Dynamic Sustainability, Resource Management, and Collective Action on Two Atolls in the Remote Pacific

Abstract

Examples of environmental transformation, the creation of sustainable lifeways, and the development of environmentally aware political forms better our understanding of how peoples build on tradition and environmental circumstance to form novel institutions. Using archaeological data, oral histories, genealogies, radiocarbon dating, and Bayesian modeling, we present a timeline of habitation and land-use patterns on Manihiki and Rakahanga, two remote atolls in East Polynesia. We track socioecological change on the atolls from the time of first colonization ca. AD 1200–1400 through to sustained European contact in the mid-1800s. The findings document and temporally anchor collective action-based processes of landscape transformation, the development of a system of cyclical mass migration aimed at sustainable resource use, and the implementation of a novel dual-chiefdom political system. This demonstrates that new levels of political “complexity” manifest as patterns of shifting hierarchy and novel forms of political and ecological management, and can arise in relation to specific social and ecological challenges in systems of any size. The perpetuation and adaptation of aspects of these traditional institutions can help to maintain the sustainability of populations today in the face of climatic and social change.

1. Introduction

The islands of Polynesia have long been a focus of research regarding human-induced ecological change and sociopolitical development (e.g., [1,2,3,4,5]). These complexities have been documented in large-island groups such as Hawai’i, the Society Islands, and the Marquesas Islands (e.g., [3,6,7]). Many coral atolls, small islands, and small-island groups, however, lack the data to formulate a deep-time understanding of the development of political and ecological systems in these settings. This has led to interpretations of past atoll-dwelling peoples as developing less “complex” social structures based on immediate survival rather than the planned production of surplus [4]. However, atoll-dwelling groups often alter landscapes, and develop creative solutions to the inherent limitations of coral-island environments [8,9]. These strategies often represent collective action-based approaches that provide opportunities for long-term sustainability through political flexibility and creative resource management.

Collective action research seeks to understand the varied ways in which individuals participate in group behaviors at different scales as the respective costs and benefits of participation shift ([10], pp. 37–38; [11], p. 100). Sustained collective action efforts are key to the development of institutions, that is, the “organizations of people that carry out objectives using regularized practices and norms, labor, and resources” ([10], pp. 40–41; [12,13]). Here, we explore the case of Manihiki and Rakahanga, a dual-atoll cluster in the Northern Cook Islands of East Polynesia (Figure 1). On these small and remote atolls, ancestral Polynesian practices engendered the development of innovative institutions, including a cyclical dual-chiefdom political structure and a population-level migration-based system of resource management. The case of Manihiki and Rakahanga demonstrates how such institutions, which rely on flexible leadership, creative resource management, and population-wide cooperative efforts, can aid in long-term cultural and ecological sustainability.

Throughout our reconstruction of the history of these atolls, we frame our discussion around the ideas of collective action and institution-building (e.g., [10]). We return to these points in our conclusions to consider how the people of Manihiki and Rakahanga built enduring, multifaceted, and resilient institutions that allowed for the long-term habitation of what are, on the whole, challenging places to live. Our main point is to illustrate the varying circumstances under which people accept institutions and engage in collective action. As Bondarenko ([14], p. 7) notes, people “accept institutions because they are products of joint action.” Of course, there is always the possibility that individuals will break with convention and that such actions will hamper collective action (see [15]). However, we suggest that the people of Manihiki and Rakahanga were acutely aware of how important these institutions were to surviving on these atolls and were heavily invested in their creation and persistence. We argue that, in part, the continuation of the settlements on these islands the and sustainability of local lifeways required collective buy-in for their success.

Located in the Northern Cook Islands, the small equatorial atolls of Manihiki and Rakahanga have a combined landmass of 9.5 km² and an elevation that only reaches about four meters above sea level at its highest point [16,17]. Today, both atolls are covered in coconut groves, with Rakahanga having large swamp taro pit-fields. Both atolls have numerous stone-walled fish traps located in the shallow channels that connect the lagoons to the ocean. Together, Manihiki and Rakahanga comprise a geographically secluded coral cluster [18] consisting of two atolls 350 km from the nearest neighboring atoll and more than 900 km from the nearest high island. Many of the people on Manihiki make their living through the farming and sale of black pearls from the black-lipped pearl oysters (Pinctada margaritifera) cultivated in Manihiki’s large lagoon. While Rakahanga lacks pearl farms, the islanders are known for their beautiful coconut and pandanus weavings.

The Northern Cook Islands are comprised of five coral atolls (Pukapuka, Penrhyn (or Tongareva), Suwarrow, Manihiki, and Rakahanga) and one sandy cay (Nassau) spread across roughly 200,000 km² of the Pacific Ocean. Manihiki and Rakahanga are roughly 10° South of the Equator and have an equatorial climate, with a wet season that lasts from November to April and a dry season lasting from May to October. Historic records for Manihiki report an average of 240.64 cm of rainfall per year in the mid-1900s [19]. Cyclone season in the Northern Cook Islands is between November and April [20]. Cyclonic winds and storm surges can be devastating to low-lying atolls. The most recent cyclonic event was Cyclone Martin, which hit Manihiki on 1 November 1997 causing the loss of many lives and damaging the infrastructure of the atoll [21].

By the time Christian missionaries arrived in AD 1849, the people of Manihiki and Rakahanga were organized in a form that resembled other Polynesian systems of chiefly hierarchy in certain respects, but widely diverged in others [22]. Oral histories and ethnohistoric accounts (e.g., [22,23]) suggest that Manihiki and Rakahanga were inhabited by a single people living in one village on a tiny Rakahangan islet called Te Kainga, literally meaning “the home.” Hiroa ([22], p. 59) stated that the people were divided into four lineage groups. Two Ariki, or chiefs, oversaw these groups in a dual chiefdom (or Arikiship), and the people mitigated resource depletion on the small atolls through a system of cyclical migration. When resources ran low on Rakahanga, the entire population voyaged to Manihiki, where they split into multiple villages. After a time, they would return to Rakahanga and reunite in the single village ([22], pp. 65–66). Maori authors later referred to these migrations as the Tûmutu [24,25].

Components of the pre-missionization fission–fusion system of the dual chiefdom and Tûmutu migration institutions seem to be local developments. However, there are parallels with other traditions in the region and around the world. These similarities and differences with other systems make Manihiki and Rakahanga an interesting case study by which archaeological interpretations of dispersed and aggregated settlement patterns can be evaluated.

Here, we present the first archaeological timeline for social and ecological changes spanning the entire human occupation of Manihiki and Rakahanga. We integrate AMS 14C dating, historical documentation, and oral histories in order to trace changes from first colonization to Christian missionization while providing timeframes for the formation of the dual Arikiship and the Tûmutu. Through this research, three specific research questions are addressed: (1) what is the timing and nature of habitation on Manihiki and Rakahanga?; (2) how do patterns of land and resource use on Manihiki and Rakahanga relate to the creation of the dual Arikiship?; and (3) how do patterns of land and resource use on Manihiki and Rakahanga relate to the development of the Tûmutu? In Section 5, we consider what these deep histories of human settlement, cyclical migration, political reorganization, resource use, and cooperation mean for creating resilience and sustainability in this socioecological landscape. We demonstrate that despite the limitations of life in atoll environments, collective action, social cohesion, political flexibility, and community buy in are vital components to sustainable lifeways both past and present.

1.1. Coral Atoll Life

Though highly variable in their age, size, composition, and remoteness, coral islands are arguably among the most precarious and marginal landforms for humans on the planet [26,27,28,29,30]. Comprised of topographically low-lying islets with poor coral soils, atolls lack certain raw materials (e.g., stone, clay, and large amounts of timber) and possess limited terrestrial biodiversity [18,29]. Due to the sparseness of resources, many atoll-based cultures relied upon either imported materials or crafted tools from available resources, such as wood, bone, shell, and coral. Surface water is absent from atolls and potable water is only accessible through rain catchment or by excavating into the Ghyben–Herzberg fresh (sometimes brackish) water lens, which sits above the dense saltwater within the porous coral of atolls ([8], p. 569).

Despite their challenges, one of the most important aspects of atolls is their range of highly productive marine environments (e.g., lagoon, reef, offshore, etc.). They often have an abundance of fishes as well as sea turtles and invertebrates. Thus, subsistence is typically focused on these marine resources. There are several crops that can be cultivated on coral islands, including tree crops such as coconut (Cocos nucifera), pandanus (Pandanus spp.), and breadfruit (Artocarpus altilis), as well as swamp taro (Cyrtosperma chamissonis). These plants provide carbohydrate- and calorie-rich foods that complement the protein-rich marine foods available on atolls. Swamp taro, known on Manihiki and Rakahanga as puraka, is grown in cultivation pits excavated into the atolls’ fresh water lenses ([31], p. 7). It is a geophyte of the family Araceae, and can grow continuously for years [32]. Giant swamp taro plants (puraka) grow today in long-abandoned cultivation pits on Manihiki, suggesting that puraka will continue to grow for years or even decades without human intervention.

Early coral atoll settlements are typically located near the center of the largest islet of the atoll ([8], p. 582; [33], p. 638). Subsistence and habitation patterns are often related to the greater configuration of the landscape with geographically proximal atolls and islands acting as extensions of the home islet or island resource base [18,34]. Thus, many of the environmental challenges posed by atoll life can be mitigated through landscape transformation, innovative and traditional subsistence practices, and the development of integrated social institutions.

1.2. Polynesia and Traditional Institutions

Roughly 3100 to 3000 years ago the Lapita peoples, known for their distinctive dentate-stamped pottery, voyaged across the 380 km stretch of sea separating the Solomon and Santa Cruz Islands and entered the previously uninhabited islands of Remote Oceania [35,36,37]. Lapita voyagers colonized much of the western Pacific, including Fiji, New Caledonia, and Vanuatu [36]. Lapita peoples reached their easternmost point with the colonization of Tonga and Samoa between 3000–2800 years ago [38,39,40]. For reasons that continue to be debated (see [41,42]), the settlement of West Polynesia was followed by a roughly 2000-year-long hiatus in eastward expansion known as the “long pause” ([43], p. 140; [44]). During this time, societal changes in West Polynesia led to the emergence of ancestral Polynesian societies, the abandonment of pottery use, and the development of the Polynesian chiefdom political structure ([43], p. 365; [45], p. 140; [46], pp. 51–52, 68).

Eastward expansion began again around AD 900, and soon thereafter the islands of central East Polynesia (e.g., the Southern Cook Islands, Society Islands, and Marquesas) were colonized [42,44,47,48]. Sea level lowering around this time exposed many of the reefs in the region, leading to the formation of habitable sandy cays and atolls, including those of the Northern Cook Islands and Tuamotu Islands ([26]; ([49], p. 92)). This environmental change created new opportunities for expansion, as habitable land emerged in large expanses of the Pacific formerly devoid of islands. Climatological changes after AD 1150 may have increased precipitation in the region, making small and remote islands more favorable for colonization [42]. Shortly after or concurrent with, these environmental shifts, there was a final voyaging push in East Polynesia in which several of the most remote islands in the world were colonized (e.g., Hawai’i, Aotearoa (New Zealand), and Rapa Nui [44]).

1.2.1. Political Structures

Many of the voyagers who colonized the islands of East Polynesia crossed larger ocean gaps than their predecessors and became increasingly isolated from their ancestral homelands. The traditions they brought to East Polynesia included a flexible, lineage-based, often-patrilineal chiefly organizational structure with a strong preference toward primogeniture ([46], pp. 31–34). These ancestral institutions were modified into new forms. This included the diminishment of chiefly power and the rise of inspirational priests in the Marquesas as well as the emergence of archaic states in Hawaii ([3]; [7], pp. 44–45). In the low-lying newly formed atolls of East Polynesia a wide variety of political institutions emerged ([4], pp. 234–237). Sahlins ([4], p. 237) has argued that this variation was due to a lack of surplus food production, as well as to the many and varied challenges found on individual atolls and among different atoll groups. Atoll-dwelling populations face a unique set of socioecological challenges, including the management of reefs, tree crops, and long-lived cultivars such as swamp taro. For example, swamp taro production requires significant up-front labor to excavate cultivation pits into the coralline sediments of atolls. The same can be said of the alteration of islet channels to create fishponds or stone (coral) fish traps. These factors likely required cooperative management from different segments of the population and played an important role in the development of atoll political structures. Numerous atoll polities developed political structures containing multiple levels of hierarchy and featuring community participation through counsels or other structures that favored consensus-building over absolute chiefly power. DeMarrais and Earle [50] have suggested that the shared need of a population may influence the development of social institutions geared towards collective action. The collective need to manage a small-scale resource base may have influenced the development of many atoll-oriented political institutions. The restricted distribution of resources in environmentally circumscribed environments likely acted as a barrier to freeriders or abusers of common pooled resources (see [51]), an effect which may have been more acutely felt on small coral atolls.

In the Tuamotu Archipelago, small-scale territorial polities united under a single “supreme chief” ([52], pp. 4–5). These polities maintained high mobility and practiced bilateral descent, an organizational form said to be related to ecological impoverishment and a high frequency of inter-polity conflict ([18], pp. 105–107). Certain groups, such as the Ana’a of the Tuamotus, may have comprised regional chiefdoms with force-based tributary structures ([53], p. 268; [54], p. 110) and strong trade-based ties to the Society Islands and the Marquesas ([7], p. 259; [54], p. 122). Conversely, the remote atoll of Pukapuka in the Northern Cook Islands developed a three-tiered chiefly hierarchy with a sacred supreme chief, two tiers of sub-chiefs, and a counsel structure that afforded political voice to adult men ([55], pp. 234–235). This political system, with its high leader-to-commoner ratio, may have managed the social and ecological concerns of a relatively dense but isolated population. Another political structure emerged on Manihiki and Rakahanga, where a small and remote population developed a dual-chiefdom political structure ([22], p. 59) and a coral-cluster resource management system ([18], p. 94) linked to cyclical migration, innovative subsistence practices, population-wide cooperation, and fission–fusion political dynamics.

1.2.2. Sustainable Resource Use

Long-term persistence can pose challenges for those that inhabit small islands [27]. Island size, resource availability, rainfall, and the choices made by the population can impact the ability or desire to survive on an island or archipelago. The sustainability of a population is defined as their ability to extract, replenish, and maintain the material and social resources needed to persist and grow [56,57] as well as the ability to prepare for and mitigate the effects of unforeseen events. This often involves protecting resources from overuse and utilizing less detrimental forms of resource production and extraction [28,58]. Sustainability is best thought of as the result of complex social and ecological process involving resource use and production as well as human agency and the desire or ability to persist in a given location [56]. A sustainable socioecological system is not one that develops a synchronic state of sustainability; rather, it is one that adapts and changes over time in order to remain viable. Achieving sustainability on islands often requires landscape transformation and the careful management of terrestrial and marine resources (e.g., [59,60]). It involves the development of strategies to extend resource bases and buffer them against risk [28]. While many Polynesian peoples developed strategies for sustainability, there are many examples of island abandonment before European arrival [61,62,63]. Thus, it is important to understand the dynamic processes through which humans transformed and managed small islands to ensure sustainability and the possible reasons for the subsequent abandonment of certain islands and not others.

The East Polynesian founding groups were undoubtedly aware of the challenges posed by the occupation of new islands. They brought with them plants and animals that enabled the transformation of new environments. These often included four domestic or commensal animals, namely, the chicken (Gallus gallus), the dog (Canis familiaris), the pig (Sus scrofa), and the Pacific rat (Rattus exulans) [64], as well as numerous plants including breadfruit (Artocarpus altilis), pandanus (Pandanus spp.), swamp taro (Cyrtosperma chamissonis), and possibly varieties of coconut (Cocos nucifera) [65]. In many cases, settlement and species introduction caused the extinction of native fauna, particularly flightless and ground nesting birds [2,66].

The settlers of Polynesia likely shared social rules and institutions regarding the management of certain ecological resources. The institution of rahui in one form or another is prevalent throughout East Polynesia [67]. Rahui is a system of ritualized resource prohibition or protection, and is typically viewed as a way to preserve fragile resources or to develop a stock of resources for emergency use ([68], p. 118). Portions of the land and sea are placed under protection by a governing body. People can be punished for breaking the rahui; however, when want or need arises, such as before a feast or during a famine, a governing body (e.g., chief or council) may lift the rahui and allow access to resources ([69], p. 141). This allows for the growth and live storage of plants and animals. Rahui continues in use today [67], and is often encoded into law. In the past, as we will discuss, people on Manihiki and Rakahanga seem to have taken rahui practices one step further by placing an entire atoll under protection, only allowing access through the Tûmutu. This system has its roots in rahui institutions, and reflects cyclical patterns of aggregation and dispersal in other societies in Polynesia and beyond.

1.2.3. Aggregation, Dispersal, and Fission–Fusion Dynamics

Processes of aggregation and dispersal are common and manifest in a wide range of settings around the globe (e.g., [70,71,72,73]). Population aggregation can take many forms, ranging from the gathering of nomadic groups around a common resource to the coalescence of dispersed villages and culturally diverse groups into larger communities [28,74,75,76]. Aggregation generally refers to the physical movement of people out of a dispersed state of settlement and into a common location [77,78]. Dispersal alternately references the fissioning or fracturing of settlements into smaller independent communities ([79], p. 169). Aggregation and dispersal are defined based on ethnographic data and archaeologically visible changes in regional homogeneity, settlement organization, house forms, mortuary practices, and trade networks ([80], p. 393).

The aggregation and dispersal of human settlements are often correlated to a number of situation-dependent social and environmental factors occurring at time scales ranging from sub-annular to supra-millennial [80,81,82,83,84,85,86]. Aggregation practices can represent cultural flexibility ([76], p. 27) and tend to be associated with intensification in resource production [81,83] and social engagement [72,76]. Aggregation of this nature does not necessarily mean integration between the population segments of aggregated communities [76]. Rather, population centers may include a number of political factions drawn together by external threats [73,85], responses to environmental stress ([87], p. 97), or to take advantage of improvements in resource production [88]. This suggests inter-polity and intra-polity cooperative effort toward a common goal.

Cyclical patterns or oscillations of settlement aggregation and dispersal can be related to patterned cycles of political change [89]. Anderson ([89], p. 1) argues that certain political structures, namely chiefdoms, are prone to cycling or cyclical fluctuations in level of authority. These fluctuations often occur in concert with population movements (e.g., aggregation or dispersal). Cyclical patterns of aggregation, dispersal, and re-aggregation do not represent exact cycles of oscillation between preset forms, but rather patterns of aggregation and dispersal related to and embedded within changing historical trajectories ([80], p. 431). Diachronic patterns of aggregation and dispersal can define the long-term social and ecological strategies of a polity, as cyclical fission–fusion processes may serve to increase social cohesion and cooperation, alleviate social pressure, or address ecological and economic stress [90,91,92]. Finally, these processes provide opportunities for political restructuring, the development of new institutions, shifting patterns of hierarchical and heterarchical power structures, the invention of new technologies, and the use of untapped resources.

Fissioning and fusing are essentially cooperative acts, and require a level of community cohesion. The size of a polity and the ability of political leaders to ensure cooperation as well as group trust and cohesion may affect the ability of a group to act cooperatively [93]. Concomitant cyclical fission–fusion processes associated with dispersals and aggregations often represent successful cooperative sociopolitical strategies capable of mitigating environmental and social stress ([10], p. 544). The fission–fusion histories of Manihiki and Rakahanga provide a case study by which to evaluate the nature of such dynamics, and the unique environmental conditions provide insight into the varied ways that such traditions were employed within the context of collective action to address challenges.

1.2.4. The History of Manihiki and Rakahanga

European accounts of the atolls begin in AD 1606 with the arrival of the Spanish explorer Pedro Fernandez de Quiros. Quiros and his crew moored off the coast of Rakahanga and interacted with the islanders. Eventually the encounter turned violent when the sailors killed numerous islanders. Despite this violence, members of Quiros’ crew recorded information about the people and environment of the atoll [94,95,96]. In AD 1849, more than two centuries after this first encounter, missionaries arrived on Manihiki and Rakahanga. In AD 1852, the Reverend W.W. Gill arrived on the atolls. Gill had a strong penchant for recording the lifeways of the populations that he meant to convert to Christianity. His works include documentation and interpretation of Indigenous life-ways, descriptions of island environments, oral histories, and genealogical and demographic information. While Gill’s descriptions provide a wealth of information regarding the populations of Manihiki and Rakahanga as well as other islands, they are biased by his efforts to bring Christianity to the South Pacific. Following Gill, the New Zealand Maori ethnologist Te Rangi Hiroa, known as Sir Peter Buck, recorded life ways on Manihiki and Rakahanga in the 1920s. Hiroa was an accomplished scholar who conducted cultural studies across the Pacific, eventually becoming the Director of the Bishop Museum in Hawaii ([97], p. 21). Hiroa documented lifeways and traditions on Manihiki and Rakahanga by recording extant practices and oral histories. This includes narratives regarding the formation of the dual chiefdom, to which he assigned a date range by tracing genealogies. He documented the foods eaten by the population as well as the function of the Tûmutu in allowing for resource sustainability. Hiroa’s work on Manihiki and Rakahanga is an invaluable resource that holds great meaning to the people of the atolls today. It should be noted, however, that Hiroa only spent three weeks on Rakahanga and two days on Manihiki before moving on to Penrhyn Atoll. This brevity of time on the atolls undoubtedly limited his ability to record detailed information.

Oral histories suggest that Manihiki and Rakahanga were colonized from Rarotonga in the Southern Cook Islands. This indicates that the original political form on Manihiki and Rakahanga may have resembled those found in the Southern Cook Islands. Ethnohistorical documents detail that Rarotonga was governed by three chiefdoms, each of which was led by an Ariki, or hereditary chief, who controlled a district ([98], p. 6). These groups are said to have originated from different Vaka, canoes carrying settlers from different origins, and thus each faction claimed descent from a different ancestral line ([68], pp. 31–34). While the Arikiships were independent of each other, they came together as a whole and cooperated on ceremonial occasions ([98], p. 6). The island of Mangaia, 200 km south-east of Rarotonga, was home to a multi-factional organization structure where multiple chiefdoms existed on the same island; the Ariki whose chiefdom proved dominant in war reigned over the whole of the island and controlled the distribution of resources until the next period of conflict arose ([4], pp. 58–59). These examples demonstrate the commonality of social organization patterns involving multiple chiefs on a single island and tentative inter-polity cooperation. If the original inhabitants of Manihiki and Rakahanga arrived from the Southern Cook Islands, then it is likely that they had knowledge of the same ancestral institutions that gave rise to the chiefdoms of the Southern Cook Islands. Hiroa ([22], p. 43) suggests that this knowledge was brought with the first settlers, who maintained aspects of the political system from their homeland, though they eventually developed new political forms.

Oral traditions collected in the late 1800s suggest that the first inhabitant of Manihiki and Rakahanga, Huku, was met by a barren terrestrial landscape “scarcely above the face of the sea…” ([23], p. 128; [99], pp. 148–149). These oral traditions stand in contrast to reports from the early 1900s that document a verdant terrestrial landscape on Manihiki and Rakahanga ([22], pp. 83–84, 92–97). The difference between these descriptions suggests that Manihiki and Rakahanga were settled shortly after the islets emerged. An estimated population of 1200 individuals (~126 per km²) dwelled on Manihiki and Rakahanga when missionaries arrived in AD 1849 ([23], p. 127). Despite the atolls’ increased productivity, the resource base of each atoll, individually was not sufficient to support this dense population in the long term.

The oral traditions of Manihiki and Rakahanga trace modern ancestry to one family. They state that following his arrival, Huku planted the first coconuts on the atolls and returned to Rarotonga. Eventually, he sent his kin Toa and Tapairu to care for the atolls [22,99]. The entire population of Manihiki and Rakahanga is said to have descended from Toa and Tapairu. Toa, as the founding male of the islands, commanded the title of Ariki that he passed to his descendants. Hiroa ([22], p. 57) indicates that within a few generations Te Kainga became a village, with the families of two of Toa’s sons Matangaro and Hukutahu forming housing clusters. The title of Ariki was held by the head of the Hukutahu line ([22], p. 57). A boundary stone was erected to mark the spatial divisions between the two groups ([22], p. 59).

The founding of the first villages on Manihiki took place at an unknown time, and there is mystery surrounding their use. Hiroa provides insight into the timing of settlement on Manihiki, suggesting that a marae, or meeting place, was built on the islet of Tauhunu during the seventh generation following Toa ([22], p. 206). By the tenth generation formalized subgroupings had developed within the two lineages, leading to further separations in settlement. This process of sub-group fissioning or ramification is typical in Polynesian societies, and is often associated with the colonization of new territories ([46], p. 32). At this time, four named subgroups called matakeinanga lived on Te Kainga in separated housing clusters under a single Ariki. In the eleventh generation, the Ariki Tautape had sons from two wives. The eldest son of the first wife would normally have claim to the Ariki title, however, two of the four subgroupings placed their support behind the son of the second wife. A compromise was struck, and the single Ariki title divided into a dual Arikiship under the control of two chiefs, each of which held both divine and secular powers ([22], pp. 22, 57).

Hiroa ([22], pp. 4, 65–66) states that in the 1800s the entire population, all four subgroupings, and both Ariki lived on the islet of Te Kainga, Rakahanga (Figure 2). However, in times of stress the islanders implemented the Tûmutu [24,25]. In this mass migration institution, the entire population relocated to Manihiki, where they dispersed into multiple villages, each led by one Ariki ([22], pp. 65–66). Once the resource base of Rakahanga regenerated, they returned to Te Kainga [22,100,101]. For the Tûmutu to function well in reducing social and ecological stress would have required polity-wide cooperation and an expert knowledge of resource limitations. Yet, the details of its origin are unclear in the extant literature.

2. Materials and Methods

Archaeological investigations on Manihiki and Rakahanga began in the 1980s and consisted of excavations on three islets and selective archaeological surveys across the atolls [100,102,103]. This research determined that Rakahanga possessed a single nucleated village on Te Kainga. The other larger islets of Rakahanga appear to have been devoted to farming. In contrast, Manihiki contained multiple villages, each located on a large islet. Several of these were adjacent to small cultivation pits. The previous teams excavated at numerous sites on Te Kainga and Rakahanga (TEK and RAK-1) as well as Porea (MNH-9) and Hakamaru (MNH-7) Islets on Manihiki (Figure 2). While the teams provided numerous radiocarbon dates for the occupation of these villages, the nature and timing of the habitation of the atolls as well as the formation of local cultural institutions remained unclear.

2.1. Archaeological Survey

Our field team, consisting of the lead author, volunteer workers, and local field assistants, renewed archaeological efforts on Manihiki and Rakahanga in 2015 and 2017 to identify, excavate, and date habitation and cultivation sites on each atoll. As suggested in reports, oral traditions, and ethnohistoric narratives, the survey indicated that Rakahanga was likely to have had one nucleated settlement, while Manihiki had many dispersed villages. Informants on Rakahanga suggest that a centralized government oversaw the management of a large communal swamp taro field/coconut grove system on islets other than Te Kainga. Tapus (laws) were in place to ensure that the lands outside of Te Kainga were preserved for the cultivation of food ([22], p. 66). These pit fields and groves were separated by water from the residential islet of Te Kainga. Informants told of strict curfews enforced by the Ariki that required people living on Rakahanga to return to Te Kainga each night; they were then only able to return to the other islets during the day. The survey documented the presence of numerous fish traps on both atolls and a single large fishpond on Manihiki. While there is currently no way to determine the antiquity of these features, it is likely that past populations used these or similar structures.

Based on the results of the survey, we tested four locations, one on Manihiki (NG001) and three on Rakahanga (NM001, NV001, TK001), with shovel-excavated test pits (Figure 2). These locations were targeted to provide evidence of past villages (TK001, NV001, and NG001) and horticultural activity (NG001 and NM001) on the atolls. We positioned larger units on Ngake islet, Manihiki (NG001) and Te Kainga islet, Rakahanga (TK001) in locations where the test-unit survey discovered well-stratified deposits [104]. The purpose of these excavations was: (1) to collect samples for radiocarbon dating; and (2) to evaluate the stratigraphic sequences. Each of these different lines of evidence, when coupled with oral histories and ethnohistoric documents, allows for the evaluation of landscape change and fission–fusion dynamics on Manihiki and Rakahanga.

2.2. Stratigraphy and Land Use Histories on Manihiki and Rakahanga

The NG001 site on Ngake Islet possessed dense surface features, including four coral-edged courts, a well, and a nearby swamp taro cultivation pit [104]. The stratigraphic sequence of NG001 mirrors those identified by Yamaguchi [103] on other islets of Manihiki and possesses relatively simple stratigraphic layering. Artifact density was low, with only two units producing artifacts; these included fishhooks and possible adornments [104]. The densest cultural deposits included an earth oven and midden deposits containing faunal materials indicating site use, however, there was no evidence of long-term or permanent habitation.

The TK001 site on Rakahanga produced numerous artifacts and extremely dense and discontinuous stratigraphic layering [104]. This is likely a result of deposition processes caused by changes in sea level, wash-over events, and differential cultural practices, including frequent periods of abandonment or lessened use. The TK001 excavation yielded a much higher density of faunal materials and artifacts than the NG001 excavations, suggesting more intense/frequent occupation [104].

2.3. Radiometric Dating of Archaeological Sites on Manihiki and Rakahanga

The previous researchers working on Manihiki and Rakahanga produced a number of radiocarbon dates that provide temporal anchoring for past events on Manihiki and Rakahanga; however, these alone do not provide a chronological narrative of site use. We compiled all available archaeological data and radiocarbon dates from the two teams that previously worked on the atolls [100,102,103,105]. We first ensured that chronometric hygiene protocols were in place in order to eliminate unreliable and imprecise dates. In doing this, we removed all dates from marine sources and those that may have required marine calibration, such as terrestrial mammals that may have had a mixed diet. We removed all dates that were not clearly associated with cultural contexts in stratigraphic excavations or were noted by the excavators to not be cultural in nature. We removed all dates with error ranges of ±100 years or more, as this uncertainty obscures the short time spans that we were interested in. Finally, we removed all dates from mixed contexts where a sample could not be reliably associated with a single context. While not as stringent as other protocols used in east Polynesia (e.g., [44]), these protocols were meant to reduce error while making the most of the existing data.

Thirteen new AMS radiocarbon dates on botanical samples from the 2017 excavations were obtained from the Center for Applied Isotope Studies at the University of Georgia following standard pretreatment and AMS protocols reported elsewhere [106]. We chose samples from disparate contexts from NG001 and TK001 (

Table 1. New and existing AMS dates from Manihiki and Rakahanga Archaeological Sites.

Laboratory Number	Unit/Layer	Dated Material	δ13C ‰	14C Age BP	Calibrated Date Range AD 68.3% **	Calibrated Date Range AD 95.4% **
	MNH-7
PLD-5831	*	UID Charcoal	-	605 ± 25	1320–1420	1320–1430
N-6149	4/I-4	UID Charcoal	-	400 ± 75	1450–1630	1420–1670
N-6145	3/I-4	UID Charcoal	-	380 ± 70	1460–1630	1430–1800
N-6148	4/I-4	UID Charcoal	-	330 ± 70	1490–1670	1450–1810
N-6147	4/I-4	UID Charcoal	-	400 ± 70	1450–1630	1430–1660
N-6146	4/I-3	UID Charcoal	-	400 ± 70	1450–1630	1430–1660
	MNH-9
N-5864	A/3	UID Charcoal	-	330 ± 75	1490–1670	1440–1810
N-5866	A/3	UID Charcoal	-	250 ± 70	1510–1940	1500–Modern
N-5867	C/2	UID Charcoal	-	200 ± 70	1660–Modern	1630–Modern
	NG001
UGAMS-35654	B-1/III	Pandanus spp. drupe	−23.77	320 ± 20	1510–1650	1500–1660
UGAMS-35655	C-1/III	Pandanus spp. drupe	−27.10	310 ± 20	1510–1660	1500–1670
UGAMS-40094	C-1/I-II	Cocos nucifera endocarp	−25.1	140 ± 25	1690–1950	1690–Modern
UGAMS-38551	C-1/II	Pandanus spp. drupe	−24.75	100 ± 20	1820–1930	1700–1930
	RAK-1
Beta-177249	II/IVb	Mixed short-lived botanicals	−25.00	290 ± 60	1500–1800	1460–1940
Beta-176304	III/IVb	Pemphis acidula	−25.00	360 ± 60	1490–1640	1450–1800
Beta-177248	III/IVb	Mixed short-lived botanicals	−26.00	380 ± 40	1480–1630	1450–1640
Beta-177250	III/IVb	Mixed short-lived botanicals	−25.00	330 ± 70	1490–1670	1450–1810
	TEK
PLD-3918	*	UID Charcoal	-	830 ± 25	1220–1270	1210–1290
	TK001
UGAMS-35651	A-1/V.1	Cocos nucifera endocarp	−24.51	100 ± 20	1820–1930	1700–1930
UGAMS-35650	A-1/III	Cocos nucifera endocarp	−24.91	100 ± 20	1820–1930	1700–1930
UGAMS-35652	A-1/V.2	Cocos nucifera endocarp	−25.02	100 ± 20	1820–1930	1700–1930
UGAMS-35649	A-1/XIII	Cocos nucifera endocarp	−24.54	210 ± 20	1670–1810	1660–1810
UGAMS-34018	A-1/XIII	Cocos nucifera endocarp	−24.61	250 ± 20	1650–1800	1640–1810
UGAMS-38552	A-1/IX-X	Pandanus spp. drupe	−23.01	110 ± 20	1710–1930	1690–1950
UGAMS-35653	A-1/IX	Cocos nucifera endocarp	−25.52	180 ± 20	1670-Modern	1670–Modern
UGAMS-36970	TU5/VIII-IX	cf. Pandanus spp. drupe	−25.20	310 ± 20	1510–1660	1500–1670
UGAMS-40095	TU8/V-VII	Cocos nucifera endocarp	−23.60	660 ± 20	1310–1400	1290–1400

* The two PLD dates are re-dates of the oldest context on Manihiki and Rakahanga, therefore, their provenience is presumed to be the deepest strata of MNH-7 and TEK and they are assumed to equate to a human presence on the atolls. ** All botanical dates were calibrated in OxCal v4.4 using the SHCal 20 calibration curve and rounded to the nearest 10 [108,109].

) in order to enable a holistic view of the complex stratigraphy of the atolls. We selected short-lived botanical samples, including charred coconut (Cocos nucifera) endocarp and charred Pandanus spp. drupes, for dating. These plant parts represent growth of one year or less, and therefore negate possible inbuilt age effects. We selected and documented samples following best practice standards as outlined by Allen and Huebert [107]. This included using only short-lived botanicals with a strong association with the particular archaeological event in question.

3. Modelling Details and Results

Bayesian chronological models are regularly employed by archaeologists to provide relative constraint to radiocarbon approximations by accounting for a-priori knowledge. This most often includes grouping dates into phases based on relative stratigraphy and placing those phases into ordered sequences [108]. All of our models were created and implemented using OxCal 4.4 [108]. All model code and modelled data are available in Appendix A (Figure A1 and Figure A2;

Table A1. Table output for the Manihiki Model.

Name	Unmodelled (BC/AD)						Modelled (BC/AD)						Indices Amodel 111.2 Aoverall 109.9
	from	to	%	from	to	%	from	to	%	from	to	%	Acomb	A	C
Curve SHCal20
Sequence
Boundary Start 1							1355	1415	68.27	1280	1435	95.45			98.1
R_Date PLD5831	1325	1415	68.27	1320	1430	95.45	1390	1420	68.27	1325	1435	95.45		102.6	99.6
Sequence MNH-7
First							1440	1490	68.27	1410	1530	95.45			99.6
R_Date N6146	1455	1625	68.27	1430	1660	95.45	1440	1490	68.27	1410	1530	95.45		93.6	99.6
Phase MNH-7 Phase 2
R_Date N6149	1455	1630	68.27	1420	1670	95.45	1470	1600	68.27	1455	1630	95.45		110.1	99.6
R_Date N6145	1460	1630	68.27	1435	1795	95.45	1475	1585	68.27	1460	1630	95.45		110.4	99.7
R_Date N6148	1495	1665	68.27	1450	1805	95.45	1495	1590	68.27	1465	1635	95.45		115.1	99.7
R_Date N6147	1455	1625	68.27	1430	1660	95.45	1470	1580	68.27	1460	1625	95.45		108.2	99.6
Last							1555	1635	68.27	1505	1640	95.45			99.6
Phase Manihikian Sites
Sequence NG001
First							1620	1650	68.27	1520	1655	95.45			99.4
Phase NG001 Phase 1
R_Date UGA35654	1510	1650	68.27	1505	1655	95.45	1625	1650	68.27	1530	1665	95.45		117.1	99.4
R_Date UGA35655	1515	1655	68.27	1505	1665	95.45	1625	1655	68.27	1530	1670	95.45		125.5	99.2
Phase NG001 Phase 2
R_Date UGA40094	1695	1945	68.27	1695	…	95.45	1695	1840	68.27	1685	1860	95.45		97.6	99.4
R_Date UGA38551	1820	1925	68.27	1700	1930	95.45	1710	1835	68.27	1695	1860	95.45		53.8	98.6
Last							1810	1840	68.27	1705	1860	95.45			99.3
Sequence MNH-9
First							1575	1675	68.27	1520	1775	95.45			99.5
Phase MNH-9 Phase 1
R_Date N5864	1495	1665	68.27	1445	1810	95.45	1570	1680	68.27	1535	1805	95.45		99	99.2
R_Date N5866	1515	1940	68.27	1505	…	95.45	1625	1775	68.27	1555	1815	95.45		116.7	99.4
R_Date N5867	1665	…	68.27	1630	…	95.45	1750	1850	68.27	1675	1860	95.45		109.2	99.7
Last							1750	1850	68.27	1675	1860	95.45			99.7
C_Date Abandoment	1835	1865	68.27	1825	1870	95.45	1835	1860	68.27	1825	1870	95.45		100.5	99.4
Boundary End							1840	1895	68.27	1830	1965	95.45			96.6

and

Table A2. Table output for the Rakahanga Model.

Name	Unmodelled (BC/AD)						Modelled (BC/AD)						Indices Amodel 67.1 Aoverall 66.2
	from	to	%	from	to	%	from	to	%	from	to	%	Acomb	A	C
Curve SHCal20
Sequence
Boundary Start							1200	1270	68.27	1115	1280	95.45			97
R_Date PLD3918	1225	1270	68.27	1215	1285	95.45	1240	1280	68.27	1220	1285	95.45		99.3	99.4
R_Date UGA40095	1315	1395	68.27	1295	1400	95.45	1315	1395	68.27	1295	1400	95.45		98.7	99.5
Phase Rakahangan Sites
Phase RAK-1 Phase 1
First							1455	1530	68.27	1445	1580	95.45			99.7
R_Date BETA177248	1480	1630	68.27	1455	1635	95.45	1480	1630	68.27	1455	1635	95.45		99.9	99.4
R_Date BETA176304	1495	1640	68.27	1450	1795	95.45	1500	1640	68.27	1455	1655	95.45		103.3	98.9
R_Date BETA177205	1495	1665	68.27	1450	1805	95.45	1500	1655	68.27	1455	1670	95.45		111.2	99.4
R_Date BETA177249	1505	1800	68.27	1460	1940	95.45	1510	1665	68.27	1460	1750	95.45		108.7	98.9
Last							1585	1670	68.27	1535	1755	95.45			99.5
R_Date UGA36970	1515	1655	68.27	1505	1665	95.45	1515	1655	68.27	1505	1665	95.45		99.6	99.5
Sequence TK001 A-1
First							1645	1755	68.27	1645	1790	95.45			99.5
Phase TK001 A-1 Phase 1
R_Date UGA34018	1650	1800	68.27	1645	1800	95.45	1650	1795	68.27	1645	1800	95.45		99.7	99.4
R_Date UGA35649	1670	1805	68.27	1660	1810	95.45	1665	1780	68.27	1660	1810	95.45		99.7	99.4
Phase TK001 A-1 Phase 2
R_Date UGA35653	1675	…	68.27	1670	…	95.45	1680	1820	68.27	1675	1825	95.45		102.6	99.6
R_Date UGA38552	1710	1925	68.27	1695	1945	95.45	1810	1830	68.27	1695	1830	95.45		65.4	99.1
Phase TK001 A-1 Phase 3
R_Date UGA35652	1820	1925	68.27	1700	1930	95.45	1815	1835	68.27	1810	1860	95.45		63.1	99.8
R_Date UGA35650	1820	1925	68.27	1700	1930	95.45	1815	1835	68.27	1810	1860	95.45		62.9	99.7
R_Date UGA35651	1820	1925	68.27	1700	1930	95.45	1815	1835	68.27	1810	1860	95.45		63	99.7
Last							1820	1840	68.27	1820	1865	95.45			99.8
C_Date Abandoment	1835	1865	68.27	1825	1870	95.45	1835	1860	68.27	1830	1875	95.45		99.6	99.8
Boundary End							1840	1900	68.27	1830	1985	95.45			96.5

). As the excavations on Manihiki had relative continuity in the stratigraphic sequences at each site, we modelled each as a sequence within an overlapping model. Due to the complex and discontinuous stratigraphy on Rakahanga, we modelled each unit separately as a sequence based on the relative stratigraphy [104]. Dates from the same context were modelled as phases. We calibrated dates using the SHCal 20 calibration curve [109] as it is argued that the new SHCal curve created for the southern hemisphere is more representative of conditions found in tropical Oceania than previous curves [109].

3.1. Manihiki

We placed all the dates from Manihiki that passed chronometric hygiene into an overlapping Bayesian model. Dates derived from the same layers or associated zones were modelled as a sequence that was based on the nature of the stratigraphic relationships among the different units and levels; the structure of the model can be observed in the probability distributions. A recent re-dating of charcoal (PLD-5831) from Manihiki, likely from MNH-7 contexts, has been reported by Yamaguchi and colleagues [105] and provides evidence of a human presence on the atoll at 605 ± 25 BP. The potential for significant old wood effects, however, suggests that these dates may be slightly older than the actual burning events. We included this date in the model to account for possible early landscape modification and other cultural activity on Manihiki. We set a terminus anti quim of AD 1850 ± 10 to provide an added constraint, as all the sites dated are thought to have been abandoned by the time missionaries arrived. The components and results of the modelled dates from each Manihikian site are described below. The results of our modeling of the dates (Figure 3) indicate agreement and returned an Amodel (111.4) and Aoverall (110.4) for the model demonstrating statistical significance and exceeding the 60-threshold established for Bayesian analysis [108,110].

3.1.1. MNH-7

The MNH-7 site (Figure 2) identified and excavated by Yamaguchi, is located on Manahiki’s Hakamaru Islet and was described as a “dwelling space” ([103], p. 36). The stratigraphy of the site consists of three main strata (I-III) divided into sublayers and a bedrock layer (IV). Yamaguchi dated each of the four layers using unidentified charcoal samples, though cultural material was only indicated in the top two strata. The date returned for the upper portion of Layer I (I-1), which was nearest to the surface, indicated that the sample was modern. Samples from the lowest portion of the stratum (I-3 and I-4), including an associated earth oven feature, returned dates of 330 ± 70 BP (N-6148), 380 ± 70 BP (N-6145), 400 ± 70 BP (N-6146), 400 ± 70 BP (N-6147), and 400 ± 75 BP (N-6149). The subdivisions of Layer II returned a wide array of dates ranging from 430 ± 70 BP to 730 ± 75 BP. However, there is no mention of cultural materials in Layer II other than a pit oven associated with Layer II-1. We modelled Layer I as a sequence with the four dates from layer I-4 as a phase within the sequence. The resultant model suggests that Layer I was in use by cal. AD 1410–1530 (95% probability). While the PLD-5831 date suggests an earlier presence on Manihiki, it is likely that intensive occupational use of MNH-7 began after AD 1400.

3.1.2. MNH-9 (Te Marae)

This site on Manihiki’s Porea Islet, excavated by Yamaguchi [103], is named Te Marae based on the traditional name for the land. While this could indicate a ceremonial purpose, the structure of the site appears to represent residential use. The site is 100 m inland and has a number of features, including wells, graves, surface structures, and adjacent puraka pits. Yamaguchi’s team excavated five units and obtained four radiocarbon dates, three on unidentified wood charcoal and one on a Tridacna sp. shell. Excavations of Unit A at MNH-9 revealed well defined stratigraphy with four layers. No cultural materials were reported in Layer 1. Layers 2 and 3 contained pit and combustion features as well as the coral gravel pavement called kirikiri. Two unidentified charcoal samples were reported from materials in Layer 3, which dated to 330 ± 75 BP (N-5864) and 250 ± 70 BP (N-5866). A charcoal sample from a pit oven feature associated with Layer 2 in Unit C was dated to 200 ± 70 BP (N-5867). Based on the relative position of the layers and features we modelled the dates as a sequence, with the Layer 3 dates in a phase. The model suggests a first occupation date of cal. AD 1520–1775 (95% probability).

3.1.3. NG001

Our team excavated the NG001 site at the south end of Manahiki’s Ngake Islet in 2017 (see [104]). We obtained radiocarbon dates for organic materials from units B-1 and C-1, which possessed similar stratigraphy. While neither unit produced cultural artifacts, both possessed a dark layer associated with intensive human activity roughly 40 cm beneath the current ground surface. In unit B-1, this layer (Layer III) consisted of a 15 cm deep earth oven feature that contained heat fractured coral, charcoal, and burnt shell. A charred Pandanus spp. drupe from B-1 Layer III (UGAMS-35654) returned an age of 320 ± 20 BP. In unit C-1, Layer III was a midden deposit which contained dark sediments and a density of faunal materials greater than anywhere else on NG001. A charred pandanus drupe (UGAMS-35655) returned a date of 310 ± 20 BP. This indicates that the B-1 earth oven and C-1 midden were contemporaneous. We obtained additional dates on samples from Layers I and II of unit C-1 to understand site use following the deposition of the midden materials. A pandanus drupe and a coconut endocarp fragment returned dates of (UGAMS-40094) 140 ± 20 BP for Layer II and (UGAMS 38551) 110 ± 20 BP for the I/II interface. The four botanical dates were modelled as two phases, with the B-1 earth oven and C-1 midden as a single phase and the C-1 layer I and II dates as a single phase; we modelled these sequentially. The resulting model (Figure 3) suggests that the most intensive occupation of NG001 (Layer III) began at cal. AD 1520–1665 (95% probability), and that less intensive land use (Layers I and II) appears to have followed.

3.2. Rakahanga

All the dates from Rakahanga came from materials excavated on Te Kainga islet. We placed all of the dates that passed chronometric hygiene into an overlapping Bayesian model sequence that was based on the nature of the stratigraphic relationships among the different units and levels; the structure of the model can be observed in the probability distributions (Figure 4). A recent re-dating of charcoal from TEK contexts on Rakahanga reported by Yamaguchi and colleagues [105] may provide evidence for a human presence on the atoll at 830 ± 25 BP (PLD-3916). Including this early date in the model suggests a human occupation of Rakahanga between cal. AD 1115–1280 (95% probability), however, old wood effects suggest that these dates may be slightly older than the actual burning events. We included this date in the model to account for possible early landscape modification and cultural activity on Rakahanga. We used a terminus anti quim of AD 1850 ± 10 to provide an added constraint, as the village on Te Kainga was abandoned shortly after missionary arrival in AD 1849 [22]. The components and results of the modelled dates from Te Kainga are described below. Our modeling of the dates indicate agreement, and returned an Amodel (67.0) and Aoverall (66.1) exceeding the 60-threshold established for Bayesian analysis [108,110].

3.2.1. TEK (Mua Marae)

This site is located at the center of Te Kainga islet, Rakahanga. The TEK site includes the 120 cm tall stone monument purported to have divided the different habitation areas on the islet. Chikimori and Yamaguchi [102,103] excavated the site and suggested that it was Mua Marae, a marae (meeting place) used by all four of the matakeinanga ([102], p. 21). Their team excavated four units and identified numerous pit features and stone alignments. While stratigraphic profiles and layer descriptions for the excavations are not available, they reported numerous radiocarbon dates on charcoal and shell associated with stratigraphic layers. It is not clear which dates are associated with cultural deposits, however, and many of the reported dates failed our chronometric hygiene protocols due to error ranges in excess of 100 years. Yamaguchi and colleagues [105] published a new date of 830 ± 25 BP (PLD-3916) for the first occupation of the site, as discussed above; however, layer and level designations were not reported for this sample.

3.2.2. RAK-1

This site is on Te Kainga, to the west of the TEK site. Di Piazza [100] excavated RAK-1 in 2002. These excavations produced large amounts of material culture and faunal remains, and contained combustion features. The stratigraphy included five main strata. Di Piazza designated Layer IVb a habitation layer, and obtained four dates from this layer on specimens that were identified as short-lived botanicals from oven features and an associated charcoal lens ([100], pp. 75–77). These dates include (Beta-177249) 290 ± 60 BP, (Beta-176304) 360 ± 60 BP, (Beta-177248) 380 ± 40 BP, and (Beta-177250) 330 ± 70 BP. All these dates originated from the same cultural layer; thus, we modelled them as a single-phase sequence. The model suggested a date range of cal. AD 1445–1775 (95% probability) for this habitation layer.

3.2.3. TK001

We dated botanical remains from three units of TK001-TU5, TU8, and Unit A-1 using AMS. We chose these units based on clear evidence of cultural activity and the presence of short-lived charred botanicals. The stratigraphy of TK001 is complex and discontinuous; thus, we treated each unit as a separate sequence. A single coconut endocarp date from layer V/VI of TU8 returned the oldest dates at 660 ± 25 BP (UGAMS-40095). A single date on a pandanus drupe from layer VIII of TU5 dated to 310 ± 20 BP (UGAMS-36970). We modelled each of these dates as a sequence. Unit A-1 produced charcoal throughout the complex stratigraphic sequence. Three coconut endocarp dates from layers III and V returned ranges of 100 ± 20 BP (UGAMS-35650, UGAMS-35651, UGAMS-35652). A coconut endocarp sample from Layer IX returned a date of 180 ± 20 BP (UGAMS-35653) and a pandanus drupe sample from Layer IX/X returned a date of 110 ± 20 BP (UGAMS-38552). Two additional coconut endocarp fragments from Layer XIII were dated to 210 ± 20 BP (UGAMS-35649) and 250 ± 20 BP (UGAMS-34018). Bayesian modelling indicates that the center portion of Te Kainga (TU5) was inhabited before the coastal margins, with TU8 having a modelled age range of cal. AD 1295–1400 (95% probability). For unit A-1, the model returned an age range of cal. AD 1645–1865 (95% probability). As the oldest dates for unit A-1 were derived from materials collected at the terminal depths of the unit, where coral bedrock was reached, this suggests that the paleoshoreline of the islet was much further inland in the past.

3.3. Genealogical Chronology of Manihiki and Rakahanga

To anchor events discussed in the oral histories from Manihiki and Rakahanga with archaeologically visible phenomena, we turned to genealogical dating methods. By counting the number of generations and assigning time spans to each, genealogical dating allows for temporal estimations of events in a people’s past, such as the founding of a new settlement. The use of chiefly genealogies to temporally place events in the past is well known in Polynesian cultures, where knowledge-keeping priests mentally recorded chiefly lineages ([1], p. 212; [22], p. 20). According to Hiroa ([22], p. 20), however, the founders of Manihiki and Rakahanga, Toa and Tapairu, brought no priests with them to record this knowledge, and estimation of generations and lines of descent were kept by various different individuals; in 1929, local knowledge keepers recorded these lineages as best they could. Hiroa [22] relied on the genealogy of a man named Kairenga, who traced direct lineage to the founder Toa, to estimate the number of generations separating the twentieth century population from the founding population of Manihiki and Rakahanga. He suggested that 22 generations separated Toa from the extant population in the 1920s ([22], p. 22). Another means of temporally situating events in the past is using genealogy is to trace parallel lineages of known figures from Rarotonga with the founders of Manihiki and Rakahanga. Gill [99] proposed a tie between Tapairu and another figure in the oral traditions, Hiro. He suggests that Hiro is the grandfather of Tapairu, and that Hiro was a voyager from Havaiki, the legendary Polynesian homeland. Hiro was of the same generation as Tangiia of Rarotonga, who is said to have lived 26 generations before 1900 ([99], pp. 149–151). The documentation of these generational counts combined with Hiroa’s [22] assertion that the dual Arikiship began in the twelfth generation provided the first recorded estimations of date ranges for the settlement of the atolls and the formation of the novel political system.

The time span assigned to each generation can vary based on the average age that an individual would produce their successor. In the case of Manihiki and Rakahanga, this was often the first son, as primogeniture of the male line often determined succession ([22], p. 43). Scholars working to date events in other parts of East Polynesia have used generational spans ranging from 20 to 30 years with a standard deviation of two years ([1], p. 213), and this number can be further refined through knowledge of average age of marriage in chiefly lines. However, in the case of Manihiki and Rakahanga the recorded genealogies do not necessarily trace chiefly descent, and it is difficult to make an informed estimation of generational span beyond the 25 years proposed by both Hiroa [22] and Gill [99]. Using a standard deviation of two years and a generational span of 25 years, Hiroa’s methods produce a date range for the generation of Toa and Tapairu of AD 1336–1424, while Gill’s methods result in a date range of AD 1198–1302. The dual Arikiship begins in generation twelve of Hiroa’s chronology (with Toa and Tapairu as generation one), between about AD 1650–1690, while Gill’s methods do not provide information on events following Tapairu’s departure from Rarotonga and therefore cannot be used to estimate events that occurred after the couple reached Rakahanga. In all, the genealogical dates align well with our radiocarbon-based chronological models and provide rough timeframes for key events in the history of the atolls. Together, these disparate dating methods can inform the description of general periods of human activity on the atolls.

4. Discussion: Social and Ecological Change on Manihiki and Rakahanga

The modelled radiocarbon dates combined with the data derived from oral histories, historic documents, and genealogical records form the foundation of a deep history of the human colonization and occupation of Manihiki and Rakahanga. The results allow for the creation of a site sequence consisting of four phases (I through IV) separated by shifts in habitation patterns (

Table 2. A summary of the four phases of habitation on Manihiki and Rakahanga.

Phase	Description	Approximate Time Frame	Habitation Pattern	Major Events
I	Arrival Period	AD 1200–1400	Nucleated settlement on Te Kainga, Rakahanga	First occupation; initial landscape transformation
II	Expansion/Dispersal Period	AD 1400–1650	Dispersed settlement on Manihiki; continued occupation on Te Kainga	Founding of multiple villages on Manihiki; “The Wave” ca. AD 1600; European contact in AD 1606
III	Dual-Ariki Period	AD 1650–1849	Migratory, with periods of aggregation and dispersal via the Tûmutu	Renewed aggregation on Te Kainga; appointment of the second Ariki; beginning of the Tûmutu cyclical migrations
IV	Three Village Period	Post-AD 1849	Dispersed settlement in three villages (two on Manihiki and one on Rakahanga)	Missionary Arrival; Abandonment of Te Kainga; cessation of Tûmutu; establishment of three permanent villages

). Taken as a whole, this sequence shows how shifting patterns of settlement, possible migration cycles, landscape use, and population-wide cooperation are articulated within a larger socioecological system. The integration of these unique institutions ultimately allowed the inhabitants of Manihiki and Rakahanga to prosper across generations on these small atolls.

4.1. Phase I: AD 1200–1400—Arrival Period

By as early as AD 600, sea level lowering began to permanently expose previously submerged reef flats in East Polynesia [49]. Manihiki and Rakahanga became habitable over the following centuries. The Southern Cook Islands were likely inhabited by AD 900 [42]. Following AD 1150, wetter conditions may have made the habitation of small and remote islands more favorable, and much of East Polynesia was colonized by AD 1300 [42]. We believe Polynesian voyagers first reached Manihiki and Rakahanga between AD 1200 and 1300, initiating the Arrival Period (Phase I) of the Manihiki and Rakahanga sequence. We currently have no evidence suggesting frequent external contacts following arrival. Oral histories suggest that the land was barren until the first inhabitants planted coconuts and began ecological transformations ([99], p. 7). Genealogical dating suggests this occurred between AD 1198 and 1424 [22,99]. Both the AMS 14C dates from our recent investigations and those of Yamaguchi and colleagues [103,105] suggest a modelled starting boundary of cal. AD 1115–1280 (95% probability). The earliest dates from Yamaguchi’s team were derived from unidentified charcoal, and thus the latter end of this range is most probable. Both teams identified the center of the islet of Te Kainga on Rakahanga as the earliest point of settlement. Yamaguchi’s team dated carbonized plant remains from Manihiki that fell into the later portion of this range, indicating that the islets of Manihiki may have come into use around cal. AD 1280–1435 (95% probability). Based on our current evidence, it is likely that during this Arrival Period, the first inhabitants of Manihiki and Rakahanga lived in a nucleated village on Te Kainga just as oral histories suggest. Anthropogenic ecological change during the Arrival Period likely included the planting of agroforests in the form of coconut and pandanus trees, as well as the early excavation of puraka pits. While this terraforming of the landscape is discussed in the oral histories [99], at this time we have no archaeological evidence of this process other than the presence of coconut materials in the deposits of this period. The earliest settlers introduced the domestic dog and the Pacific rat at this time [104]. The initially small human population of Manihiki and Rakahanga was likely able to rely on marine resources as well as imported plants and animals.

4.2. Phase II: AD 1400 to 1650—Expansion/Dispersal Period

For the second phase of the cultural sequence, we define an Expansion/Dispersal Period from AD 1400 to 1650. In this time interval, evidence of intensive habitation appears on numerous Manihikian islets [102,103]. At least three Manihikian habitation areas (MNH-7, MNH-9, NG001) were in use during this period (Table 1; Figure 3). Currently there is no archaeological evidence to indicate whether the extant villages of Tukao and Tauhunu were occupied at this time. However, Hiroa ([22], pp. 48–51, 206) noted that the people built a marae at Tauhunu in the seventh generation (between AD 1515–1575 by genealogical estimations) and that Tauhunu was occupied by the twelfth generation (AD 1650–1690 by genealogical estimations). The village at Te Kainga appears to have remained in use during Phase II based on the dates reported by Di Piazza [100]. It seems that this was a time of expansion as well as of sociopolitical and socioecological change, as the people inhabited new environments and founded new villages.

The atolls’ ecologies would have likely experienced anthropogenic change as the human population increased and spread. Each Manihikian village resembles a microcosm of the Rakahangan village and its associated landscape. The villages on Manihiki contain living areas and swamp taro pits. These puraka pits were likely dug when the villages were first founded. Accessing puraka pit antiquity is typically carried out by dating charred botanicals associated with buried A-horizons in the spoil piles or berms surrounding the pits [33]. While we sampled the berms of puraka pits at NG001 and NM001, we were unable to identify buried A-horizons with datable materials. Therefore, we have not yet established direct dates of puraka pit excavation. If these features were established on Manihiki at this time, their small size compared to the pit field networks on Rakahanga suggests short-term occupation periods, small population sizes, or both.

Political change on the atolls likely involved semi-permanent fissioning processes, including the founding and management of dispersed Manihikian villages. This pattern of habitation contrasts with the nucleated settlement on Rakahanga and suggests socioecological flexibility. Cases of locational and political fissioning can be related to efforts to reduce social and ecological stresses [90,91]. While the exact causes of this expansion are unknown, it is likely related to population in-crease on Rakahanga and need for an expanded resource base. It is probable that the growing population on Rakahanga looked to establish territory on Manihiki on a sporadic or semi-permeant basis. Alkire ([18], p. 97) estimated that the population of Rakahanga would have needed to reach 600–700 individuals before the limits of Rakahanga’s resources necessitated such an expansion. Hiroa [22] suggested that the Matangaro and Hukutahu lineage groups had created their own housing clusters on Te Kainga prior to habitation on Manihiki. By the end of this period, in the tenth generation, there were four subgroupings (matakeinanga) of housing clusters on Te Kainga. This may indicate that formalized subgroupings of families were becoming established during the time of the first intensive occupations of Manihiki and that these groups preferred to live in close proximity. Fissioning and establishing individual villages may have bolstered the autonomy of the lineage-based subgroups. It is possible that this expansion was related to building social pressure between descent groups that dwelled in close proximity on Te Kainga. As with the ecological setting, the sociopolitical structures of the Manihikian villages were likely smaller-scale versions of that found on Te Kainga.

The end of this period coincides with two potentially influential external forces acting upon the population of Manihiki and Rakahanga: environmental upheaval and European contact. Around AD 1600 by genealogical estimations, the Northern Cook Islands are said to have been impacted by a large wave that killed most of the people on the island of Pukapuka, 525 km west of Manihiki and Rakahanga [20,55]. The type of wave whether it was a cyclonic storm surge or a tsunami is unknown, as are its impacts, if any, on Manihiki and Rakahanga. However, deep deposits of light-colored sand appear in the stratigraphy of the NG001 site at this time, marking the termination of the most intensive period of occupation on Manihiki [104]. This may indicate that, as with Pukapuka, Manihiki was affected by this event. If this was the case, environmental upheaval may have influenced the aggregation seen in the following period.

The first European contact with Manihiki or Rakahanga occurred in AD 1606 when the Spanish explorer Pedro Fernandez de Quiros appears to have visited Rakahanga ([96], pp. 209–217; [112], p. 12). The two ships of the Quiros expedition anchored off the reef of Rakahanga and had numerous encounters with the islanders. Sailors’ accounts describe the village and its large voyaging canoes. The presence of these vaka may suggest regular travel between Manihiki and Rakahanga, and potentially travel to other islands or island groups. The accounts discuss sailors making landfall and killing numerous individuals. Accounts estimate that there were 500 people on Rakahanga at the time ([112], p. 12). This is less than half the size of the population suggested by Gill in AD 1852 ([23], p. 127), and suggests that either the population had yet to reach higher numbers, that the entire population of the atolls was not present at the time of Quiros’ arrival, or that the reports are inaccurate; the second scenario may have been the case if a portion of the population was on Manihiki at the time. While the names that the Spaniards called Rakahanga by varied from “Peregrina” (Pilgrim Island) to “Gente Hermosa” (Island of Beautiful People), the name given by Luis Vaes de Torres may characterize the traumatic experience that this meeting really was, as he named Rakahanga “La Matanza” (The Slaughter) ([112], p. 12). This encounter may have had lasting repercussions, as oral histories involving the formation of the dual Arikiship in the following period state that one of the two major subgroupings, the Matangaro, were affected by a sickness in the twelfth generation ([22], pp. 48–51). At this time, there is no way of knowing how severe this sickness was, when it began, or whether it was of European origin; however, it was clearly closely tied to the creation of the dual Arikiship in the following period.

4.3. Phase III: AD 1650 to 1849—Dual-Ariki Period

If Phase II was a period of dispersal, culture contact, and environmental upheaval, Phase III was a time of renewed consolidation. The Dual-Ariki Period lasted from roughly AD 1650 to 1849 and was the final period of occupation on Te Kainga. During this period, following the abandonment of dense occupation at the NG001 site we see less use of Manihikian sites and more expansion of the village at Te Kainga. Hiroa suggests that the land area of Te Kainga had increased by this time and that the population moved into these areas which had “grown up” ([22], p. 58). This expansion can be seen archaeologically in the cultural materials found in the TK001 A-1 deposits, which appear at cal. AD 1645–1790 (95% probability) and are found in basal deposits near the lagoon shore of the islet. This suggests continued sea-level fall. While it is not possible to know the full circumstance of this aggregation, we can speculate that the external threats encountered in the previous period may have encouraged population consolidation and intra-group cooperation. Oral traditions suggest that in the twelfth generation, roughly AD 1650–1690, a problematic case of succession occurred when each of the main lineages stood behind one of two half-brothers, Temu-matua and Tianewa-matua, both sons of the final single Ariki, Tautape ([22], p. 47). The single Arikiship that ruled the islands split into a dual Arikiship under the control of two chiefs, each of whom held both divine and secular powers ([22], pp. 22, 57). It was at this time that the aforementioned sickness affected the Matangaro lineage, prompting the actions of the culture hero Temu-matua, who became the first Whainga-aitu Ariki. The formation of the dual Arikiship demonstrates the power of the populace to exercise free will and alter the dynamics of the leadership structure of the atolls through the appointment of the second Ariki. This implies that community benefit and consensus was more important than the maintenance of a rigid power structure.

At least two of the Manihikian sites suggest lessening occupation in the 1600s–1800s, although surface features suggest renewed use in the years leading up to the arrival of missionaries in 1849. While it is difficult to pinpoint the beginning of the Tûmutu migrations, it is likely that the rahui institutional prohibitions on certain resources began at the time of the earliest human arrival. The complex stratigraphic layering at the TK001 site may indicate frequent periods of abandonment during the Dual-Ariki Period, suggesting that the ethnohistorically known version of the Tûmutu migrations took shape during this time. This cultural institution seems to have grown out of rahui traditions combined with the varied social and ecological developments of the prior centuries. This resulted in a system in which a single people were able to manage and maintain widespread and potentially vulnerable terrestrial and marine habitats through innovative ecological stewardship and political leadership drawn from ancestral traditions.

The frequency of the Tûmutu is unknown, and identifying a temporality of the migration cycles is beyond our current dating capabilities. The known dynamics of the Tûmutu, however, represent a pattern of community fissioning and fusion as well as a case of collective action or cluster-wide cooperation where the entire population is said to have conformed to the rules of this system, suggesting that the betterment of the population outweighed the benefit to leaders, individuals, or subgroups. Just as the Ariki imposed sanctions on those that broke the curfew on Te Kainga, sanctions likely ensured that everyone participated in the Tûmutu and rahui. This suggests that while the political system was flexible enough to allow for change, the centralized authority of Manihiki and Rakahanga was powerful enough to ensure that the population participated in the migrations. In this way, the dual chiefdom with its ability to fission and fuse may have acted to promote social cohesion and cooperation by providing autonomy and delegated power to groups when dispersed on Manihiki and then reasserting authority and strengthening familiar ties when aggregated on Rakahanga. Through this form of population-wide cooperation, the people of Manihiki and Rakahanga developed and maintained a human-mediated landscape capable of supporting a relatively large population on two small atolls. Rather than store surplus as harvested crops or food for feasts, the people of Manihiki and Rakahanga stored living resources through social institutions such as the dual Arikiship and the Tûmutu. By placing an entire island’s resource base under rahui restrictions and allowing it to regenerate, stocks of coconut, puraka, fish, turtles, and other resources were always available in time of need. Although the use of surplus as tribute to the Ariki is unclear, the ability to store food in its living state would have allowed for large harvests of fish, shellfish, and puraka upon arrival to whichever island was in disuse. This living surplus ensured the sustainability of the population while influencing the development and maintenance of the dynamic social institutions needed to preserve it.

4.4. Phase IV: Post AD 1849—The Three Village Period

In AD 1849, a process began that resulted in sustained external contact and a cascade of changes to the social and ecological systems of Manihiki and Rakahanga. During a crossing between the atolls, a group of voyagers were lost at sea, and a European ship eventually picked up these individuals. After a period of time and movement around the Southern Cook Islands, they returned to Manihiki along with Tahitian missionaries ([22], p. 8). Three years later the missionary W.W. Gill arrived. While the missionaries worked to convert the population to Christianity, they documented the dual Arikiship and the migrations between the islands. Gill suggested that the population consisted of roughly 1200 individuals, a marked increase over Quiros’ estimate of 500 in AD 1606 ([22], p. 8; [112], p. 12). More than a population increase, this may represent the coalescence seen in Phase III. In AD 1852 twenty lives were lost crossing between the islands during a storm, which led to the cessation of the migrations and the establishment of three permanent villages. The people abandoned the settlement on Te Kainga and divided themselves between the villages of Tukao and Tauhunu on Manihiki and a new village on the main islet of Rakahanga. While this was the final Tûmutu migration, the people of Manihiki and Rakahanga remained connected throughout the Three Village Period and consider themselves one people living on two atolls.

5. Conclusions

Manihiki and Rakahanga represent a case of small-scale social and ecological change in a challenging environment. Creative and innovative institutions (i.e., rahui, the Tûmutu, and the dual chiefdom) allowed people to effectively manage the resources on these atolls, demonstrating that despite environmental limitations such populations can achieve long-term and dynamic sustainability. We sought to address a number of different questions in this paper. First, through the use of new and existing radiocarbon dates coupled with Bayesian modelling we placed the first occupation of the Rakahanga between cal. AD 1115–1280 (95% probability), probably towards the latter end of this range. Oral histories and genealogical dating support this occupation date, and suggest that voyagers first arrived on the recently-emerged atolls of Manihiki and Rakahanga in the thirteenth or fourteenth century AD. The occupation of Manihiki occurred slightly after that of Rakahanga, possibly as the population grew and expanded outward. During the first few centuries of occupation the people transformed what was likely an ecologically impoverished landscape, introducing dogs and rats, planting coconuts and pandanus trees, digging pits for growing swamp taro, and building large coral-walled fish traps. While these transformations must have had profound impacts on the atoll ecosystems, they were the foundation for the development of a society that persisted and often thrived for more than seven centuries.

Our second research objective was to gain an understanding of the formation of the cycling political structure of Manihiki and Rakahanga. In a similar fashion to other Polynesian groups (e.g., [3,7]) the people of Manihiki and Rakahanga developed a political structure that suited their local needs and drew on ancestral political roots. After centuries of varied traditions of nucleated, dispersed, and possibly shifting habitation patterns, the people developed an institution that relied on a dual-chiefdom structure that enabled cyclical patterns of aggregated and dispersed habitation. The creation of the dual chiefdom and the Tûmutu cyclical migration system appear closely tied to population expansion, possible external threat in the form of Europeans, and environmental upheaval, as well as the continued need to closely manage the ecosystems of both atolls. Similar to other atoll societies (e.g., [52,55]) the institutions on Manihiki and Rakahanga made use of inter-atoll travel, had flexible descent reckoning, a multi-tiered political hierarchy, and a system where the voice of the people held importance. Archaeological evidence of habitation and abandonment of Manihikian islets, expansion of the village on Rakahanga, and oral historical accounts suggest that the formation of this formalized socioecological structure occurred during the late 15th century AD.

Our third question asked how patterns of land use were related to the development of formalized migrations between the atolls. Due to its very nature the Tûmutu migration system is difficult to date; however, the development of the Tûmutu was likely formalized during the time of the dual chiefdom. Complex stratigraphy at the TK001 site may indicate frequent abandonment and reoccupation of the site during this period. This institution was both an elaboration upon and an institution through which the people enacted traditional rahui practices. For the Tûmutu to function in the way Hiroa [22] suggested, as a system of population migration coupled with cyclical aggregation and dispersal, it must have required the cooperation of the vast majority of the population. The process of fissioning and fusing likely bolstered the autonomy of lineage groups during periods of dispersal and established communal ties and the solidarity of the population against external threats during periods of aggregation. Under the flexible management of the dual Arikiship, the Tûmutu extended and protected the resources of Manihiki and Rakahanga as a whole while encouraging population-wide cooperation.

The case study of Manihiki and Rakahanga demonstrates the importance of flexibility in leadership and collective action in promoting long-term sustainability. While the histories of Manihiki and Rakahanga are unique, we argue that similar parallel institutional structures likely exist for other island groups where public goods systems are often constrained in space and are managed by clearly defined groups of related peoples (see [10]). As demonstrated here through multiple lines of evidence, the people of these atolls created novel institutions that were dependent upon collective action to maintain a growing population while responding to a variety of challenges. These institutions allowed for the management of common pool resources, in effect temporarily abandoning an atoll in order to regenerate resources in a kind of temporary marine protected area. We see this specifically in the application of the rahui system to entire atolls. Through these practices, the people effectively stored living surplus on one atoll by allowing time for resource stocks to regenerate. This allowed them to survive, to thrive, and to grow in a human-mediated landscape. As the population increased, it is likely that flexible leadership and institutions emerged to meet the needs of the population, creating a dynamic system of socioecological sustainability.

While the people of Manihiki and Rakahanga have left the Tûmutu in the past, they continue to practice the rahui as a form of resource management. This practice, under which large portions of both islands are placed under protection for much of the year, arguably provides protection to fragile ecological populations such as coconut crabs (Birgus latro) and red-tailed tropics birds (Phaethon rubricauda) which dwell on the atolls today. As with the collective action-based Tûmutu and the associated fission-fusion habitation structure, modern rahui practice on the atolls requires collective buy-in from the community and support across both islands. By continuing to act collectively, the people of Manihiki and Rakahanga are able to protect and use the resources of their islands in a manner that promotes continued sustainability even in one of the most remote landscapes on Earth. The continuation and adaptation of traditional land-use practices aids the modern sustainability of habitation on Manihiki and Rakahanga. This suggests that while challenges differ across time and around the globe, the pursuit of sustainability benefits from knowledge achieved through long-term interaction with a particular environmental setting. Peoples from disparate environmental contexts may benefit from a retrospective look at traditional socioecological practices of the past as well as from the adaptation of those practices to meet the changing needs of our modern world.

In closing, we should note that after more than 700 years of ingenuity, resource management, resilience, and dynamic sustainability, the people of Rakahanga and Manihiki face an existential threat in the form of global warming and accelerating sea level rise. Low-lying coral atolls around the world are threatened by ongoing environmental stressors that exceed the limits of local adaptation. The effects of increased cyclonic activity, rising waters, and warmer temperatures are impacting day to day life as well as local industry [16,17]. Historically, the residents of the atolls were part of a lucrative pearl-farming industry focused on Manihiki Lagoon. However, this peaked in the year 2000; following the devastation of an oyster disease outbreak in that year the industry has declined, leaving only the most resilient farmers operating small farms in Manihiki Lagoon [113]. Large-scale outmigration, often to New Zealand and Australia, has left the islands with an ever-decreasing population. In 2011, the combined population of the atolls was 558 [16,17]. However, the actual number of full-time residents is much lower. Those who stay continue to take great pride in their island homes and continue to practice environmental stewardship through the rahui, which is only lifted in times of need or celebration.