In considering human adolescence in the context of evolutionary fitness, we considered key traits that were available from data collected in 1967-1988. These include the age of menarche, the pattern and timing of the adolescent growth spurt, the body size, and the determinants of reproductive fitness, namely the age of the first reproduction, the IBI, the number of progeny, and mortality risk . The database contains no information on birth size, which may influence the age at menarche. Extended growth and large body size in humans prompt fertility gains and reduced offspring mortality . Consequently there is a pressure for delayed reproductive onset, whereas early reproduction minimizes the likelihood of death before reproduction.
Several limitations of this approach require consideration. i. By nature of this approach, comparing population level averages and attributing differences to ecological differences may not be straightforward. It assumes that the relationships observed would be the same if examined at an individual level . In their report, Greenland and Robin construct several epidemiological examples that show that data at the ecologic level can be misinterpreted if nonlinear effects at the individual level, and other confounds, are not accounted for. Such data for the individual level were not available to us. ii. Comparative studies across societies may also suffer from problems of phylogenetic non-independence. This was addressed for these data previously by adjusting for geographical location (Africa, South America, Australia and Southeast Asia), but the effect was very weak and not significant in any of the multiple regressions .
Pubertal timing is not a univocal parameter; secular changes may occur differently for some pubertal events. Whereas the age of thelarche has reduced in the last decades, changes in menarche have been more subtle and mental maturation - deferred. Also, it is not clear that menarche is a good proxy for fecundity. The age at first reproduction in the present study strongly correlated with menarche, but the potential role of anovulatory menstrual cycles after menarche and potential non-linearity of this trend remain options. For example, it is possible that girls who differ on age at menarche may be similar with regard to age of fecundity; earlier menarche may relate to a greater number of anovulatory cycles , and indeed, regular (ovulatory) cycling was shown to follow a secular trend towards delay as opposed to early menarche .
The data show that at natural fertility, early adolescence in girls, as assessed by the age of menarche and the growth pattern, corresponds to a young age with the first reproduction occurring about four years later, as previously suggested . However, based on our analyses, we reject the two working hypotheses: reproductive fitness is enhanced in societies with early puberty  and early menarche is an adaptive response to greater mortality risk .
We focus here on subsistence-based societies because most resources are invested as somatic capital in terms of growth, body size and fertility (reproductive fitness), as opposed to stored and inherited wealth . The relative contribution of biological and behavioral factors in determining natural fertility change with the environment. Environmental factors to consider in an inter-population study include the physical environment (e.g., population density), the biological environment (e.g., food availability, disease, and other mortality risks) and social behaviors (e.g., age at marriage) . We defined reproductive fitness as a function of TFR and L15, as previously suggested ; these two variables were selected among other because information on these two parameters was available for almost all of the societies in the database. The age at menarche did not correlate with TFR or reproductive fitness. Whereas reproduction starts early in societies in which puberty occurs early, in the context of high population density , their reproductive fitness does not increase. The dwellers of the neotropical forests have a high TFR, but given their mortality risk, they have comparable reproductive fitness to the other ecology groups.
We confirm a previous assertion for greater reproductive fitness among heavier, better-nourished traditional societies [9, 24]. When considered as a whole, we found that the average adult body weight, but not height, correlated negatively with age at menarche and the age at first reproduction, and positively with reproductive fitness. The BMI may not work as well in the extremes of size; in very small or very tall populations the BMI is not as accurate as it is in average size populations. Even though, these findings provide indirect support to the hypothesis that early puberty among girls who live in affluent and developed countries is a response to a positive energy balance. Indeed, among contemporary girls in developing countries, the age at menarche among the prosperous is earlier than that of the underpriviledged .
Based on the high mortality rates of the Philippine pygmy, Migliano et al suggested that early fertility is part of the "fast" extreme of life history strategies to which the pygmy adapt , with both longevity and resource availability as limiting factors . Indeed, early life stress is associated with premature juvenility and adolescence [1, 12, 26]. The results of the present study do not confirm the fast life history theory; we found no correlation between adult height and the age of menarche with survivorship. Yet, population density correlated with the age of first reproduction, in addition to our previous assertion that body size in a traditional society was dependent upon the population density . We have previously suggested that population density acts through two pathways - nutritional constraints and juvenile mortality - at varying intensities, and can contribute to a nearly twofold range in body size across human societies . The sample of the present study includes two African pygmy groups - the Baka (West pygmy) and the Efe (East pygmy), both of whom are not consistent with the Migliano risk/early fertility model. The average age of menarche and age at first reproduction in these two societies, 14.5 and 15.5, respectively 18.5 and 19, respectively, were close to those of all of the other studied societies .
The secular trend for an early age of menarche has been rapid over the past 130 years in developed countries . This trend is rightly interpreted as a reflection of improved nutrition and health in childhood . Life history theory postulates tradeoffs of current versus future reproduction and fertility versus mortality risk. Life-history modeling predicted that a reduction in juvenile mortality reduces the age of menarche  or that low survivorship accelerates the life history . Given the close interaction between resource availability and reproduction, we anticipated that those environmental factors that determine late metabolic homeostasis, attainment of adult size and cessation of growth would interface with those influencing the timing of sexual maturation. The data do not support these predictions. Whereas body mass as a measure of resources is tightly associated with fitness in these traditional societies, the age of menarche is not. Thus, it may be that women's physiology tracks its own condition in such a way as to maximize their individual fitness.