Horticultural intensification and plant-based diets of 18th century CE Waikato Māori in Aotearoa New Zealand

Archaeologically, the earliest evidence for the first settlement of Aotearoa New Zealand (henceforth Aotearoa) places people arriving from east Polynesia in the mid-13^th century CE^1,2. Some Māori origin stories tell of an earlier date of arrival, but regardless of the exact timing, Polynesians voyagers arriving on the shores of Aotearoa would have stepped into a starkly different environment than their tropical Polynesian homeland, traditionally known as Hawaiki. Evidence from Early Period sites (ca. 1250-1450 CE) in Aotearoa shows a broad-spectrum reliance on marine and terrestrial animals^3,4,5, highly mobile populations that circumnavigated both main islands in less than 100 years^4,6, the establishment of exotic horticultural crops brought from east Polynesia^7,8, large-scale landscape modification^9,10, and hunting specializations that focused on large marine mammals and terrestrial birds, the most well-known being the now extinct moa (Aves: Dinornithiformes)^11,12. During the Transitional/Middle Period (ca. 1450–1650 CE) and subsequent Traditional Period (ca. 1650–1769 CE), there was a move toward more regionalized settlement patterns^5,13. In Te Ika-a-Maui (the North Island) and, to a lesser extent, Te Waipounamu (the South Island), this was accompanied by the intensification of horticulture^14,15,16, although practices consistent with what is now termed mahinga kai (working in concert with the natural environment for sustainable food resources) was likely important in shaping food production and evironmental use as it is to Māori today⁷ (Glossary of Māori and Moriori terms presented in Supplementary Note 1).

The Waikato Basin near the city of Kirikiriroa (Hamilton) (Fig. 1) is home to the Waikato Horticultural Complex, an area recognized in oral histories, ethno-historical accounts and the archaeological record as a region of intensive horticulture from the 16th century CE ^{reviewed in15}. The tephra-derived allophanic soils found in the middle Waikato Basin are extremely favourable for gardening because they have a high organic carbon content, are free draining, and have good aeration and nutrient exchange attributes¹³. The major root crops cultivated by Māori were kūmara (Ipomoea batatas) and, to a lesser degree, taro (Colocasia esculenta) and uwhikāho (yam; Dioscorea alata), all of which were brought to Aotearoa from east Polynesia, possibly at different times^8,15,17,18. Forest clearance by fire transformed the landscape across Aotearoa from ca. 1250 CE, although in different ways because of regional, environmental, and climactic variations^9,19,20. In the middle Waikato and elsewhere, Māori established gardens by first felling trees and burning stumps and slash to clear the land. Garden soils conducive to growing kūmara in a temperate environment were created by modifying the natural parent soils with ‘lithic material’ such as sand and gravel, sometimes in specialized garden mounds, known as puke, to improve drainage and soil temperature. Lithic material was quarried from ‘borrow pits’, deep shafts dug into the underlying subsoil^13,15,21,22.

There is unequivocal archaeological evidence for widespread and intensive kūmara horticulture in the middle Waikato Basin in the form of modified garden soils (e.g., the Tamahere soil series), borrow pits, puke and pits for the storage of cultivated tubers^15,22. However, there is a dearth of marine, freshwater and terrestrial faunal remains found in local archaeological sites, which raises questions about the overall diet and lifestyle of Māori communities in the region. Here, we investigate the lives of seven tūpuna (ancestors) dating to ca. 1700-1780 CE who were interred in a borrow pit in the middle Waikato Basin (Table 1, Supplementary Discussion, Supplementary Table 1 and Supplementary Fig. 1). We report on the burial ritual and osteological assessment of kōiwi tangata (human remains) alongside isotope and enamel peptide analyses used to determine chromosomal sex and investigate diet, weaning practices and childhood place of residence. This study is an investigation of the daily lives of Māori ancestors during the Traditional Period, immediately prior to colonization, a period of significant cultural development in art, architecture, cosmology, and expression of whakapapa (genealogy/ancestral connections)^7,13. Our results provide direct isotopic evidence that the diets of some Waikato Māori were almost entirely plant-based, underscoring not only the centrality of horticulture in Māori society at this time, but also contributing to a broader understanding of the importance of plant foods to traditional diets globally. The significant dietary changes that occurred from the time of initial East Polynesian settlement of Aotearoa to the Traditional Period reflect cultural innovation alongside environmental adaptation over a period of ~500 years. Our findings advance understanding of Māori lives in the past, including cultural dietary practices, and are of ongoing relevance to descendant communities seeking to record, revitalize, and restore traditional foodways and cultural practices.

Bone collagen stable isotope results are presented in Table 1 for the seven kōiwi. Only two individuals (Kōiwi 2 and Kōiwi 4) had permanent first molars and enamel peptide, dentine micropunch (δ¹³C and δ¹⁵N) and enamel isotope results (δ¹³C, ⁸⁷Sr/⁸⁶Sr) are presented in Supplementary Tables 2 and 3 for these tūpuna.

Enamel peptide analysis

Amelogenin peptides were recovered from both individuals, Kōiwi 2 and Kōiwi 4, who had teeth sampled for analysis. Kōiwi 4 displayed both AMEL-X and AMEL-Y peptides (i.e., chromosomally male) whereas Kōiwi 2 displayed only AMEL-X peptides (i.e., chromosomally female). Both the raw intensities and intensity ratios of the quantified peptides are presented in the Supplementary Fig. 2. All dimorphic peptides were monitored, but only SM_(OX)IRPPY from AMEL-Y and SIRPPYPSY from AMEL-X were targeted and reported. These peptides displayed the strongest intensities and showed and AMEL-Y/AMEL-X intensity ratio of >1 for chromosomally male and <0.01 for chromosomally female individuals, indicating a clear link between the observed ratios and chromosomal sex.

Bone collagen δ
¹³C and δ
¹⁵N analyses

All the bone collagen samples (n = 7) met the standard collagen quality indicators (CQI) of a %N between 11 and 16 wt%, a %C > 30 wt% and a C:N ratio between 2.9 and 3.6^23,24.

The average ( ± 1 SD) δ¹³C_collagen and δ¹⁵N_collagen values are presented in Table 2. Although the overall sample size was too small (n = 3 adults and n = 4 non-adults) for inferential statistics to be applied, there was a small difference in the average δ¹⁵N_collagen (adults 0.7‰ higher than non-adults) but there was only a 0.1‰ difference between the average δ¹³C_collagen value of the two age groups. The non-adults were aged between 5 and 15 years old (Table 1). There was a statistically significant, very strong positive correlation between the δ¹³C_collagen and δ¹⁵N_collagen values (Spearman’s rs(7) = 0.94, p < 0.01). After the correction for diet-tissue spacing was applied to the δ¹³C_collagen and δ¹⁵N_collagen values (5‰ δ¹³C for and 3‰ for δ¹⁵N values)^25,26 the results were compared to a dietary baseline for Māori Era Aotearoa (Fig. 2).

Tooth dentine micropunch δ
¹³C and δ
¹⁵N analyses

The dentine micropunch samples also met the CQI for this sample type, which is identical to the bone collagen CQI²⁴ except that the C:N ratio of bulk dentine should be between 3.1 and 3.6²⁷ (Supplementary Table 2). Tooth dentine micropunch δ¹³C values ranged from −21.1 to −20.0 for Kōiwi 2 and −20.8. to −20.3 for Kōiwi 4. Tooth dentine micropunch δ¹⁵N values ranged from 6.0 to 11.5 for Kōiwi 2 and 6.8 to 10.5 for Kōiwi 4 (Supplementary Table 2). Age alignment (assigning age mid-points) followed the method outlined in Czermak et al.²⁷ and are presented in Fig. 3.

Tooth carbonate δ
¹³C values and whole vs protein diet from tooth dentine and tooth carbonate

The tooth δ¹³C_carbonate value for Kōiwi 2 was −13.5‰ and for Kōiwi 4 was −13.9‰.

For each individual, the average δ¹³C_dentine value of the two crown dentine micropunch samples located nearest to the cemento-enamel junction (CEJ) was compared to the δ¹³C_carbonate value of tooth enamel chip sampled from the corresponding area of the crown (Supplementary Tables 2 and 3). This was to ensure the protein and ‘whole’ diet analysed from the δ¹³C_dentine and δ¹³C_carbonate represented a similar age (ca. ~ 1.9–3.5 years old)²⁸.

The δ¹³C_dentine and δ¹³C_carbonate values were correlated and plotted against regression equations developed to determine the contribution of C3 and marine/C4 foods in the diet from bone collagen and apatite²⁶ (Fig. 4). The results from the two individuals (Kōiwi 2 and Kōiwi 4) show that their childhood diet around the age of ~1.9 to 3.5 years old was a C3-based. The ‘whole’ diet (carbohydrates, lipids and protein) can be determined from the δ¹³C_carbonate of the tooth enamel of Kōiwi 2 and Kōiwi 4 by applying a correction for diet-tissue spacing ( ~ 12‰)^29,30. The corrected values (−25.5‰ and −25.9‰) are representative of foods sourced from a C3-terrestrial based ecosystem.

Tooth enamel ⁸⁷Sr/⁸⁶Sr

Both Kōiwi 2 and Kōiwi 4 displayed similar tooth enamel ⁸⁷Sr/⁸⁶Sr ratios of 0.706533 ( ± 0.000030) and 0.706611 ( ± 0.000035), respectively. An updated Aotearoa-specific strontium isoscape based on Kramer et al.³¹ was used to create probability maps of the likely regions where the individuals resided in early childhood, around the age of 1.9 to 3.5 years old (Supplementary Fig. 3). Tamahere and nearby surrounds are highlighted as likely areas where the two individuals spent their early childhood. The other potential area of childhood residency was the Rotorua Lakes District to the east of the Waikato. The young age of death of both Kōiwi 2 (7–9 years old) and Kōiwi 4 (9–15 years old) means that the most parsimonious explanation for their specific ⁸⁷Sr/⁸⁶Sr ratios is that they were likely local to the area around the Tamahere site.

Plant-based diets and regional identities

The bone and tooth isotopic results show that the Tamahere tūpuna ate primarily plant-based diets. Although a wide variety of native C3 plants were eaten during the Māori Era^32,33, kūmara, and to a lesser extent, taro and uwhikāho cultivars, would have constituted the majority of the plant foods in the diet at Tamahere. The reasons for this interpretation are twofold: (1) the archaeological evidence of the surrounding site and region shows intensive horticulture was practiced around the same time (ca. 18^th century CE) and (2) many wild plant foods were not known to have been eaten in bulk as staple foods, with notable exceptions including aruhe (bracken fern rhizome – Pteridium aquilinum), the inner leaves, aerial stems, seeds, pith and roots of introduced tī pore (Cordyline terminalis) and native tī (Cordyline spp. C. australis, C. indivisa, and C. pumilo)^34,35 and the drupe of karaka (Corynocarpus laevigatus). Tī Kouka and tī toi were an important dietary staple, especially in Te Waipounamu in regions where kūmara could not be grown (i.e., south of the Banks Peninsula)³⁵. Stands of karaka were established by early Māori around Aotearoa and Moriori on Rēkohu (Chatham Islands), where it is known as kopi. The fleshy outer fruit of the drupe is edible but the kernel is toxic and must be processed (usually by boiling/steaming, leaching, then drying and roasting) to make it safe to eat^36,37. Aruhe was eaten across Aotearoa, and was typically consumed by pulling the cooked rhizome between the teeth to extract starch, although it could be made into cakes or patties by more involved preparation techniques^34,38. Orally processing aruhe commonly resulted in an extreme and distinctive angled tooth wear pattern known as the fern-root plane, which was sometimes accompanied by tooth root displacement^34,39. Only a limited number of teeth were recovered from the Tamahere kōiwi because many had been lost after death, but the tooth wear observed was not severe and three individuals had tooth caries, supporting that their diet may have included soft, starchy and sticky foods (i.e., kūmara and taro)⁴⁰.

The stable isotope values of bone collagen and tooth dentine are primarily representative of dietary protein because amino acids from food are preferentially routed for tissue synthesis⁴¹. This is important because dietary isotope analyses from proteinaceous tissues like collagen and dentine over-represent the protein-rich foods in the diet (i.e., meat and fish) and under-represent low-protein foods in the diet (i.e., fruit and vegetables, tubers and grains)²³. At Tamahere, there is very little dietary evidence for any protein-rich foods such as meat and fish, which means most of the diet consisted of low-protein plant foods, the bulk of which were likely kūmara and, to a lesser extent, taro and uwhikāho, supplemented by wild plant foods.

The low δ¹³C_collagen and δ¹⁵N_collagen values indicate that the majority of dietary protein was from a low trophic level, terrestrial ecosystem, notably C3 plants. However, the statistically significant, strong positive association between the δ¹³C collagen and δ¹⁵N collagen values (Spearman’s rs (7) = 0.94, p < 0.01) hints that a very small amount of animal protein may have been consumed. This protein intake was likely sporadic rather than a consistent dietary component, most likely from terrestrial animals or freshwater resources. A weaker positive correlation would be expected if marine fish, shellfish, or mammals formed a greater component of the diet, as observed for the Wairau Bar kōiwi (Spearman’s rs (27) = 0.70, p < 0.01), consistent with higher marine protein consumption at that site^3,42.

The very small amount of terrestrial protein at Tamahere could have included wild birds, kurī (dog: Canis familiaris) and kīore (rat: Rattus exulans), if the latter two species only ate terrestrial foods. Dogs and rats that ate cultivated tubers could therefore have been eaten sparingly, although faunal remains are lacking to support that these animals were on the menu. The pā and gardens at Tamahere are less than 2 km from the Waikato River and close to other tributaries where riverine fish, tuna (freshwater eel: Anguilla dieffenbachii, Anguilla australis, and Anguilla reinhardtii), kōura (freshwater crayfish: Paranephrops planifrons) or waterfowl would likely have been available year round. Although the positive correlation between the δ¹³C_collagen and δ¹⁵N_collagen values suggest small amounts of terrestrial or freshwater organisms may have been eaten by the Tamahere tūpuna, these resources comprised a very minor contribution to the diet and the lack of faunal remains found at the Tamahere site complex supports the dietary interpretations of a primarily plant-based diet. The reasons for this lack of animal protein in the diet could include food preference, availability of resources, status or cultural food restrictions. Interestingly, tuna (eels) are considered an important traditional food source for Māori. However, it has recently been proposed that the mass harvesting of eels observed during the Historic period was a post-colonization development and the consumption of tuna (eels) was not widespread during the earlier Māori period for cultural reasons⁴³.

The plant-based diet of the Tamahere tūpuna is completely different to the dietary evidence from the Wairau Bar tūpuna who lived during the early settlement period of Aotearoa (ca. late 13^th century CE) and were buried in northeast Te Waipounamu near modern-day Te Waiharakeke (Blenheim) (Figs. 1 and 5, Table 2, Supplementary Data 1). Isotope analyses of the kōiwi from Wairau Bar showed that the population was highly mobile and reliant on a wide variety of marine, terrestrial and, likely, freshwater resources, including fish, marine mammals, and birds, including moa³ (Fig. 2). Broad-spectrum foraging for food across the wider landscape is emblematic of the initial settlement period in Aotearoa when small populations were exploring the virgin island environment and had not established substantial gardens for long-term food security^5,13.

The diet of the Tamahere tūpuna is also distinctly different when compared to five individuals from the Waihora site (Rēkohu/Chatham Islands) who displayed very high δ¹³C_collagen and δ¹⁵N_collagen values (Table 2, Fig. 5, Supplementary Data 1). Leach et al.⁴⁴ noted that the large proportion of fish, sea mammal meat and fat in diet of the Waihora individuals may be similar to Inuit diet⁴⁴ (Fig. 2). On Rēkohu, kopi/karaka (Corynocarpus laevigatus) would have been the most important plant food, although the proportional contribution of plant resources is difficult to determine from bone collagen when protein-rich foods comprise a large part of the diet²³. A similar dietary pattern with high proportion of fish and marine mammals was observed for a likely Early Period individual (Burial 4) from the Lagoon Flat site on the Kaikoura coast (Table 2, Figs. 2 and 5, Supplementary Data 1)⁴⁴. Four individuals from an inland Bay of Plenty site near Lake Rotoiti, thought to be from the Traditional Period, displayed δ¹³C_collagen and δ¹⁵N_collagen values representing a diet most similar to the Tamahere individuals (Table 2, Figs. 2 and 5, Supplementary Data 1), but the Rotoiti tūpuna were also likely eating freshwater resources (including waterfowl) and potentially marine foods (fish and shellfish) brought inland or from visits to coast⁴⁴.

For comparative purposes only, the δ¹³C_collagen and δ¹⁵N_collagen values of the Māori Era kōiwi (average ± SD) are visualized alongside the bone collagen and hair stable isotope values of Colonial Period pāhekā (European) and Chinese individuals who were interred between 1860 and the 1890s CE (Table 2, Fig. 5, Supplementary Data 1 and 2). Colonial immigrants to Aotearoa brought their own cultivars and domestic animals and usually had different resources and diets compared to Māori Era populations^45,46,47,48. Although only representing a short period of time before death, the hair δ¹³C and δ¹⁵N values of the Pākehā and Chinese adults are likely more representative of the local diet because the bone collagen δ¹³C and δ¹⁵N values may be averaging dietary inputs from before the time these people emigrated to Aotearoa. Both Pākehā and Chinese average δ¹³C_hair and δ¹⁵N_hair values were lower than their average δ¹³C_collagen and δ¹⁵N_collagen values, respectively. Pākehā children (all post-weaning and likely local) also displayed low δ¹³C_hair and δ¹⁵N_hair values that were similar to the δ¹³C_collagen and δ¹⁵N_collagen of the Tamahere kōiwi, indicating both these groups were eating a primarily plant-based diet from C3 terrestrial ecosystems. Near the time of death, the δ¹³C_hair and δ¹⁵N_hair values of the Pākehā and Chinese adults indicate a heavy reliance on C₃ terrestrial ecosystems, while also reflecting the consumption of animal foods from higher trophic levels, including domestic or wild animals and potentially freshwater resources such as waterfowl^45,46,47,48.

The Tamahere dietary isotope results presented here show that, around 500 years after the initial settlement of Aotearoa, Waikato Māori had developed plantation-style gardens so productive that people were harvesting and storing enough tubers to eat year-round as their main source of food. These results are in line with the archaeological evidence that identifies the middle Waikato Basin as “the most extensive area of pre-European gardening recorded in New Zealand”, with some 500 gardens, 200 pā sites (fortified settlements) and thousands of borrow pits located in the area near the Waikato River and its major tributary, the Waipā River^{13: 9}. A relationship between fortified settlements, coveted gardening areas and conflict centred on food security has been found in the Waikato and elsewhere in Aotearoa, dating from ca. 1650 CE⁴⁹. The seasonality of kūmara and the necessity to store and protect the yearly harvest from raids has been identified as an important function of pā that was intertwined with the fortifications representing status, tribal identity and the physical manifestation of military strength¹³.

Evidence for the weaning age of the tūpuna

Duration of breastfeeding, the age of introduction of supplementary foods and the nutritional quality of supplemental foods are extremely important for maternal and infant health, especially in the past^50,51. Breastmilk not only provides essential nutrients and calories, it helps with immune protection during infancy by passing antibodies directly from mother to baby⁵². Although only two individuals, both children, had first molars available for analysis, the results are the only time that patterns of breastfeeding and weaning have been investigated for Māori tūpuna. The isotopic results clearly show that for Kōiwi 2 (a female), there was a ~ 4 ‰ drop in δ¹⁵N_dentine values between birth and 2.5 years of age (Fig. 3). This pattern indicates that breastmilk gradually decreased as supplemental foods (from a lower trophic level than breastmilk) increased over this period. However, there was a further, gradual, ~2‰ decline in δ¹⁵N_dentine values between 2.5 and 5 years of age, which may suggest breastfeeding continued over this period. Alternatively, this could be related to another higher trophic level food being eaten in gradually reduced amounts at this time and/or changing nitrogen balance during periods of growth⁵³. For Kōiwi 4, there was a ~ 3.5 ‰ drop in δ¹⁵N_dentine values between birth and 3.2 years of age and no further gradual decline in δ¹⁵N_dentine values (Fig. 3), indicating complete weaning likely occurred at this age. The supplementary foods and childhood diet of both individuals were likely C3 plant foods, similar to the diet of the Tamahere adults. The breastfeeding period of the Tamahere children is in stark contrast with the observed breastfeeding and weaning trends of the European colonists who arrived in Aotearoa in the 19^th century. Incremental isotope analysis of teeth and hair from Colonial Period pākehā individuals from both rural and urban areas showed many individuals experienced very short periods or no evidence for breastfeeding, which likely had deleterious effects on both childhood and later adult health^47,48,54.

Mortuary ritual and identity

A major question is whether the people interred in the borrow pit are local to the site and representative of the wider population that lived in the Tamahere area during the 18^th century CE. The strontium isotope results for Kōiwi 2 and Kōiwi 4, both children, suggest that they were local to the area (Supplementary Fig. 3). We will likely never know for certain why the kōiwi were placed in the borrow pit, but the burial context can provide further clues about their links to the community. Although Māori-era burial traditions are extremely varied through time and across regions in Aotearoa, the burial context of the (at least) seven kōiwi interred as commingled secondary burial in an empty borrow pit is rare, but not unique. The practice of secondary burial is a well-documented Māori funerary tradition, recorded in oral histories, ethnohistorical accounts, and archaeological evidence⁵⁵. Secondary burial is part of a mortuary rite referred to as the hahunga⁵⁶ and these types of concealed burials were common during the Traditional Period (1650– ~1800 CE)¹³. Secondary burial was commonly undertaken to hide kōiwi in caves, rock shelters and underground so enemies could not steal the mana (lifeforce) of an ancestor¹³ and, as a result, secondary burials are not commonly found in archaeological contexts. Hapū approval to undertake the current research provided a rare opportunity to analyse the lives of tūpuna who were interred in a secondary burial context.

No perimortem trauma was observed on the kōiwi, which suggests these individuals were not victims of ritual killing or inter-tribal warfare. However, cut marks and other postmortem damage was found on the bones, which would have occurred before the final interment in the borrow pit. The location and appearance of this postmortem damage indicated that it occurred when the bone was not fresh, potentially during exhuming, disarticulating and handling the kōiwi as part of the mortuary ritual. Taphonomic analysis showed that some bones had been exposed to the outside environment for a period of time before the final burial at Tamahere and that at least some of the kōiwi were likely initially buried, then exhumed before select bones were placed in the borrow pit. The bones of men, women and, from the peptide analyses, a boy and a girl, were all placed into the secondary burial, indicating care in the selection and reburial of bones from many people of different ages and sexes.

The association between death and burial with a horticulture site at Tamahere may, at face value, seem in opposition to traditional Māori tikanga (customs) that create a separation between sites for burial (tapu- sacred/restricted) and areas for food-related or everyday occupation activities (noa- free from spiritual restrictions). However, we do not know if the area surrounding the borrow pit was actively used for horticultural purposes at the time of burial and AMS dates indicated a slightly later period for burial compared to the surrounding gardens. Also, culture is not static. Although tangihanga (funeral rites) is considered to be one of the most enduring aspects of Māori culture⁵⁶, is has been suggested that the conceptual division between burial and more mundane daily activities did not exist in the same manner during earlier periods when burial would often occur in or near settlements⁵⁷.

Interestingly, a layer of uncooked pipi (Paphies australis) shells, a marine bivalve, was found associated with Kōiwi 1. The shells were possibly placed as ritual items during the burial, maybe from a mortuary practice known as haehae, the laceration of the body with sharp objects (including pipi shells) as part of the mourning rites⁵⁶. The presence of marine shells in the grave is interesting considering that we now know the tūpuna were not regularly eating marine foods. Marine shells were only found in the burial context at Tamahere, and not as food refuse (i.e., midden) anywhere else in the larger site complex. In tandem, the evidence that (1) bones from men, woman and children of both sexes were carefully selected for secondary burial; (2) the mortuary ritual likely involved exhuming, disarticulating and handling the kōiwi before reburial; and (3) uncooked pipi marine shells were likely placed as ritual items in the grave, all point to a careful and thoughtful secondary burial of likely local people, possibly put there to protect the area or render it tapu⁵⁵.

Although we will never be certain of the exact reasons for the secondary burial of these (at least) seven kōiwi, the isotopic results from the Tamahere tūpuna show that this group of men, women and children were eating a primarily plant-based diet of cultivated root crops, likely mostly kūmara. Whether this was for cultural reasons (i.e., status) or a response to limited or lack of access to marine, riverine and terrestrial protein-rich foods (such as shellfish, fish and meat) is impossible to determine from the current evidence, but does raise questions about the apparent lack of inland/coastal trade of food, in addition to the importance of, and access to, riverine resources during this period. This is different from the diet of the tūpuna from the inland site near Lake Rotoiti in the Bay of Plenty who also date to the Traditional Period. Isotopic analysis showed that the Rotoiti individuals were likely eating C3 plants, such as kūmara, but also eating freshwater resources and potentially marine foods traded or brought from the coast⁴⁴.

It is known that the intensification of horticulture in the middle Waikato Basin began in the 16th century CE. Within 150–230 years, tuber crops, mostly kūmara, became the main dietary staple, at least for the people interred together in the borrow pit a Tamahere. Although the sample size was small, our findings provide direct evidence of dietary practices and cultural behaviours which have previously been inferred only indirectly from archaeological contexts (e.g., storage pits, garden soils) and provide direct evidence for patterns of breastfeeding and weaning of Māori tūpuna. How widespread these dietary adaptations were remains to be seen, but the extensive evidence for intensive horticulture in the Waikato and elsewhere across Aotearoa does naturally lead to questions if plant-based diets were common for some Māori Era communities, especially during the Traditional Period. The stark contrasts between the diets of the first East Polynesian settlers at Wairau Bar and Lagoon Flat, the later Moriori at Waihora, the Traditional Period Māori near Lake Rotoiti and the Tamahere tūpuna reveal the extraordinary dietary diversity and cultural changes that took place within 500 years across Aotearoa and Rēkohu. Future research employing compound-specific stable isotope analysis of amino acids could provide deeper insights into these past dietary practices.

Permissions and permits

This research was developed in direct collaboration with, and with support from, mana whenua (customary indigenous landowners) (Supplementary Note 2). The research was undertaken at the request of four iwi (tribes) and hapū (sub-tribes), Ngāti Maahanga, Ngāti Wairere, Ngāti Koroki Kahukura, Ngāti Hauā, who hold customary and territorial rights over the land. The four iwi and hapū had representatives on the Tangata Whenua Working Group (TWWG) who acted as kaitiaki (cultural monitors) for the Waka Kotahi New Zealand Transport Agency-led construction of the Hamilton Section of the Waikato Expressway. Archaeological investigation of the Hamilton Section of the Waikato Expressway was undertaken by S. Keith on behalf of Waka Kotahi New Zealand Transport Agency under archaeological authority 2015/958. No specific permits were needed in Aotearoa for this bioarchaeological research. All approvals, sampling and destructive analyses followed the Heritage New Zealand Pouhere Taonga (HNZPT) Archaeological Guidelines Series Kōiwi Tangata Human Remains and were undertaken in accordance with the Heritage New Zealand Pouhere Taonga Act (2014). HNZPT is the New Zealand government’s statutory authority responsible for administering heritage protection laws and regulating archaeological work in Aotearoa.

Enamel peptide analysis

Enamel peptide analyses are used for chromosomal-linked sex assessment, which is especially helpful to provide sex determinations for non-adult individuals and incomplete skeletal remains⁵⁸. The method is based on identifying isoform-specific peptides of amelogenin, a protein involved in enamel formation that is encoded as different isoforms on the X and Y chromosomes (AMEL-X and AMEL-Y)^59,60.

Enamel peptide preparation followed established methods^59,60 at the Centre for Protein Research, Research, Infrastructure Centre, University of Otago, Ōtepoti (Dunedin). First, the enamel was washed in 3% H₂O₂ for 30 s and quenched in MilliQ H₂O for 30 seconds. The enamel chip was then conditioned in 60 μL of 5% (vol/vol) of a 35% (w/w) HCl solution for 2 minutes. The HCL conditioning solution was removed and discarded. Another 60 μL of the 5% (vol/vol) HCl was added to the enamel chip and incubated for 2 min for peptide extraction. The HCl extraction solution containing the peptides was then recovered and centrifuged for 5 min at 30,000 × g to remove any particulates. The extracted peptides were then purified by solid phase extraction on a ZipTip with 0.6 μL C₁₈ resin (Merk Millipore, Mass. USA) essentially following the manufacturer’s instructions. In brief, after rinsing of the C₁₈ resin in pure acetonitrile (ACN), the ZipTip was equilibrated in 0.1% formic acid (FA) in water before loading the sample of extracted peptides. The bound peptides were then washed on-tip with 0.1% formic acid (FA) in water and eluted in 60% ACN, 0.1% formic acid (FA) in water. The eluate was dried using a centrifugal vacuum concentrator and stored as dried samples at −80 °C.

For peptide analysis by untargeted liquid chromatography-coupled tandem mass spectrometry (LC-MS/MS) samples were reconstituted in 20 μL of 5% ACN, 0.1% FA in water of which 5 μL were loaded into an Ultimate 3000 nano-flow uHPLC system inline coupled to a LTQ-Orbitrap mass spectrometer (Thermo Scientific, Mass. USA). Peptides were separated on a 75 µm ID silica emitter tip (CoAnn Technologies, Wash. Calif. USA) column that was in-house packed with Aeris 2.6 µm PEPTIDE XB-C18 100 Å bead material (Phenomenex, Calif. USA) on a length of 20 cm. The LC gradient between mobile phase A (0.1% formic acid in water) and mobile phase B (0.1% formic acid in 90% aqueous ACN) was developed from 5% B to 25% B over 13 min followed by an increase to 40% B over 3 min. and 90% B over 2 min. at a flow rate of 300 nl/min. The LTQ-Orbitrap mass spectrometer was operated in an untargeted data acquisition mode. The precursor ion scan was performed in a m/z window of 400 to 2000 at a resolution of 60,000. The precursor ion scan was followed by 9 data-dependent collision-induced dissociation (CID) fragment ion spectrum acquisitions, of which the first three scans were set up as semi-targeted acquisitions of pre-specified precursor masses if present in the precursor ion scan at a threshold intensity of at least 5000 counts. The following 6 CID acquisitions were set up as data-dependent scans of the 6 most intense ions. The peptide targets included in the precursor ion list were three peptidoforms of the AMELY encoded amelogenin isoform (SMIRPPY at m/z 432.23, SM_(OX)IRPPY at m/z 440.22 and SM_(OX)IRPPYS at m/z 483.74), one peptide specific for the AMELX encoded isoform (SIRPPYPSY at m/z 540.28) and one peptide common to both isoforms (MPLPPHPGHPGYINF at m/z 837.42).

Raw data were analysed by both software-assisted sequence database searches and manual spectrum interpretation. The Proteome Discoverer software (version 2.5, Thermo Scientific, Mass. USA) was used for sequence assignment using the Sequest HT program to search the human reference and UniProt amino acid sequence databases. Enzyme cleavage was set to non-specific and methionine oxidation was included as a variable modification. Precursor and fragment ion intensities (area under the curve) were extracted manually using the Qual Browser program of the Xcalibur software package (version 2.0.7, Thermo Scientific, Mass. USA). Proteomic data are presented in Supplementary Table 3 and Supplementary Fig. 2.

Stable isotope background information

Stable isotope analysis of human bone collagen and tooth dentine is a well-established method to investigate diet^61,62. Carbon and nitrogen stable isotope values are measured by parts per thousand (‰) relative to a standard (VPBD for carbon and Air for nitrogen) and denoted by the δ symbol. The δ¹⁵N values of bone collagen are only representative of dietary protein as carbohydrates and lipids do not contain nitrogen⁶³. The δ¹³C values of bone collagen are primarily representative of dietary protein because amino acids from protein are routed for collagen synthesis, but some carbon from carbohydrates or lipids may be used, especially in cases where there is low protein in the diet⁴¹. Carbon stable isotope values are primarily used to assess the consumption of plants with differing photosynthetic pathways (C3 vs. C4 plants) and to help determine the proportion of marine and terrestrial foods in the diet. C4 plants and marine systems display higher δ¹³C values compared to C3 and terrestrial systems⁶⁴. Importantly, there very few native and no cultivated C4 terrestrial plants in Aotearoa during the Māōri-era⁴⁴ and C4 cultivars only arrived after European colonization. Nitrogen stable isotope values are used to understand the trophic position of an organism because δ¹⁵N values increase approximately 2–6‰ with every trophic step, although stepwise increase for omnivores is likely in the lower threshold^65,66. Marine and freshwater systems also typically display higher δ¹⁵N values compared to terrestrial systems because there are more trophic steps in aquatic ecosystems^64,67. Used in conjunction, bone collagen δ¹³C and δ¹⁵N values can provide information about the consumption of plant, meat, marine and freshwater organisms^41,62,67.

Trophic differences in δ¹⁵N values of infant and adult female bone collagen or microsamples of dentine from an individual’s tooth are commonly used to determine patterns of breastfeeding and weaning in the past^47,68,69. Infants move up a trophic level when they consume the breastmilk of their mother, resulting in the enrichment in ¹⁵N of infant collagenous tissues, including bones and teeth⁷⁰. This is signalled by a 2–5‰ increase in infant δ¹⁵N values and a ~ 0–2‰ increase in δ¹³C values, relative to maternal isotopic values^69,71. If a child is fed supplemental weaning foods that are from a lower trophic level than breastmilk (i.e., similar food to the mother’s diet), the δ¹⁵N values (and, potentially, δ¹³C_dentine values) of the child’s tissues will lower to reflect this dietary change at the rate of tissue formation (for teeth or new bone) or remodelling (for bone and keratinous tissues like hair and nails)^69,72.

Bone collagen δ
¹³C_collagen and δ
¹⁵N_collagen analysis

All the bone samples were prepared at the Stable Isotope Prep Lab, Archaeology Programme, University of Otago, Ōtepoti, using a modified Longin method, described in Kinaston et al.⁴². Specifically, bone fragments were cleaned using aluminium oxide air-abrasive equipment (Bego Easyblast) and demineralised in 0.5 M HCl at 4 °C for several days. Samples were then rinsed in MilliQ H₂O until a neutral pH was reached. Collagen was gelatinised at 70 °C in a pH 3 solution for 48 h, followed by filtration using 5–8 µm Ezee® mesh filters (Elkay Laboratory Products) to remove reflux-insoluble residues. The resulting solution was subsequently ultrafiltered using Millipore Amicon Ultra-4 centrifugal filters (30 kDa NMWL) to retain molecules larger than 30 kDa. Prior to ultrafiltration, Millipore Amicon Ultra-4 centrifugal filters were pre-cleaned by centrifuging Milli-Q H₂O through the units three times at 4000 × g.

The purified collagen was analysed at IsoAnalytical (Cheshire, UK). The reference material used for δ¹³C and δ¹⁵N analysis of the collagen samples was IA-R068 (soy protein, δ¹³C _V-PDB = -25.22 ‰, δ¹⁵N _AIR = 0.99 ‰). IA-R068, IA-R038 (L-alanine, δ¹³C _V-PDB = -24.99 ‰, δ¹⁵N _AIR = -0.65 ‰), IA-R069 (tuna protein, δ¹³C _V-PDB = −18.88 ‰, δ¹⁵N_AIR = 11.60 ‰) and a mixture of IAEA-C7 (oxalic acid, δ¹³C _V-PDB = −14.48 ‰) and IA-R046 (ammonium sulfate, δ¹⁵N_AIR = 22.04 ‰) were run as quality control check samples during analysis of the collagen samples. IA-R068, IA-R038 and IA-R069 were calibrated against and traceable to IAEA-CH-6 (sucrose, δ¹³C_V-PDB = −10.43‰) and IAEA-N−1 (ammonium sulfate, δ¹⁵N _AIR = 0.40‰). IA-R046 was calibrated against and traceable to IAEA-N−1. IAEA-C7, IAEA-CH-6 and IAEA-N-1 are inter-laboratory comparison standards distributed by the International Atomic Energy Agency, Vienna. All measured values for the quality control standards were in the range of the accepted values. Analytical error, calculated from replicate measurements of samples, was routinely ±0.1‰ for δ¹³C values and ±0.2‰ for δ¹⁵N values (1 SD, n = 7). Descriptive and inferential statistics for the Tamahere and other individuals were calculated in Excel (version 16.16.27) and R (version 4.5.1) and these are presented in Table 2, Fig. 5 and in the text.

Bulk tooth dentine micropunch sampling and δ
¹³C_dentine and δ
¹⁵N_dentine analysis

Teeth were sectioned at the Bioarchaeology Laboratory, Department of Anatomy, University of Otago, Ōtepoti. Each tooth was cleaned by manual abrasion with a toothbrush to remove any contaminants. Following this, the tooth was partially embedded in gypsum (Victor® Casting Plaster) in a silicone mold and cut with a Buehler Isomet low-speed saw. The mesial or distal root of the tooth was not embedded, which helped to direct the blade when cutting the tooth. The tooth was secured into a chuck and then longitudinally cut to obtain a 2 mm slice from the centre of the tooth. The slice and the rest of the tooth were removed from the gypsum using MilliQ water.

The dentine was prepared at the Stable Isotope Prep Lab, Archaeology Programme, University of Otago, Ōtepoti. The 2 mm wide tooth section was demineralized in 0.5 M HCl for approximately 7–10 days until the dentine was flexible, but still retained its original shape, then rinsed three times with Milli-Q H₂O. The tooth was sampled sequentially from the crown cusp to the apex of the root along the mesial or distal side using a KAI Medical 1 mm diameter biopsy punch needle with a plunger. Depending on tooth size, between 10 and 11 micro-samples per tooth were obtained using this method.

The 1 × 2 mm cylindrical microsamples were then transferred to 1.5 ml eppindorf microtubes and weighed. The microtubes were labelled numerically in a sequence from crown to root apex, indicating their anatomical sampling location, allowing age alignment in later analyses. Each sample was frozen, before being lyophilised for 48 h.

The dentine micropunch samples were analysed at IsoTrace, Dunedin, New Zealand. Analytical precision was determined through repeated analyses of laboratory standards EDTA-OAS and IAEA MA-A-1 (copepod) and comparing the results against accepted values. All measured values for the quality control standards were in the range of the accepted values. All dentine micropunch stable isotope data are presented in Supplementary Table 2.

The age (midpoint of age range) of formation for each dentine micro-punch sample was estimated using the age-alignment method from Czermak et al.²⁷, which considers differences in the rate of dentine formation between anatomical regions of the tooth (crown, neck, upper and lower root). The age range of formation for each dentine micro-sample was dependent on the anatomical region it was obtained from. By dividing the average time of formation for an anatomical region²⁸ by the number of dentine samples obtained from that region, the age of formation for dentine samples from each anatomical zone of the tooth were estimated and visualized using Excel (version 16.16.27). Midpoints of each section were then calculated (Fig. 3).

Enamel carbonate δ
¹³C analysis

Carbonate is present in biological apatites that form the inorganic part of bones and teeth, especially tooth enamel⁷³. As carbonate is formed using CO₂ within the blood, the δ¹³C values of biological apatite are a reflection of the total the macronutrients (e.g., carbohydrates, proteins and lipids) in an individual’s diet^25,73,74. The diet-tissue spacing of δ¹³C values between diet and bone/enamel carbonate has been found to range between ~9‰ – 14‰, and 12‰ is commonly used for humans^29,75,76.

Enamel samples were cleaned by abrasion of surface contaminants, using a Dremel® rotary tool with a clean diamond cutting blade at the Stable Isotope Prep Lab, Archaeology Programme, University of Otago, Ōtepoti. An enamel chip of approximately 10–20 mg was taken from each tooth directly next to the cemento-enamel junction (CEJ). Any residual contaminants, including dentine, were removed using the Dremel^® with a diamond cutting blade. The samples were then weighed. To remove organic and diagenetic contaminants, the enamel chips were treated by soaking in 1% NaClO for 24 h, then rinsed three times by Milli-Q H2O, soaked in 0.1 M acetic acid for 24 h and rinsed three more time with Milli-Q H2O and dried and powdered.

Carbon and oxygen stable isotope analyses from the carbonate was undertaken at Isotrace Research Laboratory, Dunedin, New Zealand using a Thermo Delta Plus Advantage linked to a Gasbench II via a GC PAL autosampler. The dried, homogenised sample were weighed in an 8 × 5 mm tin (Sn) capsule for analysis. Carbonate values were normalized by a 3-point calibration. Normalization was made using International Standards (NBS-18 [δ¹³C = -5.01][δ¹⁸O = −23.01]; IRU-Marble [δ¹³C = 2.04][δ¹⁸O = −2.58]; NBS-19 [δ¹³C = 1.95][δ¹⁸O = −2.20]). Delta values were normalized and reported against the international standards Vienna Pee Dee Belemnite (VPDB). Analytical precision was checked by comparing results from analysed quality control standard IRU-Marble against accepted values. All measured values for the quality control standards were in the range of accepted values. Carbonate stable isotope data are presented in Supplementary Table 3 and Fig. 4.

Strontium isotope analysis

Strontium isotope analysis has been used to assess the mobility and migration patterns of humans around the world^77,78 but this method has only been reported once previously to investigate mobility in Māori-era individuals³. The method is based on the premise that the biologically available strontium a human or animal ingests during the time of tooth mineralization is retained within the enamel, which is generally resistant to diagenetic alteration in the burial environment. Underlying rocks exhibit variable ⁸⁷Sr/⁸⁶Sr ratios depending on lithology, age, and original rubidium (Rb) content. Erosion of the underlying bedrock, airborne dust (loess), sea-spray, atmospheric deposition, groundwater, and stream water are all contributors to the bioavailable strontium. Strontium purified from enamel should be representative of the biogenic strontium available during the time of tooth mineralization^79,80. The ⁸⁷Sr/⁸⁶Sr ratio of tooth enamel reflects the bioavailable strontium isotope signature of the food and drink a person consumed during the time of tooth mineralization⁷⁸.

For radiogenic strontium isotope analysis, 10–20 mg enamel samples from the crown region adjacent to the CEJ of the first molars of Kōiwi 2 and Kōiwi 4 were chemically prepared within the Class-10 (ISO 4) workstations of the Class 100 (ISO 5) clean laboratory suite of the Centre for Trace Element Analysis, Department of Geology, University of Otago, Dunedin, New Zealand. Prior to transferring to the clean lab, the enamel was cleaned through abrasion with a sonicated Dremel® reinforced diamond cutting wheel to remove possible surface contaminants. Any additional adhering materials, such as organic matter or dentin, were removed and the enamel sample was sonicated in MilliQ H2O for five minutes. Samples were then transferred to the clean lab and an appropriate mass of each sample was weighed into an acid-cleaned perfluoroalkoxy alkane (PFA) vial (Savillex Ltd, USA) and digested in 2 ml of 3 M HNO₃ overnight at 110 °C. The digests were then evaporated to dryness then reconstituted in 2 ml of 3 M HNO₃ and subsampled for the determination of strontium concentration by ICP-MS using a 7900 instrument (Agilent Technologies, USA) after appropriate dilution.

Strontium was separated from the sample matrix and purified using established protocols in ion exchange chromatography^81,82. The column eluant containing the strontium was evaporated to dryness, reconstituted in 2% (v/v) HNO₃, then analysed for its ⁸⁷Sr/⁸⁶Sr composition using a Nu Plasma-HR MC-ICP-MS instrument (Nu Instruments Ltd, UK). All ⁸⁷Sr/⁸⁶Sr data were first corrected for instrumental mass fractionation and isobaric interferences from Kr and Rb across the Sr mass range using standard procedures, then normalized to the composition of the NIST SRM 987 primary standard (⁸⁷Sr/⁸⁶Sr = 0.710248). An additional in-house HPS standard was used to independently monitor the accuracy and external precision of the measurements. All ⁸⁷Sr/⁸⁶Sr results for HPS were in excellent agreement, within error, with the long-term mean value of 0.70762 ± 0.00003 (2 SD, n = 189). Procedural blanks were run with each batch of 48 samples and all yielded negligible Sr levels of <250 pg. Strontium isotope data are presented in Supplementary Table 3 and prediction maps using an Aotearoa isoscape (detailed below) are presented in Supplementary Fig. 3.

Isoscape interpretations

A strontium isotope map, or isoscape, is a model that predicts the spatial distribution of ⁸⁷Sr/⁸⁶Sr values across the landscape by incorporating the multitude of strontium sources that contribute to the bioavailable strontium ‘pool’ of a region⁸³. Country and region-specific isoscapes are developed using a machine learning approach (a random forest model) that considers a plethora of atmospheric and geological variables (precipitation, temperature, soil depth, silt/sand/gravel content, underlying bedrock, etc.) ^e.g.,.^84,85,86. The random forest model predicts the expected strontium isotope value for every location (1 × 1 km²) across the targeted regions using the strontium isotope values from real-world plant, animal, and soil samples and the known environmental variables. If there is no real-world sample information for a location, the random forest model will predict the strontium isotope value using the nearby known environmental variables and similar geological ⁸⁷Sr/⁸⁶Sr values³¹.

The kōiwi tooth enamel ⁸⁷Sr/⁸⁶Sr results are interpreted using an updated version of the Aotearoa isoscape developed by Kramer et al.³¹. The interpretations use assignR (version 2.4.3), which operates in a semi-parametric Bayesian framework to calculate the posterior probability of the sample belonging to each cell within the isoscape raster. Maps of the North Island were created using R (version 4.5.1) and ArcGIS Pro (version 2.1) detailing the likely area of childhood residency for each individual (Supplementary Fig. 3). To facilitate visualization, maps are presented displaying the top 33% probability by area of the predicted isoscape surface for each sample with the top 33% areas coded as 1 (blue colour) and other areas coded as 0 (no colour). The top 20% and the 10% probability by area maps of the predicted isoscape surface are also presented. It has been found that there is a trade-off between accuracy and precision moving from the 33% to the 10% probability thresholds, where 33% is the highest accuracy and 10% the highest precision (i.e., least amount of areas selected)³¹. This means that, although a wider area is selected for the top 33%, the accuracy of the place of origin prediction is higher on these maps compared to the 10% and 20% maps, where fewer regions of potential origin are predicted³¹. Comparisons are made between the maps showing the predicted cells for the top 33%, top 20% and top 10% probability surfaces to assess the possible variation in accuracy and precision for these thresholds.

Dietary baseline

Human dietary isotope data should be interpreted using a dietary baseline of potential plant and animal food resources. For this study, published stable isotope data of modern and Māori-era plants and animals^3,44,87,88 were compiled to establish an Aotearoa-specific dietary baseline. Food-value ellipses were calculated in Excel (version 16.16.27) from the mean δ¹³C and δ¹⁵N values and their covariance matrix, incorporating correlation between the two isotopes. These ellipses represent 95% confidence intervals for the bivariate distribution of each food group.

Supplementary Data 3 details the data used to create the dietary baseline. To address the global decline in ¹³C following the Industrial Revolution (known as the Suess effect), baseline data for modern Pacific island plant and animal δ¹³C_collagen values were adjusted by adding +1.5 for terrestrial systems and +0.86‰ for marine systems^89,90. To interpret the edible fraction of the potential food resources, bone δ¹³C_collagen values were corrected by −3.7‰ for fish and −1.5‰ for birds and mammals and δ¹⁵N_collagen values by −0.6‰ for all species^25,30,89. Plants and animals were categorized into groups (C3 plants, terrestrial birds, freshwater organisms, waterfowl, marine shellfish, marine fish, marine birds and marine mammals) and visualized using 95% confidence ellipses.

We included only edible C3 plants that would have been available in Aotearoa during the Traditional Period. Four samples of kūmara (Ipomoea batatas) collected by B.F. Leach from an Auckland market were used as reference material⁴⁴. These kūmara were likely grown locally in Auckland, a region with rainfall ( ~ 1100–1200 mm annually) and mean temperature (15.5 °C vs. 14.0 °C) similar to Tamahere, suggesting minimal regional differences in δ¹³C values of C3 plants grown in each area. The average δ¹³C value of the kūmara (-26.7‰) aligns with global averages for C3 plants (–26‰ to –27‰), though values can range between –20‰ and –37‰ depending on aridity and canopy effects^91,92. The average kūmara δ¹⁵N values (4.0‰) are comparable to non-N2-fixing plants ranging from 0 to + 6‰ but average 3‰⁹³, although factors such as animal fertilizers and burning/swidden cultivation can increase plant δ¹⁵N values⁹⁴. The kūmara δ¹³C values used are also consistent with those of modern potatoes (Solanum tuberosum), pumpkin (Curcurbita spp.) and peas (Pisum sativum) (n = 6, −25.6‰) grown in Aotearoa, but slightly higher than the average δ¹⁵N values for the non N2-fixing vegetables (potatoes and pumpkin) (n = 4, and 2.8), which may be related to environmental factors⁹⁵. There are very few C4 plant species native to Aotearoa (exceptions being Atriplex buchananii and Zoysia spp.) and C4 plants would not have contributed to the diet of Māori until the introduction corn (Zea mays) and sugarcane (Saccharum officinarum) by European colonists in the 19^th and early 20^th centuries. A correction for diet-tissue spacing was applied to the human δ¹³C_collagen and δ¹⁵N_collagen values (5‰ δ¹³C for and 3‰ for δ¹⁵N values)^25,26,96.

Comparative human stable isotope results

Due to unethical archaeological and collection practices during the 19^th and 20th centuries, very little bioarchaeological research was conducted in Aotearoa after the 1970s, especially work involving destructive analyses of human remains^97,98. Today, kōiwi are never excavated for purely research purposes. Instead, remains are recovered only when accidentally disturbed through erosion, construction, or other activities. Once identified as archaeological (pre-1900 CE) by police and (bio)archaeologists alongside Heritage New Zealand Pouhere Taonga, decisions about analysis and reburial of the kōiwi rest with the hapū who hold kaitiakitanga (guardianship) over the rohe and tūpuna. In many cases, kōiwi are immediately reburied at the request of the hapū, or limited information is obtained and reported only in confidential reports.

As a result, comparative Māori Era isotopic studies are extremely limited. Only one large isotopic study, focused on kōiwi from Wairau Bar (site P28/21), has been published³ and is used for comparison with the Tamahere results (Table 2, Supplementary Data 1, Fig. 5). Leach et al.⁴⁴ reported on the isotopic analysis of kōiwi and kōimi (human skeletal remains from Rēkohu/Chatham Island) from four sites and one Historic-era European individual, but did not report the collagen quality indicators associated with the stable isotope results and did not report δ¹³C and δ¹⁵N values for all the individuals analysed. Taking into consideration that there is no information for the collagen quality indicators, we cautiously use the Leach et al⁴⁴ stable isotope values for comparison with the Tamahere kōiwi. We have compiled the stable isotope data for: (1) five kōimi from the Waihora site on the southwest coast of Rēkohu (Chatham Island) likely dating to the Middle/Traditional Periods; (2) one kōiwi from the Lagoon Flat Archaic site complex (site O32/31) on the Kaikoura coast north of the Conway River mouth dating to the Early Period; (3) four kōiwi from an inland Bay of Plenty site near Lake Rotoiti likely dating to the Traditional Period; (4) one 87 year old European male buried in 1876 (identification known) from the Withells Road Methodist Cemetery in Christchurch, Canterbury (included in the European comparative analyses below) (Table 2, Supplementary Data 1, Fig. 1 and Fig. 5)⁴⁴. Leach et al.⁴⁴ also analysed one individual (B41a) from Wairau Bar and their results (δ¹³C = −15.9‰ and δ¹⁵N = 16‰) were similar to those reported by Kinaston et al³ for the same individual (δ¹³C = −15.7‰, δ¹⁵N = 17.8‰) (Supplementary Data 1). Only the data from for Wairau Bar Burial 41 reported by Kinaston et al. (2013) is used in Figs. 2 and 5. Stable isotope data from a total of 37 Māori and Moriori individuals were used for comparison with the Tamahere kōiwi.

The only other isotopic research of archaeological human remains in Aotearoa has been focused on Pākehā (European) and Chinese burials from Colonial Era cemeteries in the South Island of Aotearoa. Bone collagen, dentine, and hair stable isotope values have been compiled from published research (Supplementary Data 1 and 2) and include: (1) European adults (n = 7 bone collagen, n = 3 hair) and non-adults (n = 1 dentine, n = 7 hair) from the St. John’s Anglican cemetery in Milton, Otago (majority graves dating to between AD 1860-1880); (2) European adults (n = 3 bone collagen, n = 3 hair) from the Ardrossan St. cemetery in Lawrence, Otago (ca. AD 1861-1864); (3) Chinese adults (n = 4 bone collagen, n = 4 hair) and one European adult (n = 1 bone collagen) from the Gabriel St. cemetery in Lawrence, Otago (AD 1864-late 1890s); and (4) one European (bone collagen and hair) and one likely Chinese individual (only bone collagen) from the Litany St cemetery, Cromwell, Central Otago (likely AD 1890s)^45,46,47,48 (Supplementary Data 1 and 2). A number of these individuals have been genetically analysed to confirm ancestry and, for the untested individuals, their ancestry was determined using morphometrics and associated material culture in graves⁹⁹. The enamel δ¹³C values and post-weaning dentine δ¹³C and δ¹⁵N values of the Colonial Era adult individuals were not used for comparative purposes because these values reflect the childhood diet of these people likely before immigration to Aotearoa^45,48.

For the Colonial Era individuals, both hair and bone stable isotope data were presented (Supplementary Data 1 and 2 and Fig. 5) because of the potential issue that most of the adult individuals were immigrants to Aotearoa from Europe or China. It is thought the turnover rate for cortical bone is ~10 years for adults¹⁰⁰ and therefore the stable isotope values of adult bone may be representative of the individual’s diet before they moved to Aotearoa. Hair isotope values represent the diet of an individual in the months or years before a person’s death, depending on the length of the hair and is therefore likely more representative of the NZ-based diet of these immigrants. Hair was sampled incrementally and the average δ¹³C and δ¹⁵N values of the well-preserved increments is presented for each individual (Supplementary Data 2). The stable isotope values of increments of child dentine and hair were only compiled if it represented the diet of the child post-weaning to avoid any possibly influence of trophic enrichment from breastmilk. Because hair is slightly enriched in ¹³C and ¹⁵N relative to bone collagen, we applied a conservative correction of -1‰ to the hair δ¹³C and δ¹⁵N values to make them comparable with bone and dentine δ¹³C and δ¹⁵N values^101,102. Due to the fact the dietary baseline of the Māori-Era and Colonial Era individuals was different, we only compared δ¹³C and δ¹⁵N averages ( ± SD) between the groups to address dietary trends because the diets of the colonial-era individuals were previously interpreted as being heavily reliant on C3-terrestrial based resources^45,46,47,48. The hair isotope values of one individual, Burial 12, from the Ardrossan St cemetery, were removed from the summary statistics as this person was likely eating a diet with a large proportion of C4 plant foods at the time before death. Descriptive statistics for all comparative samples were calculated using Excel (version 16.16.27) and R (version 4.5.1) and results are presented in Table 2 and Fig. 5.

Reporting summary

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