younger birds for food in times of shortage. Some birds (Mock, 1984), including the White Stork, Tawny and Short Eared Owls, Kittiwake (Lack, 1966, pp. 273-274), and even second broods of the Great Tit (Lack, 1968, p. 176), lay more eggs than can be raised in typical conditions. Upon birth the offspring aggressively compete for the parents' attention and feeding. Because the eggs are laid at intervals, the offspring are typically born at different times, and the nest typically contains birds of different ages and sizes, with the first hatched the largest. The parents preferentially feed the largest, most aggressive birds first. Only after they are satiated, will the smaller, less active birds be fed. If food is scarce, only the first hatched survive. The others die. In good times, all of the offspring survive.

By this strategy, resources are concentrated on a few offspring in poor times, but in good times there are many surviving offspring. Since there are costs to producing offspring that will not be raised, such a system is advantageous only when the environment is highly variable, sometimes permitting large clutches to be raised. In effect, the investment in excess eggs purchases an option on a large number of offspring for when food is abundant.

It is well known from financial theory that the value of an option increases with variability (Brigham & Gapenski, 1994, p. 976). Thus, it is possible that, the option value of offspring indeed leads to higher birthrates (indeed in certain birds, it seems to have) in more variable environments. If the human strategy was routinely to have more children than could be raised, and then kill, or allow to die, those who could not be supported, increased environmental variability could select for a higher birthrate, and possibly for lower average investment. These traits are referred to as r ones in Rushton's work. Thus, it must be conceded that Ankney is right conceptually, that higher variability could select for higher birth rates, and possibly even for lower average investments. Is this the human strategy?

It is unlikely that humans do purchase such an option. Whether or not an option is worth purchasing, depends on whether the option's benefits exceed its cost. The above argument shows benefits are positive, making a purchase conceivably desirable. However, the option's cost is that of raising offspring to the age when the option would be exercised. This is the age at which they would perish in poor times. For birds, this cost is likely to be small, only the cost of producing an extra egg. However, for humans the cost will be large. Human offspring have a long gestation period, followed by a long period of nursing, and rearing. This makes the option children provide very expensive. It is very unlikely that for humans the benefit is worth the cost.

The use of such a strategy in humans would be evidenced by high rates of killing or starving of excess children in poor times, and by a human emotional structure that was conducive to abandonment of such excess offspring. There is evidence that many human populations have practiced infanticide (Scrimshaw, 1984; Daly & Wilson, 1984), and in some populations infanticide has even accounted for an appreciable fraction of the total births, such as in the Ayoreo of Paraguay and Bolivia (Burgos & McCarthy, 1984). However, most infanticide appears to be due to individual level causes (killing defective individuals, twins, children unlikely to receive paternal support due to illegitimacy or cuckoldry, etc) rather than an attempt to reduce family size due to lack of the resources required to support large families. For instance, Daly & Wilson (1984) provide a list of the circumstances in which infanticide allegedly occurred for 39 out of 60 societies randomly drawn from the Human Relations Area File. Economic hardship is mentioned for only 3 of these societies, while poor infant quality was a reason in 21 cultures, twins in 14, adulterous conception in 14, mother unwed in 14, etc. In only one culture (the Yakut) was economic hardship the only reason given. Thus, the strategy of having many infants, and then killing some to reduce the number does not appear to be the human one.

If the common human strategy was one of excess births, later followed by neglecting or killing children to reduce family size to the number that could be supported, the death or killing could occur at any age. Indeed, it would probably most frequently occur at later ages. The reason is that the change in resource availability from conception to age five or ten years would be on average much greater than the change from conception to birth. Yet most child killing is at birth or shortly after (which is why it is referred to as infanticide in the literature) (Daly & Wilson, 1984, Figure 1 and Table II). Also, if the human strategy was to give birth to more children than normally could be raised, we would have developed emotions such that child killing or neglect was easy. Instead, humans soon after birth become very attached to their children and find it hard to neglect or kill them, even in times of great hardship.

A strategy of giving birth to excess children (more than can be raised in poor conditions) makes sense only if at the beginning of a period of crisis, it is possible to determine the maximum number that can be carried through the crisis. If this could be done by some signal, humans might have evolved emotional mechanisms that led them to reduce their families to this number. There is no evidence of such mechanisms. However, famine often occurs when a period of food shortage lasts much longer than anticipated (a prolonged drought which is not relieved, a winter that is followed by a much later spring, or a bad year followed by another bad year). Humans often go into these crises with some reserves. Not knowing how long the crisis will last, they spread the reserves out among the family, rather than reducing the family to what the reserves can support. By the time that it is obvious the crisis is very serious, the reserves have often been spent.

Thus, humans do not appear to purchase an option on a large family by giving birth to more children than can be supported in poor conditions, and then pruning the family size down to what can be supported. Since purchasing such options is not the human strategy, increased variability does not increase the value of the option, and does not select for increased birth rates, or for r traits. Instead, humans are more like the Great Tits in that maximization of their long run survival is obtained by keeping their family small enough so that some survive even in adverse times. With this strategy, increased variability selects for a more K strategy, the opposite of what Rushton argued.

Disease

The above argument suggests that much of human behavior is directed at obtaining the optimum family size, with the optimum heavily influenced by how many can be supported in adverse conditions. The critical period is probably several years past birth. To be specific, let us imagine the period of food shortage on average occurs five years after a typical birth. Humans would have evolved psychological drives (sex drives, anxiety levels, etc.) and biological mechanisms (probabilities of fertilization, post partum amenorrhoea etc.) designed to give the optimum family size (roughly the maximum that can survive) during famine.

Since there are some deaths between birth and this critical period, due primarily to disease, the optimal birth rates (and drives that lead to births) are higher than would be required if all children survived. Some mechanisms would be needed to vary birth rates to achieve the desired family sizes. Many species wide mechanisms probably serve to adjust birth rates to childhood mortality rates. The most important of these is the suppression of ovulation by lactation, which prevents pregnancy until the mother is relieved of the burden of lactation (Hill, 1990; Rosetta, 1990). This mechanism has the property that an early death of the nursing child stops lactation, and accelerates the birth of the next child. Psychological mechanisms may also reduce the probability of one birth when there is already a young child (possibly by making the mother feel that the strain of caring for one child makes another one undesirable).

However, the strength of selection for traits that lead to high birth rates (strong sex drives, ease of fertilization, etc.) may also depend on how often children die before reaching the age of five. Where disease rates are high, the selection for traits that increase birth rates should be greater (Ankney reminded the author of this possibility in a personal communication). Some of these are what Rushton refers to as r traits. Thus, there should be stronger selection for r traits in areas with high disease rates.

The tropics are notorious for having higher disease rates than colder areas. These higher disease rates appear to be because the activities of many disease organisms and their vectors (notably insects and snails) are reduced by cold. For instance, cool weather causes mosquitoes to quit flying, and cold weather kills them, thus protecting many cold weather populations from mosquito borne diseases. Thus, the tropics could select for those r characteristics that lead to high birth rates.

Note that the above disease effects are not fundamentally related to any unpredictability in the tropics, or predictability of colder areas. If the death of certain children was completely predictable, the selective forces would still be for high birth rates where childhood mortality was high. It is the level, not the variability or predictability, of childhood mortality that should lead to selection for differing birth rates, and for traits that affect birth rates.

Relevance to Other Species

The reader may wonder if the above argument provides a more general argument against the generalization that variable environments select for r characteristics, and stable ones for K characteristics? To some extent yes. The above argument would seem to apply to any non-colonizing species that makes heavy parental investment after birth, and divides this investment among several offspring. In good times virtually all offspring can be raised. In bad times, attempts to raise large families result in actually having a smaller number of survivors than if fewer offspring had been born. However, most species do not provide the after birth care that results in year to year variance increasing with family size.

Yet, humans, Great Tits, and many birds are faced with the problem of committing to a family size, and the resulting level of per capita investment, before it is known whether the environment will be favorable or unfavorable. In these circumstance, increasing environmental variability selects for smaller families and for greater per capita investment, unless the option value of additional births is sufficiently large, and the option's cost sufficiently small. To the author's knowledge, this exception (which applies to humans) to the r versus K generalization has not been explicitly noted in the literature on r and K selection.

When circumstances are unpredictable and the same phenotype experiences both good and bad times, the r characteristics will be selected for in good times when rapid population growth is possible, and the K characteristics in the bad times when success in competition is called for.

However, as was pointed out in the earlier paper (Miller 1993), the variability many organisms experience indeed selects for r characteristics. Many organisms survive the poor periods in another phenotype such as seeds or spores. Many of the classic r selected organisms are colonizing ones. The secret to their success is that rapid reproduction permits them to quickly colonize the vacant areas that periodically appear in variable environments. This environmental variability is not correlated across space. Thus, when one patch of resources is becoming overpopulated, another patch may be opening up. There is little competition in the new patches. Organisms with high natural rates of increase in favorable circumstances (this rate of increase is traditionally symbolized by r) can expand rapidly in these favorable areas, and will dominate them before other species move in. Such a pattern of constant change, with humans frequently moving to new areas where there are unexploited resources, does not appear to describe the human environment of evolutionary adaptation. Nor do the most important differences between the environments to which different human races evolved correspond to differences in variability, or susceptibility to a colonization strategy.

Instead, there are very obvious climatic differences between sub-Saharan Africa, Europe, and Northeast Asia. These differences seem much more likely to have had an impact than any subtle differences in unpredictability. Africans and certain other races evolved in tropical areas, while Caucasoids and Mongoloids evolved in cold climates with severe winters. The severe winters could only be survived by some combination of food storage and large game hunting. The need to store food selected for intelligence and for the ability to defer gratification (Miller 1991). The large game hunting which was critical for winter survival could only be done by males, which implied Caucasoids and Mongoloids were selected for traits which were conducive to male provisioning of females. In tropical areas, such as those where Africans evolved, females could provision themselves all year round. The males who were most successful in reproduction were those who devoted their efforts to mating, not to provisioning. The details of the argument were set forth earlier (Miller 1994b).

Today, the populations of the world are often living in circumstances which are quite different from those in which they evolved. However, the gene frequencies that prehistoric climates shaped persist.

It is not surprising that racial differences would be most apparent in those behavioral characteristics which can be plausibly tied to temperature and the severity of the winters, as in the above argument, because systems of racial classification are based on elements of appearance such as skin color and nose shape which have been influenced by climate and latitude (Baker, 1974; Brues, 1990; Coon, 1965, 1982; Krantz, 1980). In contrast, features such as skin color and nose shape appear to be uninfluenced by environmental predictability or unpredictability. Even if unpredictability does affect behavioral traits, this makes it less likely that they would be correlated with the traits that are normally used to define races, such as skin color and nose shape.

Other Applications

The insight that human family size might be influenced not by typical conditions, but by occasional adverse periods helps to explain why women seldom have as many children as they are theoretically capable of bearing. Hill (1990) concludes that the maximum (average over a population) number of children per women is 15.3 (assuming no breast feeding). Actually, in no culture do women come close to having that many children. Apparently fitness is not maximized by producing children at the maximum rate possible. Somehow, we have evolved so that, even prior to birth control, the number of offspring was held well below this. Some of these mechanisms were psychological ones that limited sexual activity, and limited sex in the absence of a pair bond. Others were physiological.

One such mechanism (admittedly stronger in some populations) is a reluctance of women to permit intercourse unless the man appears emotionally commited to them. The pair bonding mechanism (i.e. the human psychological tendency to fall in love and become attached to a member of the opposite sex) may have been most strongly selected for during famines.

Evidence is that current hunter-gather populations appear well fed when visited by anthropologists (i. e. during normal times). In times of abundant food, a female might be able to gather enough for herself and her offspring. Even if she couldn't (perhaps because she didn't hunt, and hunting was the major source of winter food) assistance from relatives and other band members probably supplied enough food. The earlier (Miller, 1994b) paper mentioned that most current hunting people normally share large game within the band. However, in poor times male assistance was probably needed to support a family. Relatives and other band members who readily help out when food is abundant give priority to themselves and their own children in time of shortage.

Thus, the benefits to the female from not having a child, unless pair bonded with a male who would help raise it, were strongest during famine. Very likely, in good times, delaying pregnancy till marriage delayed the start of reproduction, and may have lengthened the period between offspring (a sneak copulation with a non-committed male increases the probability of promptly getting pregnant).

After marriage various birth limiting mechanisms may have evolved because large families were much less successful at surviving periods of adversity than small families. The descent lines of women who had evolved mechanisms that limited their family size survived occasional famines much better than those that produced large families. During adversity both large and small families had similar resources, but dividing them among fewer members, the small families had much higher survival rates. Also, families with many children were less mobile. This made it harder for them to move in response to changing climatic conditions, or to flee from enemies. Large families are harder to conceal, and this made it harder to hide from invaders.

One of the assumptions behind the calculation of women being able to produce 15.3 children is no breastfeeding. Breastfeeding is known to suppress ovulation and menstruation through the effect of the suckling stimulus at the breast (Dunbar, 1990; Rosetta, 1990). This period of lactational amenorrhoea is normally shorter than the total period of lactation (Hill, 1990, Table 11.1). However, this lactational amenorrhoea clearly reduces the number of children women can raise. Surely anything that prevents pregnancies would have been selected against during human evolution. As is usually argued, lactational amenorrhoea prevents women from experiencing the energetic demands of both lactation and pregnancy, which could affect the health and survival of both offspring and mother.

Lancaster & Lancaster (1987, 194-195) state that, "A strong argument can be made that, among the higher primates, access to energy to support a long lactation is the principal concern of the adult female." Lactation is more energy demanding even than gestation (Anderson 1983, 27).

However, when food was abundant women should have been able to both nurse and support a pregnancy. The human digestive system or circulatory system appears able to process the quantity of food required for both lactation and pregnancy (which impose less of a caloric burden than strenuous physical activity does). Thus, in normal conditions fitness is reduced by lactational amenorrhoea.

Although humans can digest sufficient food to support simultaneous lactation and pregnancy, obtaining the food could have presented a problem, especially for unassisted women who were burdened with one or more small infants. The problem of obtaining adequate nutrition during lactation is made more difficult because nursing itself consumes time that could be spent in food gathering, and because lactating women are accompanied by a vulnerable infant whose presence may interfere with food gathering. Indeed, among the Ache hunter-gatherers of Paraguay, lactating females gather fewer calories on foraging trips than non-lactating females (Hurtado, Hawkes, Hill, & Kaplan, 1985). This is due to both lower efficiency and less effort. The lower efficiency was shown by gathering only 2601 calories per hour of palm starch pursuit versus 3524 for non-nursing females (their Table VII). Yet a nursing woman appears able to gather enough food in one hour to support herself. Thus, these numbers are such that in normal times a woman's fitness would appear to be promoted by being continually pregnant, and leaving many descendants.

However, during famine women who took on the burden of simultaneous pregnancy and lactation probably failed to do either well, or came out of the experience with their own bodies so depleted that their own survival, or the survival of earlier borne children was threatened. This led to human females evolving mechanisms by which lactation suppresses ovulation.

A similar analysis could be made of the observed relationship between female body fat and menstruation (Frisch, 1989). Very low levels of body fat delay the start of menstruation in adolescents, and suppress menstruation in older women. Fat is not needed to carry a baby to term. If there is adequate food available to the mother, a woman can support a pregnancy and the associated lactation even if she originally has very little fat reserves. That low body fat can cause amenorrhoea is usually explained as an adaptation to prevent pregnancy during famines when they might not be able to successfully complete the pregnancy, or when their own body might be adversely affected by pregnancy (Fetuses are in effect parasites on their mothers, and can extract nutrients even when the loss of the nutrients would adversely affect the mother). Thus, we see evidence of specific human adaptation to reduce family size during famines, even at the expense of reducing family size in other times. Without the need to accumulate body fat, human females could use their energy for growth and reach reproductive maturity earlier. This would permit them to leave more descendants in normal times.

Caucasoids as a Mixture of Mongoloids and Negroids

One of the more impressive of Rushton's findings was that so many characteristics were ordered Negroid, Caucasoid, Mongoloid, or the opposite. The Caucasoids were typically in between the other two major races. This is not a pattern that would emerge if racial differences on each tract were determined by separate environmental causes. Even if all gene frequency differences resulted from genetic drift, or selection for different traits on a non-systematic basis, such a pattern would not be expected. Rushton claims it is explained by his differential K theory. It is also explained by a differential paternal investment theory (Miller 1994b).

However, there is another explanation. The Europeans might be intermediate to Negroids and Mongoloids because they, or an ancestral population represented a mixture of the two groups. Cavalli-Sforza, Menozzi, & Piazza (1994, pp. 90-93) in their new book show, using both classical polymorphisms and 100 DNA polymorphisms, that several methods of tree drawing find a very short line leading to Europeans from the node where they join the tree. Statistical tests for "treeness" using the distance matrices produce results that are inconsistent with the genetic distances being produced by populations splitting and then undergoing equal degrees of genetic drift. Calculations show that the gene frequencies are consistent with a mixture of 65% Chinese with 35% Africans, occurring an average of about 30,000 years ago.

To make this somewhat surprising scenario more plausible they return to their earlier argument that European populations were heavily influenced by gene flow that occurred with the original movement of farming populations out of the Near East. The Near East is intermediate between Africa and East Asia. Continual diffusion of genes along the path from Africa to Asia could have produced a continuous gradation in gene frequencies. The Fertile Crescent areas, where agriculture is believed to have originated, would have been intermediate between Africa and Asia. This intermediate pattern of gene frequencies would have then been carried into Europe with the movement of early farmers.

One of the problems with this hypothesis is that some physical traits of the Europeans are not intermediate between the Africans and Chinese. For instance, both groups have darker skins and lesser lactose tolerance than Europeans. However, these traits may be explained by the Europeans then undergoing selection for light skins and the ability to utilize lactose. These traits maximize production of vitamin D in an area where cloudy skys limit exposure to the sun.

New Evidence

In the original paper it was argued that the critical problem for prehistoric people in the frigid North was to survive the winter. A testable implication of this statement would be that deaths were most common in the winter (or just after, while the population was still weakened). Detecting this in the archaeological record would be difficult.

One prehistoric culture where this is possible is the Windmiller native American culture (45,000 to 30,000 years ago) of the Central Valley of California, where the season of burials can be identified from the orientation of the grave. This was a culture that got most of its food from hunting. It was found that 80% of the burials occurred in the winter half of the year, with a very strong peak in late winter-early spring. "This supports a hypothesis of Windmiller late winter-early spring hardship; presumably, increased mortality was due either to direct starvation, or more likely, to the interaction of poor nutrition and poor health." (Dickel, Schulz, McHenry, 1984, p. 444). In such conditions, males (who would have done the hunting) would be strongly selected for the ability and willingness to provision their offspring, and females for the ability to induce such provisioning. Both would be selected to form strong pair bonds.

The Windmiller culture was replaced by later cultures that relied heavily on acorns and salmon for food. Stored acorns supplemented their winter food supply. While information is not available on the timing of deaths for these later cultures, the same pattern of game availability would have existed. It is plausible, even if speculative, that winter survival remained a problem and that there would have been selection for the personality traits (ability to defer gratification) and intelligence that led to storing large quantities of acorns.

Since the first paper (Miller 1994b), other racial differences have been found which can be explained by a greater emphasis on mating effort in Negroids. Based on a telephone survey of 5000 households, American blacks spend more on men's clothes in absolute dollars ($99 monthly) than American whites ($66) (Wynter, 1994). This is surprising given their lower incomes. However, their greater clothing expenditure is explainable by the importance of appearance to mating.

McShane (1983, Table 1) has found greater average skull (averaged over the occipital pole and frontal pole) thicknesses in blacks than in whites, Orientals, or American Indians. If as part of their greater competition for mates, blacks were more strongly selected for surviving injuries resulting from fighting, this greater skull thickness could be explained.

Mackey (personal communication), drawing on the data behind his book (1985) from naturalistic observations in 30 different subcultures and cultures involving over 50,000 dyads that the percent of children in men only groups among East Asian cultures was significantly above the percent in the African cultures examined. Also in a survey (Mackey, 1995) of reasons why men wanted to be fathers (among men) and why "their men wanted to be fathers" (women), he found that White and Asian-American men and women most often gave reasons of "love and emotional satisfaction" and to "continue the bloodline and name", while among Afro-American men the most common reason given was that the wife wanted children, and among Afro-American women was "accidents".

South (1993) reports that while there is virtually no difference between white men , white women, and black women aged 19 to 25 in the percentage not desiring to marry (12.6,11.2, and 12.7 respectively), significantly more black men (22.8%) report not wishing to marry. Most of the difference can be traced to black males not believing marriage will improve their sex lives.

For genetically caused differences in parental behavior to emerge, selection needs genetic variability to act on. Substantial additive genetic variability in parental behavior has recently been shown (Perusse, Neale, Heath, & Eaves, 1994). Part of the heritability appeared to be from the expression of dominant genes, which would be consistent with these traits having been subjected to natural selection.

The original argument dealt with the personality and other traits that would have been selected for in regions with different degrees of competition between males for mates (monogamy versus polygyny). The only empirical evidence that inhabitants of different regions differed in their competition for mates was the current variability in amounts of polygyny in contemporary cultures.

Yet for gene frequencies to have been affected, societies must have differed in their mating practices over tens of thousands of years. Of course, there are no records that describe prehistoric mating patterns. However, genetic patterns exist that suggest that the differences between Europeans and Africans in mating patterns are long standing.

A sharp distinction between Caucasoids and Africans has been found using markers on the Y chromosome.The Y chromosome is inherited only in the paternal line. "A haplotype (A1C0D0) absent in Caucasians and present in the majority of the Africans examined has been identified" (Torroni, Semino, Scozzari, Sirugo, Spedini, Abbas, Fellous, & Santachiara-Benerecetti, 1990). Indeed, if the mutations that gave rise to this haplotype had occurred only once, it appears that the majority of the West Africans are descendants of the male in which this mutation first occurred.

Studies of the same populations had also been conducted for mitochondrial DNA for restriction length fragment polymorphism. These studies had found a mitochondrial DNA type peculiar to Africans. Overall the Africans (Senegalese) were more heterogeneous than the Italians (Torroni, et al. 1990, p. 295), as shown by lower values of a statistic for measuring homogeneity (Fs= the sum of the squares of the frequencies for all of the haplotypes) that was .390 for Italians versus .237 and .232 for the Wolof and Peuls of Senegal respectively). Yet the studies of the Y chromosome showed a much higher degree of homogeneity in the Africans (.478) than in the Italians (.118). How is this to be explained?

The authors note "This discrepancy could be, at least partially explained by polygamy which is in use among these people. Only certain men (the richest) can have several women and thus contribute to the next generation more than do other men." Thus, the number of different men who have left descendants in the sampled populations in Africa is much less than the number of different women. Hence, through the wonders of modern molecular genetics we now have evidence that differences between Africans and Caucasoids in the level of competition for women (or strictly speaking in the success of certain men in mating with women) have existed for very long periods of time.

Additional evidence that at least many Africans have been polygynous for several thousand years comes from linguistics. Linguistic reconstruction of proto-Bantu, spoken 3,000 years ago, shows a specific term for "taking a second wife", indicating that polygyny was common (Polome, 1977). The Roman writer Tacitus reports that the Germanic tribes were monogamous. Thus, the African-European difference in mating patterns appears long standing.

New Research on Intelligence

Recently, the new controversial book by Herrnstein & Murray (1994) provides some new evidence relative to racial differences. Tabulating intelligence scores from the National Longitudinal Survey of Youth, they find that the white (non-Latino) intelligence quotient is 1.21 standard deviations above that for blacks (p. 278). This is a little larger than the one standard deviation usually reported in the literature.

Black mothers of average age had a 62% illegitimacy rate versus 12% for Whites (p. 331). After controlling for intelligence, the figures are 51% for blacks and 10% for Whites.

Likewise, they show that the percentage of men in the study interviewed while in prison (indicating they had been convicted of a serious crime since the start of the study) was higher for Blacks (13.1%) than for Whites (2.4%) holding age constant (p. 339). For men of average age and IQ the figures are 2% for Whites versus 5% for blacks (p. 338). The difference is reduced, but not eliminated by controlling for intelligence and socioeconomic class. What explains these remaining differences in illegitimacy and crime rates may be partially differences in frequencies of behavior relevant genes. My paternal investment theory predicts higher Negroid illegitimacy and crime rates.

Herrnstein & Murray show that Black mothers of average age have 10% low-birth-weight babies versus 3% for whites (p. 334). After controlling for IQ the difference shrinks to 6% for Blacks, and 3% for whites. This strong effect of controlling for IQ is somewhat surprising since most researchers tend to think of these variables as unrelated. The authors suggest that maybe intelligence affects the rates of behavior that may be adverse to fetal development such as alcohol, tobacco, and drug use or obtaining adequate prenatal care.

However, another possibility exists. In older children and adults brain size is related to intelligence (Jensen & Johnson, 1994; Jensen, and Sinha, 1993; Rushton, 1995), and head size at birth is related to IQ at age 7 (Broman, Nichols, Shaughnessy, & Kennedy, 1987). The brain accounts for a relatively large part of the weight of a new born. A larger brain would require a larger heart, lungs, digestive system etc. to support it. Thus, the mother's IQ low-birth-weight correlation may reflect the presence of genes that affect brain size in both mother and child. Just possibly the racial differences in weight at birth which Rushton interprets in r/K terms reflect partially an evolved difference in brain weights. If so, racial differences would be a byproduct of selection for intelligence. This is of course highly speculative.

A very comprehensive new book (Cavalli-Sforza, Menozzi, & Piazza, 1994) on gene geography has appeared. While Time magazine (Subramanian 1995) has described it as refuting The Bell Curve, the genes mapped are not known to affect intelligence. However, as pointed out in a recent review article (Miller, 1994c), one perusing the tabulations and maps in this book is likely to notice that there is no polymorphic gene discussed that has a uniform frequency around the globe. Nor does the theory of population genetics (which describes how gene frequencies drift apart when one population separates from another) predict a uniformity in gene frequencies (Cavalli-Sforza & Bodmer, 1971, give the theory). It follows that if there exists a polymorphic gene that affects a behavioral trait (such as intelligence, criminality, or sexual behavior) one should expect its frequency to vary among populations and races. Traits such as intelligence (Hewitt & Last, 1984; Plomin & Loehlin, 1989; McCartny, Harris, & Bernieri, 1990; Bouchard, Lykken, McGue, Segal, & Tellegen, 1990; Bouchard, 1993; Herrnstein & Murray, 1994; Plomin, Chipuer, & Neiderhiser, 1994), criminality (Mednick & Kandel, 1988; Wilson & Herrnstein, 1985), personality (Eaves, Eysenck, & Martin, 1989; Loehlin, 1992), and sexual behavior (Hamer, 1994) are now known to be subject to genetic influences (Miller, 1994a,b; Rowe 1994; Rushton, 1995). It follows that racial differences are to be expected in the frequency of such genes. Of course, if the trait is affected by many genes, and is not subject to selection, one group may be high in some favorable alleles and low in others, causing the genetic variance in the trait to be small, and not socially significant. However, there is no reasons to presume this result.

In the original paternal investment paper (Miller 1994) no paternal investment related explanation for racial differences in intelligence was offered, although it was explained by adaptation to climate and the use of food storage as a strategy for surviving the winter (as in Miller 1991). Since Rushton did claim to be able to explain the observed racial differences in intelligence by his differential K theory, this remained as the one area where his theory appeared to explain something my theory did not. I have since realized that differences in the importance of paternal investment for children's survival should lead to differences in intelligence.

It is standard sociobiological theory that resources provided by men are important to female reproductive success and that women select mates partially on the basis of the male's ability and willingness to provide such resources (Symons, 1979; Hrdy, 1981). This induces men to try to persuade women that they have such resources and will provide them to the woman and her children. Buss (1994) emphasizes how often in human mating deception is used. For instance, men try to convince women that they have, or will have resources, and will devote them to the well-being of a particular woman and her children (and not squander them on other women and their children), while women try to convince men that they will be sexually faithful to them (while possibly seeking better genes from other men). Buss states (p. 155) "Because the deceived can suffer tremendous losses, there must have been great selection pressures for the evolution of a form of psychological vigilance to detect cues to deception and to prevent its occurrence. The modern generation is merely one more cycle in the endless spiral of an evolutionary arms race between deception perpetuated by one sex and detection accomplished by the other. As the deceptive tactics get more subtle, the ability to penetrate deception become more refined." With northern women more dependent on male assistance than tropical women, selection would have been stronger in the northern regions for the intelligence needed to recognize deception.

Men are also vulnerable to deception. They can waste much in resources if they are cuckolded and devote resources to raising another man's offspring. Women try to deceive men as to their loyalty. The more dependent a man's reproductive success is on accurately directing his provisioning to his biological offspring and to those females who have or will bear his children, the stronger the selection for male intelligence. In the northern climates where paternal provisioning has the greatest impact on offspring survival, selection for intelligence will be greatest.

Differences in the importance of paternal investment may even explain differing patterns of abilities among the races. Detecting deception in a mate calls for a high level of reasoning, which would show up in high g. Merely impressing potential sex partners with conversation and song would call for high verbal skills, memory, and verbal fluency. This does resemble the observed pattern of racial differences (Jensen & Reynolds, 1982; Jensen, 1985).

Since my theory predicts genetic differences in intelligence among races, the obvious question is are there any? There is little professional dispute about the existence of racial differences in intelligence, but more about the extent to which the causes are genetic. The extensive literature cannot be reviewed here, but several recent books and articles provide reviews (Lynn, 1991a, 1991b, 1992; Herrnstein & Murray, 1994; Rushton, 1995).

However, there are several important recent findings that have not yet found their way into the reviews. Jensen & Johnson (1994) document not merely that black children have smaller head sizes than white children, that black children have lower IQ's, and that within racial and sexual groups that head size correlates with intelligence, but that controlling for IQ eliminates the racial differences in head size within pairs of siblings. In another study, Jensen (1994) showed that for 17 diverse mental tests the mean difference between black and white children is related to the tests loading on g and the tests' correlation with head size. These two findings are most simply explained by their being racial differences in the frequency of certain genes that affect both head size and intelligence.

Levin (1994) and Lynn (1994) in commenting on the Minnesota transracial adoption study both point out that black and white adopted children raised in environmentally favorable Minnesota (a state with relatively little racial prejudice) white homes display an IQ difference (17 points) slightly exceeding that observed when the children are raised in homes of their own race (about 15 IQ points), with those adoptees of mixed race displaying an IQ midway between that of the white and black race. The adoption design, by holding most aspects of the environment constant while varying the genotype, provides evidence that the observed differences are primarily genetic, although the study's authors disagree (Waldman, Weinberg, & Scarr, 1994).

Conclusions

Evolutionary success is determined by the geometric mean of fitness in successive generations, not the arithmetic mean. The genotype that becomes dominant is often that which is best at surviving adverse periods. In a species, such as Homo sapiens in which limited resources are divided among a small number of offspring, having few offspring and high average paternal investment contributes to evolutionary success. This reduces the variance over time in reproductive success, as is illustrated by the evidence from the Great tit. While in theory the option value of additional offspring could lead to greater variability selecting for more offspring, as in some birds, this appears not to happen with humans who make very large investments early.

The argument presented here may explain why humans of all races have evolved drives that lead to relatively low birth rates compared to biological maxima. Small families survived the occasional periods of famine or adversity better, and hence had higher geometric mean fitness.

An alternative to climate based evolutionary theories for explaining why Caucasoids appear in so many way to be between Mongoloids and Negroids is that they are simply the result of a mixture of two stocks, or the beneficiary of gene flows from both Africa and Asia.

Finally, several new findings have been discovered which in some way support paternal investment theory. Evidence from one ancient northern prehistoric culture has been found that deaths were indeed greater in late winter and early spring. A new finding that Blacks spend more on male clothing than Whites is explainable as greater mating effort, but not as easily explained by other theories. Climatic differences in the importance of male provisioning for offspring survival make having sufficient intelligence to discover deception more critical for offspring survival in northern climates than in tropical ones. Finally, evidence indicating that polygamy has been more common in Africa than in Europe for the thousands of years required to affect gene frequencies comes from linguistic research and research on Y chromosomes and with mitochondrial DNA alleles.

References

Anderson, P. (1983). The Reproductive Role of the Human Breast. Current Anthropology, 24, 25-32.

Baker, J. R. (1974). Race. Oxford: Oxford University Press.

Bouchard, T. J., Lykken, D. T., McGue, M., Segal, N. L., & Tellegen, A. (1990). Sources of human psychological differences: the Minnesota study of twins reared apart. Science, 250, 223-228.

Boyce, M. S. (1984). Restitution of r- and K selection as a model of density-dependent natural selection. In Johnston, R. F. (Ed.), Annual review of ecology and systematics (Vol. 15, pp. 427-447).

Boyce, M. S. & C. M. Perrins (1987). Optimizing great tit clutch size in a fluctuating environment. Ecology, 68, 142-153.

Bouchard, T. J., Jr., (1993). The genetic architecture of human intelligence. In Vernon, P. A. (Ed.). The biological basis of intelligence, Norwood: Ablex, 33-94.

Brigham, E. F. & Gapenski, L. C. (1994). Financial management. Fort Worth: Dryden Press.

Broman, S., Nichols, P. L., Shaughnessy, P., & Kennedy, W. (1987). Retardation in young children. Hillsdale: Lawrence Erlbaum.

Brues, A. M. (1990). People and races. prospect heights, Il.: Waveland.

Burgos, P. E. & McCarthy L. M. (1984). Ayoreo Infanticide: A Case Study. In Infanticide: comparative and evolutionary perspectives, Hausfater, G. & Hrdy, S. B., New York: Aldine, 503-520.

Buss, D. M. (1994). The Evolution of Desire. New York: Basic Books.

Cavalli-Sforza, L. L., & Bodmer, W. F. (1971). The genetics of human populations. San Francisco: W. H. Freeman.

Cavalli-Sforza, L. L., Menozzi, P., & Piazza, A. (1994). The history and geography of human genes. Princeton: Princeton University Press.

Coon, C. S. (1965). The living races of man. New York: Alfred A. Knopf.

Coon, C. S. (1982). Racial adaptations. Chicago: Nelson-Hall.

Daly, M. & Wilson, M. (1984). A sociobiological analysis of human infanticide. In Infanticide: comparative and evolutionary perspectives, Hausfater, G. & Hrdy, S. B., New York: Aldine, 487-502.

Dickel, D. N., Schulz, P. D., McHenry, H. M. (1984). Central California: Prehistoric subsistence changes and health. In Cohen, M. N. & Aremelagos, G. J. (Eds), Paleopathology at the origins of agriculture (pp. 439-461). New York: Academic Press.

Dunbar, R. I. M. (1990). Environmental and social determinants of fecundity in primates. In Landers, J. & Reynolds, V. (Eds), Fertility and resources (pp. 5-17). Cambridge: Cambridge University Press..

Eaves, L. J., H. J. Eysenck, & N. G. Martin.1989. Genes, culture, and personality. London: Academic Press.

Flynn, J. R. (1989). Rushton, evolution and race: An essay on intelligence and virtue. The Psychologist, 9, 363-366.

Frisch, R. E. (1989). Body fat, menarch, fitness and fertility. In Frisch, R. E. (Ed.), Adipose tissue and reproduction.(pp. 1-26). Basel: Karger.

Gillespie, J. H. (1977). Natural selection for variances in offspring numbers: a new evolutionary principle. American Naturalist, 111, 1010-1014.

Gross, B. R. (1990). The case of Philippe Rushton, Academic Questions, 3, 35-46.

Hamer, D. H. (1994). The science of desire: the search for the gay gene and the biology of behavior. New York: Simon & Schuster.

Herrnstein, R. J. & Murray, C. (1994). The bell curve: Intelligence and class structure in american life. New York: The Free Press.

Hewitt, J. K., & Last, K. A. (1984). Twin studies of intelligence: Recent data, new questions. In Turner, C. J., &, Miles, H. B. (Eds.). The biology of human intelligence, Nafferton: Nafferton Books Limited.

Hill, A. G. (1990). Understanding recent fertility trends in the Third World. In Landers, J. & Reynolds, V. (Eds.), Fertility and resources (pp. 146-163). Cambridge: Cambridge University Press.

Hrdy, S. B. ( 1981). The women that never evolved. Cambridge: Harvard University Press.

Hurtado, A. M., K. Hawkes, K. Hill, & Kaplan, H. (1985). Female subsistence strategies among Ache hunter-gatherers of Eastern Paraguay. Human Ecology, 13, 1-28.

Jensen, A. R. (1985). The nature of the black-white differences on various psychometric tests: Spearman's hypothesis. Behavioral and Brain Sciences, 8, 193-219.

Jensen, A. R. (1994). Psychometric g related to differences in head size. Personality and Individual Differences, 17, 597-606.

Jensen, A. R. & Johnson, F. W. (1994). Race and sex differences in head size and IQ. Intelligence, 18, 309-333.

Jensen, A. R. & Reynolds, C. R. (1982). Race, social class, and ability patterns on the WISC-R. Personality and Individual Differences, 3, 423-438.

Jensen, A. R., & Sinha, S. N. (1993). Physical correlates of intelligence. In Vernon, P. A. (Ed.). The biological basis of intelligence. Norwood: Ablex, 139-142.

Krantz, G. S. (1980). Climatic races and descent groups. North Quincy: Christopher Publishing House.

Lack, D. (1966). Population studies of birds. Oxford: Clarendon.

Lack, D. (1968). Ecological adaptations for breeding in birds. London: Methuen.

Lancaster J. B. & Lancaster, C. S. (1987). The watershed: change in parental-investment and family-formation strategies in the course of human evolution. In Lancaster, J. B. & Lancaster, C.S. (Eds.), Parenting across the life span: biosocial dimensions (pp.187-206). New York: Aldine de Gruyter.

Leslie, C. (1990). Scientific racism: Reflections on peer review, science and ideology. Social Science in Medicine, 31, 891-912.

Loehlin, J. M. (1992). Genes and environment in personality development. Newbury Park: Sage.

Levin, M. (1994). Comment on the Minnesota transracial adoption study. Intelligence, 19, 13-20.

Lewontin, R. C. & Cohen, D. (1969). On population growth in randomly varying environment. Proceedings of the National Academy of Sciences, 62, 1056-1060.

Lynn, M. (1989). Race differences in sexual behavior: A critique of Rushton and Bogaert's evolutionary hypothesis. Journal of Research in Personality, 23, 1-6.

Lynn, R. (1991a). Race differences in intelligence: a global perspective, Mankind Quarterly, 31, 254-296.

Lynn, R. (1991b). The evolution of racial differences in intelligence, Mankind Quarterly, 32, 99-121.

Lynn, R. (1992). Intelligence: ethnicity and culture. In Lynch, J., Modgil, C., Modgil, S. (Eds.), Cultural diversity and the schools. Washington: Falmer Press.

Lynn, R. (1994). Some reinterpretation of the Minnesota transracial adoption study. Intelligence, 19, 21-28.

MacArthur, R. H. & Wilson, E. O. (1967). The theory of island biogeography. Princeton: Princeton University Press.

Mackey, W. C. (1985). Fathering behaviors: The dynamics of the man-child bond. New York: Plenum.

Mackey, W. C. (1995). Motivations towards fathering: A preliminary comparison of the perceptions of men and women from three ethnic backgrounds. Working paper: West Burlington, Iowa: Southeastern Community College, North Campus.

Markowitz, H. M. (1959). Portfolio selection: Efficient diversification of investments. (New York: John Wiley & Sons).

McCartny, M., Harris, O. J. & Bernieri, F. (1990). Growing up and growing apart: a developmental meta-analysis of twin studies. Psychological Bulletin, 107, (2), 226-237.

McShane, D. (1983). Neurocranial form: differentiating four ethnic populations using a simple CT scan measure. International Journal of Neuroscience, 21, 137-144.

Mealey, L. (1990). Differential use of reproductive strategies by human groups?, Psychological Science, 1, 385-387

Mednick, S. A. & Kandel, E. (1988). Genetic and perinatal factors in violence.In Moffitt, T. E. & Mednick, S. A. (Eds.), Biological contributions to crime causation. Dordrecht, Martinus Nijhoff Publishers. 121-132

Miller, E. M., (1991). Climate and intelligence. Mankind Quarterly, 32 127-132.

Miller, E. M, (1993). Could r selection account for the African personality and life cycle? Personality and Individual Differences, 15, 665-676.

Miller, E. M., (1994a). Optimal adjustment of mating effort to environmental conditions. Mankind Quarterly, XXXIV, 297-316.

Miller, E. M, (1994b). Paternal provisioning versus mate seeking in human populations. Personality and Individual Differences, 17, 227-255.

Miller, E. M, (1994c). Tracing the genetic history of modern man. Mankind Quarterly, 35 (Winter 1994) No. 1-2, 71-108.

Mock, D. W. (1984). Infanticide, siblicide, and avian nesting mortality. In Hausfater, G. & Hrdy, S. B. (Eds.) Infanticide: Comparative and evolutionary perspectives (pp. 3-30). New York: Aldine.

Pearson, R. (1991). Race, intelligence and bias in academe, Washington: Scott-Townsend.

Perusse, D., Neale, M. C., Heath, A. C., & Eaves, L. J. (1994). Human parental behavior: Evidence for genetic influence and potential implications for gene-culture transmission. Behavior Genetics, 24, 327-335.

Pianka, E. R. (1970). On r- and K-selection. The American Naturalist, 104, 592-597.

Plomin, R., Chipuer, H. M., & Neiderhiser, J. M. (1994). Behavior genetic evidence for the importance of nonshared environment. In Hetherington, E. M., Reiss, D., & Plomin, R. (Eds.) Separate social worlds of siblings: The impact of nonshared environment on development. Hillsdale: Lawrence Erlbaum.

Plomin, R., & Loehlin, J. C. (1989). Direct and indirect IQ heritability estimates: a puzzle. Behavior Genetics, 19, (3), 331-342.

Plomin R., McClearn, G., Smith, D., Vignetti, S., Chorney, M., Venditti, C., Kasarda, S., Thompson, L., Detterman, D., Daniels, J., Owen, M., & McGuffin P. (1994). DNA Markers Associated with High Versus Low IQ: The IQ Quantitative Trait Loci (QTL) Project. Behavior Genetics, 24, 107-118.

Polome, E. C. (1977). The reconstruction of proto-Bantu culture from the lexicon. In Bouquiaux, L. (Ed.) L'Expansion bantoue, 2:779-791. Centre national de la recherche scientifitique.

Roberts, J. V. & Gabor, T. (1990). Lombrosian wine in new bottles, Canadian Journal of Criminology, 32, 291-313.

Rosetta, L. (1990). Biological aspects of fertility among Third World populations. In Landers, J. & Reynolds, V. (Eds.) Fertility and resources (pp. 18-34.). Cambridge: Cambridge University Press.

Rowe, D. C. (1994). The limits of family influence: Genes experience and behavior. (New York: Guilford: 1994).

Rushton, J. P. (1985). Differential K. theory: The sociobiology of individual and group differences. Personality and Individual Differences, 6, 441-452.

Rushton, J. P. (1987). Towards a theory of human multiple birthing: Sociobiology and r/K reproductive strategies, Acta Genetica Medical Gemellology, 36, 289-296.

Rushton, J. P. (1988). Race differences in behavior: A review and evolutionary analysis. Personality and Individual Differences, 9(6), 1009-1024.

Rushton, J. P. (1989a). The evolution of racial differences: a response to M. Lynn, Journal of Research in Personality, 23, 7-20.

Rushton, J. P. (1989b). The evolution of race differences in sexuality and their correlates: Another look and physiological models, Journal of Research in Personality, 23, 35-54.

Rushton, J. P. (1990). Race and crime: A Reply to Roberts and Gabor, Canadian Journal of Criminology, 32, 315-334.

Rushton, J. P. (1991). Do r-K strategies underlie human race differences? A reply to Weizmann et al., Canadian Psychology, 32(1): 29-41.

Rushton, J. P. (1995). Race, Evolution and behavior: A life history perspective. New Brunswick: Transaction Publishers.

Rushton, J. P. & Ankney, C. D. (1993). The evolutionary selection of human races: a response to Miller, Personality and Individual Differences,15, 677-680.

Rushton, J. P. & Bogaert, A. F. (1987). Race differences in sexual behavior: Testing an evolutionary hypothesis. Journal of Research in Personality, 21, 529-551.

Sacher, G. A. , Staffeldt, E. F. (1974, September-October). Relation of gestation time to brain weight for placental mammals. The American Naturalist,, 108(963), 593-615.

Schaeffer, W. M. (1974, November-December). Optimal reproductive effort in fluctuating environments. The American Naturalist, 108, 783-790.

Scrimshaw, S. C. M. (1984). Infanticide in human populations: Societal and individual concerns. In Hausfater, G. & Hrdy, S. B.(Eds.) Infanticide: Comparative and evolutionary perspectives (pp. 439-462). New York: Aldine.

Silverman, I. (1990). The r/K theory of human individual differences: Scientific and Social Issues, Ethology and Sociobiology, 11, 1-9.

South, S. J. (1993). Racial and ethnic differences in the desire to marry. Journal of Marriage and the Family, 55, 357-370.

Stearns, S. C. (1976). Life-history tactics: A review of the ideas: The Quarterly Review of Biology, 51, 3-47.

Stearns, S. C. (1977). The evolution of life-history traits: A critique of the theory and a review of the data. In R. F. Johnston (Ed.). Annual review of ecology and systematics (Vol. 8, pp. 145-169).

Subramanian, S. (1995). The story in our genes: a landmark global study flattens The Bell Curve, proving that racial differences are only skin deep. Time, 145, January 16, 54-55.

Symons, D. (1979). The evolution of human sexuality. New York: Oxford University Press.

Torroni, A., Semino, O., Scozzari, R., Sirugo, G., Spedini, G., Abbas, N., Fellous, M., & Santachiara-Benerecetti, A.S. (1990). Y chromosome DNA polymorphisms in human population: differences between Caucasoids and Africans detected by 49a and 49f probes. Annals of Human Genetics, 54, 287-296.

Waldman, I. D., Weinberg, R. A. & Scarr, S. (1994). Racial-group differences in IQ in the Minnesota transracial adoption study: a reply to Levin and Lynn. Intelligence, 19, 29-44.

Weizmann, F., Wiener, N. I., Wiesenthal, D. L., & Ziegler, M. (1990). Differential K theory and racial hierarchies. Canadian Psychology, 32: 1-13.

Wilson, J. Q. & Herrnstein, R. J. (1985). Crime and human nature. New York: Simon & Schuster.

Wynter, L. (1994). Business and Race. The Wall Street Journal. June 15, B1.