Dietary strategies of Pleistocene Pongo sp. and Homo erectus on Java (Indonesia)

During the Early to Middle Pleistocene, Java was inhabited by hominid taxa of great diversity. However, their seasonal dietary strategies have never been explored. We undertook geochemical analyses of orangutan (Pongo sp.), Homo erectus and other mammalian Pleistocene teeth from Sangiran. We reconstructed past dietary strategies at subweekly resolution and inferred seasonal ecological patterns. Histologically controlled spatially resolved elemental analyses by laser-based plasma mass spectrometry confirmed the preservation of authentic biogenic signals despite the effect of spatially restricted diagenetic overprint. The Sr/Ca record of faunal remains is in line with expected trophic positions, contextualizing fossil hominid diet. Pongo sp. displays marked seasonal cycles with ~3 month-long strongly elevated Sr/Ca peaks, reflecting contrasting plant food consumption presumably during the monsoon season, while lower Sr/Ca ratios suggest different food availability during the dry season. In contrast, omnivorous H. erectus shows low and less accentuated intra-annual Sr/Ca variability compared to Pongo sp., with δ13C data of one individual indicating a dietary shift from C4 to a mix of C3 and C4 plants. Our data suggest that H. erectus on Java was maximizing the resources available in more open mosaic habitats and was less dependent on variations in seasonal resource availability. While still influenced by seasonal food availability, we infer that H. erectus was affected to a lesser degree than Pongo sp., which inhabited monsoonal rain forests on Java. We suggest that H. erectus maintained a greater degree of nutritional independence by exploiting the regional diversity of food resources across the seasons. Trace element ratios (strontium/calcium) in teeth of Pleistocene Homo erectus and fossil orangutans (Pongo sp.) reveal different dietary strategies and contrasting adaptations to seasonal food resources. H. erectus but not Pongo sp. was able to buffer against seasonal food oscillations by exploiting more varied food sources.

During the Early to Middle Pleistocene, Java was inhabited by hominid taxa of great diversity. However, their seasonal dietary strategies have never been explored. We undertook geochemical analyses of orangutan (Pongo sp.), Homo erectus and other mammalian Pleistocene teeth from Sangiran. We reconstructed past dietary strategies at subweekly resolution and inferred seasonal ecological patterns. Histologically controlled spatially resolved elemental analyses by laser-based plasma mass spectrometry confirmed the preservation of authentic biogenic signals despite the effect of spatially restricted diagenetic overprint. The Sr/Ca record of faunal remains is in line with expected trophic positions, contextualizing fossil hominid diet. Pongo sp. displays marked seasonal cycles with ~3 month-long strongly elevated Sr/Ca peaks, reflecting contrasting plant food consumption presumably during the monsoon season, while lower Sr/Ca ratios suggest different food availability during the dry season. In contrast, omnivorous H. erectus shows low and less accentuated intra-annual Sr/Ca variability compared to Pongo sp., with δ 13 C data of one individual indicating a dietary shift from C 4 to a mix of C 3 and C 4 plants. Our data suggest that H. erectus on Java was maximizing the resources available in more open mosaic habitats and was less dependent on variations in seasonal resource availability. While still influenced by seasonal food availability, we infer that H. erectus was affected to a lesser degree than Pongo sp., which inhabited monsoonal rain forests on Java. We suggest that H. erectus maintained a greater degree of nutritional independence by exploiting the regional diversity of food resources across the seasons.
The Pleistocene hominid fossil record from the Sangiran Dome in Central Java, Indonesia, is one of the largest palaeoanthropological collections in Southeast Asia, evidencing an Early Pleistocene expansion of Homo erectus onto the Sunda Shelf [1][2][3][4] . The high morphodimensional variability of Indonesian hominid specimens led in the past to the attribution of the fossils to a variety of taxa such as H. erectus, Meganthropus palaeojavanicus, Pithecanthropus dubius or Pongo sp. and has fuelled taxonomic debates 1,5-9 . Recently, a high level of Javanese hominid palaeodiversity was revealed, which confirmed the taxonomic validity of the genus Meganthropus, a taxon that coexisted with H. erectus and Pongo 10 . Although dental macrowear and enamel thickness broadly reflect different dietary adaptations among these hominids 10 , little is known Article https://doi.org/10.1038/s41559-022-01947-0

Results
In Supplementary Table 2 we report Retzius periodicity (RP), laser track length and the corresponding time span for the analysed samples. The RP of Pongo sp. SMF-8864 was obtained through direct counts of cross-striations between two adjacent Retzius lines. The RP of H. erectus SMF-8865, given the section thickness necessary for chemical analyses and the presence of some accentuated markings, was calculated as the distance between adjacent Retzius lines divided by local daily secretion rate (DSR); the latter directly measured in areas of the section where the cross-striations were clearly visible. For H. erectus S7-37 P 4 , we report the RP calculated in ref. 21 for the S7-37 M 1 belonging to the same individual.
Elemental signals were retrieved within enamel close (<100 µm) to the enamel-dentine-junction (EDJ) because it is where environmental signals are best captured topographically during secretion and elemental overprint during enamel maturation has the least effect 17,26,[43][44][45] . For assessing postmortem diagenetic overprint, scatter plots of [Sr] or [Ba] versus [Mn] or [U] at EDJ profiles of representative samples of each trophic level were generated ( Fig. 1 and Extended Data Fig. 1). All cases show clearly positive correlations between trace elements and diagenesis-indicating element concentrations. Even though multistage diagenetic histories may be indicated by different trajectories (Fig. 1), uptake of Sr and Ba with increasing geochemical alteration is evident, which implies that the best approximations of initial biogenic [Sr]  about their life history and ecological niches. We used laser-based mass spectrometry to retrieve time-resolved information from their dental enamel about dietary diversity throughout the lives of individual early hominins, in conjunction with stable isotope analysis.
For decades, geochemical analyses-primarily δ 13 C and δ 18 O measurements-of tooth enamel have been used to retrieve palaeoenvironment, palaeodiet and life history information of extinct hominins such as Australopithecus [11][12][13] , Paranthropus 11,13-15 and Neanderthals 16,17 . This method has seen only limited application to hominins in Southeast Asia, with the exception of our own species. Consequently, we explored for the first time strontium/calcium (Sr/Ca) and barium/calcium (Ba/Ca) ratios and other trace elemental signals at high spatial/time resolution in the dental enamel of premolars and molars, to assess dietary and life history signals in Pleistocene H. erectus and Pongo sp. from the Sangiran Dome. Tooth enamel-contrary to bone and dentine-is less prone to postmortem diagenetic alteration due to its highly mineralized nature 18,19 . Moreover, it is secreted and mineralized sequentially in utero and during infancy to early adolescence and, once fully mineralized, remains compositionally and structurally stable during life. Consequently, enamel captures and preserves environmental and dietary changes that occur during the enamel development phases in an individual's life [20][21][22][23] . The incremental nature of tooth growth allows us to resolve shifts in enamel composition that relate directly to life history at near-(sub)weekly resolution. Moreover, tooth tissues in themselves contain information of potential taxonomic value 21 . The longer period rhythms in dental enamel (striae of Retzius periodicity, RP) have been used to infer taxonomic affinity in certain Far Eastern fossil specimens 24 . Indeed, elemental and isotopic analysis by laser-ablation inductively-coupled-plasma mass spectrometry (LA-ICPMS) across the incremental structures of sequentially secreted enamel provides a temporally and spatially highly resolved record of an individual's childhood. Such data allow the interpretation of diet, health, growth rates, weaning and mobility as well as changes of the environmental setting on a seasonal to weekly scale 16,17,[25][26][27] . Trace element ratios Sr/ Ca and Ba/Ca in dental enamel can record dietary signals due to the biopurification of Ca in trophic chains [28][29][30] . The higher the trophic level, the less [Sr] and [Ba] relative to [Ca] are incorporated into enamel, resulting in higher values of trace element ratios in herbivore enamel than in omnivores or carnivores 11,28,31 , although additional factors such as soil ingestion play a role 32 .
For comparison and as a trophic level reference for the Sangiran hominids, we used isolated premolars and molars of mammalian specimens belonging to different families (Felidae, Rhinocerotidae, Suidae, Cervidae and Hippopotamidae; Table 1) from the Sangiran fossil assemblage presumably co-existing with various hominid taxa such as H. erectus, Meganthropus and Pongo 10,33,34 .
All specimens were recovered from either the Early Pleistocene Sangiran Formation or from the later Early to initial Middle Pleistocene Bapang Formation, as both are fossiliferous and contain distinct faunal assemblages and taxa 4,35,36 . However, the exact stratigraphic allocation of all specimens is not documented 2,36 . The geological ages of the specimens range between 1.4 to 1.0 million years ago (Ma) and 1.0 to 0.7 Ma for specimens from the Sangiran and Bapang Formations, respectively 4 .
We focused our study on Sr/Ca (and to a lesser extent Ba/Ca) ratios as (relative) trophic level proxies, including an assessment of how well biogenic geochemical information is preserved in Pleistocene bioapatite from (sub)tropical contexts by using elements Mn, Al, Y, Ce, U as tracers of postmortem alteration 17,31,32,[37][38][39][40][41] . Previous stable isotope analyses of H. erectus bone samples from Sangiran were not successful in obtaining palaeoecological signals due to diagenetic alteration of bone tissue 42 . Here, we include sequentially microsampled carbon (δ 13 C) and oxygen (δ 18 O) isotope analyses of dental enamel of one H. erectus permanent premolar (S7-37) to contextualize our elemental results and obtain additional dietary/environmental information. Sr/Ca but more ambiguous ones for Ba/Ca ( Fig. 2 and Extended Data Fig. 7). As a result, we focus more on Sr/Ca results but also note that Ba/Ca can indicate reliable results in case of well-preserved samples (for example, Pongo SMF-8864, see below). The Sr/Ca ratio boxplots of faunal and hominid specimens (Fig. 2) show carnivorous Felidae with the lowest Sr/Ca ratio in the faunal assemblage (~8.4 × 10 −4 ), following the expected trophic level trend towards lower Sr/Ca ratios relative to omnivores (~1.1 × 10 −3 ; represented by Suidae) and different herbivore groups (1.6 × 10 -3 to 4.0 × 10 −3 ). Rhinocerotidae exhibit a Sr/Ca level about two times higher than all other herbivores and a broad Sr/Ca variability. The three H. erectus dental specimens yield Sr/Ca ratios between those of the Felidae and Suidae. The Pongo sp. specimen SMF-8864 shows the largest variation in Sr/Ca distribution among all taxa and has many distributional outliers towards higher Sr/Ca values (Fig. 2). Its median value fits well within the Hippopotamidae and Cervidae central distributions.    Fig. 3a). The consistency of the chronologies is attested by the high correspondence of the Sr/Ca signals between the two EDJ and prisms profiles. Pongo sp. SMF-8864 exhibits stark intratooth variability with three distinct peaks characterized by up to sixfold Sr/Ca and about eightfold Ba/Ca increases. This sixfold Sr/Ca change for the first peak (1.8 × 10 −3 to 10.7 × 10 −3 ) decreases for the second and third peaks to threefold and twofold values, respectively. The influence of the Sr/Ca attenuation along prisms towards outer enamel 26 is discernible but partly compensated for in, for example, prism 3 by the strong biogenic signal (Fig. 3b). On the buccal side, three hypoplastic defects and four accentuated lines (AL) are present (Fig. 3c), yet these non-specific growth disturbances 47 are not coincident with the Sr/Ca (or Ba/Ca) trends. The interval between the midpoints of two consecutive peaks on the buccal aspect approximates 1 yr, namely 364 and 324 d between peaks 1 and 2 and peaks 2 and 3, respectively. The duration of these peaks is 95, 118 and 90 relative days for the first, second and third peaks, respectively, approximating an overall duration of 3 months each.
The Sr/Ca profiles of the three H. erectus samples display low [U] and [Mn] and thus acceptable preservation, apart from localized peaks indicating spatially restricted diagenetic alteration (Fig. 4). Comparative elemental profiles for the lingual and buccal aspects of two H. erectus specimens presented in Extended Data Figs. 2 and 3 illustrate that enamel of the same tooth may be variably preserved yet we used the better-preserved domains. Limited intersample Sr/Ca variation ranges between 0.7 and 1.4 × 10 −3 , while intraprofile Sr/Ca variability is 20-30%. These H. erectus Sr/Ca values are thus always below those in Pongo sp. SMF-8864, which is even more pronounced for the intrasample variability (20-30 versus 200-600%). The temporal spacing between broad Sr/Ca troughs and/or peaks in all samples lies between 340 and 380 d, consistent with approximately annual cyclicity. As it is uncertain which of the apparent minor Sr/Ca fluctuations are indicative of variable food intake or minor cryptic diagenetic overprint, we refrain from attributing unwarranted importance to small-scale variability. Despite the uncertain molar position for H. erectus SMF-8865, the stability of the Sr/Ca ratio in the first 220 d of tooth formation suggests the absence of the breastfeeding signal 17,26 . Therefore, the tooth probably is not a first permanent molar, which starts to form earlier in life.
We report sequentially microsampled stable carbon and oxygen isotope compositions of enamel derived from H. erectus S7-37 P 4 (n = 3; Fig. 5 and Supplementary Table 1). The samples correspond to three distinct portions of the dental crown representing three partially overlapping life time moments. The δ 13 C values range from −4.9‰ to −2.4‰ (average = −3.9 ± 1.4‰ (1 s)), suggesting a diet which ranged from a dominated C 4 plant consumption to a mixed C 3 /C 4 plant consumption (54-72% C 4 fraction in the diet, calculated after ref. 48). The δ 18 O values remain stable with only very little variation between −6.7‰ and −5.9‰ (average = −6.3 ± 0.4‰).

Hominid Retzius periodicity
RPs of 7-9 d for our sample of H. erectus teeth are typical of these Pleistocene hominins. They are similar to the periodicities reported previously for H. erectus/ergaster molars and premolars (7-8 and 9 d, respectively) 21 but this apparent tighter distribution of values differs from the wider range of periodicities between 6 and 12 d characteristic of larger samples of living humans 49 . An 8 d periodicity for the Pongo sp. lower molar SMF-8864 is slightly lower than the 9-12 d periodicity reported for fossil Pongo from Sumatra and mainland Asia 50 but lies within the range of values (8-11 d) reported for living Pongo 51 .

Hominid trophic levels at Sangiran
Trophic levels portray the relative position of species in a food web and are important for ecosystem functioning 52 . Fossil teeth of Carnivora (Felidae), Perissodactyla (Rhinocerotidae) and Artiodactyla (Suidae, Cervidae and Hippopotamidae) from the Sangiran Dome with known trophic levels were used to establish an underlying relative trophic level framework for Sangiran. The ordering of fossil faunal taxa from Sangiran according to their enamel Sr/Ca ratios (Sr/Ca carnivores < Sr/Ca omnivores < Sr/Ca herbivores ) reflects trophic level differences that are in good agreement with their expected dietary habits (Fig. 2) 11,53 , suggesting reliable trophic level determination based on enamel Sr/Ca.
The Pongo sp. lower molar SMF-8864 exhibits a high intratooth variability, caused by cyclical Sr/Ca peaks (Fig. 3) along the EDJ profile covering the whole range of other herbivorous specimens in this study. The average Sr/Ca ratios between the peaks is closer to the Sr/Ca ratio of herbivorous animals such as Hexaprotodon sp. and Axis lydekkeri 54 (Fig. 2), suggesting an omnivorous diet with a certain degree of meat consumption for H. erectus on Java.

Comparison of Sr/Ca patterns in H. erectus and Pongo sp.
The biogenic Sr/Ca peaks in Pongo sp. SMF-8864 occur nearly annually (Fig. 3).

Diet of Pongo sp. reflects high seasonal food variability
The cyclical pattern of Sr/Ca and Ba/Ca peaks in Pongo sp. SMF-8864 with higher ratios occurring on an essentially annual basis gradually decreases within the ~3 yr of life represented by the tooth (Fig. 3). The repeatedly high Sr/Ca and Ba/Ca signals in this sample probably reflect annual periods with an increased intake of plant-based food resources, probably linked to a higher food availability during monsoonal periods, with a variation of the peak heights also linked to different food intake 60 . The duration and availability of food resources during the monsoon can fluctuate from season to season depending on monsoon intensity.
This might be the reason for the oscillation of the amplitude of the Sr/Ca and Ba/Ca peaks. Studies of palaeosols and the occurrence of palaeovertisols in the Sangiran Dome strongly suggest that Java was a monsoon region in the Early Pleistocene, with an annual dry season 3 . Monsoonal rain forest was probably the preferred habitat of Pongo sp. on Java. Indeed, palaeoenvironmental reconstructions propose that Java was dominated by a mix of savannah, open woodlands and monsoonal rain forests during the Early to Middle Pleistocene 3,61-63 . Besides differences in food intake, the gradual decrease in Sr/Ca amplitude across the life time of the individual might also be influenced by the geometry of the tooth cusp to cervix, which might alter the expression of Sr and Ba relative to Ca. Increased maturation overprint, which is inversely proportional to enamel thickness, where Sr and Ba signals reduce towards the thinner cervical enamel of the tooth, was observed in a previous study 26 . One recent study suggested a causal relationship to a cyclical nursing pattern, resulting in a cyclical increase of Ba concentration in teeth (through the increased intake of mothers' milk) 64 . However, the synchronous up to sixfold increase in Sr/Ca and up to eightfold increase in Ba/Ca are unlikely to reflect a breastmilk signal because breastmilk is Sr-depleted through epithelial discrimination within the mammary glands 17,26,65,66 . Recent studies on dentine and cementum in modern Pongo revealed that regions of [Sr] enrichment and depletion relate to both regular and irregular fluctuations in diet and Sr ingestion rather than to cyclical breastfeeding and may continue for as long as 20 yr into permanent canine tooth formation 67,68 . Caloric intake in orangutans is two to three times greater during supra-annual masting events where several fruit and other plant food sources happen to ripen at the same time 60 . Masting events are often then followed by periods of low fruit availability during dry periods, compensated in turn by orangutans burning fat reserves stored during mast-feeding 69 . Sr/Ca and Ba/Ca signals might also be enhanced during episodes of mast-feeding because of geophagic behaviour, that is the deliberate ingestion of soils enriched in trace elements, which absorb toxins and tannins and which appear to alleviate gastrointestinal upsets 68 . This behaviour was previously observed to be 'routine' in free-living orangutans 70 . The sharp and relatively higher Sr/Ca and Ba/Ca signal of the first peak may perhaps represent an occasion where the supra-annual masting event coincided with the monsoon.
It has been shown that non-specific stress enamel markers such as ALs can be correlated to variations in barium concentrations in dental tissues of primates 71 . In Pongo sp. SMF-8864 four ALs, occurring between the first and the second peaks (Fig. 3c), show a weak or absent correlation with elemental variations. However, the position of the ALs outside of the peaks' regions provide possible evidence of seasonal effects, as they might reflect stress events occurring during the first identified dry season. Hypoplastic defects on the tooth crown as a further sign of physiological stress do not correlate with elemental variations too (Fig. 3c) and indicate more complex, still-to-be-defined developmental deficiencies 17,47 .
Orangutans have the slowest life histories of any non-human primate with the latest weaning age of any mammal at around 7 yr but with relatively low levels of nutrient transfer during breastfeeding 64,[72][73][74] . Consequently, solid foods are supplemented in the infant's diet between 1 and 1.5 yr of age, to compensate additional nutritional demands 64,73 . Infants can forage solid foods independently from the age of ~1.5 yr, whilst the mother is not decreasing her lactation efforts 73 . Dry seasons with low food availability are compensated by extending weaning ages for infants leading to low growth and reproduction rates and solitary lifestyles 69,[75][76][77] .

Dietary strategy of H. erectus
The three H. erectus specimens show distinct Sr/Ca cycles with a duration of ~1 yr (Fig. 4). In contrast to the results from Pongo sp., the yearly Sr/Ca cycles in H. erectus are of low amplitude (20-30%), which are much smaller than the seasonal changes observed in Pongo sp. SMF-8864. For H. erectus, these might reflect the consumption of specifically selected animal or plant resources, which were available in the regional context of a highly diverse ecosystem. Our δ 13 C data show that the analysed H. erectus individual consumed a C 4 -dominated diet at the start of P 4 mineralization and then changed to a consumption of a mix of C 3 and C 4 biomass in the later stages of tooth development (Fig. 5). H. erectus probably inhabited an open mosaic setting with the C 3 signal indicating use of woodland/forest-edge habitats or gallery forests along rivers. The more C 4 -dominated diet suggests a tendency towards grasslands during the earlier period of tooth formation, possibly reflecting seasonal adaptations. The small variation of the relatively low δ 18 O values (Fig. 5) of the analysed H. erectus indicates that the individual had access to a water source with only small fluctuations in δ 18 O during the whole time of P 4 tooth formation. Therefore, H. erectus possibly exploited regionally available resources and consumed water and/or aquatic foods from, for example, rivers. Nearly 70 km east of Sangiran, at the site of Trinil, where H. erectus was first discovered and described 78,79 , it was suggested that members of this species probably consumed aquatic resources like shellfish, indicating a high level of food resilience 80 . In general, a high adaptive versatility is assumed for early members of the genus Homo 81 and dental microwear traits in Sangiran H. erectus teeth also confirm an opportunistic omnivorous dietary strategy 82,83 .

Conclusions
The main outcome of the present study is the demonstration that both Pongo sp. and H. erectus at Sangiran had cyclical food resource availability with an annual periodicity. However, distinct differences in their chemical patterns point to dietary and life history differences of Pleistocene Southeast Asian Pongo sp. and H. erectus, both reacting to seasonal resource variations differently. While Pongo sp. consumed contrasting plant-based food resources during the wet (monsoonal) season presumably available in monsoonal rain forests, H. erectus was more versatile and exploited a broader range of high-diversity food resources along open mosaic habitats possibly with a tendency towards grasslands, as suggested by the carbon isotopic data.
We demonstrate the effective use of histologically controlled time-resolved LA-ICPMS elemental analyses of hominid dental fossils to retrieve biogenic signals at subweekly time resolution. Our results show time-resolved geochemical analyses on H. erectus from the Sangiran Dome, which showcases the importance of geochemical analysis of fossil dental enamel of early humans to reconstruct past dietary behaviours and life histories in an evolutionary developmental perspective.

Methods
Overall the methodologies used here follow those in refs. 17,26 and only a brief summary is given here below. Article https://doi.org/10.1038/s41559-022-01947-0

Enamel thin sections
Preparation, imaging and histological analysis of enamel thin sections 84,85 were carried out at the Museo delle Civiltà in Rome. Sectioning was performed using a Leica high-precision diamond blade (Leica AG) and IsoMet low-speed diamond blade microtome (Buehler). Sections were ground with Minimet 1000 Automatic Polishing Machine (Buehler) using silicon carbide grinding papers with two grits (1,000 and 2,500; Buehler). Sections were polished using a Minimet 1000 Automatic Polishing Machine (Buehler) with a microtissue damped with distilled water and diamond paste (Diamond DP-suspension M, Struers) containing 1 µm monocrystalline diamonds. Thickness of the faunal thin sections was 130-150 µm, depending on the preservation and visibility of the enamel microstructure. The hominid section thickness varied between 250 and 400 µm, thus facilitating the geochemical analysis but ensuring sufficient readability of the enamel microstructures.

LA-ICPMS analyses
LA-ICPMS analyses were carried out at the Frankfurt Isotope and Element Research Centre (FIERCE), Goethe University (Frankfurt).
Histologically controlled tracks were determined on the enamel micrographs with Photoshop (Adobe). Sampling included continuous laser ablation tracks in enamel <100 µm parallel to the EDJ following the tooth growth direction 26 .
The LA-ICPMS system includes an 193 nm ArF excimer laser (RESOlution S-155; now Applied Spectra (ASI)) coupled to a two-volume laser ablation cell (Laurin Technic) 26,86 . The laser ablation system is connected to an ICPMS Element XR (ThermoFisher Scientific) using nylon6 tubing. Thin sections were ultrasonically cleaned with methanol and fixed in the sample holder together with a series of primary and secondary standards. The micrographs with premarked laser tracks were uploaded in GeoStar µGIS Software (Norris Scientific) and retraced before LA analyses. LA-ICPMS data acquisition was performed in continuous path mode due to the benefits of a two-volume LA cell with fast signal washout and constant signal response 26,86 .
Before analysis, laser tracks were cleaned with a bigger spot size (40 µm), higher repetition rate (20 Hz) and scan speed (varying between 16.7 and 30 µm s −1 , depending on the size of teeth) to remove surface residues, which could alter the results 87 . Analyses were carried out with a spot size of 18 µm, scan speed of 5 µm s −1 and a repetition rate of 15 Hz. The time signal obtained from the ICPMS can be directly transferred to distance along the LA tracks via the constant scan speed of the laser X-Y stage; no time delays of the X-Y stage exist at waypoints of composite tracks 26 . Between the LA system and the ICPMS, a signal smoothing device (squid) was included 86 .
The ICPMS (Element XR) detected the following isotopes from the ablated sample material (m/z): 25 Mg, 27 Al, 43 Ca, 44 Ca, 66 Zn, 86 Sr, 88 Sr, 89  Secondary standards with known concentrations and a matrix broadly similar to apatite (STDPx glasses) were analysed to assess accuracy and precision: STDP3-150, STDP3-1500, STDP5 (Ca-P-(Si) glass standards) 91 , KL2-G (basalt glass) 92  The compositional profiles displaying the concentration of elements relative to distance d −1 along the EDJ profile were smoothed with a locally weighted polynomial regression fit, with its associated standard error range (±2 s.e.) for each predicted value 95 . The software R (v.4.0.4) and the packages 'lava', 'readxl', 'shape' and 'tidyverse' were used for all statistical computations and generation of graphs.
Elemental data were matched with odontochronologies of the H. erectus and Pongo sp. specimens by determining the chronology of each EDJ track after LA-ICPMS analysis (Extended Data Fig. 8) and directly assessing the enamel DSRs. DSR, in other words, the speed at which the ameloblast (the enamel forming cells) move towards the outer surface of the tooth is expressed in µm d −1 along the prisms 96,97 , in the 100 µm region close to the EDJ. Carefully chosen histologically defined (EDJ) profiles facilitate the correlation between odontochronological and geochemical signals at a very high time resolution (<1 week).

Isotopic ratio mass spectrometry analyses
Stable carbon and oxygen analyses of S7-37 (right P 4 ) were performed at the Goethe University-Senckenberg BiK-F Joint Stable Isotope Facility Frankfurt, Germany. For each sample, 2.9-3.8 mg of enamel powder was retrieved with a hand-held diamond tip dental drill. To produce sufficient sample material, drill holes were expended along to the growth axis of the enamel.
To remove organic matter and potential diagenetic carbonate, enamel was pretreated with 2% NaOCl solution for 24 h followed by 1 M Ca-acetate acetic acid buffer solution for another 24 h and thoroughly rinsed with deionized water (modified after ref. 98). Typically, enamel pretreatment resulted in ~60% mass loss. Then, 950-1,100 µg of pretreated enamel powder were reacted with 99% H 3 PO 4 for 90 min at 70 °C in continuous flow mode using a Thermo Finnigan 253 mass spectrometer interfaced to a Thermo GasBench II. Analytical procedure followed the protocol of ref. 99. Final isotopic ratios are reported versus V-PDB; overall analytical uncertainties are better than 0.3‰ for δ 13 C and 0.05 for δ 18 O.

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability
Data of elemental analyses and corresponding contextual information obtained in this study are available as Supplementary Data.  Corresponding author(s): Julide Kubat, Alessia Nava, Wolfgang Muller Last updated by author(s): Oct 13, 2022 Reporting Summary Nature Portfolio wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Portfolio guidelines for submitting code & software for further information.

Data
Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: -Accession codes, unique identifiers, or web links for publicly available datasets -A description of any restrictions on data availability -For clinical datasets or third party data, please ensure that the statement adheres to our policy The dataset generated during and/or analyzed during the current study are available in the manuscript and the supplementary information files. Tick this box to confirm that the raw and calibrated dates are available in the paper or in Supplementary Information.

Ethics oversight
Ethical approval and guidance was not required, as we sampled fossil collection specimens only.
Note that full information on the approval of the study protocol must also be provided in the manuscript.