The last interglacial sea-level record of New Zealand (Aotearoa)

This paper presents the current state-of-knowledge of the New Zealand (Aotearoa) last interglacial (MIS 5 sensu lato) sea-level record compiled within the framework of the World Atlas of Last Interglacial Shorelines (WALIS) database. Seventy-seven total relative sea-level (RSL) indicators (direct, marine-, and terrestrial-limiting points), commonly in association with marine terraces, were identified from over 120 studies reviewed. Extensive coastal deformation around New 15 Zealand has resulted in a significant range of elevation measurements on both the North Island (276.8 to -94.2 msl) and South Island (173.1 to -70.0 msl) and prompted the use of RSL indicators to estimate rates of vertical land movement; however, indicators lack adequate description and age constraint. Identified RSL indicators are correlated with MIS 5, MIS 5e, MIS 5c, and MIS 5a and indicate the potential for the New Zealand sea-level record to inform sea-level fluctuation and climatic change within MIS 5 (sensu lato). The Northland (North Island) and Otago (South Island) regions, historically considered stable, have 20 the potential to provide a regional sea-level curve in a remote location of the South Pacific across broad degrees of latitude. Future work requires modern analogue information, heights above a defined sea-level datum, better stratigraphic descriptions, and use of improved geochronological methods. The database presented in this study is available open-access at this link: http://doi.org/10.5281/zenodo.4056376 (Ryan et al.,

research questions such as paleoclimate) and away from attempts to resolve sea-level fluctuations or eustatic sea-level history (e.g. Ghani, 1978;Hesp and Shepherd, 1978;Pillans, 1983;1986;Bull and Copper, 1986;Ward, 1988a;Suggate, 1992;Berryman, 1993;Ota et al., 1996;Rees-Jones, 2000;Begg et al., 2004;Kim and Sutherland, 2004;Litchfield and Lian, 2004;Alloway et al., 2005;Cooper and Kostro, 2006;Wilson et al., 2007;Claessens et al., 2009;Clark et al., 2010;Oakley et al., 2018). However, this heralded the practice of correlating a presumed last interglacial terrace with a generic age (e.g. 120,000 85 ka) and height of the highstand (e.g. 5 m), usually derived from the marine oxygen-isotope records, with little effort to provide precise locations, elevations, or stratigraphic descriptions of sea-level indicators. While the development of geochronological methods assisted in constraining the age of marine terraces to the appropriate marine oxygen-isotope stage, common practice to estimate sea level continued to be the use of a eustatic sea level-height determined from various marine oxygen-isotope records. Another unfortunate outcome of previous studies is the disparate nomenclature for marine terraces and their sediments 90 across physical space and time; the latter referring to both terrace age and research publication date (Table 1; Section 4).
In what may be considered a benefit from the prevalence of uplift along the New Zealand coast, numerous terrace sequences have been identified and correlated to the sea-level peak of MIS 5e, and subsequent interstadials MIS 5c and MIS 5a, or a combination of two of the three. Such sequences can be, and have been, used within New Zealand to study sea-level and climate fluctuation within MIS 5 sensu lato (e.g. Brothers, 1954;McGlone et al., 1984;Bussell, 1990;Berryman, 1992;95 Shulmeister et al., 1999); however, the majority of these studies were palynological with a focus on resolving climatic fluctuations and not sea level. Two regions of New Zealand have historically been considered tectonically stable (Gage, 1953;Pillans, 1990a;Beavan and Litchfield, 2012): the Northland Region of the North Island and the Otago Region of the South Island. Although these regions provide the best opportunity to identify a New Zealand sea-level record that can contribute to the discussion of MIS 5e sea level in a global context, only a few RSL indicators have been identified in these areas (Section 100 4).
This review of the LIG sea-level indicators in New Zealand, like the preceding reviews by Gage (1953) and Pillans (1990a), found considerable lack of detail in the published literature; including that published post-1990. Many locations are referred to broadly, with the only indication of place a large-scale location map from which a field site cannot be precisely determined.
Marine terrace descriptions are often limited to a range of altitude heights in meters without a defined sea level datum -105 commonly the reader is left to assume above mean sea level. In many instances where a stratigraphic description is provided, it is logged as meters depth of burial and the height above sea level cannot be determined. The use of altitude to determine age, whether above an undefined sea level datum or between successive terraces, is prevalent. On the North Island, tephrochronology is commonly used both as a regional stratigraphic marker and to assist in constraining marine terrace age (e.g. Pullar and Grant-Mackie, 1972;Chappell, 1975;Pain, 1976;Iso et al., 1982;Ota et al., 1989;Berryman, 1993;Wilson et 110 al., 2007;Claessens et al., 2009). For the purposes of determining last interglacial age the tephras most commonly referred to are the Rotoehu Tephra at ~47 ka (Danišík et al., 2012;Flude and Storey, 2016), providing a minimum age, and the Hamilton Ash at ~340 ka Lowe et al., 2001) providing a maximum age (Section 5.3.2). The most commonly applied geochronological methods for determining numerical age are calibrated amino acid racemization and various techniques of luminescence dating, both of which have complicated histories in New Zealand (Section 5.3). In summary, in New Zealand 115 there are few well-described and constrained RLS indicators in proportion to the sizable quantity of literature published concerning the physical record of last interglacial (sensu lato) sea level.

Types of sea-level indicators
The majority of LIG RSL indicators identified within New Zealand are associated with a marine terrace. The marine terrace terminology used here follows that established by Pillans (1990a; (Figure 1). Briefly described, a marine terrace is a gently 120 sloping, sub-planar landform, consisting of a basal, sub-horizontal shore platform (a wave-cut surface) overlain by cover beds https://doi.org/10.5194/essd-2020-288  Suggate, 1965 Upper Terrace (?121.9-106.7 m) ** Ota et al., 1984 Kemps Hill Terraces. Upper (240-220 m) Lower (170-150 m) ** Ota et al., 1996 Tarapuhi Terrace ( Chappell, 1975**Chappell, 1970**Nichol, 2002Richardson, 1985Chappell, 1975Brothers, 1954Pullar & Grant-Mackie, 1972 Chapman-Smith & Grant-Mackie, 1971Sewell, 1991**Palmer, 1988**Fleming, 1972 Cowie, 1961 Leamy, 1958Heine, 1982Begg et al, 2004**Mildenhall, 1995 Kaikoura Peninsul a Clarence River Chappell, 1975Pain, 1976**Pillans, 1983Chappell, 1975Dickson et al, 1974 Alloway  Yoshikawa et al, 1980Chappell 1975  intersection of the wave-cut platform and the base of the fossil sea cliff, which serves as the best indicator of the height and 125 timing of peak sea-level; although it is seldom exposed. The height of the base of the inland terrace riser or the terrace surface is often used as an alternative; however, because the thickness of cover beds is often unknown these elevations are not necessarily a good indicator of sea level (Pillans, 1983). The seaward terrace riser identifies the sea cliff of a subsequent sealevel highstand. In this review, depending on whether the elevation for the marine terrace RSL indicator came from the basal platform, marine cover beds or terrestrial cover beds, the RSL indicator is identified as either a marine-limiting point, a direct 130 indicator, or terrestrial-limiting point, respectively. Marine sediments for which the depositional environment hasn't been constrained and/or are located an undefined distance from the inland terrace riser, are considered marine-limiting. Marine sediments described as beach deposits are considered to have been deposited between the storm wave swash zone and breaking depth of waves; i.e. within the upper and lower limits of the indicative range of a marine terrace (Rovere et al., 2016).
New Zealand RSL indicators have also been identified at depths below modern sea level. These indicators, from within 135 sediment and well cores, are described as sediment packages that vary by location. The depositional environment with which those sediment facies were correlated dictate how the datapoint is identified in this review; i.e. a marine-or terrestrial-limiting point or a sea-level indicator (sensu stricto).
Modern analog information, used to determine indicative meaning (position relative to sea level at the time of formation) of a direct RSL indicator, was provided in only one study from the Northland Region (Nichol, 2002). For all other indicators, the 140 indicative meaning was quantified using the IMCalc tool (Lorscheid and Rovere, 2019).

Location and elevation measurements
The latitude and longitude for over half of the RSL indicators was determined using Google Earth to match locations from a 145 publication map. The location for each well core described by Brown et al. (1988) was acquired from the Canterbury Regional Council well database (www.ecan.govt.nz/data/well-search; Environment Canterbury, 2020), which also provides accuracy of well elevation measurement above mean sea level. This value was incorporated into the elevation uncertainty provided by Brown et al. (1988). The other primary method used in publications for assigning location information has been New Zealand https://doi.org/10.5194/essd-2020-288  Wilson et al. (2007) measured elevation using real-time kinetic GPS, locations were provided as map grid coordinates.
Three publications (Rees-Jones et al., 2000;McGlone et al., 2004;Kennedy et al., 2007) referred elevation data to a tidal datum other than mean sea level (msl). All other publications either referred to present or mean sea level or did not define the 155 sea-level datum and only referred to sea level, in which case mean sea level has been assumed. The mean sea level definition has not been differentiated in the below identification (Section 4) of RSL indicators, but is shown in the database (http://doi.org/10.5281/zenodo.4056376 Ryan et al., 2020a). The most commonly used method of elevation measurement has been altimetry, followed closely by metered tape or rod. In almost all cases of altimetry, the accuracy of measurement is considered to be within <5 m. The accuracy of the metered method, most commonly applied to core sediments, ranges from 160 within <0.1 m to <5 m (Brown et al., 1988;Shulmeister et al., 1999). Less common means of determining elevation include differential GPS (Wilson et al., 2007;Oakley et al., 2017), total station (Kennedy et al., 2007), and a combination of topographic maps and digital elevation models (Wilson et al., 2007). In one instance, depression of the sea horizon in relation to the location of measurement was used (Ghani, 1978).
Similar to the sea level datum, in many instances the elevation measurement method is not stated. The intended approach for 165 assigning elevation uncertainty with entry of an RSL indicator into WALIS was to determine elevation uncertainty by calculating the square root of the sum of the sea level datum error and adding a percentage of uncertainty of the elevation measurement depending upon the precision of the method used. If the sea level datum error was not provided, then the elevation uncertainty would be determined as 20% of the elevation. However, these approaches were not possible due to the lack of stated sea level datum and elevation measurement method in many studies. Furthermore, given the variable rates of uplift 170 along the New Zealand coast, applying 20% of the elevation would result in uncertainty in the range of tens of meters in some instances. Instead, if an uncertainty for a described elevation method was provided in the original publication, this was accepted and applied with entry of the RSL indicator into WALIS. All other instances were assessed on a case-by-case basis and uncertainty was defined dependent upon the quality of the indicator description provided within the original publication. None of the elevations provided here have been adjusted for the effects of uplift, GIA, or dynamic topography.

Relative sea-level indicators
The approach here for describing RSL indicators will continue the practice of dividing New Zealand between the North and South Islands, with further subdivision based roughly upon the government regions and the dominant tectonic regimes ( Figure   2). This approach was chosen for clarity in discussion and in recognition that coastal deformation in New Zealand is driven by the tectonic regimes resulting from the position of the archipelago over the active boundary between the Australian and Pacific

North Island
The North Island features distinctive tectonic and geomorphic domains manifested by the continental collision processes associated with the subduction of the Pacific Plate along the Hikurangi Margin ( Figure 2). The entire eastern North Island constitutes the Hikurangi Margin forearc that is compressing and uplifting (Nichol et al., 2017). The forearc structure from east to west is composed of an accretionary wedge (outer forearc, mostly located offshore), a forearc basin (inner forearc), and

200
the Axial Ranges (frontal ridge) bisected by the North Island Dextral Fault Belt (NIDFB) (Berryman, 1988 (Anderton, 1981). The western and northern North Island is a zone of backarc extension that normal faults have segmented into reticular blocks of uplifted basement rock and sedimentary 205 basins. The backarc region also features extinct volcanic arcs west of the TVZ left relict by the clockwise rotation of the Hikurangi subduction system (Wallace et al., 2004).
The North Island is subdivided into nine government regions ( Figure 2). Much of the north and west coasts are found within the Northland, Auckland, and Waikato regions. The northeast and east coasts are found within the Bay of Plenty, Gisborne, and Hawke's Bay regions with the southernmost coast of the North Island located within the Wellington region. The 210 government region of Whanganui-Manawatu extends to both the southern and eastern coastline of the North Island, with the eastern coastline colloquially referred to as the Wairarapa within New Zealand. This practice is applied here to avoid confusion over reference to geographic location within the review.

Northland and Auckland regions
Portions of the Northland and Auckland regions have been considered relatively tectonically stable through the late

215
Quaternary (Gage, 1953;Gibb, 1986;Pillans, 1990a;Beavan and Litchfield, 2012). Arguably the best described and constrained RSL indicator in New Zealand is the regressive nearshore, beach, and foredune sequence exposed at One Tree Point on the southern shoreline of Whangarei Harbour (Figure 3 WID 34; Nichol, 2002). Nichol (2002) traced the contact between the beach and foredune facies (delineated by a heavy mineral sand interpreted as the hightide swash deposit) for a distance of 3.4 km. The contact elevation decreases seawards from a maximum height of +6 m to +3 m above mean sea level 220 (amsl), reflecting relative sea-level fall. Ground penetrating radar across the sequence revealed the swash lamination of the beach face and supports the interpretation of a prograding barrier sequence. Thermoluminescence (TL) samples of overlying foredune sand indicate deposition through MIS 5 (substages 5e to 5a). Given the height distribution of the swash deposit, and under the assumption that no tectonic uplift has occurred, Nichol (2002) argued for deposition during late MIS 5e. The height of the swash deposit associated with the TL sample providing MIS 5e age (Sample OTP1, 115 ± 19 ka) is 4.6 ± 0.92 m amsl.

225
The One Tree Point RSL indicator is the only one identified in the Northland and Auckland regions. Other remnants of MIS 5 sea level have been recorded, but lack sufficient detail to be used as RSL indicators. A series of Master's theses investigated the Quaternary geology and geomorphology of the Aupouri and Karikari Peninsulas in Northland, where possible MIS 5 marine terraces and estuarine sediments were identified (Goldie, 1975;Ricketts, 1975;Hicks, 1975). However, the correlations were considered tentative in appreciation for a lack of dating methods at the time and the dangers of distant terrace correlation 230 or altimetry to determine age.
The west coast of Northland and Auckland has been a location of extensive sand deposition throughout most of the Quaternary and large volumes of these sands are subject to movement by nearshore currents and longshore drift with considerable erosion and deposition documented in historical record (Ballance and Williams, 1992;Blue and Kench, 2017). It is possible that remnants of the MIS 5 highstand may not exist outside of the more sheltered harbours. On the South Kaipara Peninsula, within 235 the Kaipara Harbour and protected from the high energy coastline, Brothers (1954)   made by Chappell (1975 ; Table 1). Richardson (1985) identified the Shelly Beach Formation on the southern shoreline of the Pouto Peninsula (on the northern side of the harbour), often occurring with a 40 m terrace. Chappell (1975) and Richardson

250
(1985) determined the MIS 5 age of the Shelly Beach Formation by correlating it with the Ngarino Terrace in the Whanganui Basin to the south (Section 4.1.2). However, the Ngarino Terrace was later correlated with MIS 7 (Pillans, 1983;1990b).    Chappell, 1970), providing a terrestrial-limiting point (Figure 3 WID 35). Remnants of these terraces, identified as North Taranaki 3 and North Taranaki 2 (Waiau A and Waiau B, respectively) in the northern Taranaki region, are used to correlate the west coast terraces to the marine terraces of the Whanganui Basin further south (Chappell, 1975; Table 1).
The south Taranaki coast, located on the western margin of the Whanganui Basin, is experiencing ongoing, gentle uplift due 270 to crustal flexure as the basin depocenter migrates south (Anderton, 1981). The movement has allowed preservation of a globally significant shallow marine sequence, spanning the entire Quaternary, that has been mapped and described by multiple authors (Dickson et al., 1974;Chappell, 1975), but most extensively by Pillans (1983;1990b;2017). Marine terraces, formed over the past ~700 ka, provide a robust framework of Late Quaternary sea-level fluctuations. The marine terraces consist of a basal wave-cut shore platform overlain by up to 15 m of marine sediments, grading upwards into non-marine sediments. The

275
total thickness of terrace cover beds generally increases with terrace age and in the westward direction towards the Taranaki volcanoes due to greater tephra and lahar cover bed thickness; however, there can be considerable local variation due to sand dunes. The MIS 5 terraces are designated as the Rapanui Terrace (120 ka, MIS 5e), the Inaha Terrace (100 ka, MIS 5c), and the Hauriri Terrace (80 ka, MIS 5a). The uplift, which allowed preservation of the terraces, has also resulted in shore-parallel deformation of the terraces. The inferred strandline elevations (Rapanui Terrace <30-70 m, Inaha Terrace 24-40 m, Hauriri

280
Terrace 16 m) must be viewed with additional caution due to the unknown thickness of the cover beds.
Age constraint for the Whanganui Basin sequence was provided by Pillans (1983;1990b) using a combination of tephrochronology, specifically the Rangitawa Tephra, and amino acid racemization of wood samples (see Section 5.3.3 for discussion of AAR). RSL indicator elevations for each of the MIS 5 terraces were derived from type sections exposed on the modern coast. Most type sections are kilometers seaward of the inland terrace riser and indicate only minimum sea-level 285 estimates for their respective highstands. The shore platform of the Rapunui Terrace (MIS 5e) is found within a cliff section at Castlecliff at 29 ± 0.3 m amsl and is directly overlain by ~5.5 m of marine cover beds (Bussell et al., 1992;Figure 3 WID 38). An additional 7.5 m of terrestrial cover bed includes dune sand correlated to MIS 5c. It is unclear which depositional environment the marine sediments should be correlated with; therefore, the shore platform is considered as a marine-limiting point. The Rapanui Terrace is again exposed 7.5 km west of Hawera, with the shore platform measured to 12 ± 0.3 m above

295
(MIS 5a) is exposed at 2 ± 2 m amsl near Waverly and overlain by ~2 m of fossiliferous marine sand with a basal conglomerate (Pillans, 1990b;Figure 3 WID 245). Again, it is unclear which depositional environment the marine sediments should be correlated with and the shore platform is considered a marine-limiting point. Chappell (1970; correlated the marine terrace sequences from the Whanganui Basin to northern Taranaki, Waikato, and Auckland Regions prior to any numerical age constraint and largely depended upon terrace elevation and tephrochronology.

300
Correlation of the Waiau A and Waiau B Formations to the Whanganui sequence was reinforced through use of the Mioceneage Kaawa Shellbed as a chronostratigraphic marker and similarities in climatic and sea-level change signals perceived within the Whanganui sequence and the Kaihu Group (Chappell, 1970). In summary, Chappell (1975) Chappell (1975) used the current heights of the Ngarino and Rapanui Terrace surfaces and estimates of uplift rates to correlate the terraces with the Huon Peninsula (Papua New Guinea) sea-level curve and assign an age of ~120 000 years for MIS 5e (Chappell, 1974;Bloom et al., 1974). Pillans (1983;1990b) later correlated the Ngarino Terrace to MIS 7 and the Rapanui Terrace to MIS 5e. The accuracy of Chappell's terrace correlation across most of the west coast of the North Island otherwise remains untested and has been used for geochronological constraint in subsequent 310 publications, e.g. Alloway et al. (2005).

Bay of Plenty and Gisborne regions
The central portion of the Bay of Plenty is defined by a subsiding backarc rift occupied by the Taupo Volcanic Zone ( Figure   2). The coastline marginal to the rift is experiencing moderated uplift. To the east, the Raukumara Peninsula, a northwards projection of the Axial Ranges, is subject to steady aseismic uplift, with some intermittent coseismic uplift events (Clarke et 315 al., 2010). Chappell (1975) attempted to correlate the marine terraces of the west coast of the North Island with new and previously described terrace remnants across the Bay of Plenty and Raukumara Peninsula, identified as Bay of Plenty 2 (BOP2) and Bay of Plenty 3 (BOP3; Table 1; Kear and Waterhouse, 1961;Selby et al., 1971;Chapman-Smith and Grant-Mackie, 1971;Pullar and Grant, 1972). Briggs et al. (1996; showed that the Pleistocene terraces identified in the western Bay of Plenty, with 320 which Chappell (1975) attempted correlation, are non-marine in origin. The marine terraces on the Raukumara Peninsula were first identified as the Otamaroa and Te Papa Members of the Rukuhanga Formation, described at a location near Cape Runaway (Chapman-Smith and Grant-Mackie, 1971). The terrace cover beds are comprised of marine sandstones and conglomerates overlain by tephras. After subtracting 3 m for tephra cover bed thickness, the inner margins of the terraces were measured to 60 m amsl (Otamaroa Terrace) and 30 m amsl (Te Papa Terrace) (Pullar and Grant-Mackie, 1972). The tephra overlying the

325
Otamaroa Terrace was tentatively correlated to the Hamilton Tephra and the tephra overlying the Te Papa Terrace was identified as Rotoehu Tephra. Both tephras were used to argue for the formation of the terraces during interstadials equivalent to the European Brörup (MIS 5c) and later Gottweig interstadials, respectively. Chappell (1975)

incorporated the Te Papa
Terrace into his Bay of Plenty 2 Terrace, which was correlated with west coast MIS 5 terraces in the Auckland and Waikato regions. The Otamaroa Terrace was incorporated in the Bay of Plenty 3 Terrace and correlated to MIS 7. Per Wilson et al.

330
(2007), the Otamaroa and Te Papa Terraces were later correlated by Yoshikawa et al. (1980) to MIS 5c and MIS 5a, respectively; however, the publication is in Japanese and we were not able to confirm the correlation. and 5c terraces are not present along the northern Raukumara Peninsula. Wilson et al. (2007) found these results unsatisfactory.
The MIS 5a and MIS 3 correlation was considered unlikely given the higher relative sea levels during the earlier highstands 345 https://doi.org/10.5194/essd-2020-288 Open Access Earth System Science Data Discussions Preprint. Discussion started: 15 October 2020 c Author(s) 2020. CC BY 4.0 License. and the consistent uplift of the peninsula. They also argue development of the terraces within MIS 5 is consistent with regional loess chronology and geomorphological characteristics of Pleistocene marine terraces. It is noted that, IRSL methods have developed significantly since this publication and it is likely the methods used for this work produced underestimated ages (Section 5.3.4).

350
Mahia Peninsula, in the Hawke's Bay region, is subject to coastal deformation and uplift due to its position within the accretionary wedge of the Hikurangi Margin forearc and its proximity to the Lachlan Anticline, the axis of which is directly offshore ( Figure 2). Berryman (1993) provided a comprehensive description of the peninsula and its eight marine terraces, of which three were correlated to MIS 5: Mahia III (5e), Mahia II (5c), and Mahia I (5a). All three terraces are identified on the northeast part of the peninsula. Mahia III is also found along the southwest coastline and across Portland Island directly to the 355 south. The terraces are composed of shore platforms cut into underlying bedrock with overlying marine sand cover beds. The marine sands grade upward into aeolian dune sands, in turn overlain by sequences of tephra and loess. The cover beds are described as varying in thickness from approximately 4 m for the lower terraces to 20 m for the higher terraces, but which terraces are considered lower and higher is not stated. Berryman (1993) correlated Mahia III to the MIS 5e peak at 124 ka, citing an amino acid minimum age for a wood sample (Pillans, 1990a) and palynological evidence (personal communication

360
with Matt McGlone). The stratigraphy of the tephra-loess cover beds was used to determine the age of the remaining terraces.
Mahia I, II, and III were differentiated from the five other marine terraces on the peninsula by the composition of their cover beds, which include three loess units with intervening deposits of Kawakawa Tephra and Rotoehu Tephra. Berryman (1993) mapped numerous spot altitudes of shoreline angle positions, which were determined by subtracting a mean, undefined, thickness for terrestrial cover beds. Due to the deformation of the peninsula, the elevation measurements from each terrace  (1990), King (1932) provided insufficient information to define a RSL indicator, the remaining Kingma (1971) and unpublished

375
Masters Theses (Kamp, 1978;Hull, 1985) were not made available in the course of this work.

Wairarapa and Wellington regions
The Wairarapa coast, south of Hawke's Bay, is subject to uplift and deformation through folding and faulting as it forms the southern portion of the accretionary wedge of the Hikurangi Margin forearc ( Figure 2). Further south, deformation of the Wellington coast is driven by the northeast-trending strike-slip faults of the North Island Dextral Fault Belt that have uplifted 380 basement blocks to form the Axial Ranges with intervening basins developed through differential subsidence (e.g. Wellington Harbour-Port Nicholson and the Hutt Valley) (Begg and Johnston, 2000;Lee et al., 2002). Uplifted marine terraces and subsurface marine sediments identified in well cores have been correlated to MIS 5 in multiple locations; however, few have been sufficiently described to qualify as an RSL indicator. preserved. The terraces are described as consisting of a wave-cut surface with overlying beach deposits and loess of undefined thickness. The age of each terrace is constrained by the number of "soil-units": two soil-units are found on the lower terraces

390
(EA and EB, 80 ka and 84 ka, respectively), three on EC (100 ka), and four on ED (125 ka); however, the two oldest soil-units are not consistently present and are limited in use for spatial correlation of the terraces. Mapping of the terraces was done from ground-truthing of features identified in aerial photographs. Primary spot heights were measured by determining the depression of the sea horizon. Secondary spot heights were determined from vertical angles and distances from 20-chains-to-an-inch maps [1 in = ¼ mile]; overall spot height uncertainty was determined by Ghani (1978) to be ±3 m. The spot heights were then used 395 to map the stranded shorelines and contour the outcrop pattern of each terrace from which vertical crustal movements were derived. None of the spot heights were added to the database due to concerns for precision of location and elevation, unknown thickness of cover beds, method of age constraint, and the extensive deformation of the coastline -numerous faults, anticlines and synclines run parallel and perpendicular to the coast. Between the four sequence locations, terrace surface elevation for each terrace varied in the tens of meters; however, the only surface to exceed 200 m amsl is ED. Heine (1974; 1979; 1982) identified numerous terrace surfaces in and around Wellington City. In the latter two studies, Heine used a method described as, "a micro-survey of the topography to identify any systematic pattern in the elevations" (Heine, 1979 p. 379), to deduce any relationship between coastal terraces and local tectonics. Over 350 elevation surfaces above mean 415 sea level were identified within the city of Wellington using altimetry and from topographic maps with the only criteria that the surface was "near to level". The spot heights were then subdivided into 56 discrete 'levels' within which ten 'major' terrace levels were identified and tentative correlation to other New Zealand marine terraces were made on the basis of terrace altitude.
Heine ( Two terraces, described as of "probable" marine origin underlain by gravels, were correlated by height to European Main

425
Monastirian and Late Monastirian, recognized as two stages of the Last Interglacial (sensu lato; Monastirian terrace surface was measured to 5.1 m amsl. These datapoints were not added to the database due to the uncertainty of their origin (fluvial or marine) and age. Webby (1964) described additional sections around the harbour identifying "redweathering horizons", of which some were correlated with the Last Interglacial. Estuarine sediments underlying a 60 ft [18.3 430 m] marine terrace correlated with the Last Interglacial are identified in section; however, the stratigraphic column is a composite with the elevation of the sequence surface described as varying between 3.0 and 26.5 m and the location was not included as a RSL indicator. Offshore and slightly to the south of the Porirua Harbour mouth, multiple terraces on Mana Island have been described as marine in origin (Williams, 1978). Although they are attributed to Pleistocene high sea levels, no other age constraint has been provided.

Southern Manawatu-Whanganui and northwestern Wellington region
West of the Axial Ranges, the onshore eastern margin of the Whanganui Basin begins as a narrow coastal plain at Paekakariki and broadens northward into the southern Manawatu-Whanganui region to form the Horowhenua lowlands (Begg and Johnston, 2000). Spatially variable deformation within the region is associated with faulting of numerous basement faults of the North Island Dextral Fault Belt and overall regional uplift of the ranges (Sewell, 1991;Begg and Johnston, 2000). Early 440 studies of the region led to the identification two deposits of interest: the Otaki Formation (Oliver, 1948) and the Tokomaru Marine Terrace (Cowie, 1961), which were not explicitly correlated until 1988 (Palmer et al.). Oliver (1948) formalized the Otaki Formation (originally Otaki Sandstone) designation for Late Pleistocene marine, beach, and dune sands deposited in the coastal region between Paekakariki and Palmerston North (Figure 3). The flat surface overlying the Otaki Formation was attributed to post-depositional erosion and the surface does not appear to be recognized as a marine 445 terrace until Palmer et al. (1988). An exposure within the outer, seaward terrace riser allowed for a detailed lithological description of the Otaki Formation by Te Punga (1962); however, with limited paleoenvironmental interpretation and no effort for lateral correlation of the sequence with other deposits in the region. Fleming (1972) summarized the sequence as consisting of a wave-cut shore platform upon which is emplaced a sequence of transgressive marine beach gravel and sands (basal Otaki) that transition to beach-derived, micaceous dune sands (upper Otaki) deposited as sea level fell. Awatea Lignite was deposited 450 by swamps ponded within the dunes. The Otaki Formation was correlated to the Last Interglacial at ~80,000 to 120,000 years based upon the minimum radiocarbon age of the lignite (>45 ka; N.Z. 14 C, No. 65) and regional geology (Te Punga, 1962;Fleming, 1972). As described, it is difficult to identify the transition of marine to terrestrial sediments; however, a sponge spicule bearing unit ('d') is considered marine-limiting point at 22.6 ± 2 m amsl ( Figure 3, WID 413).
The Tokomaru Marine Terrace has been described as extending north from Levin to Palmerston North and identified within 455 the intervening Manawatu Valley (Cowie, 1961;Hesp and Shepherd, 1978). Hesp and Shepherd (1978), described the terrace sediments as partly marine and mapped the terrace as cutting into the Rapanui Formation. The terrace was correlated with MIS 5 on the basis that the "upper beds" had been tentatively correlated with the Oturian (MIS 5) Stage by Fleming (1971).
Although Hesp and Shepherd (1978) make no mention of the Otaki Formation, it is near certain that the correlation referred to is that of the Otaki Formation section described by Te Punga (1962) and interpreted by Fleming (1971;1972), who does not 465 (Pillans, 1983;1990b) implies the mapping by Hesp and Shepherd (1978)  and Otaki dune sand to the north and south of Otaki, inclusive of the terrace riser and sediments described by Te Punga (1962).
A cross-section of the inner terrace riser, exposed by the Otaki River, provides the location for the fossil marine cliff, which would have been cut by peak sea level during MIS 5e; however, the shoreline angle is obscured by Otaki dune sand. Height 470 constraint for the Otaki beach sand is not provided other than described as ~30 m amsl for the area and therefore, the location of the fossil marine cliff is identified as direct sea-level indicator but assigned a large uncertainty, 30 ± 5 m amsl (Figure 3, WID 571).
Detailed stratigraphic and lithological analysis of the Otaki Formation accompanied extensive mapping in the region between Otaki and Tokomaru (Sewell, 1991). Sewell (1991) retained use of the Tokomaru Marine Terrace designation for the wave-475 cut platform underlying the sediments of the Otaki Formation and the fossil marine cliff cut at the peak of MIS 5e sea level.
The extensive mapping confirmed the presence of two later terraces, designated Post Tokomaru Marine Terrace (PTMT) 1 and PTMT 2, correlated with MIS 5c and MIS 5a respectively (Palmer et al., 1988;Sewell, 1991  within the Tasman region were dated to MIS 5e; however, it was concluded that the cave (and terrace) predated the Last Interglacial (Williams, 1982).
Terrace sequences of the South Island are less continuous and more fragmented than the North Island and attempts at broad correlation of terraces within South Island and New Zealand are fewer. Identification of paleo-shorelines and features on the west and south coast of the South Island is made difficult by proximity to the Alpine Fault (driving coastal deformation) and 495 by dense vegetation (Wellman and Wilson, 1964;Suggate, 1992). These regions also remain remote and difficult to access. A large portion of the Canterbury coast is subject to subsidence and the northeast coastline is positioned within the Marlborough Fault System (Rattenbury et al., 2006).

West Coast and Southland
Extensive mapping of the northern West Coast Region in the mid-20 th Century identified numerous marine terraces and 500 associated formations (Suggate, 1965;. In one of the earliest studies Suggate (1965) described terrace sequences for three separate segments of the coastline (Hokitika to Greymouth, Greymouth to Fox River, and Charleston to Westport) and designated MIS 5 marine sediments the Awatuna Formation. The best-preserved terrace sequence of the northern West Coast is found between Charleston and Westport, where four marine terrace surfaces are identified; two of which Suggate (1965) correlated to MIS 5. Nathan (1975)  and Waites Formation (also 6 m to 15 m amsl), respectively. In later reassessment of the northern West Coast, Suggate (1992) https://doi.org/10.5194/essd-2020-288  estimated the age of the terraces from each of the three coastal segments previously described using the stratigraphic 515 relationship of the paleo-shorelines to glacial deposits, previously published radiocarbon ages (the majority of which provide minimum ages), and individually calculated uplift rates for each sequence (with consideration for differential uplift along the coast) to correlate terraces with paleo-sea levels derived from the Huon Peninsula sea-level curve (Chappell and Shackleton, 1986). The Awatuna Formation was subdivided and correlated with MIS 5e and MIS 5c; the younger deposits retaining the Awatuna designation and the older MIS 5e deposits renamed Rutherglen Formation (Table 1; Suggate, 1985;. The Virgin 520 Flat and Waites Formations (Nathan, 1975) were correlated with the Rutherglen Formation (Suggate, 1985); however, To the south, between the Moeraki and Haast Rivers, Nathan and Moar (1975) identified and briefly described three terraces:

530
Sardine-2, Sardine-1, and Knights Point. The Sardine-1 Terrace is interpreted as fluvial and its surface is found at an intervening height (56.4-59.4 m amsl) to the marine Sardine-2 (24.4-32 m) and Knights Point (140-147 m) Terraces. The results of radiocarbon analysis (mostly of wood) were considered ambiguous with a mixture of ages implying contamination by either older or younger carbon; therefore, age was determined primarily by correlation with the glacial-interglacial sequences described in the north by Suggate (1965). Sardine-2 was correlated broadly to MIS 5, Sardine-1 to MIS 7 or possibly 535 MIS 6 due to the harsh climatic conditions indicated by pollen within the sediments, and Knights Point Terrace was considered to pre-date MIS 7. The Sardine-2 Terrace forms an ~400 m-wide strip on the north side of Ship Creek and dips gently to the south. The marine sands are described as ilmenite-rich, typical of beach and near-shore sand found along the modern coast to the north, and are considered a direct RSL indicator (Figure 4, WID 50); however, elevation constraint is poor and the uncertainty on the actual elevation of these sediments is large, 27.4 ± 7.87 m amsl.

540
Due to the rate of vegetative growth, the sections described by Nathan and Moar (1975) were quickly overgrown, but reanalysis of the Knights Point Terrace was made possible by road-widening early in the new century (Cooper and Kostro, 2006). Cooper and Kostro (2006)  were collected, both from Facies 7. The resulting age from KP-01-TL of 123 ± 7 ka was accepted. KP-03-TL had an older age 550 (146 ± 8.4 ka) but the sample was reported to be in radioactive disequilibrium and was considered likely to be an overestimate of age. However, based on the SAR measurement procedure used at the time, it is possible that these ages suffer from 'anomalous fading' (Wintle, 1973) and should be viewed as minimum ages only. Although, correlation to other marine terraces in New Zealand is discussed, the ramifications of the MIS 5e age of Knights Point Terrace for the ages of the Sardine-1 and Sardine-2 Terraces is not (Cooper and Kostro, 2006). Given the uncertainty of the Knights Point ages, the original interpretation 555 by Nathan and Moar (1975) with MIS 5 at ~ 24 m to 32 m amsl, retains validity. Bull and Cooper (1986) claimed to have identified the remnants of numerous marine terraces in the Southern Alps on either side of the Alpine Fault, inland of the West Coast locations described immediately above. Terrace age was determined by correlation to the Huon Peninsula terrace sequence in Papua New Guinea (Chappell, 1974). Terraces correlated with MIS 5 were identified by morphology: notched spur ridges associated with sea cliffs and shore platform remnants overlain by well-560 rounded quartz pebbles and cobbles interpreted to have been formed within a beach environment. The average altitude of these terraces is between 686 m and 899 m amsl. This work was refuted by Ward (1988b) on the basis of terrace morphology, the origin of the quartz pebbles, and the terrace altitudes. Ward (1988b) argued terrace morphology is not preserved as described, the pebbles could be moa gizzard stones, and cites concerns regarding correlation to the distant Papua New Guinea sequence and the implications of the inferred uplift rates, suggesting that any semi-regular sequence of terrace altitudes could be 565 correlated with the Huon Peninsulapoints refuted by Bull and Cooper (1988). Pillans (1990a) also considered the evidence for a marine origin of the terraces ambiguous and demonstrated, not only that the concerns of Ward (1988b) regarding the altitude correlation with Huon Peninsula were justified, but also that the terraces are as likely to be ridge crest notches https://doi.org/10.5194/essd-2020-288 controlled by drainage density, slope, and uplift rate. It is the opinion of these authors that the argument remains unresolved and further research would be necessary to conclude whether the features are marine terraces.

570
Fiordland, forming the southwest corner of Southland and the South Island, is extremely rugged and largely inaccessible.
Although numerous marine terraces have been identified, they have only been marginally described and studied (Wellman and Wilson, 1964;Bishop, 1985;Ward, 1988a, Kim andSutherland, 2004). Wellman and Wilson (1964) identified a marine bench at 500 ft [152 m] cut across a penultimate glaciation moraine within Hollyford Valley and continuing to the coast where it forms a notch in the headland on the south side of Martins Bay. Bishop (1985) identified eleven levels of marine "surfaces",

575
inclusive of the present-day intertidal reef to the highest at ~1000 m amsl, between Preservation Inlet and Knife and Steel Harbour from which he inferred uplift rates. The terrace sediments are poorly exposed and not described. The age of the Pleistocene terraces was determined by comparing the terrace sequence morphology to that of the terrace sequence at Taranaki on the North Island (Pillans, 1983). The terrace surface correlated with 120 ka, designated h6, is extensive with an average altitude of 370 m amsl at the strandline. Terraces h5 (300 m) and h4 (210 m) are correlated with MIS 5c and 5a respectively.

580
There are localized outcrops of rounded pebbles and cobbles and fossil islands and stacks rise above the terrace surface south of Preservation Inlet. However, no precise localities or descriptions are provided in the publication and a suitable RSL indicator was not identified. Ward (1988a), working between the outlet of Big River and Te Waewae Bay, slightly overlapping the study area of Bishop, described a sequence of at least 13 marine terraces reaching up to 1000 m altitude. Finding a lack of suitable material for any 585 geochronological method, terraces were matched to late Quaternary oxygen isotope stratigraphy of deep-sea cores using a "simple uplift model". The terraces are described in general terms: the lower terraces, inclusive of the last interglacial, consist of loess overlying several meters (up to 20 m locally) of marine gravel and sand on bedrock. Terraces 2, 3, and 4 are correlated to MIS 5a, 5c, and 5e respectively. The height of the inner margin of each terrace above mean sea level was estimated from a combination of 1:63,360 topographic maps (NZMS1 S173 & 174), oblique air and ground photographs, and limited ground 590 truthing with an altimeter: Terrace 2 at 60 m, Terrace 3 at 90 m, Terrace 4 at 140 m. These heights are distinctly different from the height of the terraces correlated to MIS 5 (sensu lato) by Bishop (1985) for the adjacent coastline to the west (Table 1). Ward (1988b) correlates the 370 m terrace in his study to MIS 9. Although the regions described by Bishop (1985) and Ward (1988a) are bisected by the Hauroko Fault, both authors recognize that there is no observable displacement of the terrace surfaces due to the fault and the difference in terrace heights is due to different interpretations of terrace age within each study.

595
Kim and Sutherland (2004) applied two methods of cosmogenic nuclide surface exposure dating ( 10 Be and 26 Al) to date shore platform erosion surfaces in bedrock located to the west of the Bishop (1985) and Ward (1988a)  derived by Ward (1988); although, in the east the terrace described at 60 m is assigned to MIS 5a. The cluster of ages reported by Kim and Sutherland (2004) are better correlated to MIS 5c with a peak sea level at ~109 ka, which would result in better agreement between the studies. However; later work (Putnam et al., 2010;Kaplan et al., 2011) to calibrate in situ cosmogenic 10 Be production rates indicate lower production in the Southern Hemisphere and cosmogenic exposure ages published prior to 2010 are likely to be too young, probably by at least 12% (Williams et al., 2015). Due to these discrepancies, the terrace is 610 https://doi.org/10.5194/essd-2020-288 which is at similar elevation to Terraces 3 (90 m) and 4 (140 m) correlated by Ward (1988a) with MIS 5c and 5e respectively, but all results were considered anomalously low and no age determination for that surface was made.

Otago Region
The Otago Region is the only region within New Zealand, other than Northland, that has historically been considered 615 tectonically stable or to have had little tectonic movement (Gage, 1953;Gibb, 1986;Pillans, 1990a, Beavan andLitchfield, 2012); however, relatively few studies of shorelines have been completed there since the early review of Gage (1953 Later work at Taieri Beach, in the region south of Brighton, was completed by Rees-Jones et al. (2000) and Litchfield and Lian (2004). The coastline south of Taieri Mouth (Waipori River) is on the upthrown (east) side of the reverse Akatore Fault. Rees-Jones et al. (2000) revisited a raised beach deposit identified by Bishop (1994) as 'h2' and sampled sediments for IRSL 625 determination. The terrace surface is described as occurring at 8-11 m above "high sea level" with cover beds consisting of two sand units upon a wave-cut platform capped by loess. The lower sand unit, ~1 m thick, is a coarse, well-sorted, reddishorange sand and gravel, and is considered a direct RSL indicator of a beach deposit at 5.4 ± 2.2 m above "high sea level" MIS 5e. However, the uncertainties provided are at 1-sigma deviation and the ages are consistent with any time over a span of ~50 ka. Furthermore, anomalous fading was observed in the IRSL sample (TBE1) indicating a potential underestimated age, but because age was consistent with the TL sample (W2857), the fading was considered insignificant and no correction was made. There are valid concerns regarding the luminescence methods at the time both studies were completed (Section 5.3.4) and the ages must be considered with skepticism.

640
The movement along the Akatore Fault, which runs roughly sub-parallel to the coast for ~22.5 km before moving offshore near the mouth of the Tokamairiro River to the south, had been recognized in the earlier work of Cotton (1957). He cited the mapping of the 40 ft surface (Ongley;1939) to argue that the movement appeared to be localized to the coastal regions Directly landward of the boulders is the well-sorted and laminated marine sand, which continues landward and is visible on both sides of the point. The boulders and sand are overlain by two distinct loess units of ~2 m total thickness. The distance from the inner margin is not stated and an accurate thickness of the marine sand is not provided; therefore, the shore platform 675 at 7.4 ± 0.25 m above mean low water level is considered marine limiting (Figure 4, WID 51). The sand directly behind the boulders was dated by IRSL analysis of the Na-feldspar component (WLL181), providing an age of 81.9 ± 11.7 ka. The overlying loess deposits were also analyzed and indicate two younger phases of deposition, 78.6 ± 4.2 ka and 28.9 ± 4.4 ka.
The apparent MIS 5a age of the marine sands is used to constrain the age of the boulders and argue for deposition by a tsunami wave not exceeding 3 m. However, given the likely location of sea level during MIS 5a (-10.5 ± 5.5 m; Creveling et al., 2017) 680 and concerns for the optical dating of Na-feldspar (Section 5.2.3), this interpretation must be regarded with skepticism.
Sections of the MIS 5 marine Hillsgrove Formation have been described on the southern coastline of Cape Wanbrow, located immediately south of Oamaru (Grant-Mackie and Scarlett, 1973). The two sections consist of nearshore marine sediments, likely deposited within a few meters depth of water, as indicated by the marine and terrestrial mollusc and avian fossil assemblages. The section at the north end of the South Oamaru Beach is described for height above the base of the cliff backing 685 the modern beach; constraint above modern sea level is not provided and a relative sea level indicator cannot be derived. The section to the east is constrained to height above the present shore platform with the inner margin designated '0 m'. The provided elevations are considered as heights above mean higher high water (MHHW) because the inner margins of modern shore platforms are typically identified at that height in relation to mean sea level (Rovere et al., 2016). sea level, and a minimum radiocarbon age (undescribed) are used to correlate the marine sediments to the Last Interglacial (sensu lato; Grant-Mackie and Scarlett, 1973). Only the basal gravel is considered marine following the later description of 695 the overlying sands as dunes (Worth and Grant-Mackie, 2003) and is designated a direct RSL indicator (Figure 4, WID 768) at 4.5 ± 1.1 m above MHHW.

Canterbury and Marlborough Regions
The Canterbury Plains is a coastal plain stretching 160 km from Timaru to the Waipara River. The plain is formed of a series of coalescing alluvial fans, emerging from the Southern Alps and exceeding 50 km in width, that has been subsiding under the The maximum depth of the Bromley Formation is 79 m below mean sea level (bmsl) and the minimum depth is 5 m bmsl.
Heights were also constrained for WALIS entry by use of the Canterbury Regional Council well database (www.ecan.govt.nz/data/well-search; Environment Canterbury, 2020).

710
The well cores investigated by Brown et al. (1988) are predominantly clustered within Christchurch. Outside of the city, the Bromley Formation was identified at 33 m to 56 m bmsl within a well (M36/1251, Figure 4, WID 72) on the shore of Lake Ellesmere, south of Banks Peninsula (Brown et al., 1988). An additional core from the nearby Gebbies Valley was analyzed close proximity (<5.5 km) to the Lake Ellesmere core reported by Brown et al. (1988), comparison of the core interpretations is difficult without age constraint or detailed stratigraphic description from the latter.

North of the Canterbury Plains and into the Marlborough Region, last interglacial marine terraces are uplifted by the North
Canterbury fold-and-thrust belt, with terraces most prominent on the limbs of actively growing anticlines (Oakley et al., 2017).
This coast was subject to extensive description by both Jobberns (1928) and Suggate (1965). One distinct terrace and multiple 730 smaller, higher and more dissected terraces, occur at varying elevations intermittently along the coast. The marine terrace surfaces are in many locations obscured by later, glacial-period gravel fans. Additional studies (Powers, 1962;Fleming and Suggate, 1964)  North of the Waipara River, terrace remnants are continuous along a vertically-steep coast for a distance of ~6 km. The terraces are then absent for ~10 km until the coastline broadens at Motunau Beach (Jobberns, 1928;Jobberns and King, 1933;Suggate, 1965). Yousif (1989), using remote sensing, mapped the terrace surfaces following the nomenclature and age correlation of Jobberns (1928) and Carr (1970); although, these designations were often inclusive of multiple terrace surfaces at different 740 altitudes within one interglacial. The most extensive terrace immediately north of the Waipara River is the Tiromoana Terrace (Carr, 1970); at Glenafric the surface tilts upwards to the northeast from ~45 m to ~80 m amsl (Oakley et al., 2017). At Motunau Beach the most extensive terrace has been alternatively identified as the Motunau Coastal Plain or Motunau Terrace and extends offshore to include Motunau Island (Jobberns, 1928;Jobberns and King, 1933;Suggate, 1965;Carr, 1970;Oakley et al., 2017). Jobberns and King (1933)  The Tiromoana Terrace is correlated with MIS 5c indicating that higher terrace remnants in close proximity are of MIS 5e and 755 MIS 7 age. Oakley et al. (2018) correlate the Bob's Flat Terrace, a less extensive terrace located to the southwest and northeast of the Tiromoana Terrace, with MIS 5e. However, this age interpretation appears to be reliant on the higher elevation of the terrace as no AAR or IRSL samples or data for this terrace are provided in Oakley et al. (2017). The IRSL and AAR results suggest at least two phases of deposition at Motunau Beach, a possibility first proposed by Suggate (1965). The distribution in ages indicates partial reoccupation, or incision with deposition, of the seaward edge of an earlier terrace during the MIS 3 760 highstand. The earlier, more extensive portion of the terrace, is correlated to MIS 5a, with the likelihood that the more minor terrace remnants at higher elevations are of MIS 5e and MIS 7 age. Oakley et al. (2017) argue that wave erosion explains the minimal record older terraces in the area as well as the lack of a MIS 5c terrace above the Motunau Beach Terrace and MIS 5a terrace below the Tiromoana Terrace. The apparent higher elevation of the younger MIS 5a Motunau Beach Terrace to the older MIS 5c Tiromoana Terrace is not discussed by Oakley et al. (2018), but is likely due to the variable uplift rates along the 765 coastline.
Further north the Haumuri Bluffs (alternatively spelled "Amuri") have been subject to many studies (Jobberns, 1928;Fleming and Suggate, 1964;Suggate, 1965;Ota et al. 1984;1996;Oakley et al. 2017;. Ota et al. (1984)  The Tarapuhi Terrace retains a diverse molluscan fossil assemblage and collections were described by both Fleming and Suggate (1964) and Ota et al. (1996). The fossil molluscs were used by Ota et al. (1996)  Kemps Hill Lower, and Amuri Bluff Terraces were determined by best fit to the Huon Peninsula sea-level curve (Chappell and Shackleton, 1986) constraining formation within MIS 5 and MIS 4. Additional AAR analysis of the fossil mollusc assemblages within the Tarapuhi and Amuri Bluff Terraces, as well as IRSL analysis of the surrounding marine sediments, 780 strengthened correlation of the Tarapuhi Terrace to MIS 5c and indicated formation of the Amuri Bluff Terrace within MIS 5a (Oakley et al., 2017). Formation of the two Kemps Hill terraces is attributed to a double peak in sea level at the beginning of MIS 5a (90.6 ± 2.0 ka and 84.0 ± 1.6 ka) based upon the sea level curves derived from coral terraces (Lambeck and Chappell, 2001) and a δ 18 O curve (Siddall et al., 2007;Oakley et al., 2018). RSL indicators are derived from the Tarapuhi Terrace inner margin (173.1 ± 2 m amsl; Figure 4, WID 107; Oakley et al., 2017; and an overlying fossiliferous marine unit (163 ± 785 2.2 m amsl; Figure 4, WID 102; Ota et al., 1996) and from the elevation of the Amuri Bluff Terrace inner margin (40.7 ± 5 m amsl; Figure 4, WID 106; Oakley et al., 2017; and the contact between the shore platform and overlying marine gravels (32 ± 3 m amsl and 50 ± 5 m amsl; Figure 4, WIDs 63 and 64; Ota et al., 1984). The Kemps Hill terraces have not been subject to direct geochronological analysis and indicators for these terraces are not identified. Ota et al. (1984) includes over a dozen stratigraphic sections for the terraces; however, specific measurements above sea level are not provided and description in text

790
can differ from the drawn section (e.g. thickness of the deposit and height above sea level) introducing uncertainty.
Furthermore, the surface that underlies the terraces undulates by tens of meters. Only two marine-limiting points (WIDs 63 and 64) for the Amuri Bluff Terrace are included from this publication.
Approximately 20 km to the northeast of Haumuri Bluff, the Kaikoura Peninsula forms a prominent headland. Suggate (1965) identified four distinct surfaces, of which the third highest and most extensive surface (170 ft to 200 ft [~52 m to 61 m] 795 altitude), could be traced on the mainland to the west. Ota et al. (1996) identified five marine terraces, in addition to Holocene surfaces, deformed by numerous folds and faults, which in general cut perpendicular across the peninsula, resulting in a downtilt to the northwest, parallel to the long axis of the peninsula. The terraces are labelled in decreasing elevation and age as I, II, III, IV, and V. Terrace III is the extensive terrace identified by Suggate (1965). Representative sections of each terrace were drawn from auger cores and AAR analyses of fossil mollusc from Terrace I provided geochronological constraint. A total of  Terraces; however, the complexity of the tectonics in the intervening distance recommends against comparing the two locations (Duffy, 2020).
The two remaining marine terrace sequences on the northeast South Island coast are located between Clarence River and Woodbank Stream in north Canterbury, and Long Point and Boo Boo Stream in the Marlborough Region (Ota et al., 1996).
The main marine terrace within these coastal sections is designated "MM". Below MM, at lower elevation is a later Holocene Clarence River and Woodbank Stream, the MM Terrace is covered by thick (often >10 m) slope-wash deposits. The elevation of the underlying shore platform decreases from 143 m amsl in the south to 105 m amsl in the north along a distance of 5 km, indicating significant down-tilting to the north. Detailed stratigraphic descriptions were derived from five exposures of the 820 terrace sediments (Ota et al., 1996), allowing for the identification of multiple direct sea level indicators at elevations, from south to north, of 144 ± 2.2 m amsl, 117 ± 3.6 m amsl, 112 ± 2.2 m amsl, 105 ± 2 m amsl, 107 ± 2.8 m amsl (Figure 4, WIDs 94 to 90). Beach deposits are identified in each exposure, inclusive of bored boulders abutting a bedrock cliff at Location 10, indicating exact shoreline position. Four of the five exposures show the beach deposits directly overlying the shore platform, but at Location 7 (WID 91), the beach deposits are found overlying estuarine silt, indicating that the marine terrace sediments 825 record a transgressive rise in sea level.
Between Long Point and Boo Boo Stream, the surface of the MM terrace is well defined with the best preservation to the north and increasing dissection by streams to the south (Ota et al., 1996). The elevation of the inner margin is measured between 55 and 80 m amsl, with the highest elevations at either end of the terrace. The marine sediments and terrestrial cover beds of the MM Terrace are exposed along the former sea cliff between it and the younger Holocene terrace and four direct RSL indicators Stream, a fluvial terrace merges with the MM Terrace. A TL age from loess overlying the fluvial terrace provides minimum age constraint in support of the MIS 5e designation. Ota et al. (1996) also argue that the number of loess layers (typically 3 to 840 4) overlying the MM terrace is consistent with loess stratigraphy overlying other last interglacial deposits in the northern South Island and southern North Island. The UM Terrace is correlated to MIS 7 due to its higher elevation.

Summary of New Zealand RSL indicators
This work identified and, using the WALIS database, standardized 77 unique RSL indicators (direct, marine-or terrestrial- (sensu lato). The remainder are correlated with the interglacial peak MIS 5e (12), warm interstadials MIS 5c (6) and MIS 5a (11), and indicators from one marine terrace on the Kaikoura Peninsula were correlated with sea-level fall from MIS 5a into MIS 5b (3). The most common methods for age determination are terrace correlation, luminescence methods, and amino acid racemization.

855
The North Island contains eighteen RSL indicators, of which the majority are marine-limiting points (12)  Peninsula are most likely MIS 5e and MIS 5a in age, in contrast to the luminescence results within Wilson et al (2007). We 860 agree with the reasoning of Wilson et al. (2007) that although the peninsula is uplifting, it is unlikely that the MIS 5a and MIS 3 record would be preserved and the MIS 5e would not. The IRSL luminescence method at the time was not fully developed and the results most likely represent minimum ages (Section 5.3.4). On the west coast, the Shelly Beach Formation, younger undifferentiated Hawera sediments, and correlated Waiau A and B Formations and northern Taranaki marine terraces need reassessment. These formations and terraces would likely be sources of multiple RSL indicators, but have not been described 865 in detail since Chappell (1975) and have not been analyzed by any absolute geochronological method. Reassessment of the age of these features is important because the use of terrace correlation by Chappell (1975)  In summary, in agreement with Gage (1953) and Pillans (1990a), the New Zealand record of last interglacial sea level lacks quality description, measurement, and age constraint even amongst those records produced since 1990. The following section will provide in greater detail the sources of uncertainty. However, the literature indicates robust relative sea-level indicators are present, not only for MIS 5e, but also the following MIS 5c and 5a interstadials. Better constraint on these records would 885 assist in the understanding of sea-level fluctuation throughout MIS 5 (sensu lato). Furthermore, there is potential for the development of a regionally-specific New Zealand sea-level record for MIS 5e from the Northland and Otago Regions; although care must be taken within the latter region to ensure no influence from minor tectonic movement.

Sources of uncertainty
This section provides a summary of the issues contributing the greatest amount of uncertainty to the New Zealand marine 890 terrace records: inconsistent terrace terminology and nomenclature, coastal deformation, and methods of geochronological constraint.

Terrace terminology and nomenclature
The disparate application of terminology and nomenclature used when discussing marine terraces in New Zealand contributed to uncertainty in interpreting the meaning of potential sea-level indicators within our database. The example used by Pillans

895
(1990a, pg. 221) in his argument for the use of standardized terminology for marine terraces was the prevalence for authors to refer to a terrace by altitude, e.g. the "100 metre terrace". The questions being: which feature of the terrace is at 100 metre and at what position within the terrace? The work of numerous authors over the decades has also contributed to an inconsistency https://doi.org/10.5194/essd-2020-288 Terrace designation refers to the shore platform, the inner margin fossil marine cliff, and the terrace surface but the cover beds 900 are identified as the Otaki Formation (Section 4.1.6).
Uncertainty is also introduced through age correlation (in absence of a numerical age) to either distant marine terrace sequences or local stage names. The early 20 th Century practice of correlating marine terrace sequences to Mediterranean stages (Monastirian, Tyrrhenian, Milazzian, Sicilian), which at the time also lacked numerical age and have since been redefined, precludes the correlation of any potential RSL indicator to MIS 5 with certainty. The youngest formally accepted Pleistocene 905 stage within New Zealand is the Haweran Stage, which encompasses the past 0.340 Ma (Raine et al., 2015). The stage boundary is identified at the base of the Rangitawa Tephra -the oldest bed of the Hamilton Ash Formation (Section 5.3.2). Local stage names for interglacial/glacial periods within the Pleistocene have been proposed (e.g. Suggate, 1965;, but these were often developed from local sediment sequences without numerical age constraint and have not been formally adopted.
Within our database we have endeavored to adapt any inconsistent marine terrace terminology to that recommended by Pillans

910
(1990a; Figure 1) and provide clarity for terrace nomenclature. To avoid confusion and to assist in understanding the global context of the New Zealand sea-level record, we have chosen not to use local New Zealand stage names for the Pleistocene and have given preference for marine oxygen-isotope stages (MIS) designations, an internationally recognized scale.

Coastal deformation and GIA
The position of the New Zealand archipelago straddling the active boundary of the Australian and Pacific plates has produced 915 a coastline subject to variable rates of vertical land movement (VLM) due to complex tectonics and displacement associated with earthquakes. Although displacement by an earthquake can have dramatic effect, interseismic deformation (deformation between earthquake events) is more likely to influence long-term trends in VLM (Beavan and Litchfield, 2012). Estimates of long-term VLM in New Zealand have been consistently derived from marine terraces or below-surface marine deposits (e.g. Chappell, 1975;Pillans, 1983;Bishop, 1985;Suggate, 1992;Berryman, 1993;Begg et al., 2004;Wilson et al., 2007;Oakley 920 et al., 2018).
While the marine terrace sequences have proven useful for estimating long-term (beyond Holocene) rates of uplift or subsidence (e.g. Beavan and Litchfield, 2012), the significance of Holocene ice-volume change (outside and within New Zealand) on glacial and hydro-isostatic adjustment (GIA) signals on coastal deformation and relative sea level have not been quantified for New Zealand (King et al., 2020). GIA encompasses all deformational, as well as gravitational, and rotational-925 induced changes to relative sea level in response to the buildup and retreat of ice sheets with residual and variable affect along coastal sections depending upon their proximity to former glaciers, ice caps and sheets (Arctic and Antarctic, Simms et al., 2016). In other words, the magnitude and wavelength of the solid Earth response to ice-and water-load history varies with time (in relation to glacial maxima) and geographical location producing a gradient in relative sea level that is modulated by mantle rheology. As a result, sea-level change around New Zealand is not uniform. For example, the Holocene highstand peaked in 930 the North Island at ~2.65 m apsl between 8.1 to 7.2 cal ka BP, whereas in the South Island, the highstand peaked at no more than ~2 m apsl between 7.0 to 6.4 cal ka BP (Clement et al., 2016).
GIA models for the late Pliocene (Grant et al., 2019)  Ice Sheet volume variations. Accordingly, New Zealand is an interesting location, both in terms of fingerprinting and ice-940 volume constraints; although it should be combined with other relevant locations such as South America (most sensitive to WAIS), southern Africa, and the northern hemisphere. Finally, the lack of consideration for GIA in the assessment of VLM rates in New Zealand is concerning. Neglecting GIA on active coastlines has been shown to lead to overestimated uplift rates at an average of 40%, but also up to 72% (Simms et al., 2016).

945
Marine terrace ages have been constrained in New Zealand using multiple different approaches. Correlated-age determination of a terrace at a known elevation, either with distant terrace sequences or a marine oxygen-isotope sea-level curve, is the most consistently applied method to determine age. Three additional age constraint methods are core to New Zealand marine terrace chronology: tephrochronology for terraces proximal to volcanic centers in the North Island; amino acid racemization (AAR); and luminescence techniques. The limitations of these methods and the consequent impacts on age constraint and terrace 950 correlation are briefly discussed here.

Marine terrace correlation
Muhs (2000) summarized three common global correlation methods that have been used in New Zealand to determine marine terrace age (and subsequently uplift rates): 1) assumption of a constant uplift rate on a shore-normal terrace sequence; 2) relation diagrams for shore-parallel terrace sequences; and 3) unique altitudinal spacing of terraces assuming a constant uplift 955 rate. Methods 2 and 3 were developed within New Zealand, the former was used by Pillans (1983) to assist with age determination of the Whanganui Basin sequence, and the latter by Bull (1985) to determine the age of marine terraces along the Alpine Fault. Major assumptions of each of these methods are that the uplift rate has been constant over time and/or that there has been no gradient of uplift normal to the coastline; although shore-parallel variation in uplift rate is expected in method 2 (Muhs, 2000). All of these methods also suffer from the circular problem of assigning a glacio-eustatic sea level to a terrace 960 to estimate its uplift rate, which is then used to determine age of additional terraces.
Global correlation methods largely rely upon a chosen distant terrace sequence (e.g. Huon Peninsula) for paleo sea level and/or a marine oxygen-isotope curve to assist in determining age (Table 2). However, these records are unlikely to accurately reflect the timing and height of sea-level highstands that formed New Zealand marine terraces. Distant terrace sequences are expected to have a local peak sea level differing in both height and timing due to their own GIA signal and other localized processes 965 (e.g. steric effects, Creveling et al., 2015). Marine oxygen-isotope records represent a convoluted signal of ocean mass and sea temperature variation resulting in a paleo global mean sea-level estimate with considerable uncertainty (Rovere et al., 2016) and can also include their own tectonic uplift correction (Simms et al., 2016). Murray- Wallace et al., 2016). This record was preferred because southern Australia and the South Island lie within 5° of latitude, a comparable distance from the Antarctic ice cap peripheral bulge, and because Australia and New Zealand display similar records of the mid-Holocene highstand and subsequent sea-level fall (Sloss et al., 2007;Clement et al., 2016;Duffy, 975 2020). However, the Australian record can be expected to differ from that of New Zealand because the North Island extends into much lower latitudes and due to South Island ice volume changes (e.g. Golledge et al., 2012;James et al., 2019;Carrivick et al., 2020); the role of Holocene glacial mass on GIA requires further investigation (King et al., 2020). The difference in the https://doi.org/10.5194/essd-2020-288 to expect a similar difference during the Last Interglacial; although, given the current resolution of available geochronological methods, it is unlikely that difference can be discerned within the LIG or deeper time.
An additional concern is the correlation of paleo-coastlines to the correct MIS. As pointed out by Litchfield and Lian (2004), the correlation of the Taieri Beach to either MIS 5e or MIS 5a has significant implications for derived uplift rates. Furthermore, estimates of the height and timing for each sea-level highstand within MIS 5 (sensu lato) has changed, particularly for the 985 interstadials, significantly through time (Table 2).

Tephrochronology
On the North Island, two tephras have been used consistently as stratigraphic markers to constrain and identify MIS 5 terraces and associated sediments: the Rotoehu Ash and the Hamilton Ash (Pullar and Grant-Mackie, 1972;Chappell, 1975;Pain, 1976;Iso et al., 1982;Ota et al., 1989;Berryman, 1993;Wilson et al., 2007;Claessens et al., 2009). The Rotoehu Ash, the 990 basal member of the Rotoiti Tephra Formation, erupted from the Taupo Volcanic Zone and serves as an important regional stratigraphic marker in the North Island. Numerous age estimates, ranging from c. 61 ka to 45 ka, have been published (see comprehensive summary by Flude and Storey, 2016), but the most recently published age estimates for the Rotoehu Ash (derived from a combination of 14 C-acclerator mass spectrometry, (U-Th)/He, and 40 Ar/ 39 Ar geochronological methods) place deposition at 47.5 ± 2.1 ka (Danišík et al., 2012;Flude and Storey, 2016). The age of the Rotoehu Ash, although well-

995
constrained, does only provide minimum age constraint to Pleistocene marine terraces.
It should be noted that loess cover bed stratigraphy has also been used to assign relative age to marine terrace sequences in

1015
New Zealand, e.g. the Mahia Peninsula (Berryman, 1993). However, loess stratigraphy, as described in the publications reviewed, is generally localized and lacking in regional (or greater) spatial correlations. Furthermore, without numerical age constraint to the sedimentation history of the location, the resolution of such an approach is questionable (Muhs, 2000).

Amino Acid Racemization (AAR)
Amino acid racemization has been applied to both wood and marine mollusc shell to produce numerical ages for marine 1020 terraces in New Zealand (http://doi.org/10.5281/zenodo.4056376 Ryan et al., 2020a). The first numerical age constraint was derived from alloisoleucine-isoleucine D/L values of fossil wood fragments overlying marine terraces of the Whanganui Basin terrace sequence analyzed on a modified Technicon amino acid auto-analyaer. The fossil wood D/L values were calibrated with the Rangitawa Tephra, which overlies numerous mid-Pleistocene marine terraces, the youngest of which is the Ararata Terrace (Pillans, 1983;Pillans, 1990b;Pillans and Kohn, 1981;Pillans et al., 1996). The age of the tephra, at c. 370 ka, was 1025 determined by fission-track dating of both zircon and glass components, and used to constrain the age of the Ararata Terrace to c. 400 ka, MIS 11. Calibration of the fossil wood retrieved from the Ararata Terrace lignite cover bed, closely underlying the Rangitawa Tephra, allowed numerical ages to be derived from the D/L values of fossil wood samples from the younger Rapanui Terrace using the integrated rate equation.
The minimum age derived for the Rapanui Terrace was 110 ka and the terrace was correlated with the MIS 5e transgression 1030 cycle, culminating at c. 120 ka with peak sea level between 5 and 8 m apsl (Chappell and Veeh, 1978;Pillans, 1983). An uplift model, developed using the Ararata and Rapanui Terraces as anchors, was used to calculate the ages for the other marine terraces, which were correlated with every interglacial (odd MIS) from MIS 17 to MIS 3. The only marine isotope stage to have multiple sea-level peaks represented is MIS 5, the Rapanui Terrace (MIS 5e, 120 ka), the Inaha Terrace (MIS 5c, 100 ka), and the Hauriri Terrace (MIS 5a, 80 ka) (Section 4.1.2). This geochronological framework has not only served as a basis 1035 for determining regional uplift rates of the marginal Whanganui Basin in the mid-late Quaternary, but also underpins many of the marine terrace age correlations within New Zealand since (Bishop, 1985;Ward, 1988a;Bussell, 1990;Pillans, 1990a; et al., 1996). Ota et al. (1996) (Beu and Edwards, 1984;Beu et al., 1987;Shackleton et al., 1990;Bassinot et al., 1994;Pillans et al., 1994). The reassessment by Bowen et al. (1998), although not resulting in any change in MIS correlation, provided more direct constraint for the marine terraces as the previous analysis by Pillans (1983) was of wood fragments of unknown genera in lignite beds overlying marine sediments. Furthermore, there are methodological concerns regarding the application of AAR to wood that have never been resolved; namely the kinetics of 1050 racemization within wood and the presence of internal sugars (e.g. arabinose), which interact with amino acids to form melanoidin polymers (Zumberge et al., 1980;Blunt et al., 1987;Rutter and Vlahos, 1988). Additional causes for concern are wood degradation and the seemingly increased sensitivity of wood to internal and external environmental factors.
The aminostratigraphy developed by Bowen et al (1998) does not appear to have as broad application to determining marine terrace age as the former framework developed by Pillans (1983 where an exponent is applied to either the D/L value or t, time (Goodfriend et al., 1995), with aspartic acid D/L values used for calibration (Oakley et al., 2017). These results were used in combination with infrared stimulated luminescence to develop a new chronology for North Canterbury marine terrace development, resulting in significantly different chronologies in some

1065
locations than earlier studies where elevation or degree of fluvial dissection of the marine terrace had been used to assess age (e.g. Carr, 1970;Yousif, 1987).
AAR has also been applied to fossil shell of the warm-water estuarine bivalve Anadara trapezia found in multiple locations around North Island associated with interglacials ranging in age from MIS 11 through MIS 5e (Beu and Maxwell, 1990;Murray-Wallace et al., 2000).  Wallace et al., 2000).
In summary, although early age constraint for the Whanganui Basin marine terraces was derived from AAR analysis of wood fragments (Pillans, 1983), an unreliable method, AAR of mollusc shell has proven useful in building an aminostratigraphic https://doi.org/10.5194/essd-2020-288  Murray-Wallace et al., 2000;Oakley et al., 2017). It has been shown to be complimentary to luminescence 1080 methods, providing more certainty where results converge, with the ability to assist in resolving discrepant results in the latter method (Oakley et al., 2017). The relatively new (Allen et al., 2013) Bayesian method to determine numerical age from D/L values has been proven successful when applied to a New Zealand Pleistocene dataset (Oakley et al., 2017). The applicability of AAR to mollusc and foraminifer tests for resolving relative age at Pleistocene timescales, identifying reworked contributions to deposits, and developing useful aminostratigraphic frameworks, as well as its complimentary nature to luminescence 1085 techniques, has been proven in numerous locations globally (Hearty et al., 1992;Murray-Wallace and Belperio, 1994;Murray-Wallace et al., 2010;Wehmiller et al., 1995;Wehmiller, 2013;Kaufman et al., 2013;Ryan et al., 2020b). These capabilities of AAR, and its relatively low cost (in comparison to other geochronological methods), recommend it for continued and future use resolving geochronology in New Zealand.

1090
Both thermoluminescence (TL) and infrared stimulated luminescence (IRSL) have been used in the dating of New Zealand last interglacial sediments; predominantly of multi-grain aliquots of fine polymineral silts (4-11 μm), and rarely, coarser sandsize grains of quartz and feldspars (http://doi.org/10.5281/zenodo.4056376 Ryan et al., 2020a). The application of luminescence dating to New Zealand quartz grains has been found to be unsuitable because the quartz grains generally have a very dim optically stimulated luminescence (OSL) signal intensity, are adversely affected by thermal transfer of charge 1095 between single-aliquot regenerative-dose (SAR) measurement cycles, and display unpredictable sensitivity changes across these cycles (Preusser et al., 2005;. These characteristics are, at least in part, attributable to the young sedimentary history of the quartz grains, but may also be a characteristic of the geological provenance of these grains influencing their intrinsic brightness (Preusser et al., 2006). Even if these hurdles could be overcome, the relatively high dose rates in New Zealand sediments (e.g. in comparison to Australia) would more than likely lead to the saturation of the quartz OSL signal 1100 well within 100 ka , meaning OSL analysis of quartz sediments dating to the last interglacial would more than likely result in minimum ages only. These numerous issues effectively remove quartz OSL from the chronological toolkit for dating Last Interglacial sediments in New Zealand.
Greater success in dating LIG sediments may be gained using K-feldspar grains; although these too have impediments. The primary concern is 'anomalous fading' (Wintle, 1973), which is the leak of electrons from unstable traps in the grains over the 1105 period of burial resulting in underestimated ages. Significant developments in the procedures used for measuring K-feldspar grains allow for this fading to be either reduced to negligible levels or accounted for (Thiel et al., 2011;Buvlaert et al., 2012;Rui et al., 2019). However, the stability of the non-fading component of the IRSL signal results in a much slower IRSL decay rate (Buylaert et al., 2012;Li et al., 2013;Smedley et al., 2015). Thus, a longer period of sunlight exposure is required to fully reset the signal and/or a residual dose measurement must be made to account for this prior to age determination. Alternatively,

1110
the measurement of single-grains of K-feldspar (rather than multi-grain polyminerals) would enable the adequacy of the previous resetting to be tested, which the use of multi-grain aliquots cannot as reliably disentangle (Jacobs and Roberts, 2007).
Finally, unlike quartz, the onset of saturation of K-feldspars occurs at much higher doses enabling the accurate dating of LIG and older sediments (Aitken, 1998;Huntley and Lamonthe, 2001;Li et al., 2014).
The lack of critical information regarding how the luminescence signal/s was measured in the studies reviewed here does not 1115 instill confidence in the reported ages. Given the methodological advances in luminescence measurement procedures since many of the studies reviewed here were completed, with the exception of the recent work by Oakley et al. (2017), the ages presented above represent, at best, a starting point. This uncertainty in absolute age of the associated marine terrace sequences and correlation with the appropriate sea-level peak within MIS 5 has significant implications for estimates of uplift rates in https://doi.org/10.5194/essd-2020-288 some locations (e.g. Taieri Beach, Section 4.2.2). Advances in method, particularly in using the elevated temperature post-1120 infrared IRSL (pIRIR), have been shown to be effective in overcoming the fading issues in very young New Zealand coastal sediments (Madsen et al., 2011) and indicate the ability to produce ages with a higher degree of accuracy. The use of single feldspar grains, instead of polymineral aliquots, has the potential to assist in the identification and removal of grains that are 'poorly behaved' prior to age estimation, identify whether or not the deposits contain a reworked component, and the extent to which the previous resetting of electron traps was completed (Jacobs and Roberts, 2007). The applicability of the single 1125 grain method depends upon the grains having a sufficiently bright signal for accurate equivalent dose measurement -New Zealand quartz is often reported as rather dim, K-feldspars as much brighter. We therefore recommend a complete and systematic reassessment of all previously reported ages. Furthermore, in future all luminescence data (e.g. individual dose rate components, fading factors, equivalent dose distribution patterns and associated statistics, and age model selection) needs to be reported completely. Such transparency will enable future reviewers to make a much more informed judgement about the 1130 quality of the data.

Further details: Pleistocene and Holocene sea-level fluctuations and climatic change
The Whanganui Basin, in addition to the Rapanui (MIS 5e), Inaha (MIS 5c), and Hauriri (MIS 5a) terraces (Section 4.1.2), retains a shallow marine basin sequence recording all high sea-level marine oxygen-isotope stages of the last 2.6 Ma of which the past 0.7 Ma are represented by marine terraces. This record is important not only for its completeness, but also because it 1135 offers a detailed paleoenvironmental record from an isolated part of the South Pacific (Pillans, 1991;see Pillans, 2017 and references therein) and has been subject to an extensive variety of geochronological and stratigraphical methods. The stratotype section and point of the four stages representing Quaternary New Zealand are defined by the fossiliferous marine sediments within the Whanganui Basin. A paleovegetation and paleoclimatic record spanning much of the Haweran Stage (0.340 Ma to present) has been developed from the terrace cover beds. It also retains a long record of tephra (to c. 2.17 Ma) and loess (to c.

1140
0.50 Ma) deposition, which provide opportunity for regional correlation in the North Island.
Although the Whanganui Basin sequence is surely the longest and most complete record of sea-level fluctuation and climatic change within New Zealand, it is not the only one to extend beyond MIS 5. The vertical land movement along several sections of the New Zealand coastline has allowed for the preservation of paleo records of marine, coastal and terrestrial environments extending not only to the present but also farther into the Pleistocene; e.g. Mahia Peninsula, North Island (Berryman, 1993) 1145 and the Fiordlands, South Island (Bishop, 1985;Ward, 1988a). The sediments preserved along the New Zealand coastline have proven valuable sources of proxy data (commonly in the form of fossil marine mollusc assemblages or fossil pollen) useful for biostratigraphy and paleoclimatic reconstructions (e.g. Grant-Mackie and Scarlett, 1973;Dickson et al., 1974;Moar, 1975;McGlone et al., 1984;Mildenhall, 1985;1995;Moar and Mildenhall, 1988;Bussell, 1990;Berryman, 1992;1993;Ota et al., 1996;Shulmeister et al., 1999;Murray-Wallace et al., 2000). However, lacking the scrutiny of the Whanganui Basin,

1150
any sea-level record (or associated climatic record) derived from older Pleistocene marine terraces and sediments is likely to suffer the similar problems of insufficient description as the MIS 5 record, leading to large uncertainty in age and interpretation.
A recent study (Clement et al., 2016) found spatial and temporal variation in New Zealand's Holocene relative sea-level change may be influenced by a number of different mechanisms. A north-south gradient in RSL may be a result of the position of the archipelago within the intermediate field around Antarctica across broad degrees of latitude. Continental levering could have 1155 a significant effect on the timing and magnitude of sea-level change at a regional to local scale as driven by glacial meltwater loading and width of the adjacent continental shelf. Potentially significant drivers of relative sea-level change are regional and local effects of tectonic regime, wave climate, and sediment regimeall of which require further research to characterize (Clement et al., 2016). to both global studies of sea-level and ice-volume change and to New Zealand estimates of long-term vertical land movement and paleoenvironment. A New Zealand regional sea-level curve, derived from Northland and Otago regions, would provide a valuable record within the remote South Pacific, which could assist in understanding the eustatic sea-level response to ice mass 1165 change in Antarctica and would allow for better assessment of coastal deformation and improved estimates of long-term vertical land movement around New Zealand. In recognition that most RSL indicators are poorly described, not related to a defined sea-level datum, lack numerical age control, and that most existing luminescence-derived numerical ages were derived from outdated methods, greater accuracy would also require reassessment of nearly all RSL indicators identified to present.
The apparent extensive preservation of MIS 5e, 5c, and 5a marine and coastal sediments and the terrestrial sediments of 1170 intervening MIS 5d and MIS 5b substages provides opportunity for study of sea-level fluctuations and climatic changes throughout MIS 5 (sensu lato) and changes to long-term rates of vertical land movement.

Data Availability
The database is available open access and kept updated as necessary at the following link: http://doi.org/10.5281/zenodo.4056376 (Ryan et al., 2020a). The files at this link were exported from the WALIS database 1175 interface on 23 September 2020. Description of each data field in the database is contained at this link: https://doi.org/10.5281/zenodo.3961543 (Rovere et al., 2020), that is readily accessible and searchable here: https://walishelp.readthedocs.io/en/latest/. More information on the World Atlas of Last Interglacial Shorelines can be found here: https://warmcoasts.eu/world-atlas.html. Users of our database are encouraged to cite the original sources in addition to our database and this article.
1180 9 Author Contributions D.D.R. was primary author, responsible for all entries into WALIS, and the conceptualization, development and writing of manuscript, as well as providing expert review of amino acid racemization data. A.C. contributed to the structure and writing of the manuscript, and assisted D.D.R. in developing figures. N.R.J. provided expert review of luminescence data, contributing significantly to that section of the manuscript as well as more minor contributions elsewhere. P.S. contributed significantly to 1185 the discussion of coastal deformation and GIA in New Zealand.

Competing Interests
The authors declare that they have no conflict of interest.

Acknowledgements
The data presented in this publication were compiled in WALIS, a sea-level database interface, developed with funding from D.D.R. wishes to thank David Lowe for clarifying aspects of New Zealand tephrochronology and providing useful publications towards that purpose. The authors also thank Alessio Rovere for constructive comments on various drafts of this manuscript.

References
Aitken, M.J., An introduction to optical dating: the dating of Quaternary sediments by the use of photon-stimulated luminescence. Oxford University Press, Oxford. 280 pgs., 1998.