Developing a South Pacific tephra framework: Initial results from a Samoan Holocene sequence

Tephra preserved in sediments form useful isochronous marker layers, linking disparate geological, palaeoenvironmental and archaeological records. The application of tephrochronology is greatly enhanced through the detection of macroscopically invisible tephra (cryptotephra). Here, we identify two discrete cryptotephra in Samoan lake sediments, the first identification of cryptotephra in the region outside of New Zealand. Geochemical data suggest one ash layer is from a local Samoan source, providing the first data on an eruption of this age, adding to knowledge of the local volcanic record. The second has a distinctive rhyolitic glass composition, which matches either that of Raoul Island in the Kermadec Arc (1800 km south of Samoa), or two currently submarine volcanoes in the Tongan Arc, ‘Volcano F’ and Lateiki/Metis Shoal (550 and 700 km south of Samoa, respectively). In all possible source cases, this points to a regionally significant eruption of a Kermadec–Tongan volcano at ca. 10 000 a bp. The study marks the first step in the establishment of a South Pacific tephra framework that can be used to answer questions about the synchronicity of changes in hydroclimate, vegetation and early Polynesian migration patterns, as well as providing more information on the volcanic history of Pacific island volcanoes.


Introduction
The recent eruption of the Hunga Tonga-Hunga Ha'apai volcano on 15 January 2022 demonstrates the impacts that large explosive volcanic eruptions can have on the South Pacific islands but also further afield. Ashfall and the associated tsunami damaged all the islands of Tonga with the entire Pacific Rim being placed on tsunami watch (Zuo et al., 2022). The eruption has also caused scientists to begin to re-evaluate the hazards posed by submarine volcanoes, which are often poorly monitored (Poli & Shapiro, 2022;Witze, 2022).
In addition to being volcanically active, the South Pacific Ocean plays an important part in the global climate system (Brown et al., 2020) and the archipelagos of Polynesia in the South Pacific are believed to be the last habitable places on Earth colonized by humans (Sear et al., 2020). The late colonization of many Pacific islands makes them ideal locations to study natural environments prior to the arrival of humans and their subsequent impact. However, to understand the effect of colonization, there is an urgent need to determine the background context of the late Holocene climate variation in this region. In the tropical South Pacific, this requires a detailed reconstruction of the South Pacific Convergence Zone (SPCZ) over centennial to multi-millennial timescales (Brown et al., 2020). The SPCZ may have influenced human behaviour and migration routes (Anderson et al., 2006). New work is attempting to refine this, using lake sediment archives, hydroclimate reconstructions from algal lipid and leaf wax biomarkers, and carefully constructed chronologies (Sear et al., 2020). While lake sediments preserve evidence of these competing influences (e.g. Gosling et al., 2020), to date there is a paucity of Holocene palaeoclimatic data for this region. In the Eastern and Southern Pacific only a few islands have been studied in detail, and researchers have demonstrated that they contain high-resolution environmental records, notably in the Galápagos (Conroy et al., 2008), Rapa Nui (Roman et al., 2021;Saez et al., 2009) the Cook Islands (Hassall, 2017), 'Uvea (Wallis and Futuna: Maloney et al., 2022), Samoa (Gosling et al., 2020;Hassall, 2017) and Vanuatu (Sear et al., 2020). However, constructing chronologies to link sediment records from widely separated islands remains challenging. Dating individual records from islands relies on often limited radiocarbon dating, and the inability to synchronize and compare records precisely compromises a full appraisal of the interactions between palaeoenvironmental, palaeoclimate and archaeological research.
Tephrochronology has the potential to overcome some of the problems of synchronizing sedimentary and vegetation records from South Pacific islands, as well as providing insights into the volcanic history of the region. Volcanic ash (tephra) is rapidly dispersed and deposited over wide areas following an explosive eruption and can be correlated to eruptions based on their chemical composition (Lowe, 2011). Such tephra layers provide isochronous marker horizons that can be used to connect disparate archives and, where independent age estimates for an eruption exist, to contribute additional chronological information (e.g. Bourne et al., 2015). This approach has been greatly enhanced through the detection of macroscopically invisible volcanic glass particles (cryptotephra), which can extend the application of tephrochronology up to thousands of kilometres from source volcanoes (Davies, 2015). Cryptotephras have resolved questions regarding the timing and dispersal of early Humans in Europe and the North Atlantic region (Lowe et al., 2015), and the potential exists to apply a similar approach in the volcanically active South Pacific region, where the presence of cryptotephras may be particularly useful as chronological controls beyond the limit of radiocarbon as the region lacks other suitable materials for dating in this time period.
The South Pacific has been volcanically active throughout the Quaternary, with many volcanic eruptions originating from a range of geochemically distinct sources (Cole-Dai et al., 1997). However, with the exception of New Zealand (Hopkins et al., 2021) relatively little is known about tephra layers from other South Pacific volcanoes throughout the Holocene (the time period of interest in this study) (Fig. 1). Table 1 summarizes the available published information about tephra layers in the region. This study presents the first step in the development of a Holocene South Pacific tephra framework with a tephrochronological record from Lake Lanoto'o, Upolu, Samoa.

Regional setting
The Samoan archipelago The Samoan archipelago lies within the central Pacific Ocean (13-14°S, 170-173°W), with two main islands, Savai'i and Upolu, making up 96% of the landmass (total 2934 km 2 ; Fig. 1). Samoa lies at the northern termination of the subduction volcanism of the Tongan Arc, but it contrasts by erupting ocean island basalt magmas, more akin to Hawaiiantype volcanism Hawkins, 1976;Kear, 1967;Kear & Wood, 1959). The islands are remnant basaltic shields that were heavily eroded during the Pliocene and early Pleistocene, and are now buried by late Pleistocene lava flow deposits and scoria cones (Stearns, 1944). The last known volcanic eruptions in Samoa were in AD 1760, 1902and 1905-1911on Savai'i (Anderson, 1912Németh & Cronin, 2009;Venzke, 2013).
The climate of Samoa is principally controlled by the SPCZ and its interplay with the Trade Winds (Fig. 1). The most noticeable effect of the SPCZ and its location is on precipitation, which varies from >400 mm (January) to <150 mm (July) (Hassall, 2017). Precipitation characterizes seasonality in Samoa, as temperatures are relatively constant throughout the year (mean monthly minimum 23-24°C and maximum 29-30°C (Gosling et al., 2020).
Upolu Island has a tropical climate, with a mean annual temperature of 26.6°C and mean annual precipitation of 2800 mm (Lagomauitumua, 2011); the southern, more windward side of the island receives slightly more rain due to the location of the southeast trade wind belt. Whilst the island has no distinct dry season, November and April are slightly warmer and wetter with increased frequency of tropical cyclones (Mueller-Dombois & Fosberg, 1998). Different El Niño Southern Oscillation (ENSO) phases lead to strong year-to-year climatic variation due to movement of the SPCZ; the El Niño phase leads to drier conditions and the La Niña phase delivers greater wet season rainfall (Lagomauitumua, 2011). Sedimentary archives of past climate on Samoa have been reconstructed over the Holocene and show abrupt changes in precipitation proxies interpreted to indicate regional changes in the position and intensity of the SPCZ (Hassall, 2017).

Samoan volcanism
There has been active volcanism in Samoa from the Pliocene to present times (Kear & Wood, 1959;Natland, 1980). Shield growth in Savai'i is inferred to have been between 5.29 and 2 Ma, starting with ocean island basalt and terminating with evolved, trachytic lavas (Koppers et al., 2008). Rejuvenated volcanism followed the shield stage from~386 ka, including several sites of activity in the last 2000 years, and erupting as recently as 1905-1911(Fepuleai, 2016Németh & Cronin, 2009;Workman et al., 2004). Kear and Wood (1959) mapped (Western) Samoa lavas into six formations named after villages and districts where these lavas are well exposed. From the oldest to the youngest formation, they are the Fagaloa Volcanic Formation (Pliocene to Middle Pleistocene), Salani Volcanic Formation (Middle to Late Pleistocene), Mulifanua Volcanic Formation (Late Pleistocene), Lefaga Volcanic Formation (Early Holocene), Puapua Volcanic Formation (Middle to Late Holocene) and Aopo Volcanic Formation (Historical) (Fig. 2a). Subsequently, with additional radiocarbon ages and mapping, Nemeth and Cronin (2009) and Fepuleai (2016) noted that these formations are time-transgressive. Fepuleai (2016) suggested that three formations represent active volcanism during human occupation in Samoa, including the Aopo Formation in northern Savai'I, Lefaga Formation in southwest Upolu, as well as the Puapua and Mulifanua Formations in cones within broad rift zones across Samoa's main islands (Fig. 2). Upolu cones are aligned along a central rift along the long-axis of the island.
Volcanism coeval with human occupation is characterized by low-volume scoria and spatter cone building along with lava flows over weeks to several years of total eruptive time (Németh & Cronin, 2009). The majority of these eruptions exhibit mainly Hawaiian-style fire fountaining activity with associated eruption columns probably less than a few kilometres in height (Németh & Cronin, 2009). There is very little radiometric dating of volcanism on Samoa. Many morphologically young scoria cones occur, but systematic morphometric study of the Samoan cones similar to those elsewhere (Kereszturi & Németh, 2012) has not yet been performed.

Study site
Lake Lanoto'o (171°50′W, 13°54′S) is a 0.11-km 2 volcanic crater lake at the centre of Upolu island~760 m above sea level (asl) with a catchment area of 0.23 km 2 . Today the lake is 400 m wide with a maximum depth of 17.5 m (Fig. 1); in the surface layers lake water is pH 7, with a temperature of~27°C, oxygen saturation of 105% and a conductivity of 15 μs cm -1 (measured in September 2014). At~10 m depth a thermocline results in a relatively abrupt change to cooler (23°C) anoxic conditions (dissolved oxygen 10%), pH 4, and increased conductivity (21 μs cm -1 ) (Gosling et al., 2020;Hassall, 2017).

Sediment recovery
A sequence of overlapping cores was obtained from the deepest region of Lake Lanoto'o in September 2014 ( Fig. 1 and Gosling et al., 2020). A UWITEC gravity-type corer was used to recover the upper 60 cm. A Geocore cam-modified piston corer was used to retrieve overlapping sediment cores to a depth of 262 cm below the mud-water interface, below which point the sediments were impenetrable with this coring system.
All cores were stored intact in airtight tubes and kept in cold storage (+4°C) at the University of Southampton. Sub-samples were subsequently extracted from the cores for radiocarbon dating and tephra analysis.

Age vs. depth model
Radionuclide dating was used to generate an age model for the Lake Lanoto'o sediments by Gosling et al. (2020). A combination of 137 Cs and 210 Pb measurements from the surface core and 17 radiocarbon ( 14 C) measurements from bulk (13) and organic plant material (four) samples were used to create an age vs. depth model for Lake Lanoto'o in BACON 2.2 Bayesian modelling software (Blaauw & Christen, 2011). BACON highlighted two samples (LAN14-2-2 11-12 cm and LAN14-1-3 17-18 cm) that are outliers, the former being older than expected and the latter being younger than expected; the remainder of the dates are stratigraphically consistent throughout the sequence. Re-analysis of the core correlation since Gosling et al. (2020) has resulted in a revised updated age-depth model that we present in this paper. The result is a relatively small reduction in master core length compared with the Gosling et al. model. The revised age-depth model for the Lake Lanoto'o core is given in the Supporting Information, and for the last 2000 years in Maloney et al. (2022).

Tephra extraction
Contiguous 5-cm-long sub-samples were extracted from throughout the entire core sequence and sieved to recover all sediment particles between 80 and 15 µm in size. The sieve mesh was changed regularly to avoid cross-contamination. This fraction was then immersed in sodium polytungstate of prepared density using a cleaning float of 2.00 g cm −3 and an extraction float of 2.55 g cm −3 (Blockley et al., 2005). The supernatant of the extraction float was mounted on glass slides in Canada Balsam and scanned for the presence of glass shards under an optical microscope fitted with cross-polarising filters. In addition, the >2.55 g cm −3 residue was also mounted on glass slides to check for the presence of basaltic glass particles. The numbers of shards were then counted and concentrations per gram of sample (dry weight) calculated. Where tephra shards were identified the sample resolution was refined to 1 cm.

Labelling of tephra layers
Individual discrete tephra layers were assigned a unique code; this refers to the composite depth in the profile at which the peak glass shard concentration was detected in the sequence.

Geochemical analysis of glass shards
In tephra layers selected for geochemical analysis, glass shards extracted from the layers were mounted in Steurs Epofix epoxy resin. Mounts were sectioned and polished and examined using reflected light microscopy. Chemical analysis of vitreous material was undertaken using wavelength-dispersive spectrometry electron probe microanalysis (WDS-EPMA). Analysis was carried using a Cameca SX100 microprobe housed at the Department of Geosciences, University of Edinburgh. This operated using a defocused 5-μm beam size, a 15-kV voltage and a 2-nA current (for Na, K, Si, Al, Mg, Fe and Ca) or an 80-nA current (for F, Cl, S, Mn, Ti and P) (Hayward, 2012). The machine was calibrated using modified standard blocks supplied by the instrument manufacturers while a combination of internally assayed Lipari and BCR_2G Jochum & Willbold, 2006) were used as secondary standards.
The data were screened for non-glass material and outlier values and samples with analytical totals <95% were excluded and then remaining data were normalized to 100% (Hunt & Hill, 1993).The geochemical results, including data obtained from standards, are provided in the Supporting Information.

Results
Four cryptotephra layers were identified at 5-cm resolution: LAN14_13cm, LAN14_114cm, LAN14_249cm and LAN14_ 254cm (Fig. 3). It is clear that despite Lake Lanoto'o being a volcanic crater lake, there is no continuous background of glass shards (e.g. from erosion events on crater walls); instead, discrete cryptotephra layers are identified. In all cases the volcanic glass is fresh and shows no evidence of alteration, as described by Fepuleai (2016) from volcanic lake Lanoto from eastern Upolu (not to be confused with lake Lanoto'o in this study) (Fepuleai et al., 2018).

LAN14_13cm
A cryptotephra layer comprising 70 brown shards per gram was identified at a composite depth of 13 cm; the age model dates it to approximately 1887 AD (52.7-73.3 cal a BP). Geochemical data classify is as a shoshonite trachybasalt with SiO 2 concentrations of 48.03-48.73 wt% and K 2 O values of 4.14-4.20 wt% (Fig. 4).

LAN14_114cm
A cryptotephra comprising 55 clear shards per gram was originally identified in the 5-cm-resolution samples at a composite depth of 114 cm (Fig. 3); this equated to six tephra shards present on the slide itself. However, when increasing the sampling resolution to 1 cm, no distinct peak in glass shards could be identified; instead shards were dispersed over the sample, with one or two shards present in a couple of centimetres, and therefore this layer was not subject to further analysis.

LAN14_249cm and LAN14_254cm
Glass shards are present at the base of the core (262 cm) and rise to a significant peak of over 10 000 clear shards per gram at a composite depth of 254 cm; this is coincident with a magnetic susceptibility spike of 0.0004 κ (Kappa) (Fig. 3). These clear shards continue for a further 5 cm but in the 249-cm sample brown shards are also present, warranting classification of LAN14_249cm as a separate layer. Geochemical data from the clear shards in both LAN_249cm and LAN14_254cm are indistinguishable and the layers are classified as tholeiitic rhyolites with SiO 2 concentrations of 71.62-76.29 wt% and K 2 O values of 0.75-1.00 wt% (Fig. 4). Unfortunately, geochemical data could not be obtained from the brown shards originally present in the LAN14_249cm sample. These layers are associated with a large magnetic susceptibility spike and the age model dates the LAN14_254cm layer to approximately 9498-10 495 cal a BP.

Discussion
This study demonstrates that discrete cryptotephra layers can be preserved within lake sediments in Samoa. Despite the islands' volcanic origin, there is not a large amount of volcanic material within the lake sediments. This, combined with the fact that the geochemistry of the local volcanic products hinders their direct dating, highlights the potential of using tephrochronology as a stratigraphic and correlation tool in this region.
The geochemistry of the LAN14_13cm layer indicates it is consistent with a local source in the Samoa Volcanic Field (Fig. 4a). The chronology of Holocene volcanic events on Samoa is very poorly resolved. Nemeth and Cronin (2009) provide a radiocarbon age of 230 ± 70 14 C a BP from a lava flow and possible volcanic ash layer in the Lalofasa area on Savai'i island. Calibration of this radiocarbon date gives an age of 1629-1896 AD (Hogg et al., 2020) which is contemporaneous with LAN14_13cm (1887 AD from the age model), but no geochemical data exist to confirm this. There are many broadly contemporaneous eruptions in Tonga (e.g. Niuafo'oũ 450 km away, Tofua~700 km away and Hunga~800 km away and several Vanuatu centres >2000 km away) (Venzke, 2013) (Table 1). The limited geochemical data available for these contemporaneous eruptions do not match well with LAN14_13cm ( Fig. 5a,b).
Very little information about early Holocene eruptions from South Pacific volcanoes has been published (Crosweller et al., 2012;Venzke, 2013) (Table 1). However, we have ascertained that the composition of the glass in layer LAN14_254cm lies well outside the Samoan Volcanic Field (Fig. 4a), suggesting a more distal Pacific volcanic source. All known radiocarbon-dated tephra sequences in the Tongan arc extend only as far back as~6500 cal a BP (Cronin et al., 2004;Cronin & Cashman, 2007;Cronin et al., 2008) and compositions of comparatively younger large-scale eruptives from Hunga, Tofua and Fonunelei volcanoes are all more mafic than LAN14_254cm (cf. Brenna et al., 2022;Caulfield et al., 2011;Turner et al., 2012). Other possible sources of large-scale eruptions in the central to northern Tongan arc (500-700 km from the study site) include the submarine calderas of Home Reef, Late'iki (Metis Shoal) and Volcano F (Brandl et al., 2020;Yeo et al., 2022). These have normally only erupted as pumice rafts in recent history, but with lower sea levels at the start of the Holocene they may have produced subaerial tephra. Recent data obtained from sediment samples from these shoals show compositional similarities to LAN14-254cm (Fig. 5c,d).
Further south, at >1300 km from the study site, the Kermadec arc also produces tephra with some compositional affinities to LAN14-254cm. Shane and Wright (2011) identified high-K and low-K Late Quaternary tephras from marine cores around the Kermadec arc, suggesting the low-K rhyolites are probably sourced from Raoul island (1800 km from the site). Two eruptions from New Zealand (~3000 km south) have contemporaneous ages, Rotoma from Okataina (9472 ± 40 cal a BP) and Opepe (Unit E) from the Taupo Volcano (9906 ± 246 cal a BP) (Lowe et al., 2013). However, these tephras are compositionally distinct from LAN14_254cm (Fig. 5c,d) being calc-alkaline rhyolites.
The LAN_254cm data plot within the low-K rhyolite field of the Kermadec arc and similar to some of the historically erupted Late'iki and Volcano F compositions (Fig. 5c,d). Highlighted on Fig. 5c and d are two layers which may be similar in age to LAN14_254cm; the 4/18 has an age <9.7 ka BP and the 10/19 layer is dated to <10 ka BP (Shane and Wright, 2011). While a clear match cannot be made, the more proximal Tongan sources seem the most likely to disperse tephra to Lake Lanoto'o, based on distance, but the Kermadec arc provides a better geochemical match. Whilst a correlation to a specific tephra layer cannot be made, the distinctive composition of this layer means that it could be a useful isochron, if it can be traced in other sequences in the region. It also highlights many unanswered questions about potential distal tephras erupted from Kermadec-Tongan arc volcanoes that are largely small islands or submarine and the implications these eruptions may have had on the environment at the time. Unfortunately the locations of LAN14_13cm and LAN14_254cm lie outside the depths analysed for pollen and charcoal analysis by Gosling et al. (2020) (see Fig. 3) so no comparison of the impact of these potentially large eruptions can be made at this time. The present study also highlights how little is known about the preservation of ash layers in the region outside of New Zealand beyond 2000 years ago (Table 1), therefore demonstrating the potential of further work on cryptotephra layers in this volcanically active region.

Conclusion
This study demonstrates that discrete, far-travelled cryptotephras are preserved in lake records from the volcanically active island of Samoa and highlights the potential for cryptotephra analysis to be expanded in the South Pacific region. Two cryptotephras are dated, described and geochemically characterized. The youngest layer (LAN14_13cm) is indicative of a local Samoan source, suggesting an eruption occurred around 1887 AD. This study shows the potential of adding information to local volcanic histories through the use of cryptotephra layers. The older layer (LAN14_254cm, 9498-10 495 cal a BP) potentially correlates to a source in the Kermadec-Tongan arc. This highlights the potential for identifying tephras from currently submarine volcanoes and whether this represents different volcanic patterns in the past (i.e. with sea-level change) or changes in volcano form with growth and collapse cycles (as observed in Hunga-Tonga-Hunga-Hapa'ai). Whilst neither eruption can be correlated to known eruptions in the region, Figure 5. (a, b) LAN14_13cm geochemical data compared to data from contemporaneous eruptions. The Niuafo'ou Field data are from the Lau Basin Lavas (Regelous et al., 2008), the Ambrym Field data are from glass shards of the Hospital-tuff ring of the Ambrym 1913 AD eruption (Németh and Cronin, 2011) and the Hunga Field data are glass data from the Hunga 1100 AD eruption (S. Cronin, pers comm.) and is included in the Supporting Information. (C) LAN14_249cm and LAN14_254cm geochemical data compared to data from contemporaneous eruptions and Tongan geochemical data. The Rotoma field data are from Hopkins et al. (2020) and Smith et al. (2006), the Opepe (Unit E) data are from Hopkins et al. (2020), Lowe et al. (2008) and Smith et al. (2005), and the Kermadec Arc low-K rhyolite, 4/18 and 10/19 data are from Shane and Wright (2011). The Volcano F data are from the 2019 eruption (Brandl et al., 2020) and the Late'iki field data are from the 2019-2020 eruption (Yeo et al., 2022). Two-sigma errors based on the standard data are shown. [Color figure can be viewed at wileyonlinelibrary.com] they form an important first step in developing a South Pacific tephra framework and improving local (e.g. Tongan) volcanic chronologies through the analysis of other sediment cores in the region, work which is currently underway. The study demonstrates the potential for identifying cryptotephra throughout this highly volcanically active region and its value in the identification of large-scale events, which in turn would have had huge implications for palaeoclimate, past vegetation changes as well as influencing human settlement.