Sigatoka sand dunes archaeological context and chronology issues

The Sigatoka Sand Dunes archaeological site, located on the southwest coast of Viti Levu, Fiji Islands (Fig 1), is of great importance in defining Fijian and Pacific prehistory [58]. The site (VL16/1) is part of an extensive parabolic dune system, covering an area over 5 km long and 20 to 60 m high, east of the mouth of the Sigatoka River. Since 1944, it has yielded cultural remains of various forms, preservation and nature, ranging from potsherd scatters, fireplace scatters, fire-broken rocks, animal and human bone scatters, to earth oven features, postholes and formal burials protected by cairns [59–62]. Often exposed by erosion, the remains have been recovered in three stratigraphic levels (Levels 1, 2, 3) associated with periods of dune stability [63,64]. These levels represent stages of the whole sequence of the Fijian prehistoric chronology, from the Lapita to the historical periods. Their dating is generally discussed in relation with the discontinuous processes of dune formation and erosion.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Geographical map showing location of site Sigatoka (Viti Levu, Fiji) and other Southwest Pacific comparisons sites cited in text.

The map was created with Shuttle Radar Topographic Mission coordinates available from CIAT–CSI SRTM website [65].

https://doi.org/10.1371/journal.pone.0300749.g001

Using optical and radiocarbon data, Anderson et al. (2006) [58] showed consistency between radiocarbon and optical ages for Level 1, with dates around 2,800–2,600 cal BP, and Level 3, with dates in the 600–500 cal BP range. They also observed a time lag between the natural formation of Level 2, which starts around 2500–2300 cal BP and human occupations, which extends from 1,700 to 1,300 cal BP [58]. On the other hand, diversity of vessels forms and others characteristics were used to established a four and then five phase ceramic chronology [59,61,66,67]. While some authors have identified the transition from Fiji Plainware to Navatu around 1800 years ago using data from various sites across the Fiji Islands [68–70], this transition appears to occur later at Sigatoka, during the process of formation of Level 2 [59]. Excavations at the easternmost end of the dune revealed two stratigraphically separate but superimposed domestic occupations corresponding to the Fijian Plainware and Navatu phases respectively, with a village occupation dated to 1,433–1,298 cal BP and a subsequent occupation focused on sea salt production dated to1,330–1,266 cal BP [59,71–73]. The change from Fijian Plainware to Navatu is seen as an abrupt cultural modification, interpreted by some as indicating interactions with external population(s) (e.g. [71]].

Human burials have been identified in Levels 2 and 3. Marshall and collaborators [62] have inventoried a total of 119 burials discovered during surveys or archaeological rescue investigations during the 20^th century (up to 1998). Their spatial distribution is interpreted as forming three burial grounds, together with several smaller groups of burials or isolated burials scattered across the dune [62: 44–53]. Possible Burial Ground 3, located in the east part of the mapped section of the dune, in Level 3 and above Burial Ground 2, consists of at least 18 interments in deep pits probably related to the last 500 years. Burial Ground 2, which is associated with Level 2, consist of at least 13 poorly preserved burials in an area measuring 25 by 15 m located at about 200 m west of Burial Ground 1. Excavation at Burial Ground 2 did not reveal any coral or basalt fragments covering the burials, some of which consisted of multiple inhumations containing two to four individuals in flexed position following an east–west orientation [52: 110–114,62,74]. Burial Ground 1, where the individuals studied here were found, is the largest burial area. Located in the eastern most part of the dune, it has yielded at least 61 burials associated with Level 2 [62], including 55 (52 labelled and 3 fragmentary) individuals buried in 35 graves excavated by Best [50,57]. A further six interments were subsequently uncovered on the edge of Burial Ground 1 [62: 48–49].

Some scholars have associated the Level 2 burials with the 1,700–1,300 cal BP interval (radiocarbon dates on wood charcoal) [58]. Burley [59,71] raises the possibility that Burial Ground 1 is more closely associated with the Fijian Plainware village occupation found in the easternmost part of the dune and dated to 1,433–1,298 cal BP [31,32,73]. So far, the contemporaneity between the village and Burial Ground 1 has not been established using radiocarbon determinations. Only one burial was radiocarbon-dated with a result outside the chronological range proposed for the village (burial FC1, 1,870±70 BP (Wk–996b), [50]). This result is regarded as questionable due to inadequate bone pre-treatment for radiocarbon dates and unknown diet [75], leaving us with a lack of direct dates for the mortuary contexts.

Sigatoka Burial Ground 1 graves and individuals

The 55 burials excavated by Best in Burial Ground 1 were in an area measuring 23 by 15 m [57] (Fig 2). They display a consistent pattern of positioning and orientation. The majority of individuals, male, female and children, were buried on the back with the lower limbs flexed to varying degrees, following an east–west orientation. The burial treatment of at least some individuals may have been more complex, comprising a stage of defleshing as indicated by the sharp–force trauma observed on several bones and individuals [75]. Associated artefacts were rare, one stone adze and few shell artefacts, and found in only four burials (B10a, B17, B19, FC1). Most of the graves (n = 20 out of 35) were capped with a cairn of coral or basalt blocks (Figs 2 and 3). These cairn graves contained equal numbers of males and females, and, children in few cases. Graves without a cairn (n = 15) contained mainly females and children (n = 13 out of 15), only one or two were identified as male [28,76]. Some of these cairn graves (n = 8) were single inhumations, while others (n = 12) consisted of group interments of two to four individuals. In one case (B21/22), individuals (two females and an adolescent) were buried successively in the same location (Fig 3), and perhaps in a second case corresponding to the interment of a young child (6–7 years, B9b) found in the cairn of a female (B9a). In the 10 others, individuals were buried as cadavers during a single mortuary event involving several individuals (Fig 3). This practice of burying individuals together at the same time has been interpreted as reflecting a social or familial link, that bound the deceased during their life [57]. This practice can also be seen as an evidence of contemporaneity between the individuals involved.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Map of Sigatoka Burial Ground 1.

Grave distribution function of topography, burials B10a,b,c are at the highest point (modified from [57] by Hemmamuthé Goudiaby). Reprinted from [57] under a CC BY license, with permission from Simon Best, original copyright 1989.

https://doi.org/10.1371/journal.pone.0300749.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Examples of Sigatoka Burial Ground 1 graves.

Left: B21/22 illustrate the single case of successive inhumations at the same location, under cairn. Right: B4 illustrating multiple simultaneous inhumations under cairn (Photos and drawings courtesy Simon Best, modified and assembled by Hemmamuthé Goudiaby). Reprinted from [57] under a CC BY license, with permission from Simon Best, original copyright 1989.

https://doi.org/10.1371/journal.pone.0300749.g003

The spatial organisation of the graves, following the crest of the dune, does not show a specific pattern in the form of strict alignments or rows. Rather the spatial pattern seems to correspond to localised clusters of graves [57] (Fig 2). Their location in relation to the topography of the site appears to be significant. Cairn graves containing males (B10a, b, c), in one case with a number of artefacts (B10a), were found at the highest point of the site, while graves containing females and other individuals were at a lower elevation. Female and child burials, which lacked superstructures and less systematic in their position and orientation, were located toward the northern fringes of the cemetery. This pattern has been interpreted as reflecting social differentiation within Sigatoka community [28: 170–173], with individuals buried under cairns at higher elevations having a higher status in society [57]. It has also been speculated that the spatial separation of Burial Grounds 1 and 2 may reflect the spatial segregation of different segments of the same community [62: 110–111].

Inter-islands contacts may have influenced Sigatoka population lifestyles and dietary practices, with migrants introducing new resources and expertise. Some scholars, using morphological data from skeletal remains from Burial Ground 1 and conducing comparative analysis, suggest that early inter–islands contacts between Viti Levu population and Melanesian region populations already occurred by the beginning of the first millennium CE [80]. As cranial elements exhibit morphological affinities with populations from New Caledonia and Loyalty, in particular, others argue for a later movement of population from North Melanesia [28,77]. Indeed, some levels of population movement between Fiji, Vanuatu, and North Melanesia are recognized during the last millennium CE [78]. Migrants identified through strontium analysis, mainly associated with the Vuda Phase (450–350 year ago) of the Fijian chronology [32] were buried in the Sigatoka Sand Dune. Interestingly, although controversial, the results of an early ancient DNA (mtDNA) study showed that the two individuals tested from Burial Ground 1 display the presence of two copies of the 9 b.p. repeat, a feature common in the mitochondrial genome of present–day populations from the Melanesian region [79].

Assessment of health status points to general good sanitary conditions and adequate nutrient intake during both childhood and adulthood [28,76,80,81]. Low rates of linear enamel hypoplasia (LEH), absence of cribra orbitalia were recorded in both children and adults from Burial Ground 1. Signs of infection, seen in three males (B10a, B5c) and female (B18c), and healed trauma, seen in six males (B4c, B10b, B10a, B17b) and females (B4a, B14), are rare. Osteoarthritis and occupational skeletal changes, which occur in both males and females, are also rare, particularly in the postcranial skeleton [76]. Marked joint changes were recorded in the glenoid fossa of some males (n = 3) and females (n = 2). Oral health is good, with the exception of high levels of tooth wear (dentin exposure accounts for about 70% of observations, [76] and antemortem tooth loss [28,76,81]. There is limited evidence of dental caries and periodontal disease and an unusual wear patterns in 6 females (B2c, B3b, B3d, B4b, B6a, B9a) and one male (B8), suggesting non–dietary use of teeth for at least some of the individuals [28,80,81]. Sexual differentiation in dental wear, antemortem tooth loss and temporo–mandibular joint arthritis patterns of co–variation points to sex specific activities with males chewing kava [80] and females perhaps processing fibrous plant material for fishing nets and lines [81].

Radiocarbon dates

Thirteen human bone samples from Sigatoka Burial Ground 1 were tested for AMS radiocarbon determinations at the Waikato Radiocarbon Lab (Hamilton, New Zealand) and at the Poznan Radiocarbon Laboratory, Poznan, Poland (S1 Table). All dates were calibrated using OxCal 4.4 software [82] and the SHCal20 calibration curve, with correction for the marine contribution to human diet [83,84]. For the correction, we use the linear extrapolation between terrestrial (−21 ‰) and marine (−12 ‰) δ¹³C endpoints to obtain an approximation of the marine contribution to diet, as proposed in Petchey study of early Pacific series [20,85,86]. ΔR and its uncertainty were calculated using the weighted average of the 10 nearest points of the site using the Marine20 database [87].

Age at death and sex variation analysis

We have focused here on age at death indications provided by subadult mortality variation in order to detect anomalies of representation and structure in the age distribution of the Sigatoka Burial Ground 1 series. For this analysis, 41 adults (over 20 years old) were combined into a single group and 11 subadults were distributed into five age categories [0] year, [1–4] years, [5–9] years, [10–14] years, [15–19] years using data presented in Pietrusewsky and collaborators [76]. Sigatoka Burial Ground 1 mortality tables and curves were reconstructed from the raw data by calculating mortality quotients (aQx) for each age category. These were then compared to our model of natural mortality and to two mortality profiles of series that experienced epidemics and famine. Our non–sex specific model of natural mortality is established using Lederman life tables [88] for pre–vaccination populations with life expectancy at birth between25 to 35 years, for which Lederman provides means values and standards deviation that allow the calculation of quotient confidence interval at 95% (Réseau 100, Tableau 100 MF, p 52) [89,90]. Epidemic and famine profiles were constructed using European series related to the 16^th century plague resurgence (Belgium, Dendermonde, archaeological context, [91,92]) and 19^th–century famine context (Ireland, Workhouse (1841–1851), register, data in [93]). Sex ratio was calculated for individuals over 20 years old using sex estimations provided in Pietrusewsky and collaborators original report [76].

Isotopic analysis: Sampling strategy, dietary assessment approach

Our sample differs from that of Phaff et al. (2016) [31] in several respects. It includes individuals and paired bone–tooth samples not represented in that publication and for consistency we propose here a new set of analyses performed with the same protocol and analytical measurements, rather than combining Phaff et al.’s data and ours. Our isotopic study uses teeth (all molars, 24 M3, 4 M2 and 2 M1) and bone samples collected on 38 adult individuals among the 41 labelled at Burial Ground 1. These 38 individuals are represented by either a bone or tooth samples, 30 of which represented by paired bone–tooth. The later stage of adult life [94] was documented by bone samples taken preferentially from the skull (n = 13) and long bones (n = 11) as well as the mandible (n = 7) and other skeletal elements (vertebra, scapula and phalanx, n = 7) (S2 Table). The third molar was preferentially selected (n = 24), followed by the second molar for four individuals (B3d, B8a, B10c, B21/22b) and the first molar for two (B10a, B10b). As described in AlQahtani and collaborators’ chart [95], the half distal part of the first molar root captures biological information corresponding to an age of 6.5–9.5 years, the second molar to 10.5–13.5 years and the third molar to the period between 15–20 years. Ten bioarchaeological and mortuary markers: sex, morphology [96], bucco–dental lesions related to physiological stress (LEH, caries, premortem tooth loss, and evidence for periodontal disease, data in Pietrusewsky and collaborators [76,81]), and graves characteristics [57] were used to analyse intra–individual isotopic variations. Funerary, biological and health data for each individual are reported in S2 Table. The isotopically studied sample comprises 16 males and 22 females, with 8 individuals are aged below 40 years old and 30 above, based on original data from Pietrusewsky and collaborators [76].

In order to better evaluate the dietary habits and food procurement strategies of the individuals buried at Sigatoka Burial Ground 1, we compared our data with those from other earlier and contemporaneous series from the Southwest Pacific. To this end, we selected data from 112 individuals in 11 publications (Table 1). These include data from coastal localities in Palau [23], Vanuatu [13,17,18,21], New Caledonia [85], Fiji [19,27,85,97,98] and Tonga [24,85]. Table 1 summarises the contextual information and dietary interpretations provided in the original publications and Fig 1 shows their geographical locations. The data selected for our purposes represent individuals older than 5 years with collagen integrity criteria meeting Ambrose’s recommendations [99] (S3 Table). These were divided in three temporal groups using direct dates of the human skeletal remains and/or controlled stratigraphic information available in the original publications (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Chronological and dietary contexts of Southwest Pacific Islands individuals selected for comparison with Sigatoka Burial Ground 1 individuals (individuals and sites with insufficient chronological data [22] have been omitted).

https://doi.org/10.1371/journal.pone.0300749.t001

The interpretation of human isotope values aims to identify the dietary resources that were preferentially consumed. The most reliable modelling of the human diet requires knowledge of the isotopic variability of foods available in the local environment of the population under study This baseline is now well–described [24,49,107] thanks to more than 400 stable carbon and nitrogen isotope ratios showing the wide isotopic variability of the local food resources [24]. To infer the consumed dietary resource, we used a theoretical isotope fractionation between food and consumer tissue, of 3 to 6 ‰ for nitrogen, and 0 to 2 ‰ for carbon, at collagen level. We include a +5 ‰ offset due to fractionation during carbon incorporation from fresh food (plant, flesh) into collagen tissue [4,108] and a correction for modern food (+0.86 ‰ for marine ecosystems and +1.5 ‰ for terrestrial ecosystems [13,49]. Based on the Pacific baseline, described elsewhere [24: 312], it is possible to differentiate individuals consuming food of different trophic levels within the foodweb (low trophic level: terrestrial C3 and C4 plants and inshore resources such as shellfish and algae; intermediate trophic level: coral reef fish, non–reef fish and freshwater fish; high trophic level: marine mammals and turtles and terrestrial animals) and individuals consuming terrestrial or marine resources.

Isotopic analysis: Protocols and statistical analysis

Methods of pre–treatment protocol, isotopic measurements and validation criteria are reported in S1 Text. All archaeological and biological criteria were explored with statistical tests performed with R version 3.6.1 with RStudio [109] with the package Rmisc for descriptive analysis. The non–parametric tests Kruskal Wallis and Mann–Whitney tests are used, when differences are detected, post–hoc comparisons are conducted using the FDR correction [110]. A 0.05 probability (p < 0.05) is considered significant. Data are visualized using ©Excel and the package ggplot2 and ggpubr in RStudio.

AMS radiocarbon dates

Ten direct AMS radiocarbon dates were obtained from the 13 tested human bone samples from Burial Ground 1 burials. The dates, presented in S1 Table, range from 1,888 to 1,272 cal BP. This range includes the date previously obtained for burial FC1 (1870±70 BP Wk–996, [57]) and overlaps the middle of the first millennium CE (Fig 4). The dates overlap only partially the range of Level 2 radiocarbon dates known to be 1,700–1,300 BP [58] or 1,750–1,550 BP [62: 71], and they are consistent with the range given by optical dating for the start of formation of this level [58]. The individuals buried at Burial Ground 1 represent thus a population that can be considered to be derived from the earliest inhabitants of the area associated with the Sigatoka/Lapita phase [67]. Our results place the burials after the Fijian Plainware Navatu transition as defined by Best and Clark [57,68,70] and before the Navatu phase as defined by Burley and Edinborough (1,330–1,266 cal BP). Some overlap the period associated with the Fijian Plainware village occupation (1,433–1,298 cal BP) [73].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. AMS radiocarbon results.

Calibrated using using OxCal 4.4 software [82] and SHCal20 calibration curve [83] and hypothesised time of Fijian Plainware/Navatu transition according to Best [57,68] and Clark [70] in green and Burley [73] in red [see Cochrane [67] for a comparison of different ceramic chronologies).

https://doi.org/10.1371/journal.pone.0300749.g004

Age at death and sex composition results

Our analyses of the sex ratio and distribution by age at death tend to support Visser [28] and Pietrusewsky and collaborators [76] views. The ratio of males to female is almost the same (sex ratio = 0,77; χ² test is not significant (χ² = 3.389 p = 0.066). We also found that quotients of children aged between birth and 4 years old are significantly lower than expected while that of individuals aged between 19 and 5 years are consistent with our model of natural mortality (S4 Table). A deficit of young children, and particularly those under 1year of age, is a feature of catastrophic mortality, including epidemics of plague and famine (e.g. [92,93,111]). However, other characteristics of mortality crises do not co-occur in the Sigatoka studied sample, suggesting rather the possibility of a cultural selection of the individuals buried at Burial Ground 1, with those buried representing only part(s) of a larger community. This hypothesis is consistent with Marshall and collaborators suggestion that the spatial separation of Burial Grounds 1 and 2 represents a spatial segregation of different segments of the same community [62: 110–111].

Isotopic results

Quality check and general results.

Of the 30 teeth sampled, seven did not provide sufficient collagen samples for measurements and four collagen samples did not meet quality criteria. Of the 38 bone fragments sampled, five did not provide sufficient extraction yield for isotopic measurements and seven collagen samples had criteria of %C, %N and C:N atomic ratios outside the validation ranges for reliable isotopic data. At the end, 25 bones and 19 teeth have well preserved collagen sample for isotopic analysis and paired bone–tooth measurements are available for 15 individuals (S2 Table).

Teeth δ¹³C values range from −17.2 to −14.0 ‰, and δ¹⁵N values from 9.5 to 11.2 ‰ (mean ± 1SD, −15.9 ± 0.8 ‰; 10.2 ±0.4 ‰, n = 19). Bone δ¹³C values range from –17.2 to –14.5 ‰, and δ¹⁵N values range from 5.4 to 10.8 ‰ (−15.6 ±0.7 ‰; 9.2 ± 1.1 ‰, n = 25) (Fig 5). Mann Whitney tests show that teeth δ¹⁵N values are statistically higher than bone values (p = 0.000; Δ¹⁵N_teeth–bone = 0.9‰), whereas differences between teeth and bone δ¹³C values are not significant (p = 0.270) (Fig 5). For the 15 paired bone–teeth individuals, teeth δ¹³C values range from −17.2 to −14.0 ‰, and teeth δ¹⁵N values from 9.5 to 11.3 ‰ (mean±1SD, −15.9 ± 0.9‰; 10.2 ± 0.5 ‰, n = 15) while bone δ¹³C values range from –17.2‰ to –14.5‰, and bone δ¹⁵N values from 8.2 to 10.8 ‰ (−15.5 ± 0.8 ‰; 9.5±0.7 ‰, n = 15). Similarly, Mann Whitney tests show that teeth δ¹⁵N values are statistically higher than bone values (p = 0.006; Δ¹⁵N_teeth–bone = 0.7‰), while differences between teeth and bone δ¹³C values are not significant (p = 0.119). Globally, δ¹³C and δ¹⁵N bone values present a wide dispersion, 2.7 ‰ and 5.4 ‰, respectively, spanning more than one trophic level. Thus, our results indicate the presence of several dietary behaviours among the Burial Ground 1 individuals.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Carbon and nitrogen isotopic ratios of bone and teeth samples for Sigatoka Burial Ground 1 individuals.

(Mann-Whitney test between bone values of North Melanesian (NM) and South Melanesian (SM) morphology, p-(δ¹³C) = 0.0031, p-(δ¹⁵N) = 0.016 respectively).

https://doi.org/10.1371/journal.pone.0300749.g005

Isotopic values and criteria analysis.

We evaluated the relationship between isotopic values and ten bioarchaeological and mortuary criteria for all available individuals for each type of tissue (regardless their number) (Fig 6, Table 2). Only carbon isotopic ratios show one significant difference, with δ¹³C values higher in bone than in tooth for females, suggesting an increased intake of enriched in ¹³C foods, such as marine items, during their life course (Δ¹³C_teeth–bone = 0.6 ‰, Mann–Whitney test, p = 0.039).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Carbon and nitrogen isotopic ratios of bone and teeth samples for Sigatoka Burial Ground 1 individuals according to archaeological and biological criteria.

(See Table 1 for statistics; Without: single individual without cairn; With: several individual under a cairn. For biological criteria: <40: below 40 years old; >40: above 40 years old; NM: North Melanesian; SM: South Melanesian. For health criteria: 0: no lesion; 1: presence of lesion; LEH: Linear enamel hypoplasia; Calc.: calculus; Roll: rolled rim; Res.: alveolar resorption; ATM: premortem tooth loss).

https://doi.org/10.1371/journal.pone.0300749.g006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Sigatoka Burial Ground 1 intra–individual analysis.

Descriptive statistics of isotopic data according to tissues (B = bone, T = teeth) and bioarchaeological and archaeological criteria.

https://doi.org/10.1371/journal.pone.0300749.t002

Nitrogen isotopic ratios show systematically higher values in teeth than in bones (Fig 5). Females have significantly different median nitrogen values, with tooth value being 1.1 ‰ higher than bone value (p < 0.001); males do not show such a difference. Individuals of North Melanesian morphology show a weak difference between bone and teeth suggested by a lack of overlap in values. Individuals of South Melanesian morphology show no significant difference. Interestingly, individuals with and without oral lesions show higher isotopic nitrogen values in teeth than in bone. Median nitrogen values are significantly higher in tooth than in bone by +0.8 ‰ for individuals without hypoplasia, +1.1 ‰ for those without calculus, +1.2 ‰ for those without premortem tooth loss and by +1.0‰, +0.9 ‰ and +0.4 ‰ for those with caries, rolled rim and premortem tooth loss, respectively (Table 2). There is also a significant difference in median nitrogen values of +1.1 ‰ and +0.7 ‰ between bone and teeth for both individuals died below and above 40 years old (p = 0.015, p = 0.017). Finally, individuals buried under a cairn (grouped or isolated) display significant difference between bone and teeth (Δ¹⁵N_teeth–bone = +0.8 ‰, Mann–Whitney test, p < 0.001). Individuals buried without cairn display no difference between tissues.

Paired bone–teeth and patterns.

Focusing on the paired bone–teeth isotope values, we found that 13 of 15 individuals, documenting the end of adolescence (between 15–20 years, M3 sampling), have higher tooth nitrogen isotopic ratios than the two others, documenting childhood (B8–M2, B10a–M1) (Fig 7A). We also observe that (i) 14 out of 15 individuals present a difference in nitrogen isotope ratios greater than +0.4 ‰ and (ii) 12 out of 15 show significant ¹⁵N depletion over the individual’s life course (Δ¹⁵N_bone–teeth < –0.4 ‰) (Fig 7A). Carbon isotopic differences between paired bone–teeth show significant enrichment in ¹³C for seven individuals (B1a, B4a,b, B5c, B6a, B14, B15: Δ¹³C_bone–teeth > +0.4 ‰) and depletion of ¹³C for two individuals (B4c, B13a: Δ¹³C_bone–teeth < –0.4‰) (Fig 7B). Nine show a significant difference for both carbon and nitrogen isotope values. The isotopic shifts are different for each element (carbon or nitrogen) and do not affect the direction of the shift between paired bone–teeth in the same way (Fig 7C). Thus, the isotopic differences between the two tissues display six different isotopic patterns that can be interpreted as representing different dietary or physiological modifications during life (Fig 7C):

• Pattern 1 corresponds to an increase (> 0.4‰) of δ¹³C values and a decrease of δ¹⁵N values from teeth (childhood) to bone (adult). It is the most recurrent pattern, affecting six of the 15 individuals (B25, B6a, B14, B1a, B4a,b). The opposite isotopic changes of the two elements (Δ¹⁵N_bone–teeth < –0.4 ‰, Δ¹³C_bone–teeth > +0.4 ‰) suggest an increasing dependence on low trophic level marine resources such as shellfish and seaweed and/or a decrease of animal protein intake from adolescence to adulthood.
• Pattern 2 shows a decrease of δ¹⁵N values of four individuals (B21/22b, B13/1, B5c, B1c). The negative Δ¹⁵N_bone–teeth values (< –0.4‰) indicate a greater reliance on terrestrial plant resources over the course of their life.
• Pattern 3 is characterised by a significant decrease in both carbon and nitrogen isotopic ratios from teeth (childhood) to bone (adult), observed in two individuals (B13a, B4c). With negative values for both Δ¹⁵N_bone–teeth and Δ¹³C_bone–teeth, these two individuals would have had a lower terrestrial and marine animal protein intake in adulthood than in adolescence.
• Pattern 4 is characterised by an increase in both carbon and nitrogen isotopic ratios visible in only one individual (B10a), suggesting a greater terrestrial/marine animal protein intake in adulthood than in childhood.
• Pattern 5 only shows an increase of δ¹⁵N values in one individual (B8). Although the pattern is different, a similar dietary interpretation to Pattern 4 can be proposed, with a greater reliance on animal proteins in adulthood than in childhood.
• Pattern 6 shows no isotopic change for either carbon and nitrogen, and involves only one individual (B18a), which would support the hypothesis of a similar diet throughout her life.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7.

Paired bone–tooth isotopic values for Sigatoka Burial Ground 1 individuals (A: Paired bone–tooth nitrogen isotopic ratios. B: Paired bone–tooth carbon isotopic ratios. C: Δ¹³C_bone–teeth (δ¹³C_bone–δ¹³C_teeth) and Δ¹⁵N_bone–teeth (δ¹⁵N_bone–δ¹⁵N_teeth), a negative Δ value indicate a higher collagen isotopic value in teeth than in bone, corresponding to an impoverishment of heavy isotopes over the life). Labels in Bold: South Melanesian morphology, Labels in Bold–Italic: North Melanesian morphology).

https://doi.org/10.1371/journal.pone.0300749.g007

We also note that nine of the 15 individuals with paired bone–tooth (B1a, B4a,b,c, B6a, B10a, B13a, B14, B25) with significant isotopic shift during life, for both δ¹⁵N and δ¹³C, are dispersed across the cemetery (Fig 2: 1/4 NE: B13a; SE: B1a; SW: B6a; NW: B14, B25; SW: B4a,b,c, B10a). Similarly, these individuals do not show any specific dietary trends/patterns in relation with their grave form or mortuary treatment. Of the six individuals with Pattern 1, four were buried with others at the same time under the same cairn, one was interred alone under a cairn and another without a cairn. Contemporaneous individuals (buried simultaneously soon after death) display different patterns like for grave B4 or grave B1. For example, in grave B4 (Figs 3 and 7C), the two females (B4a,b) show a decrease in trophic level associated with an increase in marine food over their lifetime (Pattern 1) whereas the male (B4c, Pattern 3) is characterised by a shift towards a more terrestrial plant protein intake.

Finally, we observe that the five females of Pattern 1 (B4a,b, B6a, B14, B25) and the female of Pattern 2 (B13/1) show a decrease in high trophic level foods from tooth to bone whereas the seventh (B18a, Pattern 6) shows no dietary shift (Fig 7). Six of the eight males (B1a,c, B5c, B21/22b, B4c, B13a) present a similar trend to females, with a decreasing trophic level of food over the life course and two of them (B4c, B13a) also show a marine dietary shift. The last two (B8 and B10a) display a higher consumption of ¹⁵N enriched food over the life course (positive Δ¹⁵N_bone–teeth values > +0.4‰). Their condition illustrates a shift occurred during the earlier periods of growth (sampling on M1, M2) while the others males and females represent a later period of growth (sampling on M3).

Chronological and geographical comparisons.

Comparison of bone collagen isotopic ratios of individuals from Sigatoka Burial Ground 1 and other individuals from Southwest Pacific highlights several trends that may be a function of chronology and geography (Table 3, Fig 8, S3 Table). Sigatoka individuals δ¹³C median value is similar to that of the more ancient individuals considered in this study (with respective mean ± 1σ −15.7 ± 1.2 ‰, n–G1 = 55 and −15.8 ± 0.9 ‰, n–G2 = 40) but is significantly higher than that of the more recent individuals (+2.3 ‰ to G3, p = 0.001) (Fig 8A). In contrast, Sigatoka individuals δ¹⁵N median value is significantly lower than that of the more ancient individuals studied (−2.9 ‰ to G1 and −1.0 ‰ to G2, respectively 12.1 ± 1.0 ‰, n–G1 = 55 and 10.3 ± 1.0 ‰, n–G2 = 40, with for both groups p < 0.001) and is similar to that of the more recent individuals considered in this study.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Comparison between Sigatoka Burial Ground 1 individuals δ¹³C and δ¹⁵N values.

https://doi.org/10.1371/journal.pone.0300749.g008

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Descriptive statistics of isotopic data (bone collagen sample) of Southwest Pacific series dated between 3,000 and 1,400 BP.

(Isotopic data were selected in published data considering preservation criteria, age above 5 years, see Table 1 for references and see S3 Table for individual data).

https://doi.org/10.1371/journal.pone.0300749.t003

Comparisons between archipelagos show that (i) Sigatoka individuals δ¹³C and δ¹⁵N median values are significantly higher than those of the most recent individuals of Vanuatu included in the study (G3–Vanuatu n = 15: +2.2 ‰ in ¹³C, −17.8 ±1.8 ‰, p < 0.001 and +0.2 ‰ in ¹⁵N, 9.0 ± 0.7‰, p = 0.036), (ii) only the Sigatoka individuals δ¹⁵N median value is significantly lower than that of the earlier Tongan individuals studied (−1.5 ‰, 10.7 ± 0.7 ‰, p < 0.001, n = 17) and (iii) there is no differential isotopic pattern between the Sigatoka and the early Fijian individuals included in the study (G2–Fiji) (Table 3, Fig 8B).

Comparisons between Fijian sites are shown for series of more than 5 individuals. Sigatoka median values for both elements are similar to those at Olo (2,800–2,500 years ago, n = 13). Sigatoka individuals δ¹⁵N median value is significantly lower than that of the more ancient studied individuals from Teouma (−2.9 ‰, 12.1 ± 1.0 ‰, p < 0.001, n = 55). Individuals from Sigatoka Burial Ground 1 have a significantly higher δ¹³C median value than the more recent individuals of both Teouma 7C and Uripiv sites (respectively +3.8 ‰ −19.5 ± 0.3‰, p = 0.003, n = 6 and +1.4 ‰; −17.1 ± 0.5 ‰, p = 0.003, n = 8) (Table 3, Fig 8B).

Chronology

The individuals from Sigatoka Burial Ground 1 constitutes a skeletal series unique in the region for the first millennium CE, with the exception of the two isolated burials from Naselala in North Fiji [98) and Pea in Tonga [85,112] and an unpublished series of graves uncovered also in a dune system in the Bourail region in New Caledonia [113–115]. Our ¹⁴C dates indicate that the interments took mainly place in the first half of the first millennium CE, with an age range of 1,888 to 1,272 cal BP. Such an age range places 70% of the burials before the Fijian Plainware period (1,433–1,298 cal BP) as defined by Burley and Edinborough [73] (Fig 4). Burials and Fijian Plainware village occupations do not appear to be fully contemporaneous, in contrast with what was previously recognized [31,32]. This temporal difference may account for the inconsistency between the limited abundance of remains of terrestrial fauna and coastal remains recorded for the village locus and the isotopic reconstruction of human diet produced by Phaff and collaborators [31] rather than any other explanation. The range of variation in mortuary practices observed at Burial Ground 1 is probably not related to the transition from Fijian Plainware to Navatu, regardless of the date of that transition (1,800 BP or 1,300 BP) [67] (Fig 4). Our dates indicate that the funerary events were not concentrated in a short period, but rather occurred over a longer period of about 600 years, albeit with no discernible spatial pattern, with older graves scattered throughout Burial Ground 1. It can be noted, however, that the older individuals (B21/22b, B10a, Fig 3) are either a secondary deposit associated with burial B21/22a and a burial (B10a) located at the highest point of the site, that is perhaps foundational of Burial Ground 1 [57] (Fig 2). With 34% of the funerary events involving contemporary deaths of multiple individuals distributed over a long period of time combined with a cultural selection of the interred individuals, the examined skeletal series reflect a discontinuous representation of the population who once inhabited the Sigatoka Sand Dune area.

Isotopic changes during the lifetime

Higher δ¹⁵N values in teeth (n = 19) than in bone (n = 25) suggest either a diet enriched in ¹⁵N during dental tissue formation or a depletion of ¹⁵N in the adult diet (Figs 5 and 6). Given the isotopic variability of food resources in the Southwest Pacific and the trophic enrichment between food and consumer tissues, the observed dietary change over the life course could be related to a differential quantitative contribution of both terrestrial and marine protein and both high and low trophic level resources. However, considering the potential for body protein recycling in the context of physiological stress, such as starvation [53,54], this increase in dental tissue may also indicate a period of stress during childhood.

Dietary changes over the life span have been discussed in terms of physiological stress for other Southwest Pacific individuals, such as in Southeast Solomon, at the Namu site (c. 440–150 cal BP) [45,46], in American Samoa (1,200–100 cal BP) [47] and in Fiji, at the Bourewa site (740–150 cal BP) [48]. It is interesting to note that regardless the broad geographical and chronological contexts covered by these archaeological sites, teeth appear to be regularly enriched in ¹⁵N compared to bone. Such a broad distribution and inter–site effect is also observed at the intra–site level at Burial Ground 1. Isotopic values were obtained from individuals whose dates of death are widely distributed over time, spanning some 600 years of cemetery use. It is thus difficult to conceive that such a repetitive pattern is consistent with a single physiological process associated with a "starvation effect" for all individuals, a similar process that would have occurred over such a long period of time and with sufficient impact to be recorded in the growing tissues of children and adolescents. The hypothesis of a physiological stress remains complicated to consider as a robust explanation for the systematic elevated δ¹⁵N values in dental tissue compared to bone tissue, because bone tissues are more sensitive to environmental changes than dental tissues [116], though physiological stress, pathology and metabolic syndrome could also have affected the body nitrogen balance of certain individuals [117]. This would require future investigation using modern tooth samples of individuals whose dietary and health information can be monitored. Moreover, as the age of death of individuals, before or after 40 years, is not associated with higher teeth δ¹⁵N values compared to bones, other criteria must be considered to explain the isotopic trend observed over the life course.

A dietary change in protein trophic level over the life course is apparent in 14 of the 15 Sigatoka Burial Ground 1 individuals represented by paired bone–tooth data. Only one individual (B18a, Pattern 6) would have maintained a similar diet throughout life (Fig 7C). The dental isotopic signal documents the period between 15–20 years for 12 of the individuals (M3, n = 12) while the dental i signal of the three others refers to the periods between 6.5–9.5 years (M1, B10a) and 10.5 to 13.5 years (M2, B8a, B21/22b), respectively. These results suggest that the dietary change generally occurred in a short time, between the end of adolescence and adulthood. It could finally support another interpretation of the intra–individual isotopic changes for the Sigatoka Burial Ground 1 individuals. Firstly, these individuals display a pattern suggesting good sanitary condition [71]. Pietrusewsky and collaborators [81] do not report any pathological lesions that could be related to metabolic disorder and our mortality profile, in agreement with the palaeopathological evidence, show no evidence of famine. Second, the observation that the nitrogen isotopic ratios are systematically higher in M1 and M2 dentine than in adult bone (whatever the archaeological, social or health criteria) does not support a physiological stress induced by a poor health status during dental tissue formation (Table 2, Fig 6). The pattern is more consistent with the hypothesis of dietary changes than with health disruption and physiological stress, a hypothesis proposed by other authors for other Southwest Pacific series [45,46,48]. Interestingly, B10a individual, an old male who has suffered from extreme dental antemortem loss and moderate dental wear during his adulthood, displays no evidence of dietary stress that could have occurred during infancy. He is the only individual to present an isotopic pattern consistent with a dietary shift towards more marine fish ingestion at the end of his life. Perhaps explaining their originality, their grave, very close to the grave of two other males at the highest point and amongst the earliest of the site (Fig 2), is considered to be the foundation of Burial Ground 1 [57].

Our results do not show clear connection between isotopic changes and the location of the graves in the cemetery, nor with the burial treatment. The exception is grave B4, which contains three contemporaneous bodies buried together under the same cairn (Fig 3). B4a and B4b, two females, experienced a similar dietary change during their life, with a decrease in trophic level associated with an increase of marine food (Pattern 1). The third individual, a male (B4c, Pattern 3), is characterised by a shift towards a more terrestrial plant protein intake (Fig 7). Their association at death may reflect a similar life history as suggested by Best [57], with the two females having similar dietary behaviours at different stages of their life. Interestingly, these two females share the most common dietary pattern (Pattern 1), which shows a change towards more shellfish and algae intake over the life course. Our results clearly emphasise sexual variation, which was expected on the basis of ethnographic information [118–120], but not discernible in the analysis of the adult bone alone [31]. They are therefore consistent with the hypothesis of a specific use of coastal and reef resources by female adolescents and women, as already put forward by isotopic data measured on other Fijian series [16,121]. For now, if a higher marine resource intake has existed at Sigatoka, it would have distinguished mainly the female diet over life course. Could this be an illustration of a social difference? If so, contrary to what was reported previously [31,32], this would not be reflected in the funerary practices, but rather in the daily life activities. Dental markers analysed for individuals from Burial Ground 1 show sex–specific trends in wear, antemortem tooth loss and temporo–mandibular arthritis [80,81]. Our isotopic results suggest that this male/female social distinction existed at Sigatoka since the childhood, whereas Fijian ethnohistoric sources mainly inform on lifeways of male of high rank [16,119,120].

Temporal and geographical variations across Southwest Pacific

Focusing on the bone isotope data of individuals from Sigatoka Burial Ground 1, one individual (B20) has the lowest human δ¹⁵N value of all the Southwest Pacific data used in this study (5.4 ‰). Their health condition and layout of their grave, which is similar to that of other individuals found at the site, do not allow us to propose any hypothesis other than the influence of a particular diet to explain this super–low value. The elemental carbon and nitrogen composition of this bone sample confirms a well–preserved collagen, and to our knowledge, no physiological stress that induces a negative ¹⁵N balance has been described in the literature. The only alternative to identify which ¹⁵N–depleted foods could have been consumed by this individual would be to reinforce the local isotopic baseline. Other individuals (B13/1, B13a, B23a, B25), with a North Melanesian morphology and buried in the north part of Burial Ground 1, also present low δ¹⁵N values (5.4 to 9.0 ‰) (Fig 5). Their δ¹³C and δ¹⁵N values are significantly lower than those of individuals with a South Melanesian morphology dispersed in other parts of the cemetery (8.9 to 10.5 ‰) (p = 0.0031, p = 0.016 respectively) (Fig 2), a particularity also noted for their mortuary and health characteristics [28,57,59]. Interestingly, these individuals (B13/1, B13a, B23a, B25) present similar δ¹⁵N values as post-Lapita individuals from Teouma 7C and Uripiv (8.1 to 9.6 ‰) (Figs 5 and 8, Table 3).

Based on the Southwest Pacific isotopic baseline and the diet–consumer expected trophic level enrichment (see 3.3 section), the differences in δ¹³C and δ¹⁵N values between individuals from Burial Ground 1 and individuals dated to 3,000 to 2,000 years ago suggest (i) a decrease in the consumption of foods from trophic levels (lower δ¹⁵N values) over time (Fig 8A, Table 3) and (ii) the persistence of a marine protein intake similar to that observed for individuals dated to 3,000 to 2,500 years ago. The dietary pattern described would be consistent with a mixed marine and terrestrial diet as defined for Lapita and Late Lapita individuals [17,24,107] but (i) with consumption of very low trophic level foods (shellfish, seaweed, algae), and (ii) without the contribution of terrestrial C3 plants observed in the individuals dated to 2,500 to 2,000 years ago [18]. In other words, our data suggest that individuals from Sigatoka Burial Ground 1, whose radiocarbon age range from 1888 to 1272 cal BP, have not yet undergone a dietary transition towards the consumption of horticultural products, except for a few individuals (B13/1, B13a, B23a, B20, B25) characterized by more vegetarian diets. Indeed, diachronic dietary changes appear to be uneven across the Southwest Pacific region. A different timing of acquisition of the horticultural component in diet has already been suggested for the case of Tongatapu island, Kingdom of Tonga, with a date at the beginning of the second millennium CE [25].

The site–to–site comparisons indicate that individuals from Sigatoka Burial Ground 1 have higher nitrogen isotopic ratios than individuals dating to 2,500–2,000 years ago in Vanuatu (Uripiv and Teouma 7C) and are similar to the early Fijian individuals from Waya (Olo) island. Inter–site comparisons also show an overlap in isotopic values between the Sigatoka individuals and (i) the 2,800–2,000 years ago individuals from Moturiki (Naitabale), Lau (Nayau), and Lakeba (Qaranipuqa), all from Fiji, and (ii) the two contemporaneous individuals from Naselala in Northern Fiji and Pea in Southern Tonga. By showing that the Sigatoka population is closer to other Fijian populations than to populations from other archipelagos, these results suggest a pattern of dietary consumption of marine resources linked to geography, which in turn may be linked to environmental and/or social factors. It has already been observed that island characteristics, including size, climate, and soil, are important factors that can shape subsistence strategies at a local scale [19]. The example of individuals from Bourewa (south coast of Viti Levu), dated to around 740–150 cal BP, shows a strong dependence on marine resources [48], though the adoption of horticultural practices is widely attested at that time in other Fijian island and locations. The influence of a social and cultural factor remains a plausible explanation for the diversity of dietary practices observed in the Pacific over time.

On the other hand, based on terrestrial sedimentary and soils records of biomass burning, Roos and collaborators [122] suggest an anthropogenic transformation of the landscape around Sigatoka and the use of swidden practices around 1,600–1,800 cal BP [122], a hypothesis that apparently contradicts with our interpretation of isotopic data from Sigatoka Burial Ground 1. A horticultural/agricultural production, perhaps not yet efficient enough to provide a sufficient amount of food to sustain the inhabitants of the region could explain why marine and coastal resources remained largely consumed by the majority of the individuals until the middle of the first millennium CE. This situation contrasts with that at the Teouma7C and Uripiv sites, where the consumption of horticultural products was adopted from about 2,500 years ago [18,21]. The changes in dietary behaviour observed at Teouma between 3,000 and 2,500 years ago could be related, at least in part, to the arrival of new comers whose presence in the archaeological record has been shown through morphometric and genetic analyses [123–126]. Conversely, the trend observed at Sigatoka, which clearly suggests more stable dietary habits over a longer period of time, could be associated with a close intergenerational transmission of cultural practices. Cultural evidence tends to confirm continuity, with no change in ceramic data and in Sigatoka Level 2 paleosol before 1,350 years ago [71].

Whether the lack of diachronic dietary change demonstrated by individuals from Sigatoka Burial Ground 1 is actually due to sociocultural or environmental factors, or a combination of both, remains an open question. Sociocultural factors may have been played a key role as our analysis of the composition per age of the series suggests a sociocultural selection of the individuals buried at Burial Ground 1 from a larger group. A lack of pressure on marine coastal is another possibility, as suggested for the Talasiu individuals (Tongatapu, Tonga) [24]. The continued consumption of marine resources implies either that the size of the living group was in line with the resources available or that the group may have been able to effectively manage their marine resources in a sustainable manner. Settlement contraction or demographic decline during the first millennium CE has been suggested for several small islands of Western and Eastern Fiji [68,127–129]. Such a reduction of population number may have created a positive retroactive balance between population need and available food resources. This is supported by the good health status and the lack of physiological stress in individuals from Burial Ground 1 [81] and our isotopic results.