Sex ratios at birth in African populations: A review of survey data

Human Biology, Dec 2002 by Garenne, Michel

In South Africa, sex ratios also tend to be low. For instance, in the Agincourt population laboratory (Tollman et al. 1999), the sex ratio of all births recorded prospectively from 1992 to 2000 was 0.997 (95% CI, 0.964-1.030). The average sex ratio of three sample surveys, the October Household Surveys of 1995, 1997, and 1998, was 0.995, based on 235,063 births (95% CI, 0.987-1.003).

Comparison with Registered Births. If birth registration remains very low in sub-Saharan Africa, there is no reason to suspect that completeness varies with sex. In some cities virtually all births are registered, whereas in rural areas few are, or are not counted or not published. Among the few published data a similar range of variation was found for the sex ratio, from 0.881 in Rwanda and 0.943 in Angola, to 1.128 in Togo and 1.154 in Cote d'Ivoire (Table 2). More important, the same opposition was found between populations of Bantu origin and others. In the first group (Bantu), the average sex ratio at birth was lower than average (0.973), and in the other it was higher (1.052). These findings validate the findings from the sample surveys, and underline the fact that there is no evidence of differential completeness by sex in African vital registration.

Discussion

As already noted in other instances, the sex ratio at birth in sub-Saharan African populations appears on the average lower than that of European and Asian populations. The mean value found in the DHS surveys (1.033) is consistent with other estimates for African populations: 1.033 in the United States (Khoury et al. 1984) and 1.043 to 1.027 in the United Kingdom (James 1984).

Furthermore, this preliminary investigation shows that sex ratios in sub-Saharan Africa are at least as variable as elsewhere, if not more. The range of variation seems to go from below 1.00 to above 1.08, with possible values lower than 1 in some countries of Southern Africa. In particular, Bantu populations seem to have lower sex ratios (possibly around or below 1.00), whereas West African populations seem to have average sex ratios (close to 1.04). Some other populations, such as Nigeria and Ethiopia, could have sex ratios as high as 1.08 or higher.

In the sample of surveys investigated, three main patterns were identified. However, the sample of 1.13 million births remains limited compared to the estimated 745 million births which occurred in sub-Saharan Africa from 1950 to 1999 (United Nations Population Projections, 1998 Revision). It is therefore likely that many other patterns are prevalent in national populations of Africa, not counting the possible local variations by ethnic group or specific areas.

It is beyond the scope of this paper to try to explain the low value of the sex ratio of Bantu populations and the variations found in Africa. However, recent modeling offers an opportunity to propose hypotheses. Kumm et al. (1995) show that in the absence of mortality differentials between sexes, evolution leads to a long-term equilibrium reached for a balanced sex ratio at birth and a balanced sex ratio in the adult population. In case of mortality differentials, a population may evolve in diverging directions, depending on whether excess mortality of one sex leads to replacement fertility or not. The argument proposed by Kumm and colleagues is directed towards behavioral factors (discrimination against females in particular), which tend to be unstable over time and across cultures, but it can be used in a similar way for more stable biological factors accounting for mortality differentials, in particular for infectious diseases. Demographic evidence in high mortality populations of the 19th and 20th century shows an overwhelming excess male mortality in infancy and lower probability of survival to adulthood for males. In natural fertility situations, that is, those without family limitation, excess male mortality in infancy should lead to replacement through the straightforward mechanics of the birth intervals, therefore tending to increase the sex ratio in the long run, according to the model developed by Kumm and colleagues. This could explain why in European populations, from which the mortality data are derived, the sex ratios evolved to be somewhat higher than 1. The same argument in a situation of prolonged excess female mortality could also have led to sex ratios at birth lower than one. In terms of the long-term evolution of humans, one could argue that earliest African populations, to which the Bantus are probably the closest, could have evolved earlier towards a natural equilibrium of a balanced sex ratio at birth adapted to the local disease situation. European and Asian populations could have diverged from this model while being exposed to other diseases, especially in high density populations of Far Eastern Asia, Western Europe, and the Mediterranean area. The time scale of 2000 generations proposed by Kumm and colleagues is that of Homo Sapiens-Sapiens, so that the pressure of evolutionary forces could have had a full impact on contemporary populations. Of course, this hypothesis is speculative at this point, since we know virtually nothing about the pattern of diseases in prehistoric times. Furthermore, it is also possible that evolutionary forces started to have an impact in the Neolithic period (about 360 generations), and that we are still in the fluctuations shown in the early phases of Kumm and colleagues' model. In fact, we do not know whether the situation of high sex ratios in Eastern Asia, the Mediterranean area, or Nigeria and Ethiopia have been stable for a long time, or whether they have changed over the past centuries. On the contrary, if the cultural hypothesis holds, emphasizing sex-selective infanticide and child neglect leading to higher female mortality, selection could also produce an increase in the sex ratio over time assuming no replacement (sex-selective abortion is a very recent phenomenon). In this respect, it is striking to note that areas with higher population densities (Eastern Asia and Nigeria), in which population pressure is higher and motivation for infanticide could therefore be higher, also have higher sex ratios. These observations call for more research in this complex area.