Secular movements in sex ratios of adults and of births in populations during the past half-century

William H. James

The Galton Laboratory, University College London, Wolfson House, 4 Stephenson Way, London NW12 HE, UK


    Abstract
 Top
 Abstract
 Introduction
 Study 1. The adult...
 Study 2. Maternal age-specific...
 Study 3. Sex ratios...
 Further research
 References
 
There is some evidence for a small overall negative correlation across populations between sex ratio (proportion male) at birth and adult sex ratio. There seems to be no systematic correlation within populations across time of sex ratio at birth with adult sex ratio during the past 50 years. So even if adult sex ratios play some part in determining the overall level of the sex ratio at birth, they apparently have played little role in the recent widespread secular changes in sex ratio at birth. It is shown here that there is a strong cohort effect in adult sex ratios: if a woman is in a marriage squeeze (i.e. in a cohort with a relative abundance of women) at age 15 years, she will remain in such a squeeze for the rest of her reproductive life. In England and Wales, the maternal age-specific sex ratios at birth moved roughly in parallel across time during the years 1950–1995. This suggests that sex ratio at birth is not a cohort phenomenon (as it would be if it were affected by adult sex ratio) but is subject to some agents which change with time and affect women (parents) of all ages roughly equally.

Key words: adult sex ratios/cross-country comparisons/secular trends/sex ratios at birth


    Introduction
 Top
 Abstract
 Introduction
 Study 1. The adult...
 Study 2. Maternal age-specific...
 Study 3. Sex ratios...
 Further research
 References
 
It is well established that population sex ratios at birth move up and down slowly across time (Gini, 1955Go). Though statistically significant as judged by trend tests, these movements remain unexplained. However, they have recently become the subject of considerable interest. Thus concern has been expressed at the recent significant declines in sex ratios at birth in many countries, e.g. Canada 1970–1990 (Allan et al., 1997Go), Germany 1950–1960 (Bromen and Jockel, 1997Go), England and Wales 1970–1990 (Dickinson and Parker 1996Go), Japan since 1966 (Minakami and Sato, 1998Go), Denmark since 1950 (Moller, 1996Go), The Netherlands since 1950 (Van der Pal-de Bruin et al., 1997Go), Sweden, Norway and Finland (Moller, 1998Go) and several Latin-American countries since the 1970s (Feitosa and Krieger, 1992Go). Apprehension was expressed by some of these authors that such declines might have been due to environmental endocrine disruptors. However, declines were not universal: in particular, sex ratios increased in Australia 1921–1995 (Lancaster and Day, 1998Go) and Italy slightly since 1960 (Ulizzi and Zonta, 1995Go).

Parazzini et al. (1998) reviewed the data worldwide and confirmed many of the above findings. In addition they noted decreases since the 1950s in Greece, Hungary, Poland, Rumania, Portugal and Mexico; and increases in France, Ireland, Spain and New Zealand. Lastly, in the USA 1969–1995, the live birth sex ratio increased in black births and decreased in white births (Marcus et al., 1998Go, 1999Go). Thus, though decreases have been more common during the past half-century, there have been some increases. In this note I shall consider whether these movements in sex ratio at birth may be (i) related to movements in adult sex ratios, and/or (ii) a cohort phenomenon, and/or (iii) the result of external environmental agents (e.g. some form of pollution). These three hypotheses are not mutually exclusive. Three studies will be described.


    Study 1. The adult sex ratio
 Top
 Abstract
 Introduction
 Study 1. The adult...
 Study 2. Maternal age-specific...
 Study 3. Sex ratios...
 Further research
 References
 
Introduction
Perret (1990) reported that female mouse lemurs mated after being housed with other females produced litters with an excess of male offspring: in contrast, those mated after being housed singly produced litters with an excess of female offspring. The difference between these two sex ratios was highly significant (P < 0.00002). A rather similar phenomenon has been reported in human beings. Polygynous women housed together in a harem reportedly produce more sons whereas those housed singly produce more daughters: accordingly I suggested that mammalian offspring sex ratios may be under some form of stabilizing homeostatic control (James, 1995Go). Adaptive argument [summarized by Lummaa et al. (1998)] tends to the same conclusion, and these authors cite evidence of adaptive variation in mammals, including humans. They reported the sex ratio at birth and the `operational' sex ratio [i.e. of reproductive-aged adults (15–50 years)] in 21 Finnish parishes 1775–1850. They found a significant negative correlation such that in the majority of the parishes more sons were produced when adult males were rarer than adult females. These authors wrote: `This suggests that humans adjusted the sex ratio of their offspring in response to the local operational sex ratio to maximize the reproductive success of their progeny' (because a biased operational sex ratio is expected to lead to lower mean fecundity in individuals of the more abundant sex). It seemed worth testing this finding more generally by examining the relationship of adult (operational) sex ratios with sex ratios at birth in populations across the world by correlation analysis.

Material
Most countries publish their numbers of births by sex in annual volumes of vital statistics. In addition, most countries hold censuses of their populations and these are enumerated by sex and age every decade. These two sorts of data are reproduced in the annual Demographic Yearbooks of the United Nations. In regard to data from the USA, the numbers of black residents by age and sex at each census date since 1930 have been estimated (Preston et al., 1998Go). I am grateful to Professor Preston (of the University of Pennsylvania) for advising me that these estimates may be subtracted from official totals to derive reasonable estimates of comparable white adult sex ratios at those time points. The numbers of births by sex and race in the USA are published in the annual Natality volumes of the US Vital Statistics.

Methods
It is hypothesized that adult sex ratios somehow exert a stabilizing homeostatic control over offspring sex ratios with the ultimate result of keeping the adult sex ratio close to an optimum value (i.e. around 1:1 male:female). If this hypothesis were true, then the movements across time of these two parameters should, in general, show a negative correlation with one another. It is not required to estimate that correlation in any given population, but simply to test whether it is negative in the majority of populations. Since censuses occur every 10 years in many countries, it was convenient to choose five time points, i.e. one for each of the five census times in the past half-century (i.e. since the perturbations in sex ratio at birth associated with the Second World War). For this purpose Spearman's rank correlation coefficient, {rho}, was calculated for each of 25 of the 29 populations (Parazzini et al., 1998Go); and for eight Latin-American populations (Argentine, Chile, Costa Rica, El Salvador, Guatemala, Panama, Puerto Rico and Venezuela); for five former British colonies (Bahamas, Barbados, Bermuda, Hong Kong, and Mauritius) and for Cape Verde Islands, Guadeloupe and Iceland. In accordance with demographic usage, the adult sex ratio was taken as the number of resident males per hundred resident females. The age range was arbitrarily chosen as 15–44 years.

Moreover, if the association between sex ratios in adulthood and at birth were causal, then (depending on the causal nexus) one might wonder whether those populations with high mean adult sex ratios would have low mean offspring sex ratios and vice versa. In other words, one would expect a negative correlation not simply within populations but across populations. For this purpose, mean values of the two parameters were calculated and subjected to correlation analysis in respect of 23 (mainly) Caucasian populations [i.e. Austria, Belgium, Bulgaria, Denmark, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Netherlands, Norway, Poland, Portugal, Spain, Sweden, Switzerland, England and Wales, Canada, United States (whites), Australia, New Zealand]. The reason for not including the other populations in this analysis is that sex ratio at birth varies by race (James, 1987Go).

Results
Correlations within populations
Of the 41 correlation coefficients, 23 were negative and 18 positive. Table IGo gives the distribution of these correlations. Many of them were statistically significant. However, the lack of consistency of sign (+ or –) of the correlations is powerful evidence that the temporal trends in the two parameters are (largely) causally unrelated.


View this table:
[in this window]
[in a new window]
 
Table I. Spearman correlations between sex ratios at birth and of adults 1950–1997, selected countries
 
Correlation across populations
Spearman's {rho} for the 23 pairs of mean values was –0.393 (t = 1.96 with 21 df, P = 0.064 two-way).

Comment
Thus one may conclude that if secular movements in the adult sex ratio have any effect on the secular movements in the sex ratio at birth, this effect is very small in large populations. In general, it seems that the well-established secular variations in sex ratio at birth must have some other cause. There are one or two caveats to this generalization. For instance one may wonder whether changes in adult sex ratios played any part in the (very small but persistent) rise in sex ratio at birth in Australia during this century; and in the curious rise and fall respectively in the sex ratios at birth of black and white births in the USA across the years 1969–1995. These examples are not chosen randomly: they all occurred in the presence of movements of the adult sex ratio in the opposite direction. So they are consistent with the hypothesis that movements in sex ratios at birth are partially controlled by movements in adult sex ratios. To test this hypothesis in regard to these three populations, the data need to be submitted to analysis of a kind advanced in the following section. This will suggest that secular changes in the sex ratio of adults played little part in the changes in sex ratio at birth in England and Wales since 1950.


    Study 2. Maternal age-specific changes in sex ratios at birth and in adulthood in England and Wales since 1950
 Top
 Abstract
 Introduction
 Study 1. The adult...
 Study 2. Maternal age-specific...
 Study 3. Sex ratios...
 Further research
 References
 
Introduction
It is useful to contrast movements of maternal-age-specific sex ratios at birth with movements of corresponding age-specific adult sex ratios. This is so because (as will be seen) the movements of adult age-specific sex ratios are substantial and show a strong cohort effect. In other words, if a woman is in what demographers call a `marriage squeeze' at age 15 years, she is likely to remain in a marriage squeeze for the rest of her reproductive life (if she does not emigrate). A marriage squeeze exists where the available marriage partners of one sex outnumber the available marriage partners of the other sex. The more numerous sex is said to be in a squeeze. If sex ratios in adulthood played a role in correcting such squeezes (by exerting homeostatic control over sex ratios at birth) then there should be a cohort effect in sex ratios at birth: women in a squeezed cohort should, ex hypothesi, have a tendency to produce boys. In contrast, if some sort of (external) environmental agent(s) were responsible for the widespread secular declines in sex ratio at birth, then the maternal-age-specific sex ratios should move in parallel across time. (It is assumed that such agents operate at the time of conception, regardless of maternal age.) Figure 1Go gives the quinquennial movements of the maternal-age-specific live birth sex ratios (proportions male) in England and Wales 1950–1997.



View larger version (19K):
[in this window]
[in a new window]
 
Figure 1. Five yearly maternal age-specific sex ratios at birth in England and Wales 1950–1997.

 
Figure 2Go shows the annual movements of the corresponding adult sex ratios, here taken as



View larger version (33K):
[in this window]
[in a new window]
 
Figure 2. Adult age-specific sex ratios in England and Wales, 1950–1997, offset by 5 year intervals. Displaying the data in this fashion demonstrates the permanent nature of the ratio within a given cohort. Data are not available for the years 1974, 1978, 1979, 1986 and 1987.

 
no. of resident men aged (x + 5) to (x + 9)/

no. of resident women aged x to (x + 4),

where x = 15, 20 . . . 35. (The ages of men and women are different to take account of the mean age difference at marriage.) Table IIGo gives the age-specific sex ratios arranged by cohort, the unweighted means of these, and the unweighted means of the corresponding adult sex ratios.


View this table:
[in this window]
[in a new window]
 
Table II. Maternal 5 year age-specific live birth sex ratios (proportions male) and adult sex ratios (males per 1000 females) arranged by cohort, England and Wales 1950–1979
 
Results
It will be seen that there is no persuasive (negative) correspondence between the movements of the sex ratios at birth and the adult sex ratios. Both show trends that are statistically significant – but these trends are apparently independent. The movements of the 5 age-specific adult sex ratios show a remarkable similarity when offset by 5 years (Figure 2Go). In other words, the adult sex ratio of a cohort of adults in the reproductive years remains stable throughout those years.

In contrast, Figure 1Go suggests that secular movements of the age-specific sex ratios at birth are (very roughly) in parallel. Table IIGo also lends weight to the suggestion that any cohort effect in sex ratio at birth must be very small. For instance, of the maternal cohorts considered, the one that was least squeezed (with the greatest abundance of men of marriageable age) was that aged 15–19 years in 1970–1974. In contrast the following cohort (that aged 15–19 years in 1975–1979) experienced a deficit of marriageable men. Yet the latter cohort produced a higher proportion of daughters than the earlier cohort. These data suggest that the decline in sex ratios was the consequence of a period effect rather than a cohort effect.


    Study 3. Sex ratios at birth in isolated island populations with biased adult sex ratios
 Top
 Abstract
 Introduction
 Study 1. The adult...
 Study 2. Maternal age-specific...
 Study 3. Sex ratios...
 Further research
 References
 
If the adult sex ratio per se had an appreciable power to affect sex ratio at birth, then one would expect biased sex ratios at birth in those island communities where the adults contain excesses of either men or women. The UN Demographic Yearbooks give data on a number of such communities. Table IIIa,bGo shows data from islands selected for imbalance of their numbers of adult men and women. There seems little evidence in these data for any postulated effect of adult sex ratio on sex ratio at birth.


View this table:
[in this window]
[in a new window]
 
Table III. Sexes at birth in isolated island communities with excesses of one sex or the other
 
Overall comment
This paper presents evidence that adult population sex ratios play little role in determining population sex ratios at birth. Other forms of analysis are needed to explore the hypothesis further, e.g. (i) more elaborate multivariate methods might tease out a relationship, and (ii) it seems likely that if sex ratio at birth were affected by parental perceptions of adult sex ratio, then these perceptions would be based on local–rather than national–conditions. Therefore the hypothesis should be better tested by locally based data (like those of Lummaa et al., 1998)

One last question arises: if mean sex ratio at birth and mean adult sex ratio are (negatively) correlated, why are they? Is it because some factor affects both (rather than that one directly affects the other)? In this context, one may note that communities differ in their propensity to violence and warfare. War and murder certainly reduce the adult sex ratio. Moreover there can be no reasonable doubt that regardless of the causal details, human aggression is correlated with male and female testosterone concentrations. Archer (1995), for instance, wrote: `Overall the correlation between various measures of human aggressive behaviour and testosterone level is about 0.38'. Now according to my hypothesis, high parental concentrations of testosterone are associated with high sex ratios at birth (James 1996Go). So in principal it is possible to envisage a proximate explanation for such a correlation. If the above reasoning were correct, the surplus males produced at birth in violent societies would die through murder and warfare. The point may be illustrated by the Yanomamo Indians. About 30% of deaths among adult males of this tribe in one region are due to violence: most victims are males, and about 44% of men aged >=25 have killed someone (Chagnon, 1988Go). The sexes of all reported Yanomamo births were 383 males and 295 females (Chagnon et al., 1979Go). This sex ratio, 0.565 (with a standard error of 0.019) is substantially and significantly higher than the sex ratios reported in industrial societies (James, 1987Go). Thus an adaptive explanation would be available for a negative correlation between mean sex ratios at birth and in adulthood. But the mere availability of an adaptive explanation is far from a guarantee of its truth.


    Further research
 Top
 Abstract
 Introduction
 Study 1. The adult...
 Study 2. Maternal age-specific...
 Study 3. Sex ratios...
 Further research
 References
 
More data are needed on the sex ratios of the offspring of polygynous unions. Do such data confirm that domiciliary arrangements (wives housed together or separately) affect offspring sex ratios in the way suggested by the reports cited in James (1995)?

It should be possible to identify further samples in which adult sex ratios are substantially skewed in one direction or the other. For instance, what is the sex ratio of births in feminist communes? Or of births to women prisoners (or men prisoners) following conjugal visits? In this context, it may be wondered whether the well-established rise in sex ratio in wartime is due to a low domestic adult sex ratio (men being engaged in war elsewhere). But that phenomenon is susceptible to other interpretations and seems not a useful topic for further speculation here.

If it were correct that offspring sex ratios are affected by the sexes of the adults with whom the mothers cohabit around the time of conception, then one might expect that among children born outside marriage, those whose births are registered by the mother alone would have a different sex ratio from those whose births are registered by both parents (the latter suggesting more stable domiciliary arrangements at the time of conception and subject to a more permanent adult male presence prior to conception).

It would be particularly interesting to see attempts to confirm the finding of Lummaa et al. (1998). Data are presumably available in the reconstructions by historical demographers from parish records of pre-industrial France and Quebec.

The States of the USA have differing adult sex ratios. It would be interesting to see whether this is reflected in their sex ratios at birth.


    References
 Top
 Abstract
 Introduction
 Study 1. The adult...
 Study 2. Maternal age-specific...
 Study 3. Sex ratios...
 Further research
 References
 
Allan, B.B., Brandt, R., Seidel, J.E. and Jarrell, J.F. (1997) Declining sex ratios in Canada. Can. Med. Assoc. J., 156, 37–41.[Abstract]

Archer, J. (1995) Testing Mealey's model: the need to demonstrate an ESS and to establish the role of testosterone. Behav. Brain Sci., 18, 541–542.[ISI]

Bromen, K. and Jockel, K.-H. (1997) Change in male proportion among newborn infants. Lancet, 349, 804–805.[ISI][Medline]

Chagnon, N.A. (1988) Life histories, blood revenge and warfare in a tribal population. Science, 239, 985–992.[ISI]

Chagnon, N., Flinn, M.V. and Melancon, T.F. (1979) Sex ratio variation among the Yanomamo Indians. In Chagnon, N. and Irons, W. (eds), Evolutionary Biology and Human Social Behavior. Duxbury Press, North Scituate, MA, pp. 290–320.

Dickinson, H.O. and Parker, L. (1996) Why is the sex ratio falling in England & Wales? J. Epidemiol. Commumity Hlth, 50, 227–228.

Feitosa, M.F. and Krieger, H. (1992) Demography of the human sex ratio in some Latin American countries, 1967–1986. Hum. Biol., 64, 523–530.[ISI][Medline]

Gini, C. (1955) Sulla probabilita che x termini di una serie erratica sieno crescenti (o non decrescenti) ovvero tutti decrescenti (o non crescenti) con applicazioni ai rapporti dei sessi nelle umane in intervalli successivi e alle disposizioni dei sessi fratellanze umane. Metron, 17 (3–4), 1–41.

James, W.H. (1987) The human sex ratio. Part 1: a review of the literature. Hum. Biol., 59, 721–752.[ISI][Medline]

James, W.H. (1995) What stabilizes the sex ratio? Ann. Hum. Genet., 59, 243–249.[ISI][Medline]

James, W.H. (1996) Evidence that mammalian sex ratios at birth are partially controlled by parental hormone levels at the time of conception. J. Theor. Biol., 180, 271–286.[ISI][Medline]

Lancaster, P.A.L. and Day, P.L. (1998) Declines in population sex ratios at birth. J. Am. Med. Assoc., 280, 1139–1140.[Free Full Text]

Lummaa, V., Merila, J. and Kause, A. (1998) Adaptive sex ratio variation in pre-industrial human (Homo sapiens) populations? Proc. R. Soc. Lond., B265, 563–568.[ISI][Medline]

Marcus, M., Kiely, J., Xu, F. et al. (1998) Changing sex ratios in the United States. Fertil. Steril., 70, 270–273.[ISI][Medline]

Marcus, M., Kiely, J., Marcus, M. et al. (1999) Reply. Fertil. Steril., 71, 969–970.[ISI]

Minakami, H. and Sato, I. (1998) Change in male proportion of newborn infants in Japan. Am. J. Obstet. Gynecol., 178, 624.

Moller, H. (1996) Change in male:female ratio among newborn infants in Denmark. Lancet, 348, 828–829.[Medline]

Moller, H. (1998) Trends and sex ratio, testicular cancer and male reproductive hazards: are they connected? AMPIS, 106, 232–239.

Parazzini, F., La Vecchia, C., Levi, F. and Franceschi, S. (1998) Trends in male:female ratio among newborn infants in 29 countries from five continents. Hum. Reprod., 13, 1394–1396.[Abstract]

Perret, M. (1990) Influence of social factors on sex ratio at birth, maternal investment and young survival in a prosimian primate. Behav. Ecol. Sociobiol., 27, 447–454.[ISI]

Preston, S.H., Elo, I.T., Foster, A. and Fu, H. (1998) Reconstructing the size of the African American population by age and sex 1930–90. Demography, 35, 1–21.[ISI][Medline]

Ulizzi, L. and Zonta, L.A. (1995) Factors affecting the sex ratio in humans: multivariate analysis of the Italian population. Hum. Biol., 67, 59–67.[ISI][Medline]

Van Der Pal-de Bruin, K.M., Verloove-Vanhorick, S.P. and Roeleveld, N. (1997) Change in male:female ratio among newborn babies in Netherlands. Lancet, 349, 62.[Medline]

Submitted on October 12, 1999; accepted on February 9, 2000.