Reproductive life histories: can incremental dentine isotope analysis identify pubertal growth, pregnancy and lactation?

Abstract

Background

There are few reliable osteological indicators to detect parity or infer puberty in skeletal remains. Nitrogen (δ¹⁵N) and stable carbon (δ¹³C) isotope ratios in human tissues can be affected by metabolically unbalanced states engendered by pregnancy or rapid growth, offering potential biomarkers.

Aim

This pilot study explores the potential of incremental dentine-collagen isotope ratio analysis to identify puberty and gestation.

Subjects and methods

Incremental dentine δ¹⁵N and δ¹³C profiles were produced by analysing third molars extracted as part of dental treatment of 10 individuals living in Sudan. Demographic and anthropometric data at the time of tooth extraction was available. Medical histories were unknown.

Results

Isotopic signatures potentially related to pubertal growth, with an average δ¹⁵N reduction of 0.78 ± 0.29‰, are indicated. Six isotopic signals suggestive of pregnancy, with an average δ¹⁵N decrease of 0.48 ± 0.22‰, are also observed. The timing, speed and amplitude of post-partum δ¹⁵N patterns seemingly infer infant feeding practices and maternal nutritional status.

Conclusion

This pilot study highlights the potential of incremental dentine isotope analysis for the reconstruction of early reproductive histories in skeletal remains. However, controlled studies with a larger human cohort are needed to validate these findings, establish isotopic signals linked to puberty and lactation, and improve chronology accuracy.

Keywords:

• Nitrogen isotopes
• dentine
• puberty
• growth
• pregnancy

Introduction

Studying variables relating to reproduction has the potential to make important contributions to our understanding of societies, past and present, in terms of health, nutrition, physiological stress and culture (Sperling and Beyene 1997, p. 144). Indeed, while the age and rate at which individuals go through puberty and the length of their reproductive period are governed by established physiological factors, wide inter-individual variations linked to nutrition, growth, in utero stress and biological sex are seen (Soliman et al. 2014; English et al. 2020). In females, age at first gestation (AFG) is influenced by age at menarche -i.e. first occurrence of menstruation signalling the onset of fertility- as well as social norms and education (Marchetta and Sahn 2015). Fertility is similarly highly variable, being affected by malnutrition and cultural practices like deliberate birth spacing or infant feeding, with prolonged lactation creating a longer ovulation interval (Delgado et al. 1982). Thus, wide disparities in reproduction patterns are observed between and within societies depending on subsistence practices, food availability and cultural practices. In turn, gestation has a significant impact on the physical health, nutritional status and mortality of both mothers and offspring. Pregnancy and lactation cause a substantial increase in maternal nutritional demand, thus gestations occurring at high frequency or in challenging environmental conditions increase the risk of malnutrition and morbidity, with subsequent repercussions on population growth and health (Black et al. 2008; Victora et al. 2008).

However, these significant aspects of life are a neglected area of study for past societies, especially archaeological populations, due to the challenge of uncovering information related to puberty or parity. For past populations, few references in early historical texts exist (Fildes 1986) and the few existing skeletal indicators for pregnancy are mostly unsatisfactory.

Osteological indicators for parity include ‘pubic tubercule extension’ (PTE), which results from the increased stimulation of abdominal muscles during the later stage of pregnancy and childbirth (Cox 2000). PTE is positively associated with the number of obstetric events but does not inform on fertility, AFG or inter-birth interval (IBI) (Cox 2000). Pre-auricular sulcus and pubic pitting are often referred to as ‘parturition scars’, yet these indicators are not specific to parturition since they have multiple causes, including hormonal and physical stress on pelvic articulation experienced during childbirth (Cox 2000). In addition, the skeletal remains of women who gave birth often lack both these markers. ‘Parturition lines’, based on the recording and subsequent preservation of microstructural alterations associated with gestational metabolic changes in developing teeth, can be observed in histological dentine sections and may indicate AFG and IBI if these events occur before tooth development is completed by about 23 years (Dean and Elamin 2014). However, parturition lines do not always form during pregnancy, they can result from other stress events, and their estimated timing is inaccurate.

Puberty encompasses sexual and physical maturation, including the development of secondary sex characteristics (e.g. breast, genitals and pubic hair) and a peak growth period (Viner 2008). Puberty consists of a clear sequence of stages over time, affecting endocrine, skeletal, muscular and reproductive systems, with puberty completion taken to indicate that an individual has achieved its adult body stature and is able to reproduce (Dorn and Biro 2011). James Tanner introduced the concept of pubertal status being linked to growth patterns, as summarised by the ‘tempo of growth’ (i.e. the timing and intensity of peak growth velocity) (Cole 2020). Puberty onset is marked by growth acceleration until ‘peak height velocity’ (PHV) is reached (Marshall and Tanner 1986). As growth starts to slow down, girls experience menarche and boys develop their full adult voice (Hägg and Taranger 1982). The growth spurt concludes with the fusion of long-bone epiphyses and pelvis maturation (Marshall and Tanner 1986). The physical transition into adulthood denoted by puberty may not correspond with mental and cultural maturation processes (Arnett 2007). Thus, individuals starting puberty early may appear older, and late developers may appear younger, creating a dichotomy between physical and psychosocial identities. In this paper, we refer to the transition between childhood and adulthood as adolescence, a period marked by the onset of puberty and the ability to function as an independent adult (Lewis et al. 2016).

In contemporary populations, developmental age is generally established using Tanner scales to assess secondary sexual characteristics, an intrusive and often inaccurate process (Dorn and Biro 2011); a reason proposed by English et al. (2020) to explain boys’ underrepresentation in studies on the timing of puberty in humans. Indeed, in girls, menarche provides an alternative and precise measure, a set point in time that can be self-reported accurately. Physical maturation is also established by generating individual linear growth velocity curves using longitudinal data and extracting the timing of growth landmarks like age at PHV (Cole 2020). However, these two methods are only applicable to living individuals. A third method, assessing the developmental stage of bone fusion or teeth formation using x-ray, is restricted to medical investigations of growth disorders in living individuals but has been adapted for use with archaeological skeletal remains to help establish the age when important stages of pubertal development were reached in past populations (Shapland and Lewis 2013). Nevertheless, while osteological assessment helps establish the timing of growth and therefore puberty, in archaeological populations, the method relies on studying individuals who did not survive into adulthood, and thus may not be representative of the population; and accuracy depends on skeletal preservation (Henderson and Padez 2017).

Stable light isotope analysis may offer an alternative biomarker to help identify both the pubertal growth spurt and pregnancy in skeletal remains. Indeed, δ¹⁵N of body tissues primarily reflects the diet of an individual but is also affected by their metabolic status, with anabolic states like growth and pregnancy generally resulting in reduced δ¹⁵N (Fuller et al. 2004; Waters-Rist and Katzenberg 2010; Reitsema 2013). Isotopic variations linked to growth are suggested in animal studies, yet they have not yet been investigated in incremental tissues of modern living human beings (e.g. Trueman et al. 2005; Warinner and Tuross 2010). The δ¹⁵N reduction associated with tissue accretion means that incremental dentine-collagen isotope ratio analysis (IDIA) could be a powerful tool to investigate the pubertal growth spurt -potentially establishing the ‘tempo of growth’-. The close relationship between growth and sexual development means that IDIA could also indirectly aid the study of sexual maturation in skeletal remains.

Isotopic shifts linked to pregnancy have been successfully measured in hair (Fuller et al. 2004; D'Ortenzio et al. 2015). IDIA achieves a lower temporal resolution than hair analysis, but teeth are much more commonly preserved in archaeological deposits and allow retrospective data collection beyond the last few years before death or sampling (Wilson and Gilbert 2007, p. 156; Henderson et al. 2014). Like hair, dentine in teeth contains a significant amount of protein, forms sequentially, and does not remodel (Eerkens and Bartelink 2013). Teeth are also highly resistant to taphonomic damage, and third molars (M₃) typically form between 8 and 23 years of age (AlQahtani 2009), making them a viable alternative tissue to study reproduction patterns -i.e. sexual maturation including the start of fertility, and potential subsequent pregnancies- during late adolescence and early adulthood.

This study, therefore, aims to explore the potential of IDIA to detect short-term isotopic changes from anabolic physiological events linked to either puberty or pregnancy and reconstruct early reproductive life. We do this by analysing M₃ teeth from ten contemporary individuals for whom some basic demographic, anthropometric and geo-political information is available; including eight women suspected of having experienced pregnancy during M₃ formation, although medical and obstetric histories were unknown.

Nitrogen and carbon stable isotopes

Elements naturally exist in various isotopic forms with marginally different atomic masses, which affect the rate at which each isotope enters chemical reactions (Van der Merwe 1982). Thus, as elements go through metabolic processes, the relative abundance of their isotopes changes, a phenomenon known as fractionation or discrimination (Phillips 2012). This means that the isotope ratio of an individual reflects the isotopic composition of their diet plus a fractionation factor (Phillips 2012). This principle has enabled the reconstruction of dietary histories, environments and the identification of metabolic disturbances from nutritional, pathological or physiological events, using isotope ratios.

Isotopes as biomarkers of diet

The isotopic values of body proteins like dentine-collagen predominantly reflect the isotopic composition of the protein component of the diet at the time the tissue is formed (Ambrose and Norr 1993; Jim et al. 2006). Dietary carbohydrates and lipids have a lesser influence on δ¹³C values of body proteins, contributing up to 25% of the carbon skeleton used in the de novo synthesis of non-essential amino acids (Martinez del Rio et al. 2009; Fernandes et al. 2012). The main factor influencing δ¹⁵N values of foods is their place in the food chain, with an average stepwise δ¹⁵N increase of 3.6% between each trophic level (Fuller et al. 2012). Thus, δ¹⁵N values of body proteins can help differentiate plant-based and animal-based protein intake (Petzke et al. 2005, Figure 1). The δ¹³C change with each trophic level is less pronounced, with average discrimination between diet and muscles of 0.8 ± 1‰ (Chisholm et al. 1982; Newsome et al. 2010). Still, combining δ¹⁵N and δ¹³C can also help distinguish marine resources subsistence from terrestrial-based diets since aquatic food chains are longer (Grupe et al. 2009, Figure 1). However, δ¹³C is mostly used to differentiate plants with different photosynthetic pathways at the base of the human food chain (Wright and Schwarcz 1999, Figure 1). The majority of economically important plants like wheat, rice and tubers are C₃-plants, which have δ¹³C values of −26.5 ± 4‰ (Van der Merwe 1982). C₄ plants like maize, millet, sorghum, sugar cane and other tropical grasses, display δ¹³C values of −12 ± 2‰.

Figure 1. Biplot of bulk δ¹⁵N and δ¹³C for the 10 M₃, including values during various consecutive phases of M₃ formation for individuals with indications of major dietary and/or environmental shifts.

Isotopes as environment indicators

Food isotope values are also influenced by their environmental conditions. Generally, arid, acidic, saline or anaerobic waterlogged environments are associated with increased δ¹⁵N and δ¹³C compared with wet colder climates, with particularly elevated δ¹⁵N observed in animals living in salt marshes and coastal ecosystems (Ambrose 1991; Tieszen 1991; Britton et al. 2008, Figure 1). The effect of aridity on δ¹⁵N is intensified when animals become water-stressed since inadequate fluid intake leads to increased excretion of ¹⁵N-depleted urea to help conserve water, resulting in increased δ¹⁵N of body tissues (Ambrose 1991). Severe water stress may also lead to increased δ¹⁵N in humans, although the effect has not been studied.

Isotopes as nutritional and physiological biomarkers

The isotopic composition of body proteins like dentine-collagen is also affected by fractionation associated with protein metabolism. In metabolically stable individuals meeting their nutritional requirements for protein and energy, δ¹⁵N discrimination between diet and collagen tissues (Δδ¹⁵N_{diet-collagen}) is relatively stable at 3–5‰ (Bocherens and Drucker 2003). The higher offset of 6‰ observed in well-nourished individuals in two human-based studies could result from their reported protein intake being significantly higher than their requirements (Hedges et al. 2009; O'Connell et al. 2012). Indeed, a positive correlation between Δδ¹⁵N_{diet-collagen} and protein intake above requirements exists, which may be linked to the increased rate of deamination and ¹⁴N-enriched urea excretion associated with the breakdown of amino acids in excess of requirement (Tomé and Bos 2000; Martinez del Rio et al. 2009). Thus, in metabolically stable individuals, excessive protein intake leads to raised δ¹⁵N, independently of the isotopic composition of the diet (Figure 1).

Isotopes as markers of catabolic states

Catabolic states result from insufficient or poor-quality dietary protein intake; and during catabolism, the rate of breakdown of less vital body proteins like muscles increases to provide amino acids for the synthesis and replacement of more essential body proteins (Mekota et al. 2009). Such ‘recycling’ of body amino acids can be likened to an internal trophic effect and leads to raised δ¹⁵N of body proteins. Inadequate protein intake also results in the increased contribution of dietary carbohydrates and lipids -or even body fat stores when energy intake is insufficient-, to provide carbon for amino acid synthesis (Newsome et al. 2010; Nardoto et al. 2011; Iacumin et al. 2016). Since lipids are generally depleted in ¹³C, inadequate protein intake, particularly when accompanied by energy deficiency, leads to reduced δ¹³C (Tieszen 1991).

Thus, in patients with anorexia or non-disease-related protein-energy malnutrition, catabolic states are associated with δ¹⁵N increase of up to 1.9‰ and δ¹³C decrease of up to 5.4‰ (Hobson et al. 1993; Mekota et al. 2009; Neuberger et al. 2013). The wide variations in the extent and rapidity of δ¹⁵N and δ¹³C shifts during catabolic states may be related to varying diets, the original amount of body fat and protein reserves, and the extent of weight loss and accompanying physiological states like growth, pregnancy or disease (Neuberger et al. 2013). Pathological states like systemic infections, fevers, diarrhoea, cancer or AIDS can have a similar catabolic effect, therefore isotopic responses to such conditions can mimic protein deficiency (Reitsema 2013).

Isotopes as markers of anabolic state

Anabolic states occur when body protein synthesis increases to enable extra tissue deposition during juvenile growth, pregnancy, recovery from starvation, or tissue repairs following trauma (Waters-Rist and Katzenberg 2010; Webb et al. 2015). Anabolic states are characterised by increased efficiency of protein metabolism, including enhanced direct routeing of amino acids towards protein synthesis, thus by-passing processes creating isotope fractionation like deamination and transamination, and reduced ¹⁴N excretion through decreased amino acids oxidation and increased urea salvage (Kalhan 2000). Improved dietary protein utilisation through decreased amino acid oxidation has been demonstrated in association with pubertal growth (Beckett et al. 1997).

Isotopes and growth

An inverse correlation between growth rate and δ¹⁵N enrichment of animal tissues has been shown (e.g. Trueman et al. 2005; Warinner and Tuross 2010; Reitsema and Muir 2015; Webb et al. 2016). An inverse relationship between weight gain and δ¹³C has also been observed in a small-scale controlled feeding animal study, which recorded hair isotope values about 3.9‰ lower for δ¹³C and about 0.9‰ lower for δ¹⁵N in faster-growing pigs (Warinner and Tuross 2010). A similar isotopic response, inversely correlated to growth rate, is expected during infancy or puberty growth spurts in humans, although early work by Waters-Rist and Katzenberg (2010), based on long bones from protohistoric individuals aged 7–19 years, suggests that growth has a minimal influence on the isotopic composition of human bone-collagen. Indeed, the bone turnover rate has since been shown to be too slow to record metabolic shifts in isotopic values previously recorded in hair studies (Nitsch et al. 2010; Beaumont 2020). Certainly, δ¹⁵N reductions, -unexplained by dietary or environmental changes on the basis of a lack of accompanying δ¹³C variations-, are seen in dentine isotopic profiles derived from second or third molars of both Seventeenth century Scottish soldiers and Nineteenth century Irish populations (Beaumont and Montgomery 2016, Figure 2; Millard et al. 2020). The recurrence of the isotopic phenomenon across different historical periods or locations hints at an anabolic physiological event, an interpretation reinforced by the temporal agreement of the isotopic dip with the putative pubertal growth spurt. The observed δ¹⁵N reductions vary in intensity (up to 3.1‰), timing and duration, potentially illustrating the genetically driven variety of growth patterns seen within populations, but also variations introduced by differing nutritional status (Veldhuis et al. 2005).

Figure 2. Incremental dentine δ¹⁵N profiles of Seventeenth century Scottish soldiers and Nineteenth century Irish men from Kilkenny Union workhouse famine cemetery ( adapted from Beaumont and Montgomery 2016; Millard et al. 2020).

The effect of growth on isotope values of incremental tissues has not been established in living human beings, however, an inverse correlation between weight gain and δ¹⁵N values has been noted during pregnancy (Fuller et al. 2004). δ¹⁵N values inversely correlated with weight gain were also noted in patients recovering from anorexia -i.e. underweight individuals, transferring from either a catabolic or steady-state nitrogen balance to an anabolic state, as they received treatment to gain weight- (Mekota et al. 2009). δ¹⁵N reductions of 0.6–2.2‰ were seen by Mekota et al. (2009). Re-instatement of positive energy intake during the starvation recovery also incurred a 1.5–5.7‰ δ¹³C increase, reflecting reduced reliance on adipose tissue to provide carbon skeleton for de novo protein synthesis (Mekota et al. 2009).

Figure 3 presents projections of the nitrogen isotopic changes expected during the pubertal growth spurt, based on normalised linear growth velocity patterns in European children and our understanding of the relationship between growth and δ¹⁵N values (Veldhuis et al. 2005). Based on the assumption that δ¹⁵N variations linked to pubertal growth are inversely proportional to growth rate, then the point when δ¹⁵N begins to decrease is considered to indicate growth spurt initiation, with the resultant lowest δ¹⁵N value highlighting PHV, and the restored δ¹⁵N value denoting growth completion (Figure 3). Such standardised isotopic curves, however, mask non-conformities among individual children in the timing and tempo of pubertal growth, especially for children of different ethnic origins or living in different geographic, socioeconomic or historical environments (Veldhuis et al. 2005).

Figure 3. The exemplar of incremental dentine δ¹⁵N profiles during pubertal growth (based on linear growth patterns defined by Veldhuis et al. 2005).

Isotopes during gestation and breastfeeding

During healthy (human) pregnancies, maternal δ¹⁵N values at birth are on average 0.8‰ (0.3–1.3‰) lower than δ¹⁵N values recorded at conception (Fuller et al. 2004; D'Ortenzio et al. 2015). However, no overall δ¹⁵N changes over the 9 months period have been noted in gestations associated with nutritional stress (Fuller et al. 2005). δ¹³C shifts have not been observed during pregnancy (Fuller et al. 2004, 2005; D'Ortenzio et al. 2015).

The effect of lactation on maternal isotopic composition is not yet fully understood, however, there are indications from both animal and human studies that lactation may engender isotopic changes in mothers (e.g. Kurle 2002; Fuller et al. 2006; Habran et al. 2010; Reitsema 2012). Reitsema and Muir (2015) reported that lactating rhesus macaques exhibit significantly lower δ¹⁵N values (by 0.3‰) and higher δ¹³C values (by 0.4‰) than non-lactating females eating the same diet. One explanation suggests that the excretion of milk leads to lower δ¹⁵N values in maternal tissues because milk is enriched in ¹⁵N (Kurle 2002). However, increased bulk δ¹⁵N values of milk relative to the mother are seen in marine animals (Kurle 2002), but not in humans (de Luca et al. 2012; Romek et al. 2013; Herrscher et al. 2017) or other animals (Jenkins et al. 2001; Reynard and Tuross 2015). In humans, δ¹⁵N values for milk proteins appear to match those of mothers' fingernails (Fuller 2003); instead, breastmilk ¹⁵N-depletion is thought to be linked to its relatively high concentration of ¹⁴N-enriched urea, while the relatively low δ¹³C values come from high levels of ¹³C-depleted fat rather than ¹³C-enriched proteins (Schoeller 1999; Braun et al. 2013). An alternative hypothesis is that, although ultimately lactation does not result in a positive nitrogen balance since the extra tissue produced (milk) is then excreted, lactogenesis still leads to metabolic adaptations akin to anabolic states to conserve nitrogen for milk synthesis (with the possible difference that urea excretion shifts from urine to milk), resulting in lower δ¹⁵N values of maternal tissues, as seen during pregnancy.

Figure 4 shows the modelling of δ¹⁵N variations expected in maternal dentine during pregnancy and breastfeeding, based on our understanding of the effect of those two physiological events on δ¹⁵N values from the limited data available. The projection for the isotopic effect of lactation is based on the practice of exclusive breastfeeding for the first 6 months of life and continuation until the child is 2 years or beyond, in line with recommendations from the World Health Organisation (World Health Organization (WHO) United Nations Children’s Found (UNICEF) 2003).

Figure 4. The exemplar of maternal incremental dentine δ¹⁵N profiles during pregnancy and lactation (based on Fuller et al. 2004; D'Ortenzio et al. 2015; Fuller et al. 2006; Herrscher et al. 2017).

Incremental dentine isotope analysis (IDIA)

Primary dentine grows sequentially, forming successive layers from the dentine-enamel junction in the tooth crown to the root apex (Hillson 1996). Once formed, dentine-collagen does not remodel, therefore providing a sequential record of biological signals at the time of teeth formation (Fuller et al. 2003). By cutting serial sections perpendicular to the tooth axis, the isotopic history of an individual can be reconstructed, allowing the investigation of isotopic changes over time. Since the timing of teeth formation is well-established and differs between each tooth type, a chronology of the isotopic changes can be created, spanning in utero to early adulthood life (AlQahtani 2009).

Reduction of the minimum section size required for analyses to 0.75–1mm has significantly increased temporal resolution to 8–9 months, which should enable the detection of shorter-term events like pregnancy (Beaumont et al. 2013a, 2013b). However, growth lines do not form exact horizontal lines, instead growing at an angle, which increases towards the root tip (Eerkens et al. 2011). Cutting horizontal sections thus creates an overlap between each successive sample, resulting in attenuated isotopic signals and providing isotopic trends rather than absolute time-averaged values (Beaumont et al. 2013a, Henderson et al. 2014). This is particularly significant for segments from the root, where the angle of the growth lines is more acute. Nevertheless, IDIA has been successfully applied to help identify the duration of breastfeeding and infant feeding practices (e.g Dupras and Tocheri 2007; Howcroft et al. 2012, 2014; Eerkens and Bartelink 2013; Sandberg et al. 2014), and malnutrition and recovery from food-shortages, including dietary changes associated with food relief (Beaumont et al. 2015; Beaumont and Montgomery 2016).

Diet, nutrition and reproductive health in Sudan

Sudanese diet remains traditional, based on a limited choice of local foods, with cereals like sorghum, vegetables like cassava, and milk forming the basis of the diet across the country (Food and Agriculture Organisation of the United Nations (FAO) 2010). Proteins are mostly derived from dairy products and groundnuts, with an overall relatively low intake of animal proteins like meat or eggs, although fish can make an important contribution to the diet in regions along the Nile and its tributaries, the Red Sea coast or river systems in the southern regions (Food and Agriculture Organisation of the United Nations (FAO) 2020).

Prolonged political unrest and recurring droughts have resulted in seasonal to chronic food shortages, with the first severe drought recorded in 1967–1973 in the Sahel belt, while southern regions experienced particularly severe famines throughout the 1980s and 1990s (Tambel et al. 2014). Such circumstances have resulted in substantial population displacements from the regions most affected by conflicts and famines towards Khartoum.

In Khartoum, contemporary women generally meet their overall energy and protein needs, although the risk of essential amino-acids deficiency remains since proteins are mostly derived from plant sources with lower protein quality (Nyuar et al. 2012). The likelihood of Sudanese women developing protein-energy malnutrition is increased during gestation since their food intake is generally unchanged -or even reduced as a result of food aversions or cultural food avoidances- despite the increased nutritional requirements associated with pregnancy (Karrar-Osman 1985; Harrison 1997). Furthermore, 7% of women are married before the age of 15, with nearly a quarter giving birth before 18 years of age, and women in poorer rural areas generally giving birth at even younger ages (Oxfam and Ministry of Health (MoH)) 1997; UNICEF 2016). A study by Otor and Pandey (1998) reported that first births occur on average at 19.7 years. Such relatively young AFG is likely linked to both the young age of women at marriage and limited access to contraception (Rogers and Earnest 2014; World Health Organization (WHO)) 2011). Gestation during adolescence, a period of increased nutritional requirement for the mother who is still growing, further increases health and nutritional stress during pregnancy.

Certainly, high levels of maternal malnutrition are observed outside of the capital city, especially in rural areas, with half of the women being reported as severely underweight (BMI <18.5) in the rural Red Sea region (Oxfam and Ministry of Health (MoH)) 1997). Such widespread maternal malnutrition is supported by the high prevalence of maternal morbidity and mortality associated with gestation and parity (a third of women experience complications), low-birth-weight babies (25%) and infant mortality (8%) (World Health Organization (WHO) 2015; UNICEF 2016).

In Sudan, the average IBI is 2.5–3.5 years, linked to the universal practice of prolonged breastfeeding (Tyler et al. 2014). Indeed, prolonged breastfeeding for 12–30 months is generally practiced across Sudan, with half of the babies still being breastfed at 24 months of age (Tyler et al. 2014). The risk of maternal malnutrition is further influenced by infant feeding practices since prolonged breastfeeding significantly increases requirements for both proteins and calories (Black et al. 2008). In Ethiopia, up to 1 in 5 lactating mothers are undernourished (Boke et al. 2021). However, the effect of infant feeding on the nutritional status of Sudanese mothers is lessened since many women traditionally increase their intake post-partum to compensate for loss during childbirth and enhance milk production (Karrar-Osman 1985).

Data on supplementary feeding, especially bottle feeding, is highly variable. A 1987 survey in rural Khartoum indicated that 62% of babies were introduced to bottle feeding relatively early in their life, yet in a 2014 study, as few as 17% of mothers in Khartoum supplemented breast milk with bottle feeding from the age of 4–6 months (Harrison 1997; Tyler et al. 2014). ‘Household’ supplementary food is generally introduced between 4 and 6 months of age in towns, although weaning is more likely to occur from the later age of 9 months in rural areas (Oxfam and Ministry of Health (MoH)) 1997; Harrison 1997).

Subjects and methods

Materials and subjects

A sample of 10 M₃ teeth was randomly chosen from a larger assemblage, collected as part of a separate study, alongside basic (self-reported) demographic and (measured) anthropometric information during routine dental treatment in Khartoum. The 10 selected teeth represent a cross section of the Khartoum population in terms of age, ethnicity and body mass index (BMI), although only 2 of the teeth are from males (Table 1). Medical histories were reported by some of the women who donated teeth to the larger tooth assemblage, although unfortunately, pubertal, obstetric, medical or dietary histories were not available for the teeth selected for this study. All selected teeth were free of caries or excessive wear.

Table 1. Demographic and anthropometric data of the 10 individuals.

Subject	Sex	Age (years)	Ethnicity	Height (meter)	BMI	Estimated timing for M3 formation*
A	Female	23	African	1.44	22.6	1997–2012
B	Female	26	African	1.55	23.8	1994–2009
C	Female	42	African	1.54	25.8	1978–1993
D	Female	28	African	1.70	22.4	1992–2007
E	Female	25	African	1.65	24.8	1995–2010
F	Female	24	African	1.59	18.8	1996–2011
G	Female	50	African	1.72	16.3	1970–1985
H	Female	42	African	1.54	25.8	1978–1993
I	Male	43	African	1.79	29.6	1977–1992
J	Male	50	Filipino	1.76	22.0	1970–1985

*Determined using the known age range for the formation of M₃ (AlQahtani 2009) and age at tooth extraction in 2012.

Ethical approval was granted by the Ethics Committee at the Khartoum Centre for Research and Medical Training (KCRMT, date 06-03-2012). The teeth and associated data were processed through Ethical Tissue at the University of Bradford, ensuring that conditions of patient anonymity and informed written consent were met.

M₃ teeth were chosen to investigate early reproductive history since they are the last teeth to develop, generally between the ages of 8.5 and 23.5 years (AlQahtani 2009). Teeth from this assemblage are particularly appropriate since many Sudanese women experience their first gestation(s) during this period (Otor and Pandey 1998). While they were not available for use in this study, five women who donated teeth to the larger tooth assemblage for the Dean and Elamin (2014) study are known to have experienced pregnancies during the formation of their M₃ teeth. The five women experienced their first pregnancy at the average age of 15.6 years, with the earliest recorded AFG at 12 years, and displayed an average IBI of 3.5 years (Dean and Elamin 2014). Aspects of contemporary Sudanese life such as diets containing few processed foods and the high prevalence of breastfeeding, mean that the results could also potentially be extrapolated to anthropological studies of past or traditional-subsistence populations.

Incremental dentine isotope analysis

The teeth were sectioned to obtain a full-length longitudinal section of dentine. Surface debris and identified secondary or tertiary dentine were removed by air-abrasion using a slow-speed tungsten-carbide rose-head bur. Each root was demineralised using the modified Longin method (O’Connell and Hedges 1999) and sequential dentine-collagen was obtained using method 2 in Beaumont et al. (2013a). Teeth were demineralised in 0.5 M hydrochloric acid at 4 °C. Each demineralised root was cut into approximately 1 mm regular sections using a scalpel, with each segment then placed in sealed microtubes, with pH3 acidified water at 70 °C for 48 hours, to enable denaturation of the collagen fibrils. Filtration was not carried out to prevent sample size reduction. Instead, samples were centrifuged prior to freezing and freeze-drying to concentrate contaminants to the base of the microtube.

δ¹⁵N and δ¹³C were measured by combusting 0.5 mg of each dentine sample. Samples were measured in duplicate in the University of Bradford Stable Light Isotope Laboratory in a Thermo Flash EA 1112 with the separated N₂ and CO₂ then introduced to a delta plus XL via a Conflo III interface and compared with laboratory and international standards (IAEA 600, CH3, N1) that were interspersed through each analytical run.

The C/N ratios obtained from each dentine sample are within the recommended 3.1–3.5 range, confirming good preservation of the collagen (Van Klinken 1999), which would be expected from modern teeth which have never been buried. The results are expressed using the delta (δ) notation in parts per thousand (‰) relative to the international standards Vienna-Pee Dee Belemnite for δ¹³C and AIR for δ¹⁵N. The analytical error was determined at ±0.2‰ or better.

An approximate chronological age (based on the average timing of M₃ formation) was assigned to each segment and isotope ratios profiles for each tooth were plotted using Excel software (Beaumont and Montgomery 2015). Nutritional stress, dietary and environmental shifts were identified using δ¹³C, enabling the isolation of isotopic signals related to other metabolic changes. The timing of M₃ teeth formation is notably variable (Mincer et al. 1993). M₃ formation begins on average around age 8.5 years, but can start any time between 7 and 13 years (Hillson 1996; AlQahtani 2009). Isotopic changes were therefore ‘matched’ to known extreme climate events or famines, allowing to ‘correct’ (or confirm) isotopic timelines. The estimated ages stated in the rest of this paper refer to ‘corrected’ chronologies. Demographic data available for each subject and information about modern Sudanese geo-political history and culture were used to help interpret isotopic variations for each profile.

Results

Calculated bulk isotope values, the range of incremental isotope values and the temporal resolution obtained for each tooth are shown in Table 2. Temporal sequence profiles for each tooth are shown in Figures 5–7. Results potentially relating to puberty and pregnancy are summarised in Tables 3 and 4. Analysis to define the baseline diet and environment along with the detailed interpretation of dietary and environmental changes will be published in a separate paper.

Figure 5. The proposed interpretation of δ¹⁵N and δ¹³C profiles in relation to pubertal growth, for women who display life events or stress during late childhood and adolescence.

Figure 6. The proposed interpretation of δ¹⁵N and δ¹³C profiles over the time of M₃ formation, in relation to pubertal growth, for the two men.

Figure 7. The proposed interpretation of δ¹⁵N and δ¹³C profiles in relation to pubertal growth, for women with no obvious life events during late childhood and adolescence.

Table 2. Summary of isotope results for the 10 M₃.

Subject	δ^15N (‰)		δ^13C (‰)		IDIA resolution (months)	C:N
	Bulk ± SD	Range	Bulk ± SD	Range
A	10.5 ± 0.3	0.8	−17.6 ± 0.5	1.4	14	3.1
B	11 ± 0.2	0.8	−17.8 ± 0.4	1.4	10	3.2
C	10.6 ± 0.5	1.5	−18.4 ± 0.4	1.2	11	3.2
D	11 ± 0.3	0.9	−16.3 ± 0.2	0.5	11	3.2
E	11.7 ± 0.3	1.1	−17.3 ± 0.3	1.2	9	3.3
F	10.4 ± 0.4	1.6	−16.6 ± 0.5	1.6	10	3.2
G	10.4 ± 0.8	2.9	−11.8 ± 1.3	4.2	14	3.2
H	12.1 ± 0.3	1.0	−14.8 ± 0.5	1.4	18	3.2
I	12.5 ± 0.7	2.2	−11.9 ± 1.3	4.2	12	3.1
J	12 ± 0.5	1.4	−18.7 ± 0.2	0.7	10	3.3

Bulk isotope values consist of the mean of all sections for each tooth. The range of isotope values denotes the difference between the minimum and maximum isotope ratios.

Table 3. Summary of the isotope data proposed to represent pubertal growth.

Puberty	Sex	Magnitude of δ^15N reduction (‰)	Growth spurt onset	Peak weight velocity	Completion of weight gain
			(years)	(years)	(years)
PA	Female	0.8	≤9	11.9	14.7
PB	Female	0.6	10.8	12.4	14.1
PC	Female	0.9	≤9	12.6	14.4
PD	Female	0.6	<10	10–11	12.7
PE	Female	0.5	≤9.9	11.4	13
PF	Female	(1.5)	8.8	9.6	–
PG	Female	1.4	≤9.1	11.3	–
PH	Female	(0.5)	≤11.3	–	–
PI	Male	(1.5)	≤12	(18.7)	(23.5)
PJ	Male	1.3	9.8	13.1–16.5	22

Uncertain data: the proposed puberty coincides with events which also influence isotopic values.

Table 4. Summary of the isotope data interpreted as gestational events.

Pregnancy	Estimated age at conception (years)	Magnitude of δ^15N reduction (‰)	Length of anabolic state (Number of dentine-sections)	Description of post-anabolic δ^15N
GA	15.6	0.5	3	Return to pre-gestation ‘baseline’ over 3 increments
GB	14.4	0.7	3	Return to pre-gestation ‘baseline’ over 4 increments, with further increase over 2 increments
GC	15.3	0.3	1	Return to pre-gestation ‘baseline’ over 2 increments, with further increase over 4 increments
GD	14.4	0.3	1	Return to pre-gestation ‘baseline’ over 2 increments
GF	13.0	0.3	2	Return to pre-gestation ‘baseline’ over 1 increment
GG	15.1	0.8	2	Return to pre-gestation ‘baseline’ over 1 increment

δ¹⁵N and δ¹³C temporal sequences

The dentine sections obtained represent, on average, periods of 11.9 ± 2.7 months, varying from 9 months for female E to 18 months for female H. Variations in temporal resolution between teeth are introduced by the differing root length. Such timing remains approximate because of the averaging produced by the method.

Small isotopic changes can result from the collagen preparation method and instrument imprecision, therefore only δ¹⁵N and δ¹³C variations greater than 0.3‰ are considered as potentially reflecting behaviourally meaningful events (Eerkens et al. 2016).

Dietary, environmental and catabolic events

Since δ¹³C levels vary with diet, environment and nutritional stress, but are not thought to be significantly affected by other metabolic events (e.g. anabolic states of pubertal growth or pregnancy), δ¹³C is used to isolate dietary and environmental changes. Four individuals have highly variable δ¹³C values, indicative of major events (Figures 5 and 6). δ¹³C shifts indicative of dietary and/or environmental changes are suggested for female H before she enters her teenage years; although the lower temporal resolution (linked to the shorter root length of the tooth) and some missing data (due to a misplaced sample) render full interpretation challenging (Figure 5). Male I also experiences three major δ¹³C shifts at the estimated ages of 13, 20 and 24 years (Figure 6). Both H and I display substantial δ¹³C and δ¹⁵N reductions at the estimated ages of 14 and 15 years, respectively, which corresponds with the severe 1980–1984 drought across the Sahel Belt and the subsequent 1984–1985 famine (Teklu et al. 1991; Figure 1, Tambel et al. 2014). The isotopic profiles exhibited by female H and male I at the start of M₃ formation -including the high δ¹⁵N values compared with other individuals- are therefore understood to reflect increased isotopic values of local plants and animals from water stress, with the subsequent isotope shift highlighting the end of the drought. For male I, nutritional stress, suggested by a short but substantial δ¹⁵N peak also coincides with these elevated isotopic values and could have resulted from the ensuing famine, substantiating an environmental interpretation (Figure 6). The first isotopic shift for male I is therefore interpreted as showing concurrent events in the form of decreased aridity and nutritional stress. The two further δ¹³C shifts displayed by male I also relate to two known localised droughts in 1989 and 1993; while two small δ¹³C peaks of 0.6‰ for H correspond with two recorded droughts in 1987 and 1993 (Figures 5 and 6, Tambel et al. 2014).

The overall δ¹³C and δ¹⁵N patterns for female G at the start of M₃ formation are comparable to that of male I -albeit with a terrestrial rather than fish-based diet- and correspond with a known severe drought in 1967–1973 (Figure 1, Tambel et al. 2014). δ¹⁵N levels reached after those events in early adolescence are the lowest for all the participants, highlighting a likely extremely low intake of animal proteins, an observation in agreement with the collapse of food production in the aftermath of droughts (Teklu et al. 1991). δ¹³C patterns after the drought, as G enters her teenage years, also suggest acute malnutrition (Figure 5). The absence of a δ¹⁵N peak, which is normally seen with nutritional stress, can be explained by the accompanying decreasing aridity and dietary trophic level linked to food shortages. The substantial δ¹³C reduction may also reflect increased wheat consumption from food relief (Food and Agriculture Organisation of the United Nations (FAO)) 2005). The major isotopic changes displayed by female G around age 11 years are therefore indicative of a simultaneous reduction in aridity levels and animal protein intake, increased C₃-cereals consumption, and nutritional stress. The isotopic upward shift experienced by G around the age of 18 years is thought to represent recurring increased aridity as the region entered the 1980–1984 drought .

The temporal isotopic profile for female F suggests that at about 9 years of age she experienced a catabolic event, seemingly associated with dietary and/or environmental changes and coinciding with the 1996–1997 drought and subsequent famine in 1998 (Figure 5, Tambel et al. 2014). The small but sustained δ¹³C reduction of 0.6‰ associated with an extensive δ¹⁵N decrease of 1.5‰, and the fact that F has one of the lowest bulk δ¹⁵N, point to the combination of decreased aridity, catabolic stress, and long-term dietary shift akin to both overall protein intake and trophic level reduction resulting from increased reliance on proteins from plant-foods (Table 2). The δ¹³C shift displayed by F aged 21 years is also linked to a localised drought and famine in 2009 (Figure 5, Tambel et al. 2014).

Since δ¹³C shifts from the proposed simultaneous events linked to droughts and famines exacerbate each other, while the δ¹⁵N effect of catabolic stress cancels that of dietary and environmental changes, individuals G-I display a higher range of δ¹³C variations compared with δ¹⁵N values (Table 2). Major δ¹³C changes suggestive of dietary, environmental or catabolic events are not seen in other teeth, yet the range of δ¹³C variations for females A and B is unexpectedly higher than that of δ¹⁵N since metabolic events are thought to only affect δ¹⁵N values (Table 2). Both women display small but sustained δ¹³C shifts (Figure 7). For B, a lesser event may be indicated at age 11 years, corresponding with the 1996–1997 drought (Tambel et al. 2014). However, these small δ¹³C shifts may also highlight a greater influence of physiological events on δ¹³C values than anticipated; possibly reflecting increased reliance on body fat as a source of carbon for protein synthesis during periods of increased tissue production, or body fat changes that occur in women during puberty, pregnancy and lactation (Prentice et al. 1996; Rogol et al. 2002).

Anabolic events

Women A, B and C have comparable δ¹⁵N patterns during M₃ formation, experiencing two consecutive and discernible dips of 0.6–0.9‰ and 0.3–0.5‰ from 9–11 years of age and 14–15 years of age, respectively (Figure 7). Female E similarly experiences a gradual 0.5‰ δ¹⁵N reduction from about 9 years old and displays a second smaller 0.3‰ δ¹⁵N dip, but at the later age of about 18 years (Figure 7). Some of the observed δ¹⁵N depressions are subtle, yet instrumental imprecision is unlikely since the isotopic patterns are sustained over time. Slight variations in terms of shape or intensity for the initial δ¹⁵N dip are seen, yet the δ¹⁵N decline consistently starts at around age 9–10 years for all four women (despite woman C being born 15–20 years before the others), pointing to the occurrence of a universal physiological event. Certainly, for women A, C and E, the δ¹³C changes accompanying the initial δ¹⁵N reduction do not match isotopic patterns expected from either dietary or environmental changes (Figure 8). Instead, δ¹⁵N variations are consistent with isotopic projections for the pubertal growth spurt (Figure 3). Therefore, for these four women, we propose that the initial δ¹⁵N depression represents the pubertal growth spurt. We suspect that the subsequent smaller δ¹⁵N dips relate to pregnancies; although, for woman E, a 0.5‰ δ¹³C decrease matching the δ¹⁵N reduction is seen, and corresponds with severe floods and crop failure in 2007, suggesting a possible alternative interpretation as a wetter period with a transient decrease in trophic level (Figures 7 and 8, Moszynski 2007).

Figure 8. Projected δ¹⁵N and δ¹³C shifts linked with dietary, physiological or environmental events.

δ¹⁵N reductions which could be associated with pubertal growth are also observed in the isotopic profile of other individuals. Male J displays a 1.3‰ δ¹⁵N decline over 3.5 years from the estimated age of 10 years, with δ¹⁵N recovery starting from the age of about 16.5 years (Figure 6). Female G also experiences a gradual and extensive δ¹⁵N reduction of 1.4‰ from about 11 years of age, while F experiences a sudden extensive decrease of 1.5‰ at about 9 years; although the subsequent δ¹⁵N recovery is limited for G and absent for F (Figure 5). For male I, the δ¹⁵N decrease, which coincides with the start of M₃ formation and seemingly continues for about 6 years, is interrupted by a catabolic event, while an uncertain δ¹⁵N recovery is suggested at about 23 years old (Figure 6). A δ¹⁵N reduction of 0.5‰ is also seen at the start of tooth formation for female H (Figure 5). However, the potential effect of the coinciding environmental changes on isotope ratios, combined with the missing data which corresponds with the presumed timing of puberty, means that the presence of isotopic pattern consistent with pubertal growth cannot be confirmed for H. The exception is for female D, for whom no initial δ¹⁵N reduction is seen, but instead, D displays a gradual 0.6‰ δ¹⁵N increase at the start of M₃ formation (Figure 7).

Thus, in this small group, the δ¹⁵N decrease, which we interpret as representing pubertal growth, ranges from 0.5 to 1.5‰ with a mean of 0.96 ± 0.42‰ (N = 10) (Table 3). The mean is reduced to 0.78 ± 0.29‰ (N = 6) when individuals who display major dietary or environmental changes or stress during this period are excluded. A mean of 0.7 ± 0.2‰ (N = 5) is indicated for women, while the mean for men cannot be calculated due to the limited amount of data. Overall δ¹³C values remain stable or display minimal changes; and no specific δ¹³C pattern is recognised in association with the δ¹⁵N dip which we propose represents pubertal growth. δ¹³C changes matching δ¹⁵N variations, in line with isotopic patterns seen in growing animals, are seen in only one isotopic profile (woman B) (Figure 7, Warinner and Tuross 2010).

Subsequent to the isotopic signal interpreted as representing pubertal growth -and therefore the start of fertility-, δ¹⁵N patterns, which are consistent with pregnancy, are also recognised in the isotopic temporal sequences of four other women. Women D and F experience small 0.3‰ δ¹⁵N dips at the estimated ages of 14 years, while G displays a distinct 0.8‰ dip at about 15 years of age (Figures 5 and 7). For H, the outline of a δ¹⁵N dip comparable to that of A, is also apparent at about 14 years, although the δ¹⁵N reduction is small (0.3‰) and inconsistent (Figure 5). Given the lower temporal resolution, the missing isotopic value, and the potential isotopic effect of two droughts (factors that make reliable interpretation of isotopic profiles in relation to short-term physiological events difficult), confirming the small δ¹⁵N variation seen in the H isotopic profile as a meaningful metabolic event is not possible. The start of a third δ¹⁵N reduction is also observed at about 21 years for women C and E, although instrumental influence cannot be ruled out for those two isotopic variations, since their occurrence at the end of M₃ formation results in incomplete signals (Figure 7).

In total, six δ¹⁵N dips which are consistent with pregnancies are suggested, with an overall δ¹⁵N decrease of 0.48‰ (0.3–0.8‰) (Table 4). δ¹³C patterns accompanying the δ¹⁵N changes interpreted as pregnancies -i.e. no overall changes- do not match dietary or environmental explanations, confirming a more likely physiological interpretation (Figure 8).

Following the δ¹⁵N dips interpreted as pregnancies, a variety of isotopic patterns are observed (Table 4). With the exception of suspected pregnancies G_A and G_B, δ¹⁵N begins increasing towards isotopic pre-gestation baselines immediately or very soon after the proposed gestational events; an increase which is either gradual (e.g. G_C) or very rapid (e.g. G_D, G_F and G_G) (Figures 5 and 7). For women A and B, the lower δ¹⁵N values purported as pregnancies are sustained beyond the expected 9 months of gestation, before being gradually restored to (or above) pre-gestation values (Figure 7). We suspect that the various post-gestational isotopic responses are linked to lactation and nutrition status of the mothers.

Discussion

Pubertal growth and δ¹⁵N

For women A-E, δ¹⁵N dips are noted at the beginning of M₃ formation (Figure 7). The uniformity of these depressions, including their timing around age 10 years and their amplitude ranging from 0.5 to 0.9‰, suggests that they likely represent analogous physiological events. The timing, duration and shape of these δ¹⁵N dips are also consistent with projected isotopic patterns for pubertal growth (Figure 3). Interpreting the identified δ¹⁵N dip in terms of pubertal growth suggests growth onset at 9.7 ± 0.9 years (N = 4) and completion at 13.8 ± 0.9 years (N = 5); PHV is indicated to be reached by 11.7 ± 1.0 years (N = 5) and the acceleration phase to last 2.4 ± 1 years (N = 4). In women, sexual development is closely linked to growth, with puberty onset corresponding with growth spurt initiation, menarche occurring about one year after PHV, and full fertility being achieved by the end of the growth spurt (Shapland and Lewis 2013). Therefore, in this group of women, based on the assumption that the δ¹⁵N reductions represent pubertal growth, attainment of menarche at about 12.7 years and full fertility by 13.8 years is inferred. The proposed timing for menarche is consistent with the average of 13 years for European/North American populations; although, puberty onset is early relative to the average start of 11 years for European/North American populations (Rogol et al. 2002). Earlier puberty is documented in women of African descent, and the results are consistent with puberty onset at 8.8 years seen in Afro-Caribbean populations (Boyne et al. 2010).

A distinct δ¹⁵N depression is also seen in the isotopic profile of male J (Figure 6). If this δ¹⁵N dip is interpreted in terms of pubertal growth, then growth acceleration is indicated to start at about 9.8 years, with accelerated growth lasting 6.7 years. PHV is reached at 13.1 years and growth is completed at 22 years. The proposed timing for the growth spurt onset, the longer acceleration phase and later PHV are all consistent with puberty patterns seen in male populations, including an average PHV of 14 years (Rogol et al. 2002). The δ¹⁵N profile is also reminiscent of isotopic variations seen in teeth from Seventeenth and Nineteenth century men, which also coincides with puberty (Beaumont and Montgomery 2016, Figure 2; Millard et al. 2020). Yet, the length of the δ¹⁵N depression proposed to represent pubertal growth diverges from projected patterns derived from height velocity curves; and growth completion is later than expected (Figure 3, Rogol et al. 2002). Male J (like Seventeenth century Scottish soldiers PGL21 and PGL12) seems to experience a particularly prolonged growth spurt over more than 12 years, with δ¹⁵N still increasing at the end of M₃ formation, despite J displaying an average adult height of 1.76 m (Table 1, Figure 2). If male J experienced malnutrition in earlier childhood, this extended δ¹⁵N depression could indicate catch-up growth into early adulthood (Golden 1994). Alternatively, we suspect that the discrepancy between projected and observed δ¹⁵N patterns results from the way exemplars for pubertal growth was developed. Indeed, projections are based on linear [height] growth, but δ¹⁵N changes have been found to be inversely correlated with weight gain, and linear growth differs from ponderal [weight] growth (Rogol et al. 2002; Fuller et al. 2004). During the male acceleration phase, weight gain matches height growth, and peak weight velocity (PWV) coincides with PHV; however, weight continues to increase significantly for several years after peak growth as musculature develops beyond age 19–20 years (Rogol et al. 2002). This is illustrated by peak velocity for thigh and arm circumference -which reflect soft tissue accretion- extending beyond PHV, with an increase in arm circumference continuing beyond the age of 19 years (Cole 2020). Based on patterns of soft tissue accretion in boys, reduced δ¹⁵N values can be expected to continue beyond the achievement of peak growth, as seen in male J.

In girls, weight gain also matches height gain during the growth spurt, although PWV is achieved 6 months after PHV, meaning that menarche occurs 6 months after PWV (Rogol et al. 2002). As for boys, weight gain in girls continues beyond peak growth but over a shorter duration, with adult muscle mass being achieved soon after puberty completion by 15–16 years old. Post-pubertal weight gain in girls also consists of a higher proportion of body fat than muscle, which is presumed to have a lesser impact on δ¹⁵N. Given this understanding of growth in girls, isotopic models for puberty created using ponderal growth are expected to be similar to projections established with height velocity, except for the timing for menarche. Based on the premise that δ¹⁵N reflects ponderal rather than linear growth, for women A-E, an earlier age of 12.2 years is indicated for menarche, which corresponds with the earlier menarche at age 12 years noted in Afro- Caribbean populations (Boyne et al. 2010). Completion of linear growth -and therefore fertility- may also be achieved before δ¹⁵N returns to its baseline level, although this conjecture cannot be verified without access to sexual development and growth histories.

References for the magnitude of the growth effect on δ¹⁵N are lacking in humans. However, peak growth during puberty (8–9kg per year) and recommended weight gain during pregnancy (5–18kg) occur on similar scales (Rogol et al. 2002; Institute of Medicine 2009). Given that δ¹⁵N is related to weight gain, δ¹⁵N reductions during puberty can be expected to be comparable to the decrease reported during pregnancy (Fuller et al. 2004). The observed δ¹⁵N reductions of 0.5 to 1.3‰ proposed to represent pubertal growth in our study match the range of isotopic variations of 0.3–1.3‰ observed in 13 pregnant women (Fuller et al. 2004; D'Ortenzio et al. 2015). Since peak growth during puberty is smaller in girls than in boys (8.3 kg and 9 kg per year, respectively), δ¹⁵N reductions linked to pubertal growth can also be expected to be less important in girls than in boys (Rogol et al. 2002). The proposed δ¹⁵N decrease linked to pubertal growth in our study, 0.7‰ (0.5–0.9‰) for five girls and 1.3‰ for one boy, is in line with such expectations.

Pregnancy, lactation and δ¹⁵N

Isotopic patterns interpreted as gestational events are identified in the isotopic profiles of women A-D, F and G (Figures 5 and 7, Table 4). The observed δ¹⁵N variations match expected changes created by pregnancy, including two events (G_A and G_B) with lengthy reduced δ¹⁵N, fitting projections for prolonged lactation, when intense lactogenesis is proposed to result in reduced δ¹⁵N values which gradually return to pre-gestation levels as milk production decreases during the weaning process (Figure 4). After the other four proposed pregnancies, δ¹⁵N values return relatively rapidly to pre-gestation values, perhaps indicating short-term or absence of lactation. Indeed, in a study by Fuller et al. (2006), an immediate and rapid δ¹⁵N increase after giving birth was displayed by a mother who did not breastfeed. However, given the drought which occurs after the proposed pregnancy G_G, interpretation of the potential post-gestational isotopic data is unreliable for woman G. Certainly, when isotopic effects of droughts are accounted for, a more gradual δ¹⁵N increase is suggested.

The range of δ¹⁵N reductions (0.3–0.8‰) displayed during the six proposed pregnancies compares well with the range noted in 13 American and Canadian pregnant women (0.3–1.3‰) (Fuller et al. 2004; D'Ortenzio et al. 2015). If these isotopic signals are taken to represent pregnancies, an overall AFG of 14.6 ± 0.9 years is indicated for those six women (Table 4). This AFG is comparable to the AFG of 15.6 years (derived from obstetric histories) from five separate women who donated teeth to the same assemblage (Dean and Elamin 2014). The suggestion of prolonged breastfeeding after the proposed pregnancies of G_A and G_B is also in agreement with infant feeding practices in Sudan, where half of the children are breastfed beyond the age of 2 years (Tyler et al. 2014). While prolonged nursing for 2–3 years alongside food is common in Sudan, cessation of breastfeeding by the time the baby is 12 months also takes place, therefore the suggested shorter nursing duration after at least three of the other proposed pregnancies remains within conventional Sudanese infant feeding practices (Tyler et al. 2014). Shorter lactation would also be seen when a child does not survive, therefore the suggested brief periods of breastfeeding could also reflect high infant mortality (8%) in the region (UNICEF 2016). Still, given the lack of obstetric and infant feeding history, and the limited understanding of the effect of lactation on isotope values, reliable interpretation in terms of infant feeding practices is not possible. Indeed, diverse δ¹⁵N responses to lactation have been observed in humans: reduced maternal δ¹⁵N values resulting from gestation are occasionally maintained throughout lactation, yet some nursing mothers display progressive δ¹⁵N increase towards pre-gestation levels from birth (Fuller et al. 2006; Herrscher et al. 2017). Such a variety of isotopic patterns associated with breastfeeding could be linked to differing lactation intensity, maternal nutritional status or diet. Indeed, Habran et al. (2010) observed that δ¹⁵N reductions usually observed in lactating seals were absent in fasting nursing animals, instead an insignificant upward tendency was noticed. Habran et al. (2010) inferred that nursing individuals with limited food intake are faced with a dual metabolic challenge, and the increased δ¹⁵N resulting from protein catabolism cancels the δ¹⁵N reduction associated with the anabolic event.

Isotopic effect specificity – diet, climate or physiology?

Interpretation of sequential isotopic profiles can be relatively straightforward. δ¹⁵N depressions which are not easily explained by diet or climate (due to the absence of a simultaneous decrease in δ¹³C) but can be interpreted in relation to puberty, are readily identified in the isotopic profiles of women A, C and E (Figure 7). Isotopic patterns which correspond with projected patterns for pregnancy (e.g. G_C, G_D, G_F, G_G), or pregnancy and prolonged lactation (e,g, G_A and G_B), and which are not easily accounted for by dietary or environmental factors, are also observed (Figures 4, 5, 7 and 8). Nevertheless, extracting dietary, environmental and physiological information from δ¹⁵N and δ¹³C sequences can be complex since different factors influence isotopic values in similar ways (Figure 8). For example, the δ¹⁵N dip of 0.3‰ displayed by woman E at the age of 18 years matches projected isotopic patterns for pregnancy. However, given the better temporal resolution of 9 months achieved for woman E, longer isotopic signals (relative to other isotopic profiles in this study) are anticipated; yet the identified δ¹⁵N reduction seems shorter than the six δ¹⁵N dips proposed as gestational events (Table 3, Figures 5 and 7). Such a brief signal could be explained as a lost pregnancy or preterm birth; however, the identified δ¹³C reduction is unlike δ¹³C patterns in pregnancy reported by Fuller et al. (2004). Instead, given the accompanying δ¹³C reduction of 0.5‰, the short isotopic shift and the known flood in 2007, an alternative interpretation as a seasonal wetter period with a short-term decrease in animal protein intake is indicated (Figures 4, 7 and 8, Moszynski 2007). For woman B, the δ¹⁵N dip at the age of 11 years is consistent with projected isotopic patterns for pubertal growth, and comparable to δ¹⁵N reductions interpreted as pubertal growth for women A, C and E (Figures 3 and 7). A δ¹³C reduction in agreement with isotopic patterns seen in growing animals is also observed (Warinner and Tuross 2010). Yet, dual δ¹⁵N and δ¹³C reductions are also produced by decreased aridity or trophic levels (Ambrose 1991; Tieszen 1991; Petzke et al. 2005, Figure 8). Certainly, the isotopic dip can be linked with a period of food shortages after the 1996–1997 drought and could therefore represent a short-term decrease in animal protein intake (Figure 8, Tambel et al. 2014). For woman D, isotopic variations evoking the end of a δ¹⁵N depression are also observed at the start of M₃ formation at the age of 11 years (Figure 7). Given that the timing of the isotopic shift is consistent with the chronology of puberty in girls, and that the scale of the δ¹⁵N shift (0.6‰) is comparable to the amplitudes noted for P_A, P_C and P_E, this δ¹⁵N rise could represent a truncated pubertal isotopic signal, with growth onset starting before the initiation of M₃ formation (Rogol et al. 2002, Table 3). Alternatively, the δ¹⁵N increase could represent catabolism persisting for about 2 years. The absence of δ¹³C reduction points to either a pathological origin or moderate chronic malnutrition (Robertson et al. 2014; D'Ortenzio et al. 2015). Indeed, Robertson et al. (2014) have shown that moderate chronic caloric restriction results in increased δ¹⁵N, but unlike severe acute nutritional stress from starvation, it does not affect the δ¹³C values of body proteins. Still, an isotopic signature indicating catabolic stress recovery is not obvious. The distinct δ¹⁵N dip seen in the isotopic profile for male J fits pubertal growth interpretation, and the associated δ¹³C increase is incongruent with either dietary or environmental explanations (Figure 6). However, when the δ¹⁵N reduction aged 11 years and δ¹⁵N increase aged 17 years are considered as two separate events, a possible alternative interpretation for the δ¹⁵N dip emerges as a period between two episodes of nutritional stress. Indeed, the isotopic shifts between 11 and 13 years of age match patterns seen during post-starvation weight gain, while isotopic variations at age 17 years correspond with patterns seen during nutritional stress (Figure 8).

Physiological isotopic signals and eventful life histories

Isotopic data for individuals A-E and J (as well as Seventeenth century Scottish soldiers and Nineteenth century Irish populations) provide compelling indications that δ¹⁵N variations may be useful to track pubertal growth and by association the progress of puberty (Beaumont and Montgomery 2016, Figures 2 and 7 ; Millard et al. 2020). However, isotopic patterns suggestive of pubertal growth are not as obvious for individuals F-I (Figures 5 and 6). Isotopic shifts resulting from anabolic states are small in relation to the effect of other factors like diet. δ¹⁵N differences of 1.8‰ are seen between vegans and omnivores (Buchardt et al. 2007). Starvation results in an δ¹⁵N increase by up to 1.9‰ (Mekota et al. 2009). Yet, the most important δ¹⁵N reduction noted during pregnancy is 1.3‰; and during lactation, δ¹⁵N changes of 0.5‰ or less are displayed by five American women (Fuller et al. 2004; 2006; D'Ortenzio et al. 2015). In our study, the maximum δ¹⁵N reductions during the proposed pregnancies and pubertal growth are 0.8‰ and 1.3‰, respectively (Tables 3 and 4). For our small group of Sudanese individuals, the overall isotopic shifts of 0.48 ± 0.22‰ (0.3–0.8‰) and 0.78 ± 0.29‰ (0.5–1.3) suggested for pregnancy and puberty are also relatively small compared with the overall range of δ¹⁵N variations of 1.42 ± 0.68‰ (0.8–2.9‰) (Table 2). Isotopic effects arising from anabolic states could therefore easily be masked by starvation, dietary or climatic changes when physiological events coincide with such occurrences.

Incremental isotopic data for individuals F-I, supported by known demographic and anthropometric data and geo-political information, indicate that they most likely experienced metabolic stress, dietary and/or environmental changes linked to droughts and famines during adolescence. For example, isotopic shifts observed at the start of M₃ formation for female F are associated with the end of a severe drought and proposed catabolic and dietary events; and may therefore conceal the onset of the pubertal growth spurt. Certainly, the extensive δ¹⁵N decrease accompanied by a much smaller δ¹³C reduction hints that the δ¹⁵N variations are not fully explained by diet and climate (Figure 5). For woman G, the major isotopic variations resulting from the proposed simultaneous reduction in animal protein intake, climatic changes, and nutritional stress linked to the 1967–1973 severe drought and food shortage could also have masked the expected δ¹⁵N recovery from the concurrent growth-related anabolic event. The major environmental and nutritional isotopic shift identified for male I between the estimated ages of 13–16 years, corresponds with a period of expected growth acceleration, therefore increasing the complexity of isotopic interpretation in relation to developmental maturation. Still, as for male J, hints for a relatively late peak growth velocity and completion are perceived in the δ¹⁵N profile of male I (Figure 6). Such suggested delayed growth could be linked to the nutritional stress male I experienced in early puberty (Soliman et al. 2014). The last downward isotopic shift linked to the 1993 drought could also mask a continued anabolic δ¹⁵N increase associated with growth completion.

Isotope analysis sensitivity: missing isotopic signals and ‘background noise’

For individuals F-I, concurrent diet, climate and catabolic events, therefore, appear to conceal the isotopic effect of pubertal growth. Yet, it is also possible that in those individuals, pubertal growth may not have had any visible effect on δ¹⁵N values. Indeed, physiological events do not show consistently with isotope analysis. Fuller et al. (2005) noted that pregnant women experiencing mild nutritional stress in the form of morning sickness display smaller or even minimal δ¹⁵N reduction. Habran et al. (2010) highlighted a comparable counteracting effect of malnutrition on anabolic δ¹⁵N responses in fasting lactating seals. A similar interaction between anabolic and catabolic states can be envisaged during the pubertal growth spurt. Given this suggested offsetting effect of malnutrition on anabolic δ¹⁵N shifts, as well as the increased risk of malnutrition during puberty, pregnancy and lactation due to increased nutritional requirements (Black et al. 2008), and the likely low protein intake during the time of M₃ tooth formation suggested by bulk isotopic values in our study , some physiological events experienced by our study group could have been unobservable isotopically and therefore missed. Certainly, we suspect a relatively high incidence of missed gestational events. Indeed, based on data from women who took part in a large survey across Sudan, an IBI of 2.5 years is expected (Otor and Pandey 1998); or at the most an IBI of 3.5 years, based on data from the small group of separate women from Khartoum who donated teeth to the same assemblage (Dean and Elamin 2014). Based on the proposed AFG of 15.6 years, and projected IBI of 2.5–3.5 years, we would expect to observe the second set of isotopic variations suggestive of pregnancies around the ages of 18–19 years in at least some of the isotopic profiles. Yet only one possible second gestation is suggested [by an incomplete and therefore highly unreliable isotopic signal], for woman C at the age of 21 years (Figure 7).

The issue of inconsistent isotopic effects of anabolic events may be compounded by the lower sensitivity of IDIA compared with hair analysis. Indeed, when teeth are sectioned horizontally, temporally distinct isotopic changes may be diluted by the temporal averaging that arises because of the mixing of dentine layers formed at different ages (Beaumont et al. 2013a, Henderson et al. 2014). Attenuated δ¹⁵N signals may also result from the lower temporal precision achieved with IDIA compared with the isotope analysis of hair. Indeed, Fuller et al. (2004) and D'Ortenzio et al. (2015) measured incremental δ¹⁵N values in hair, which provides a temporal resolution of about 1 month compared with the overall resolution of about 12 months obtained with IDIA in our study. δ¹⁵N shifts between conception and birth recorded by Fuller et al. (2004) and D'Ortenzio et al. (2015) were inconsistent: the reduced δ¹⁵N values were occasionally sustained over the entire 9 months, but more often δ¹⁵N reductions were limited to the last few months of gestation when the rate of protein synthesis and weight gain is highest. Given the uneven δ¹⁵N reductions during the course of pregnancy and the averaging of isotopic values over 9 months of gestation ensuing from the lower temporal resolution of IDIA, further dilution of isotopic variations resulting from pregnancies are therefore expected when measured in dentine rather than hair. Certainly, in our study, the amplitude of 0.48‰ (−0.3 to −0.8‰) for the six δ¹⁵N reductions suggested as possible pregnancies are lower than the average of 0.75‰ (−0.3 to −1.3‰) seen in 13 pregnancies in studies by Fuller et al. (2004) and D'Ortenzio et al. (2015). However, the calculated average δ¹⁵N decrease over 9 months of gestation based on the data from Fuller et al. (2004) and D'Ortenzio et al. (2015) is 0.45‰ (+0.2 to −1.0‰), a value with which the results of 0.48‰ suggested by our study compare relatively well.

The suggested attenuated δ¹⁵N variations created by IDIA imply an increased likelihood of isotopically unobservable events, but also isotopic signals being at the limit of the meaningful level and therefore difficult to distinguish from isotopic variations introduced by instrumental imprecision. Differentiating physiological events from ‘background noise’ is especially difficult for shorter events recorded over a single dentine-section (e.g. pregnancies without lactation, lost or preterm pregnancies), and the problem would be compounded when lower temporal resolutions are achieved (e.g woman H). In the studies by Fuller et al. (2004) and D'Ortenzio et al. (2015), δ¹⁵N reductions between conception and birth equal to or larger than the mean level [set at 0.3‰] were seen during all 13 pregnancies. Yet, when the calculated averaged δ¹⁵N shift (over the 9 months gestation) is considered, 29% of pregnancies display isotopic changes smaller than the mean level and would therefore be potentially imperceptible with IDIA. In our study, the magnitude of the isotopic signal associated with three of the six proposed gestations is at the limit of the accepted behaviourally meaningful level of 0.3‰ (Table 4). We deemed those small isotopic variations to represent actual events since δ¹⁵N reductions persist over two or more dentine samples, although some uncertainty of interpretation remains with such small isotopic shifts. Two short δ¹⁵N dips of 0.3‰, each over one dentine sample, seen in the isotopic profile for woman H, and two downward δ¹⁵N shifts of 0.3‰ observed over the last chronological dentine sample for women C and E, could not however be confirmed either as pregnancies or ‘background noise’ (Figures 5 and 7).

Physiological events and δ¹³C

The effect of anabolic states on δ¹⁵N values is relatively well-established (Fuller et al. 2004; Trueman et al. 2005; Reitsema and Muir 2015). However, there is a high degree of uncertainty regarding expected δ¹³C shifts in relation to physiology. Fuller et al. (2004) reported insignificant δ¹³C variations in mothers between the time of conception and birth of 11 children. D'Ortenzio et al. (2015) confirmed that gestational events result in smaller variability in δ¹³C than δ¹⁵N values (+0.4 to +0.6‰ and −1.0 to −1.3‰, respectively). By extension, anabolic events are assumed to have a negligible effect on δ¹³C values. Certainly, in relation to pregnancy, results from our study match these observations since no overall δ¹³C changes are seen in association with the six proposed pregnancies.

However, throughout the time of M₃ tooth formation, two women (A and B) display a wider range of δ¹³C variations (1.4‰) compared with δ¹⁵N (0.8‰), when no dietary or climatic events are indicated (Table 2). The effect of anabolic events like growth or lactation on δ¹³C may therefore be more important than presumed. Indeed, a small, controlled study with well-nourished macaques eating the same diet has shown δ¹³C and δ¹⁵N variations of comparable magnitude during lactation, with a significant δ¹³C increase of 0.4‰ and δ¹⁵N decrease of 0.3‰ displayed by lactating females (Reitsema and Muir 2015). Based on the Reitsema and Muir (2015) study, we expect to see increased δ¹³C alongside reduced δ¹⁵N during lactation (Figure 8). Certainly, following proposed pregnancies G_A and G_B, δ¹³C increases of 0.4‰ and 0.9‰, respectively, are seen during the suggested prolonged lactation (Figure 7). Yet, a diversity of δ¹³C responses to lactation have been observed in humans, suggesting that the interaction between lactation and δ¹³C may be more complex (Fuller et al. 2006; Herrscher et al. 2017). Indeed, in the studies by Fuller et al. (2006) and Herrscher et al. (2017), three women experienced δ¹³C elevations of about 0.6‰, while three displayed reductions of about 0.4‰. In our study, an eventual δ¹³C decrease of 0.9‰ is also seen after G_B (Figure 7). We suspect that such disparate δ¹³C responses could be linked to differing nutritional status. Indeed, the increase in δ¹³C values during lactation noted by Reitsema and Muir (2015) was in well-nourished animals. However, when protein intake is lower than requirements for both maternal maintenance and milk production, lactating mothers mobilise available nutrients from body tissues, including adipose tissues, as a source of carbon for protein synthesis (Institute of Medicine (US) Committee on Nutritional Status during Pregnancy and Lactation 1991). A similar situation is seen during starvation when the observed reduced δ¹³C values are attributed to an increased reliance on ¹³C-depleted lipids as a source of carbon for protein synthesis (Neuberger et al. 2013).

Projected isotopic patterns for growth (decreased δ¹⁵N and δ¹³C) are based on one animal case study and are therefore highly conjectural (Figure 8, Warinner and Tuross 2010). Extrapolating the results of animal studies to the interpretation of specific human situations is also complicated by their different size, physiology, metabolism and growth patterns. Generalising the results from Warinner and Tuross (2010) to healthy populations with ‘normal’ growth patterns is especially unreliable since the study compares one undernourished pig displaying faltering growth with three pigs showing ‘normal’ weight gain. Rather than simply characterising slower weight gain, the δ¹³C shift seen in the slower-growing pig may therefore represent a complex interaction between growth, protein and energy deficiency. The multi-layered relationship between δ¹³C, anabolism and nutrition proposed for lactation can be envisaged during accelerated growth. Certainly, in our study, a variety of δ¹³C patterns are seen in association with the proposed pubertal growth spurt. Women B and C display a δ¹³C decrease of 0.6‰ consistent with projected δ¹³C patterns based on Warinner and Tuross (2010) (Figure 7). However, δ¹³C elevations by 0.7‰ and 0.9‰ are seen in women E and A, respectively; while the initial δ¹³C reduction of 0.6‰ displayed by male J is followed by a δ¹³C increase of 0.6‰ (Figures 6 and 7).

Limitations

We acknowledge that given the lack of known puberty and obstetric histories, the identification of physiological events cannot be verified, and therefore interpretation of the results from this pilot study remains informed reasoning, while assessment of the potential of IDIA to reconstruct reproductive histories is tentative. The speculative nature of the interpretation is compounded by the uncertainty of the proposed projected isotopic patterns, especially for δ¹³C, resulting from the paucity of data in humans for the effect of physiology on isotope values, especially in relation to growth and lactation. The ambiguity of interpretation is also accentuated by the occasional non-specificity of isotope signals, the lower sensitivity of IDIA (compared with hair analysis), and limited understanding of the interaction between physiology, malnutrition and isotope values. The complexity of interpretation in relation to anabolic events was compounded by the complex geo-political history of Sudan, and consequent seemingly complex life histories of at least individuals F-I, including multiple concurrent events like droughts, famines, diseases and migrations, and increased risk of malnutrition. Results for those four individuals may not, therefore, be representative of other populations. The difficulty with generalising the results is especially true given that the data from only 10 teeth were analysed as part of this pilot study, including only 2 males.

Given the variability of M₃ formation and the lack of known extreme climatic events to aid in correcting the chronology for A, C and J, the reliability of absolute timings of physiological events for these three individuals is also poor (Mincer et al. 1993). Indeed, based on the identification of isotopic variations which can be linked to known droughts, delays by up to 3 years compared with the suggested onset at 8.5 years by AlQahtani (2009), are indicated for the other seven individuals. For female A, inaccurate absolute age assignment may explain the inconsistent results linking stress and puberty timing. Indeed, undernutrition, including malnutrition in utero or during infancy, is normally associated with delayed fertility, while obesity can result in early sexual maturation (Cole 2000; Rogol et al. 2002; Soliman et al. 2014). Certainly, male I, who experienced nutritional stress during the acceleration phase, displays a relatively late PWV at about 18 years, or F who experienced catabolism towards the start of puberty, exhibits the latest puberty onset for the group. Yet, A achieved her various puberty stages at the average time for the group despite probably experiencing malnutrition in early childhood.

Conclusion

This pilot study provides compelling indications that physiological events like pubertal growth, pregnancy and lactation are recorded in developing dentine; and therefore, that not only short-term physiological events should be considered when interpreting isotope ratios chronologies, but also that IDIA can help identify some events linked to reproduction and has the potential to support the study of growth progression and sexual maturation in past populations. However, some pubertal events were not observable or difficult to discern, and it seems likely that some pregnancies were missed. Extracting diet, geography and physiology information based solely on the dual analysis of δ¹⁵N and δ¹³C was also challenging on occasions. Because of limitations linked to the sample (especially the lack of medical data), the results and their interpretation remain uncertain, and the potential (and limitations) of IDIA when identifying reproductive events could not be established with confidence. The study also highlights important gaps in knowledge regarding the relationship between δ¹⁵N, muscle mass accretion, linear and ponderal growth in humans, as well as the interaction between isotope ratios, lactation (or growth), diet and nutritional status. An improved understanding of the interactive influence of physiology and nutrition on isotope values would be invaluable to reconstructing life histories in archaeological populations.

Studies including medical, growth, obstetric and life history information are needed to assess the ability and reliability of IDIA when identifying short-term physiological events. Analysing teeth that start developing earlier in life, over a more consistent timeframe overlapping with M₃ formation (e.g. canine or M₁), alongside M₃, would also help adjust the chronology and provide earlier isotope data when puberty starts before M₃ formation. Larger cohorts would help evaluate the projected isotope models for puberty, pregnancy and lactation too. Analysis of multiple isotopes or combining isotope analysis with other methods, like dental parturition lines and linear enamel hypoplasia, can also help distinguish more clearly the different factors influencing isotope values (e.g. Huss-Ashmore et al. 1982; Bol et al. 2007; Dean and Elamin 2014).

Acknowledgements

The authors are grateful to the Khartoum Centre for Research and Medical training for kindly giving us access to the teeth and associated demographic and anthropometric data, the participants who donated their teeth and the professionals who carried out the dental treatment and collected the data. We particularly thank Dr Yousra Mirghani (University of Khartoum, Sudan), Dr Israa Osman, Dr Israa Salak and Dr Nihal Abd El Azeem (Khartoum Centre for Research and Medical Training, Sudan) for their help with aspects of this study. We would also like to thank Marise Gorton and Dr Jacqueline Towers from the Stable isotope laboratory at the University of Bradford for their help with the analysis of the teeth, and the two anonymous referees for their careful reading of the manuscript and helpful comments.

Disclosure statement

The authors alone are responsible for the content and writing of the paper. The authors report no conflict of interest.