Atoll islands are areas of low, flat land, and the sustainability of habitable land in such environments is sensitive to even slight changes in sea level. The projected rise in sea level during this century may lead to the submergence of atoll islands and the widespread loss of habitable land. However, the actual time sequence of past sea level change, island emergence events, and human settlement of newly emerged islands remain poorly constrained. The results of geomorphological and archaeological surveys, combined with calibrated radiocarbon age dates, at Majuro Atoll, Marshall Islands, central Pacific, reveal that the emergence of the island, triggered by a fall in sea level, was quickly followed by human settlement. The elevation of the central body of the island exceeded high water level at 2000 years CalBP, and the complete formation of the island occurred within an interval of 100 years. The island was colonized by people shortly after emergence, at 2000 years CalBP, prior to the establishment of dense vegetation, and has been continuously settled since that time.
 Atoll islands are formed upon coral reef flats by the accumulation of calcareous skeletal sand and gravels with less than 8% of their surface exceeding 3 m above mean sea level on Pacific atolls [Woodroffe, 2008]. Such islands are now under threat of submergence, with an accompanying loss of habitable environment, because of a projected sea level rise of 18–59 cm this century [Intergovernmental Panel on Climate Change, 2007]. Indeed, Funafuti Atoll in Tuvalu has been reported to be sinking [Patel, 2006], although this submergence event has been explained by a change in land-use pattern in addition to sea level change [Yamano et al., 2007]. In this context, it is important to evaluate past interactions between human settlement and the emergence and formation processes of habitable islands.
 The formation process of atoll islands has been discussed in relation to sea level change during the middle Holocene. In Maldives, the Indian Ocean, atoll islands were formed in the course of sea level rise [Kench et al., 2005]. In the tropical Pacific, on the other hand, a high sea stand of 1–2 m above present sea levels was widely observed from 2000 to 4000 years before present, and many atoll islands were formed during the subsequent relative fall in sea level [Schofield, 1977; McLean and Woodroffe, 1994; Woodroffe et al., 1999]. Although it is debated whether island formation is triggered by sea level fall, the emergence of islands above the high water level since 2000 years BP has provided habitable land for human settlement [Dickinson, 2003].
 Archaeological studies have found that early settlement of atoll islands in the Marshall Islands, central Pacific, occurred as early as 1900 years ago [Beardsley, 1993; Weisler, 1999, 2001]. Both linguistic and archaeological evidence indicate a strong relation between the settlers of these islands and the Lapita cultural complex; thus, these remote islands were colonized from the south, probably from the southeast Solomon–Vanuatu region after 2000 years ago [Intoh, 1997; Kirch, 2000]. It has been suggested that the atoll islands were uninhabitable prior to 2000 years ago [Irwin, 1992], and a time lag of 1000 years has been inferred for human settlement following the formation of the islands, representing the time required for the islands to change from a bare sandy surface to habitable land with dense vegetation [Dye et al., 1987; Irwin, 1992]. However, the temporal relation between island formation and early human settlement on atoll islands has yet to be quantified at an individual site.
 To address this limitation, we reconstructed history of island formation and human settlement at Laura Island in Majuro Atoll (7°6′N, 171°22′E), Marshall Islands (Figure 1a). The subaerial landscape of Laura Island has been anthropogenically modified in the form of agricultural pits for wet taro (Cyrtosperma) circled by banked-up sediment excavated from the pits. These bank sediments have protected the underlying, intact stratigraphic layers and provide an opportunity to investigate the timing of island formation in combination with the anthropogenic history (Figure 1d). We excavated one of the largest banks at the center of Laura Island. The archaeological results, which were reported by Yamaguchi et al. , are reassessed in this study in relation to late Holocene sea level change and the history of island formation. To reconstruct a detailed time sequence of human settlement and island formation, it is necessary to obtain a precise calibration of the local ocean reservoir effect (ΔR). To estimate ΔR at Majuro, we compare radiocarbon age data of charcoal flecks with those of marine mammal bones from the same cultural layer.
2. Study Site and Methods
 Majuro Atoll is rectangular in shape, with a width of 40 km (Figure 1b). Laura Island, the largest of the islands upon the atoll, is triangular in shape and situated at the western corner of the atoll (Figure 1c). A cross-section of the island reveals that it consists of (from the ocean to the lagoon) a coral reef, a storm ridge, a central depression, and a lagoon beach ridge (Figure 1d).
 Topographic cross-sections and the elevations of trenches, and coral and sediment samples were measured using an auto level (AE-5, Nikon Corp., Japan) or a total station (GTS-320, TOPCON Corp., Japan). Measurements were corrected to altitude above present Mean Sea Level (MSL) with reference to tide observations (South Pacific Sea Level and Climate Monitoring Project at http://www.bom.gov.au/pacificsealevel). The levels of mean high and low water spring tide (MHWS and MLWS) are +79 and −82 cm from MSL, respectively [NOAA, 2002].
 Four cores were obtained from the reef on the ocean side of the transversal survey line, using a portable core sampler (Geoact Co. Ltd, Japan). The internal structure of the island was investigated by excavating 13 trenches along transversal (west–east) and longitudinal (north–south) survey lines on Laura Island (Figure 1c). The size of each trench was 1 m wide, 2 m long, and 1–2 m deep. A large well-preserved bank located at the center of Laura Island (Lr-2-8) was excavated (trench size: 4 m long, 2 m wide, and 3.5 m deep) to reveal the history of human settlement and the timing of island formation.
 Radiocarbon ages were measured by accelerator mass spectrometers (AMS). Age data were corrected for isotopic fractionation and calibrated to CalBP using OxCal (ver. 3.10) [Ramsey, 2009]. The ages of marine samples related to the timing of island formation and sea level change were obtained from corals in growth position in cores recovered from a reef flat and from microatolls, and from individuals of the foraminifera Calcarina in island sediment that contain fresh spines, indicating their deposition soon after death [Yamano et al., 2001]. Marine04 [Hughen et al., 2004] was used as a calibration curve. The local reservoir effect (ΔR) at Majuro is estimated to be –35 ± 25 years, based on a comparison of the calibrated ages of charcoal with those of dolphin bones obtained from the same horizon (U2/Fe3) within Trench Lr-2-8 (auxiliary material). All the dates are reported as years CalBP (calibrated years before present) and listed in Tables S1 and S2.
3. Results and Discussion
3.1. Sea Level Change and Island Formation
 Six emergent microatolls with their average surface elevation of 40 ± 5 cm above MSL occur on a reef flat open to the lagoon without oceanward reef crest nor ponding during low tide (Figure 1b). Two of them yield ages of 2100–1940 and 2260–2090 years CalBP (filled squares in Figure 2). At present, the highest level of living microatolls and branching corals is −73 ± 6 cm at Majuro; thus, the sea level is estimated to have been 113 ± 8 cm at 1940–2260 years CalBP (Figure 2). The exposure of these microatolls indicates a fall in sea level since this time. Charcoal flecks from an earth oven yield ages of 2010–1890 years CalBP (a cross in Figure 2 and U1/Fe6 in Figure 3), by which time the island surface was exposed above MHWS (79 cm above MSL). The elevation of the land surface at that time was 175 ± 5 cm above the present MSL shown by the upper surface of Layer 7 in Figure 3, and sea level would have been lower than 94 ± 5 cm above MSL. The sea level curve reconstructed from the microatolls and charcoal flecks indicates a fall in sea level at 2000 years CalBP.
 Although we have no direct evidence of sea level change before or after 2000 years CalBP, emergent microatolls −3 ± 3 cm from MSL with ages of 1470–910 years CalBP at Arno Atoll 50 km west of Majuro indicate a gradual or stepwise fall in sea level since this time (open triangles in Figure 2). Before 2000 years CalBP, the occurrence of emergent microatolls at Enewetak Atoll suggests a period of high sea stand since 4000 years ago (open circles in Figure 2) [Tracey and Ladd, 1974].
 The upper one-meter of the coral reef flat on the ocean side of the island consists of cemented bioclasts with in situ corals with dates of 4300–1900 years CalBP (Lr-I, -II, and -III in Figure 1d). The reef flat surface reaches −50 cm from MSL, and is exposed during low tide. The coral reef had formed during a period of high sea stand and became emergent following a fall in sea level since 2000 years CalBP.
 The island body consists of coral and foraminiferal sand underlain by coral gravels (Figure 1d). The foraminifera are generally shallow-water species (genera Calcarina, Amphistegina, and Family Soritidae) that are abundant on the reef flat [Fujita et al., 2009]. In terms of the timing of island formation, foraminifera from the central part of the island yield the oldest ages (Figure 1d), including four radiocarbon dates of 2280–1940 years CalBP from the transversal transect (Lr-TP2 and Lr-2-8), four dates of 2170–1770 years CalBP from the longitudinal transect (N1 and S1), and three surface samples that yield dates of 2100–1940 years CalBP. The narrow cluster of dates for foraminifera sand indicates that the central part of the island was exposed above high tide level by 2100–1940 years CalBP. Once the central part of the island had formed, it accreted in both the transversal and longitudinal directions. This process of island formation is consistent with the island-formation model of “accretion from the central core” proposed by Woodroffe et al. .
 At 2100–1940 years CalBP, the reef was exposed above the low water level due to a fall in sea level. Exposure of the reef flat resulted in the reduced effect of high-energy ocean waves, enabling the accumulation of gravels on the flat [McKoy et al., 2010] and providing a habitat for abundant foraminifera, thereby accelerating the supply of sand required to form the island [Yamano et al., 2001]. Thus, the fall in sea level forced transition of ecological process that was more conductive to island formation.
3.2. Island Formation and History of Human Settlement
 We excavated a bank in the central part of the island, between two agricultural pits (Figure 3). At this site, the naturally deposited island sediment (Layer 7) is overlain by artificially banked-up sediment rich in charcoal fragments. The banked-up sediment was dug out from the adjacent pits, and consists of several layers (Layers 1 to 6). Some of the layers overlie earth ovens (shaded layers in Figure 3) consisting of burned coral pebbles and charred flecks. Charred flecks from the lowest earth oven (U1/Fe6) yield an age of 2010–1890 years CalBP. On the other hand, foraminifera sand from the top of naturally deposited layer (Layer 7) in this section yields an age of 2100–1970 years CalBP. The ages of the earth ovens in the layers overlying the Layer 6 in Figure 3 and at other archaeological sites in Laura Island [Yamaguchi et al., 2009] distribute later than 1900 years CalBP, and thus the date of U1/Fe6 shows the oldest date of human settlement. On the other hand, the age of top surface of Layer 7 falls within the island exposure age of 2100–1940 years CalBP shown in the previous section. Therefore, the age difference between the natural island surface and the earth oven in this excavated transect indicates the timing of island formation is within 100 years of the timing of initial colonization.
 The newly exposed island would have initially become sparsely covered with monocotyledon trees, and it had been considered that it would have held little resources and been vulnerable to climate variability, making it unfavorable for human settlement. Our finding that human settlement occurred soon after emergence of the island is in contrast to previous studies that proposed a time lag of 1000 years between formation of the island and human settlement, which represents the time required for the island to change from a bare sandy surface to habitable land with dense vegetation [Dye et al., 1987; Irwin, 1992].
 The first settlers did not wait for natural vegetation succession to provide a favorable habitat: they started to make the land more habitable by themselves. The charred flecks from the bottom layer were identified as monocotyledon pieces, thereby suggesting that the island was covered with coconut and/or pandanus trees at 2000 years CalBP, which might be vegetated by the settlers [Yamaguchi et al., 2009]. On the other hand, those from the ovens in the upper layers with date range from 1830 to 1600 years CalBP contain Chinese lantern, which was used as green manure, and Pacific rosewood, as well as coconut and pandanus trees. The bank sediment was dug out from the taro pits and indicates that pit agriculture was being practiced at this time. The age reversal between U1/Fe5 and U2/Fe0 within an error range in the bank shows it was heaped within a short period, and about 200 agricultural pits had been excavated since then [Yamaguchi et al., 2009]. Human modification of the island's vegetation and landforms is evident in the bank sediment overlying the layer. Moreover, some of the ovens (U1/Fe5 and U2/Fe3) contained dolphin vertebrae, and the premaxilla and maxilla of the Melon-headed Whale, Peponocephala electra, which are used to obtain ocean reservoir dates (auxiliary material), and also indicates use of marine resources by the settlers. The settlement 2000 years CalBP might be opportunistic, but vegetation effort of monocotyledon trees even at that time, continuous archaeological evidence since then and start of pit agriculture at 1800 years CalBP suggest the first settlement represents the initiation of archaeological sequence of human habitation since this period.
 This study is the first to reveal the time sequence of island exposure and human settlement. Dickinson  hypothesized a link between island exposure, driven by sea level fall from the mid-Holocene high stand, and early human settlement in atolls of the central Pacific. The present results are consistent with this view and provide age constraints on the timing of sea level fall, island formation, and human settlement. The narrow atoll island has been continuously settled for the past 2000 years since the pioneering people migrated to colonize the bare island shortly after it emerged above sea level. Since its formation, the island has been resilient against environmental change [Webb and Kench, 2010]. The landscape of the atoll islands has been modified by the settlers since colonization, with particularly rapid change in recent years.
 We deeply appreciate the people and the government of Republic of the Marshall Islands for their support of our research. The research was funded by Environmental Research and Technology Fund of Ministry of the Environment (Projects B-15 and A-0805), SATREPS of JST and JICA (08080918) and Grant-in-Aid for Scientific Research on Innovative Areas of MEXT (20121006).
 The Editor thanks the two anonymous reviewers for their assistance in evaluating this paper.