A Global Database of Marine Isotope Substage 5a and 5c Marine Terraces and Paleoshoreline Indicators

. In this review we compile and document the elevation, indicative meaning, and chronology of Marine Isotope Substage 5a and 5c sea level indicators for 39 sites within three geographic regions: the North American Pacific coast, the North American Atlantic coast and the Caribbean, and the remaining globe. These relative sea level indicators, comprised of geomorphic indicators such as marine 10 and coral reef terraces, eolianites, and sedimentary marine and terrestrial limiting facies, facilitate future investigation into Marine Isotope Substage 5a and 5c interstadial paleo-sea level reconstruction, glacial isostatic adjustment, and Quaternary tectonic deformation. The open access database, presented in the format of the World Atlas of Last Interglacial Shorelines (WALIS) database, can be found at in the primary literature and indicative ranges equal to, or less than, ~2 m calculated by IMCalc. Coral reef terraces beach deposits with a poor quality (2) applying methods for elevation adopting an

A rich literature catalogues the legacy of mapping globally distributed MIS 5a and 5c reef tracts, wave-30 cut platforms, and other marine-and terrestrial-limiting sedimentological indicators for the purpose of measuring the local peak sea level achieved during these ice-volume minima (Griggs, 1945;Alexander, 1953;Bretz, 1960;Land et al., 1967;Mesolella, 1967;Chappell, 1974;Chappell and Veeh, 1978;Cronin et al., 1981). Here we adopt the standardized framework provided by the World Atlas of Last Interglacial Shorelines (WALIS) database (WALIS, https://warmcoasts.eu/world-atlas.html) to compile 35 the English-language publications of globally outcropping relative sea level (RSL) indicators ascribed by the primary authors as MIS 5a and 5c in age. The open-access database, which includes site descriptions, elevation and geochronological constraints, and associated metadata, is available at this 45 link: https://doi.org/10.5281/zenodo.4426206 (Thompson and Creveling, 2021). Database field descriptors can be queried at this link: https://doi.org/10.5281/zenodo.3961543 (Rovere et al., 2020). This database builds on foundational regional syntheses of MIS 5a and 5c sea level indicators for the Atlantic coast of North America and the Caribbean , Pacific coast of the United States and Baja California, Mexico (Muhs et al., 2012;Simms et al., 2016), and a subset of far-50 field localities (Creveling et al., 2017) in compiling a global dataset of 39 sites (Figure 1). The database includes site excluded from previous reviews.
The following sections include a summary of the types of geomorphic and sedimentological sea level indicators included in this review (Section 2); details on how elevation measurements, measurement 55 uncertainty, and sea level data are reported (Section 3); and an overview of the dating methods utilized in the primary publications (Section 4). The majority of this publication (Section 5) reports the current measured elevations and chronologies, along with the history of the literature, for individual sites. Section 6 summarizes future research directions. Section 7 addresses the data availability. The present elevation of MIS 5a and 5c sea level indicators reflects a number of convolved processes, 65 including, but not limited to, tectonic deformation and glacial isostatic adjustment (GIA), which require carefully applied corrections to reconstruct peak global mean sea level (GMSL). Tectonic deformation can alter the elevation of an indicator by tens to hundreds of meters (Alexander, 1953;Chappell, 1974), and glacial isostatic adjustment by similar magnitudes (Creveling et al., 2015;Simms et al., 2016). For active margins, this convolution serves as an opportunity to constrain rates of Quaternary tectonic deformation (Adams, 1984;Marquardt et al., 2004;Muhs et al., 2014) given robust assumptions of the magnitude and source of MIS 5e ice-volume melt and glacial isostatic adjustment (Broecker et al., 1968;Dodge et al., 1983;Chappell and Shackleton, 1986;Creveling et al., 2015;Simms et al., 2016). 80 Similarly after correction for tectonic uplift, discrepancies in the local elevation of globally distributed MIS 5a and 5c indicators of peak sea level retain a meaningful signal of glacial isostatic adjustment  which, in turn, allows for the reconstruction of peak global mean sea level and assessments of the sensitivity of Quaternary ice sheets to the influence of Milankovitch forcing on climate in sub-100 kyr timescales (Lambeck and Chappell, 2001;Potter et 85 al., 2004;Muhs et al., 2012;Simms et al., 2016;Creveling et al., 2017). We emphasize that this database reports measurements of uncorrected, present-day elevation of various relative sea level indicators that will enable the user to apply corrections based on the most current data-and model-based predictions.

90
The data detailed in this review (and the associated database) comprises a set of geomorphic and sedimentological indicators of past sea level, for which a comprehensive overview can be found in Rovere et al. (2016). Wave-cut marine platforms make up the largest portion of these data, particularly for the North American Pacific coast (see Muhs et al., 1992b for a detailed description of this indicator). Constructional coral reef terraces are prevalent across North American Atlantic coast, the Caribbean, 95 and the far field regions (see Chappell (1974) for an example of this indicator). Additional sedimentary features, such as eolianites, submerged beach ridges, and exposure surfaces, make up the remainder of the local inferences.

Elevation Measurements
Here we catalogue the elevation, with uncertainty (if listed), of a given indicator as reported in the 100 primary publication(s) without modification. Methods adopted to measure the present-day elevation range of reported indicators vary from hand level and altimeter surveys to mapping with modern differential GPS and Digital Elevation Models. The majority of publications summarized herein did not report the sea level datum that the elevation references and, thus, these methods are reported in the WALIS database as "Not Reported". For those sites for which primarily field workers did not report 105 elevation uncertainty, we noted this absence in the database and assigned a measurement uncertainty based upon the defined accuracy of the elevation measurement method. When available in the original publication, latitude and longitude coordinates for indicator sites are reproduced for the WALIS database; when unavailable, coordinates were interpreted from publication maps using Google Earth and noted accordingly. 110 For each site we rated the quality of the RSL elevation data following criteria established by the World Atlas of Last Interglacial Shorelines project documentation (see Relative Sea Level at https://doi.org/10.5281/zenodo.3961544). Quality assessments for RSL elevation reflect a combination of measurement precision, the specificity of the reference datum for the elevation, and the range and uncertainty of the indicative meaning (sensu Rovere et al., 2016). When these three variables constrain 130 total RSL uncertainty to <1 m or 1-2 m, then WALIS defines the RSL elevation as quality excellent (5) or good (4), respectively. If, however, uncertainties in these three variables lead to a RSL elevation of 2-3 m or > 3m, then a rating of average (3) or poor (2) applies, respectively. For sites without a specified reference datum (e.g., 3 m below sea level rather than 3 m below mean high tide), we limited the maximum quality rating to 3 and assigned a rating based on the remaining factors. Any RSL 135 indicator of poorer quality than described above receives a very poor quality rating (1). Any terrestrial or marine limiting indicator that serves only as an upper or lower bound on RSL receives a rating of rejected (0). Not all primary references report the indicative meaning of an RSL indicator, and thus for a subset of sites we calculated the indicative meaning with the IMCalc software (Lorscheid and Rovere, 2019) in order to assign an elevation quality rating. 140
For each site we rated the quality of the RSL chronology following criteria established by the World Atlas of Last Interglacial Shorelines project documentation (see Relative Sea Level at https://doi.org/10.5281/zenodo.3961544). Quality assessment of indicator age reflects how well the 160 geochronology translates to a stage versus substage assignment. An excellent rating (5) attributes a RSL indicator to a narrow window within a substage of MIS 5 whereas a good rating (4) more generally assigns an RSL indicator to a substage. If geochronology only assigns an indicator to a generic interglacial (such as MIS 5), then this warrants an average rating (3). A poor rating (2) applies to incomplete chronologic data, or data that provides only a minimum or maximum age on the RSL 175 indicator. Conflicting age assignments between marine isotope stages warrant a very poor quality rating (1). Finally, chronologic data unable to distinguish between two or more Pleistocene Epoch interglacials warrants a rejected quality rating (0).

180
Field observers first documented emergent marine terraces on the North American Pacific Coast in the early 20th century. Early mapping efforts documented terraces along much of the California coastline (Ellis 1919;Davis, 1932;Woodring et al. 1946). Alexander (1953) pioneered the interpretation of west coast marine terraces as indicators of paleo-sea level and tectonic uplift. Since then, extensive documentation of regional MIS 5a and 5c paleo sea level indicators has yielded 18 sites which are 185 included in the present review. As noted below, many such studies reported geomorphic or chronological data that applies to multiple adjacent sites. Griggs (1945) completed early work on the Oregon coast, documenting four terraces, for which we focus on the Whisky Run and Pioneer terraces.
McInelly and  provided further geomorphic cross sections for the Whisky Run and Pioneer terraces at the Cape Arago and Coquille point sites. Early chronologies for many west coast 190 sites come from Kennedy et al. (1982), which provided leucine D:L ratios for individual sites plotted against isochrons independently constrained by uranium-series ages. These sites were summarized by Simms et al. (2016), which compared tectonically corrected RSL sites to models of Pacific coast glacial isostatic adjustment.   Kelsey et al. (1996) utilized topographic maps and altimeter surveys to map and document the platform 240 elevation ranges of six emergent bedrock terraces which crop out discontinuously through the region due to faulting; we focus on the lowest two of the surveyed platforms, the Newport and Wakonda terraces. Kennedy et al. (1982) first dated the Newport terrace using AAR. The leucine D:L ratio was plotted against isochrons of other AAR ages from proximal locations, assigning the Newport terrace to MIS substage 5a. Kelsey et al. (1996)  Rock, and then crops out just above modern beach elevation between Yaquina and Alsea bays before 255 gradually descending below sea level south of Alsea Bay-was assigned to MIS 5c. Based on these age assignments, Kelsey et al. (1996) correlated the Newport and Wakonda terraces to the Whisky Run and Pioneer terraces (see 5.1.2 Cape Arago). Figures 2 and 3 include the terrace elevations reported by Kelsey et al. (1996) and Figures 4 and 5 illustrate the MIS substage assignments for the Newport and Wakonda terraces, respectively (Kennedy et al., 1982;Kelsey et al., 1996). 260 Griggs (1945) mapped four emergent wave-cut platforms along the central Oregon coastline for which we focus on the lowest two, the Pioneer and Whisky Run terraces. Adams (1984)    published altimeter surveys that revised the peak shoreline angle elevations of the Whisky Run and Pioneer terraces reported by Adams (1984) to 31 m apsl and 68 m apsl, respectively (Figures 2 and 3). McInelly and  also reported the thickness of the sediment packages overlying both terraces and revised 270 the mapped faults that displace each terrace.

Coquille Point, Oregon
This entry focuses on the Pioneer and Whisky Run terraces documented in Griggs (1945). Attribution of the Whisky Run terrace to the MIS 5a substage first appeared in Kennedy et al. (1982) based on leucine D:L ratios on Saxidomus and a single uranium-series age on a coral (Fig. 4). Subsequently, six uranium-275 series ages on corals and bryozoans, 10 amino acid ratios on bivalve mollusks Mya truncata and Saxidomus giganteus, and oxygen isotope stratigraphy on mollusk shells supported the MIS 5a age assignment for the Whisky Run terrace (Kennedy et al., 1982;Muhs et al., 1990;Muhs et al., 2006;Fig. 4). McInelly and  revisited the region and presented a representative geomorphic crosssection that documented deformation of the Whisky Run terrace by the Pioneer anticline. This survey 280 reported a Whisky Run terrace maximum elevation of 18 m apsl ( Figure 2), a small upward revision from the 17 m apsl reported by Muhs et al. (1990). No elevation for the Pioneer terrace was reported in the text of this article, though Simms et al. (2016) extracted an elevation of 70 m apsl from the crosssection of McInelly and . Based on their similar elevations, McInelly and  correlated the Pioneer terrace at Coquille Point with the MIS 5c Pioneer terrace at Cape Blanco, itself 285 dated through amino acid ratios and faunal assemblages (Muhs et al., 1990; Figure 5). Kelsey and Bockheim (1994) utilized topographic maps to document seven wave-cut platforms; the lowest two terraces, Harris Butte and Brookings, crop out at bedrock platform elevations of 30 -62 m apsl and 57 -90 m apsl (see Figures 2 and 3), respectively, south of the Whaleshead fault zone. No 290 radiometric ages constrain the age of the Harris Butte and Brookings terraces. These terraces were assigned to MIS 5a and 5c, respectively, based upon similar soil development stages to the Whisky Run 310

Brookings, Oregon
and Pioneer terraces at Cape Arago (Figures 4 and 5). Merritts and Bull (1989) surveyed the 14 marine terraces cropping out at Bruhel Point, for which the 10 m apsl and 23 m apsl terraces (surveyed from the inner edge of the terrace) have been assigned to the MIS 5a and 5c highstands (Figures 2 and 3). A leucine D:L ratio on the 10 m apsl terrace indicates 315 either an MIS 5a or 5c substage designation (Kennedy et al., 1982). In contrast, Merritts and Bull (1989) supported age assignments of MIS 5a and 5c for the 10 m apsl and 23 m apsl terraces, respectively (see Figures 4 and 5), based upon correlations made using diagrams of inferred uplift of the Bruhel Point terraces versus inferred ages (uplift-rate diagrams) to the New Guinea sea level curve (see Section 5.3.6 Tewai and Kwambu). 320 Grove et al. (2010) utilized differential GPS to map nine marine terraces across five transects for which the inner edge of the lowest terrace, of purported MIS 5a, was surveyed between 38-77 m apsl ( Figure  2). Five sediment samples overlying the lowest terrace yielded three ages per sample from three energy stimuli: optically stimulated blue-light luminescence (OSL), optically stimulated infra-red luminescence 325

Point Reyes, California
(IRSL), and thermoluminescence (TL). Grove et al. (2010) discounted the OSL ages as being too young (MIS 2-3) whereas these authors interpreted TL ages as maximum ages; the IRSL ages were identified as the most accurate estimates of the time of deposition. The sample identified as the best indicator of terrace age (PR-2) assigns the lowest terrace to MIS 5a; the other four samples yield ages ranging from MIS 3 to MIS 5a or show internal inconsistencies ( Figure 4). 330 Alexander (1953) first mapped the emergent marine terraces cropping out in the Santa Cruz, CA region, for which we focus on (often conflicting) interpretations for the age of the laterally extensive, sediment mantled terrace formerly named the Santa Cruz terrace, as well as the two higher elevation Western and Wilder terraces later documented by Bradley and Griggs (1976). Bradley and Addicott (1968) presented 335

Santa Cruz, California
four uranium-series ages on mollusks that assigned an age to the Santa Cruz terrace-referred to as the "first" terrace in their nomenclature-consistent with either MIS 5a or 5c. Bradley and Griggs (1976) utilized seismic surveys to subdivide the Santa Cruz terrace into three constituent terraces which, in ascending elevation order include: the Davenport, Highway 1, and Greyhound terraces. A rich literature discusses the chronostratigraphic assignment of the Davenport and Highway 1 terraces. Most authors 340 have concluded, on the basis of amino acid ratios, diffusion modeling of paleo-sea cliffs, and 15 Useries ages on corals, that the Davenport terrace, which crops out at 5 -10 m apsl (see Muhs et al., 2006), formed during MIS 5a (Kennedy et al., 1982;Hanks et al., 1984;Muhs et al., 2006). The Highway 1 terrace, at an inner edge elevation of 26 -39 m apsl (Bradley and Griggs, 1976), has been alternately assigned to MIS 5c using models of the diffusion of paleo-sea cliffs (Hanks et al., 1984) or 345 to MIS 5e based upon a geologic fault modeling (Valensise and Ward 1991). In contrast to the above chronology, Perg et al. (2001) presented 10 cosmogenic nuclide ages that assigned the Santa Cruz (undifferentiated), Western (87 m apsl; Figure 2), and Wilder (132 m apsl; Figure 3) terraces to MIS 3, 365 5a, and 5c respectively. Muhs et al. (2006) contested these cosmogenic ages, arguing that they date the deposition of alluvium overlying the terraces and therefore provide a minimum age for terrace formation. We report Davenport/Highway 1 and Western/Wilder terrace elevations and chronologies as separate entries in the database (see Figures 2 -5) to most accurately represent the existing body of literature. For each terrace all of the reported numeric ages and age assignments are represented, even 370 when conflicting. Hanson et al. (1992) mapped five wave-cut platforms along the San Simeon fault zone. North and south of the fault zone the San Simeon terrace shoreline angle crops out discontinuously between 5 -7 m apsl and 9 -24 m apsl, respectively ( Figure 2). While a uranium-series age on bone and a radiocarbon age 375 on detrital charcoal provide a minimum terrace age of MIS 3 , two thermoluminescence ages on sediment underlying the emergent platform place the San Simeon terrace within the range of MIS 5a to 5c (Berger and . Of these possibilities, Simms et al. (2016) elected for the MIS 5a San Simeon terrace age assignment ( Figure 4).

380
Hanson et al. (1992) mapped a flight of marine terraces across south-central California; the lowest terrace, Q1, with surveyed shoreline angles between 5 -14 m apsl, is overlain by up to 2 m of marine sediment and 15 -30 m of non-marine sediment and is cut by several reverse faults ( Figure 2). The Q1 terrace was assigned to MIS 5a based upon three uranium-series minimum ages on marine and terrestrial mammal teeth and bones found in the overlying sediment deposits ( Figure 4). We included in 385 the WALIS database two uranium-series ages on corals considered unreliable by the authors. Rockwell et al. (1992) mapped five well-expressed marine abrasion platforms using transit-stadia surveys; the lowest instance, the Cojo terrace (10 -17 m apsl shoreline angle; Figure 2), is found west of the South Branch Santa Ynez fault (SBSYF) and crosses the hinge of the Government Point 390

Gaviota, California
Syncline. Seven uranium-series ages on bone and mollusk (two of which are maximum ages), eight amino acid ratios, and the cool-water aspect of the terrace fauna assign the Cojo terrace to MIS 5a Kennedy et al., 1992; Figure 4). Rockwell et al. (1992) correlated the first emergent terrace east of the SBSYF to the Cojo terrace based upon similar amino acid ratios and terrace faunal aspect. While the elevation of the overlying second terrace at this locale was not reported by 395 Rockwell et al. (1992), Simms et al. (2016) extracted an elevation of 25 m apsl from an illustration of the shore-parallel terrace profile (see Fig. 2 of Rockwell et al., 1992; Figure 3). No radiometric ages exist for the second terrace; Rockwell et al. (1992) inferred an MIS 5c age for the second terrace using stratigraphic correlation to the 5a terrace and the cool-water aspect of the terrace fauna ( Figure 5).

California Channel Islands (San Miguel and Santa Rosa islands)
Orr (1960) first mapped marine terraces on northwestern Santa Rosa Island and later efforts by Dibble and Ehrenspeck (1998) and Pinter et al. (2001) elaborated on both terrace mapping and the geology of the island. Muhs et al. (2014) utilized differential GPS to map the flight of emergent marine terraces on San Miguel and Santa Rosa Island, for which we focus on the lowest terrace at Santa Rosa Island, with 425 a shoreline angle mapped between 5.4 -7.4 m apsl, and the lowest terrace at San Miguel Island, with a shoreline angle mapped at ~3.5 m apsl ( Figure 2). Two amino acid ratios on Chlorostoma shells assigned the lowest Santa Rosa Terrace terrace to MIS 5a ( Figure 4). The lowest San Miguel terrace is overlain by a veneer of fossiliferous cemented gravel and marine sand, which itself is overlain in areas by alluvium (Johnson, 1969;Muhs et al., 2014). Based on stratigraphic position, Muhs et al. (2014) 430 assigned the San Miguel terrace to MIS 5a ( Figure 4). Woodring et al. (1946) conducted early mapping of the 13 emergent, wave-cut platforms of the Palos Verdes Peninsula; seven early uranium-series ages, each corrected for open-system behavior, assigned the first (lowest) terrace to MIS 5a (Szabo and Rosholt 1969). Muhs et al. (1992a) showed, on the basis 435 of aminostratigraphy on Tegula and Protothaca, and oxygen isotope data on Epilucina, that the 'first terrace', as mapped by Woodring et al. (1946), instead represents highstand deposits of both MIS 5a and 5e. Muhs et al. (2006) refined map units and assigned place-based names to supplant the counting scheme of Woodring et al. (1946), redefining the lowest, horizontally continuous surface as the Paseo del Mar terrace (the 'second' terrace of Woodring et al., 1946) with an estimated shoreline elevation 440

Palos Verdes Hills, California
angle of ~45 -47 m apsl ( Figure 2). A combination of 13 uranium-series ages on Balanophyllia and extralimital northern species within the faunal assemblage assigns the Paseo del Mar terrace to MIS 5a ( Figure 4). Around Lunada Bay, Muhs et al. (2006) estimated a ~60-70 m apsl elevation shoreline angle for the 'third' terrace of Woodring et al. (1946), an unnamed intermediate terrace between the Paseo del Mar (MIS 5a) and Gaffey (MIS 5e; the 'fifth' terrace of Woodring et al., 1946); as this terrace does not 445 have geochronological control, we do not include it as MIS 5c indicator. Vedder et al. (1957) first mapped the emergent marine terraces at San Joaquin Hills. Grant et al. (1999) extended the mapping of previous unpublished surveys (see literature discussed therein) and revised the shoreline angle elevation of the first emergent terrace to 19 -22 m apsl ( Figure 2). Based on a single 450 uranium-series age on coral and correlations between terrace height and presumed eustatic sea level, Grant et al. (1999) argued that either the first terrace formed during MIS 5a and hosts reworked MIS 5c coral, or that the MIS 5a and 5c highstands occupied the same terrace ( Figure 4).

Point Loma and Oceanside, California
Ellis (1919) first mapped five marine terraces at Point Loma. Hertlein and Grant (1944) briefly revisited the lower terraces. Carter (1957) extended the mapping to include an additional terrace cropping out 485 lower than the lowest terrace surveyed by Ellis (1919), later named Bird Rock terrace (Kern, 1977). Kern (1973) documented deformation of the Point Loma terraces. Kern and Rockwell (1992) utilized hand levels to revise the shoreline angle elevation of the Bird Rock terrace to 9 -11 m apsl ( Figure 2). The outcropping of Bird Rock terrace to the west at Oceanside is also mapped at 9 -11 m apsl ( Figure  2). Ku and Kern (1974) reported two uranium-series ages on mollusks but did not utilize these to assign 490 an age to Bird Rock terrace due to secondary uptake of uranium in the mollusk shells. Uranium-series ages MH02-056-001 derive from Chutchavaran and Dutton (2021). In the absence of radiometric ages, calibrated amino acid ratios support a Bird Rock terrace age assignment to MIS 5a (Kern 1977;Kern and Rockwell, 1992; Figure 4).

495
Davis (1932) mapped three emergent marine terraces around Malibu, California, which, in ascending order, are named the Monic, Dume, and Malibu terraces, for which we focus on the westward tilting Dume terrace. Birkeland (1972) revisited the Dume terrace, of which the shoreline angle sits between ~7 and 40 m apsl ( Figure 3). Szabo and Rosholt (1969) reported seven uranium-series ages on mollusks sampled from the Dume terrace consistent with an MIS 5c age ( Figure 5), though these ages utilized an 500 open-system model to compensate for mobile uranium within shells. If correct, these radiometric ages imply that the Dume terrace correlates with San Nicolas Island terrace 2 (Birkeland 1972; see above). Szabo and Rosholt (1969) and Birkeland (1972) documented a higher, adjacent terrace referred to as terrace C/Corral terrace, which crops out between the Malibu and Dume terraces. Szabo and Rosholt (1969) utilized the same corrected uranium-series methods to assign the Corral terrace to MIS 5e. 505 Lindgren (1889) first documented Punta Banda marine terraces and Allen et al. (1960) and Rockwell et al. (1989) mapped the lowest 13 and 12 terraces in detail, respectively. The Lighthouse terrace, the lowest mapped terrace (15 -17.5 m apsl shoreline angle; Figure 2), is well preserved on the south side of the peninsula, though discontinuous on the north side (Rockwell et al., 1989). Fifteen uranium-series ages on Balanophyllia elegans assign the Lighthouse terrace to MIS 5a, and extralimital northern 525 species in the faunal assemblage support this designation (Rockwell et al., 1989; Figure 4). A fragmented, narrow second terrace crops out across the Punta Banda peninsula with a shoreline angle of 22 m apsl (Rockwell et al., 1989; Figure 3). No radiometric ages exist for the second terrace, though stratigraphic relationships imply an age of MIS 5c (see discussion in Rockwell et al., 1989; Figure 5).

530
At Punta Cabras, Baja California, Mexico, Addicott and Emerson (1959) first documented a narrow, discontinuous marine terrace with inner edge elevations of 4.5 -17 m apsl ( Figure 2). No radiometric ages exist for this terrace, though a limited number of radiocarbon ages and extralimital northern species in the faunal assemblage of overlying marine and non-marine deposits support a terrace assignment to MIS 5a (Addicott and Emerson, 1959;Mueller et al., 2009; Figure 4). 535 Figure 6. MIS 5a (a-c) and MIS 5c (d-f) sea level indicator elevation quality ratings atop the Matplotlib Basemap Shaded Relief map (Hunter, 2007). For the five Iranian subsites we plotted the lowest quality ratings of those listed in Tables 1 and 2. 540

Pacific Coast Atlantic Coast and Caribbean
Far Field

Summary
Marine terraces comprise the entirety of MIS 5a/5c relative sea level indicators cropping out along the North American Pacific coast. While in principle marine terraces can serve as excellent quality 555 indicators (see 3 and 4), Pacific coast marine terraces receive quality ratings from very poor (1) to average (3) (Figure 6a,d; Tables 1 and 2). These ratings reflect four systematic uncertainties. First, no primary reference reports a terrace's indicative range, the distance between the storm wave swash height and the wave breaking depth (Vacchi et al., 2014;Rovere et al., 2016), and this precludes a calculation of indicative meaning. Given this absence, we used the IMCalc software to quantify the 560 indicative meaning of all Pacific coast marine terraces (WALIS RSL IDs 3473-3503; Lorscheid and Rovere, 2019). The arising indicative ranges tend to vary between 4 and 7.5 m which necessitates quality ratings of poor (2) (2) Average (3) Good (4) Far Field and topographic map measurement techniques before the widespread adoption of differential GPS means that literature-reported measurement uncertainties generally exceed ~3 m (as for the MIS 5a terraces at Palos Verdes Hills and San Joaquin Hills). Fourth, many regional terraces also crop out over 595 a range of elevations due to faulting or tilting, and this range further contributes to RSL uncertainty (such as for MIS 5a and 5c terraces at Newport and Brookings; MIS 5a terraces at Point Reyes, San Simeon, Point Buchon, Punta Cabras; and MIS 5c terraces at Santa Cruz (Highway 1) and Malibu). As all four systematic uncertainties apply to many Pacific coast marine terraces, and at least one uncertainty applies to all terraces, the Pacific coast yields quality ratings of very poor (1) to average (3). 600 From this we conclude that revisiting indicator elevation measurements with modern mapping methods could better constrain Pacific coast MIS 5a and 5c terrace elevations, indicative meanings, and RSL uncertainties.
We assigned the chronologies of the North American Pacific coast marine terraces ratings from good 605 (4) to poor (2) (Fig. 7a,d; Tables 1 and 2). The two methods that conferred good quality (4) ratings for this region include: first, reproducible, high precision uranium-series ages on solitary coral skeletal carbonate (as for MIS 5a terraces at Coquille Point, Santa Cruz (Davenport terrace), Palos Verdes Hills, San Nicholas Island, and Punta Banda) and, second, high-precision luminescence ages (as for the MIS 5a Point Reyes marine terrace). For chronologic methods that yielded age uncertainty beyond the 610 bounds of an MIS 5 substage-often arising from substrate experiencing open-system diagenesis-we applied an average (3) quality rating. Examples of this include uranium-series ages on coral (the MIS 5a terrace at San Joaquin Hills and the MIS 5c terrace at Malibu,) or molluscs (the MIS 5a terrace at Point Loma), cosmogenic ages (the MIS 5a and 5c terraces at Santa Cruz (Western and Wilder)), radiocarbon ages (the MIS 5a terrace at Punta Cabras), and luminescence ages (the MIS 5a terrace at San Simeon). 615 While Muhs et al. (2012) interpreted the MIS 5c and 5e uranium-series ages on corals from San Nicholas Island Terrace IIb as an indication of terrace reoccupation, which could afford a good (4) quality rating, here we assign Terrace IIB an average (3) quality rating given that these ages span a time interval longer than either individual MIS substage. Poor quality (2) chronology ratings arise from relative dating methods, such as AAR on molluscs (the MIS 5a and 5c terraces at Gaviota, and the MIS 620 5a terraces at Newport, Bruhel Point, Santa Rosa Island, and Oceanside), and terrace counting (MIS 5a and 5c terraces at Bruhel Point; the MIS 5a terrace at San Miguel Island; and the MIS 5c terraces at Cape Arago, Coquille Point, and Punta Banda). For this region, chronologic assignment by terrace counting is especially common for MIS 5c terraces with an adjacent, well-dated MIS 5a terrace. Nontraditional methods and maximum/minimum limiting ages confer a poor (2) rating for MIS 5a and 5c 625 terraces at Newport, Brookings, and Gaviota; the MIS 5a terrace at Cape Arago; and the MIS 5c terrace at Santa Cruz (Highway 1 terrace). Likewise, minimum limiting uranium-series age on mammal teeth and bone, and unreliable uranium-series dates on coral (see Hanson et al., 1992) assign Point Buchon a poor quality rating (2). For the North American Pacific coast, MIS 5a terraces generally received higher quality ratings than MIS 5c terraces (Tables 1 and 2).  (Figure 2). 27 uranium-series ages on corals assign the terraces at these four sites to MIS 5a (Cronin et al., 1981;Szabo, 1985;Wehmiller et al., 2004; see Figure 4). Further, five electron spin resonance ages and eight amino acid ratios at the Virginia Beach site support an age assignment of MIS 5 sensu lato, without a specific substage designation (Mirecki et al., 1995). No MIS 5c-equivalent coral terraces were recognized across this region. 655 Parham et al. (2013) utilized outcrops and sediment cores to map late Quaternary beach deposits, for which we focus on a deposit primarily composed of a thin veneer of sand and laminated sand. The sequence appears discontinuously through the study site at an elevation of 9 -14 m apsl ( Figure 2): east of the Chowan river, the deposit forms a prograding spit whereas further east of the Suffolk shoreline 660 the deposit forms a seaward thickening wedge, and the deposit is not well preserved in the northern and southern portions of the study area. Shelly marine material within the deposit was assigned to MIS 5a using calibrated amino acid racemization and optically stimulated luminescence dating (Figure 4). Two uranium-series ages now support an MIS 5a age assignment for these beach deposits (Wehmiller et al., 2021a). See Wehmiller (2021b) for a comprehensive database of AAR ages for the North American 665

Pamlico Sound, North Carolina
Atlantic coast.

Freeport Rocks, Texas
Simms et al. (2009) documented a sedimentary deposit within offshore sediment cores, referred to as the Freeport Rocks Bathymetric High, consisting of barrier island facies. The deposit appears at its 670 highest elevation within the core at 18.9 m bpsl (below present day sea level; Figure 2). The authors assign the Freeport Rocks Bathymetric High to MIS 5a based upon a single optically stimulated luminescence age (Figure 4).

Fort St. Catherine, Bermuda
A rich literature describes the evolution of the stratigraphic nomenclature of Bermuda, comprised of six 675 carbonate units separated by terra rosa paleosols (Land et al., 1967;Vacher and Hearty, 1989;Hearty, 2002). The marine member of the Southampton Formation, mapped at 1 m apsl (Vacher and Hearty, 1989), was tentatively assigned to MIS 5a or 5c using amino acid stratigraphy (Harmon et al., 1983;Hearty et al., 1992) though, subsequently, 24 uranium-series ages reported across multiple studies have indicated an MIS 5a age assignment (Harmon et al., 1983;Ludwig et al., 1996;Muhs et al., 2002; see 680 Figure 4). The Pembroke unit of the Rocky Bay Formation, previously classified as the Pembroke Formation (with a proposed alternate name of the Hungry Bay Formation), has generated more debate. Amino acid ratios on Poecilozonites support a MIS 5c age assignment (Harmon et al., 1983), while 700 whole rock amino acid ratios support an age assignment of MIS 5e . Four uraniumseries ages yielded one MIS 5e coral, two MIS 5c-aged corals, and one modern coral (Harmon et al., 1983; see Figure 5) though Vacher and Hearty (1989) argued that published uranium-series ages were unable to resolve an MIS 5 substage for the Rocky Bay Formation, and instead hypothesized that this unit represents MIS 5e. For the purposes of this review, the Pembroke unit chronology is included in 705 Figure 5, with both 5c and 5e age assignments included. No elevation was reported for the Pembroke unit. Moreover, controversy exists over the classification of both the Southampton Formation and the Pembroke unit of the Rocky Bay Formation. Harmon et al. (1983) hypothesized that both units formed from storm wave activity and, thus, the deposits do not serve as indicators of a sea level highstand. 710 Toscano and Lundberg (1999) supported this hypothesis, further specifying that the units were formed in the Holocene and incorporated corals of many different ages. Alternatively, other studies argued the location of Fort St. Catherine relative to the platform margin and the narrow range of MIS 5a ages on corals found within the Southampton Formation reveal that these units represent a sea level highstand (Vacher and Hearty, 1989;Ludwig et al., 1996;Muhs et al., 2002). Here we include both units within 715 the database.

725
Skeletal eolianites comprising Eleuthera Island are interpreted as eolian dunes and are separated from the underlying formations by paleosols (Kindler and Hearty, 1996;Hearty, 1998;Hearty and Kaufman, 2000). The authors report an inferred paleo-sea level of 0 -5 m bpsl, rather the modern elevation of the deposit (Hearty and Kaufman, 2000; Figure 2). The stratigraphic position, location, and amount of diagenetic alteration of the eolianite, along with whole rock amino acid ratios, assign the unit to MIS 5a 730 (Figure 4).

Berry Islands, Bahamas
Newell (1965) reported a single uranium-series age for a coral welded with caliche to a platform on Berry Island, which assigned the coral to MIS 5a. Neumann and Moore (1975) (Figure 3) as MIS 5 in age, though the range of ages could not assign the ridge to a specific substage highstand. Creveling et al. (2017) noted that an MIS 5c age assignment for the Berry Islands corals fits the the published age uncetainty ( Figure 5). Woodring et al. (1924) first documented the geology of the Northwestern Peninsula, which was 750 followed by further research summarized by Dodge et al. (1983). Dodge et al. (1983) mapped seven constructional coral reef terraces composed primarily of Acropora palmata, for which we focus on the lowest two, the Mole and Saint terraces. Dumas et al. (2006) published altimeter surveys for the Mole and Saint terraces (also referred to as T1 and T2), revising the previously reported inner edge elevations to 23 m apsl and 37 m apsl, respectively (Figures 2 and 3). Dodge et al. (1983) and Dumas et al. (2006) 755 argued that 15 uranium-series ages on corals support the conclusion that: (i) the Mole terrace formed during MIS 5a; (ii) the Saint terrace formed during MIS 5c; (iii) both terraces host corals reworked from the adjacent MIS 5e terrace (see Figures 4 and 5). Mesolella (1967) first mapped the coral reef terraces of the Christ Church and Clermont Nose traverses. 760 Bender et al. (1979) revisited nine reef tracts at Christ Church and seven reef tracts at Clermont Nose. 28 uranium-series ages on corals, along with 12 protactinium-231 ages, assign the Worthing and Ventor terraces at both traverses to MIS 5a and 5c, respectively (Broecker et al., 1968;Mesolella, 1969;Bender et al., 1979;Bard et al., 1990;Edwards et al., 1997;see Figures 4 and 5). The elevations of the Worthing and Ventor terraces are mapped at 3 m apsl and 6 m apsl at Christ Church and 20 m apsl and 765 30 m apsl at Clermont nose (Figures 2 and 3). Schellmann and Radtke (2004) mapped additional terraces on the south coast. The two sub-terraces T-1a1 and T-1a2 were collectively mapped at 2 -3 m apsl and assigned to MIS 5a based on two averaged ESR ages and one uranium-series age (Figures 2  and 4). The sub-terraces T-1b, T-2, and T-3 were collectively mapped at 4 -16 m apsl and assigned to MIS 5c based on thre averaged ESR ages and one uranium-series age (Figures 3 and 5). 770

Summary
The North American Atlantic coast and Caribbean sea level indicators, which consist of coral reef terraces, beach deposits, and terrestrial limiting beach ridges, received elevation quality ratings from good (4) to rejected (0) (Figure 6b,,e; Tables 1 and 2). Since none of the primary literature sources report indicative ranges for the relative sea level indicators along the North Atlantic coast and 775 Caribbean, we calculated these values with the IMCalc software and reported these for WALIS RSL IDs 3504-3508, 3511-3519, 3556, and 3983-3984 (Lorscheid and Rovere, 2019). The clear reference datum for the Berry Island and MIS 5a Christ Church coral reef terraces (Neumann and Moore, 1975;Bender et al., 1979) warrants the only good quality (4)  Catherine and Berry Island). Due to the absence of material available for numerical dating methods at Eleuthera Island, the relative dating methods warrants a quality rating poor (2). We advocate continued focus on sites with average quality ratings (3)-those that have substrate amenable to geochronology yet have imprecise ages-to improve chronologic assignments. Sherman et al. (2014) utilized offshore drill cores to document and date two submerged coral reef terraces on Oahu. A single uranium-series age assigned one terrace to MIS 5a; two uranium-series ages assigned a second terrace to MIS 5c (Figures 4 and 5), though distinct elevations were not reported for the MIS 5a and 5c terraces. Instead, a range in elevations from 20 -30 m bpsl was provided for both 830 terraces (Figures 2 and 3). Marquardt et al. (2004) utilized altimeter surveys to map eight emergent marine terraces and the overlying fossiliferous terrace deposits. The shoreline angle of the two lowest terraces found above the Holocene beach is mapped at 10 m apsl and 31 m apsl (Figures 2 and 3). Uranium-series and electron 835 spin resonance ages from a neighboring region assign the two terraces to MIS 5, and their stratigraphic position, assuming uniform uplift rate, further assigns the 10 m and 31 m terraces to MIS 5a and 5c (see discussion in Marquardt et al., 2004;Figures 4 and 5). Choi et al. (2008) utilized differential GPS to map wave-cut marine platforms in the Daebo-Gori region. 875

Daebo-Gori region, Korea
The T2 terrace, with a shoreline angle mapped at 8 -11 m apsl (Figure 2), is a wide and continuous surface, covered in sediment of up to 2 m thick; the T3b terrace, with a shoreline angle mapped at 17 -22 m apsl (Figure 3), is a narrow, lower bench of the T3 terrace, separated from the upper bench by a gently sloping riser, which in places makes it difficult to distinguish the two platforms. 16 optically stimulated luminescence ages assign the T2 terrace to MIS 5a (Choi et al., 2003; see literature 880 referenced in table 2 of Choi et al., 2008; see Figure 4); eight OSL ages and two paleomagnetic ages assign the T3b terrace to MIS 5c (Choi et al., 2003;Choi et al., 2008; see literature referenced in table 3 of Choi et al., 2008; see Figure 5). Ringor et al. (2004) mapped the coral reef terraces on Pamilacan Island, for which four uranium-series 885 ages on corals have assigned the unnamed broad terrace, with a shoreline angle mapped at 6 m apsl, and the more narrow and poorly developed terrace (also unnamed), with a shoreline angle mapped at 13 m apsl, to MIS 5a and 5c, respectively. Omura et al. (2004) mapped the coral reef terraces found along Panglao Island. The lowest well-890 developed terrace is mapped at 5 m apsl ( Figure 3) and is composed of two geomorphic units. The lower unit contains one uranium-series age on Porites which dates to late MIS 5e. Four uranium-series ages on Platygyra ryukyuensis assign the upper terrace, of which the inner margin is mapped at 5 m apsl, to MIS 5c ( Figure 5). Chappell (1974) first mapped the emergent coral reef terraces along the northern coast of the Huon Peninsula, for which we focus on the well preserved, laterally tilted Va and VIa terraces. 11 uraniumseries ages on corals and mollusks assign these terraces to MIS 5a and 5c (Veeh and Chappell, 1970;Chappell, 1974;Esat et al., 1999;Cutler et al., 2003;Figures 4 and 5). Chappell and Shackleton (1986) reported the reef crest elevations of the Va and VIa reef terraces to be 260 m apsl and 338 m apsl at the 900

895
Tewai section and 117 m apsl and 160 m apsl at the Kwambu section (Figures 2 and 3). Chappell et al. (1996) reconciled Huon Peninsula terrace highstands with oxygen isotope records for the interval spanning 70-30 ka. Ota and Hori (1980) proposed an initial chronology for the six lowest Quaternary terraces around 905 Hateruma Island following early mapping efforts discussed in Ota and Omura (1992). Omura (1984) published four uranium-series ages on corals that assigned terraces V and IV to MIS 5a and 5c, respectively (Figures 4 and 5). The updated elevation surveys of Ota and Omura (1992) mapped the maximum height of the terraces V and IV at 23 m apsl and 30 m apsl, respectively (Figures 2 and 3).

Atauro Island, East Timor
The lowest two coral reef terraces of Atauro Island, termed 1a and 1b, are mapped at inner edge elevations of 20 m apsl and 36 m apsl, respectively (see literature discussed in Chappell and Veeh, 930 1978;Chappell and Veeh, 1978; Figures 2 and 3). Two uranium-series ages on corals have assigned terrace 1b to MIS 5c ( Figure 5). While no radiometric ages exist for terrace 1a, Chappell and Veeh (1978) used the position of this terrace on age-height plots to infer an age of MIS 5a (Figure 4).

Spencer Gulf, Australia
Hails et al. (1984) utilized sediment cores and sonar to study shallow water stratigraphy in the Spencer 935 Gulf, for which we focus on the Lowly Point and False Bay formations. The Lowly Point and False Bay formations occur at water depths exceeding 14 m bpsl and 8 m bpsl, respectively (Figures 2 and 3). No radiometric ages exist for either formation; the stratigraphic position of the Lowly Point and False Bay formations between the MIS 5e age Mambray formation and the Holocene age Germein formation (see literature discussed in Hails et al., 1984), along with the extent to which the depositional events flooded 940 Spencer Gulf, support age assignments of MIS 5a and 5c (Figures 4 and 5).

Robe Range, Australia
Sprigg (1952) (Figure 3). Comparison between the local stratigraphy and established deep-ocean oxygen isotope records assigns the Robe II and Robe III shorelines to MIS 5a and 5c (Schwebel, 1984). Calibrated amino acid ratios on the Robe III barrier are numerically 950 indistinguishable from the adjacent Robe II barrier which hosts numerical AAR ages that fall within MIS 5a, though the authors note that the AAR method is not able to distinguish specific MIS substages (Blakemore et al., 2015). Based on anomalously young luminescence ages, debate exists in the literature as to whether the Robe II barrier records a true interglacial sea level highstand or formed as a result of local processes (Huntley et al., 1993a; see literature discussed in Huntley et al., 1993a;Banerjee et al., 955 2003;Murray-Wallace, 2018). We include Robe II in the database as an MIS 5a indicator for completeness, along with the associated anomalously young ages (Figure 4). Four luminescence ages, in conjunction with the stratigraphic position of the Robe III unit, supports an age assignment of MIS 5c (Huntley et al., 1993b;Huntley et al., 1994;Huntley and Prescott, 2001;Banerjee et al., 2003; Figure  5). 960 5.3.11 Makran Subduction Zone, Iran (Lipar, Ramin, Gurdim, Jask sites) Harrison (1941) first documented the marine terraces along the Makran subduction zone of Iran; subsequent authors extended the initial mapping, documenting the differential uplift and lateral tilting of 975 the terraces (Page et al., 1979;Reyss et al., 1998;Normand et al., 2019). Normand et al. (2019) revisited the terraces and provided updated elevations for the terraces found at the Lipar, Ramin, Gurdim, Jask sites. The majority of the T1 terraces, dated with three optically stimulated luminescence ages and a single uranium-series age to MIS 5a (Figure 4) terrace at the Ramin site, which hosts two luminescence samples dated to MIS 5e and 5a, respectively, concluding that the chronology implies a reoccupation of the T1 terrace during multiple sea level highstands ( Figure 4). The Lipar site also contains a T2 terrace mapped at 45 m apsl (Figure 2) which hosts a luminescence sample dated to MIS 5a ( Figure 4).

The Gulf of Gabes, Tunisia
985 Gzam et al. (2016) utilized high precision echo sounders to map two submerged beach ridges, with peak elevations of 8 m and 19 m bpsl (Figure 2 and 3). From their analysis of biocalcarenite development, the authors concluded that the ridge formation could indicate a rapid sea level transgression which, along with the stratigraphic position of the 8 m and 19 m ridges, supported their respective age assignments of MIS 5a and 5c (Figure 4 and 5). At present, no radiometric ages exist for these beach 990 ridges.

Summary
Sea level indicators in the far field encompass marine terraces, coral reef terraces, and both marine and terrestrial limiting indicators (shallow water facies and beach ridges, respectively), which have quality ratings between average (3) and rejected (0) (Figures 6c,f; Tables 1 and 2). As with the previous two 995 regions, we used IMCalc to quantify the indicative meaning of each relative sea level indicator in the absence of reported modern analog data (WALIS RSL IDs 3520-3537, 3557-3562, 3982;Lorscheid and Rovere, 2019). Most primary references do not report a clear reference datum, which limits the maximum quality rating to average (3). An average (3) rating for a coral reef terrace typically reflects both a precise measurement and a narrow indicative range (less than ~2 m) as calculated by IMCalc (as 1000 for MIS 5a and 5c indicators at Pamilacan Island and Hateruma Island and the MIS 5c indicator at Panglao Island). In contrast, as marine terraces typically have larger indicative ranges, an average (3) quality rating necessitates a precise elevation measurement (e.g., the MIS 5a and 5c indicators at Daebo-Gori region and the MIS 5a indicators at Lipar and Gurdim). Poor ratings (2) for marine and coral reef terraces arise from imprecise measurements intrinsic to survey methods such as altimeters and 1005 topographic maps, or from not reporting measurement uncertainty (such as the MIS 5a and 5c indicators at Bahía Inglesa, Oahu, Atauro Island, Tewai Section, Kwambu Section and MIS 5a indicators at Ramin and Jask). All sites in Tunisia and Australia receive rejected (0) quality ratings because they only provide minimum/maximum limiting constraints on paleo-sea level. In summary, we advocate for applying modern measurement methods (with more precise error estimates) to marine and coral reef terraces with present quality ratings of average (3) and poor (2) to refine indicative meaning.
The far field chronologies quality ratings range from good (4) to poor (2)   The global database of sea level indicators dated (or assigned) to MIS 5a and 5c presented in this paper 1085 reasonably covers the Pacific coast of North America (18 field sites), the Atlantic coast of North America and the Caribbean (9 field sites), and more sparsely covers the remaining globe (12 field sites). The broad geographic spread of the data allows for an increasingly resolved reconstruction of MIS 5a and 5c GMSL sea level, with especially good coverage in the near field of the North American Ice Sheets. Future efforts could expand this database by including additional indicators reported outside of 1090 North America, particularly those reported in non0English language journals.
Two complementary research themes rely upon MIS 5 relative sea level indicators with unequivocal substage chronology and robust elevation data. One focus leverages the spatial variation in reconstructions of local MIS 5a and 5c RSL elevations to constrain global geophysical models for 1095 glacial isostatic adjustment and, hence, to refine estimates of substage global mean sea level (GMSL; e.g., Lambeck and Chappell, 2001;Muhs et al., 2012;Creveling et al., 2017). The other focus deduces regional rates of tectonic motion from the vertical displacement of (MIS 5e) RSL indicators from the sea level at which they formed (e.g., Matthews, 1973;Chappell, 1974;Wehmiller et al., 1977;Muhs et al., 1990Muhs et al., , 1992bSimms et al., 2016). Numerous field and numerical 1100 analyses highlight the entanglement of these themes. Robust efforts to deduce MIS 5a and 5c GMSL from misfit analyses between field observational data and GIA models necessitate a quantitative correction for a site's tectonic uplift history (Creveling et al., 2015;Simms et al., 2016). Inasmuch as a vertical tectonic uplift correction requires a robust paleo-sea level reference datum, this reference datum should reflect the estimates of interstadial GMSL and melt-induced spatial variations in local sea level 1105 from glacial isostatic adjustment models (Creveling et al., 2015;Simms et al., 2016). Advances in each research theme enrich the other, and both rely upon RSL indicators with high quality elevation measurements and substage-resolution chronology.
Tectonic uplift-corrected RSL indicators in the near-to-intermediate field of the North American ice 1110 complex display distinct geographic trends arising from glacial isostatic adjustment Simms et al., 2016). North American Atlantic coast and Caribbean MIS 5a highstand elevations display a north-to-south latitudinal gradient that decreases by ∼30 m elevation (Cronin et al., 1981, Szabo, 1985, Bard et al., 1990, Cutler et al., 2003, Wehmiller et al., 2004Parham et al., 2013).  demonstrated that this trend reflects the glacio-1115 isostatic disequilibrium imposed by the forebulge of the Laurentide ice sheet. After correcting Atlantic Coast and Caribbean RSL inferences for glacioisostasy,  concluded that MIS 5a GMSL peaked ∼-28 m below present (with a similar value for MIS 5c).  predicted broadly consistent magnitudes, though narrower bounds, on MIS 5a and 5c substage GMSL as Lambeck and Chappell (2001) who reconstructed GMSL of 23-37 m and 18-30 m below present, 1120 respectively, from Huon Peninsula coral reef terraces. In contrast, tectonic uplift-and GIA-corrected MIS 5a and MIS 5c RSL indicators along the Pacific coast of the U.S. and Mexico reveal an opposing latitudinal gradient in local high-stand elevations from that observed on the North American Atlantic coast and Caribbean (Simms et al., 2016). On the basis of the North American Pacific coast geographic gradient, Simms et al. (2016) concluded that MIS 5 and 5c peak GMSL reached up to ~-15 m and ~-1135 10 m below present sea level, a conclusion in agreement with that of Muhs et al. (2012) who reconstructed peak GMSL elevations of -16 m and -9 m during MIS 5a and MIS 5c, respectively, based on tectonic uplift-and GIA-corrected RSL indicators at San Nicolas Island, California; the Florida Keys; and Barbados. 1140 The opposing latitudinal gradients in MIS 5a and 5c peak highstand elevations imposed by the peripheral bulge of the North American ice complex do not find reconciliation with conventional '1-D' glacial isostatic adjustment models that assume a depth-varying but laterally homogenous viscoelastic structure (Creveling et al., 2017). Notably, embedding an upper mantle viscosity in a GIA models to reconcile the highstand latitudinal gradient from one geographic region (i.e., the Pacific or Atlantic 1145 coast of North America) exacerbates the misfit of GIA predictions to the RSL indicators of the other region. Hence, GIA analyses that focus on a regional subset of global data produce GMSL estimates with systematic errors (hence the conflicting GMSL predictions of  versus Simms et al. (2016)). Creveling et al. (2017) promoted the adoption of a sensitivity analysis between globally distributed RSL indicators and GIA predictions that adopt viscosity models that honor the 1150 complexity in (North American) upper mantle viscosity. The resulting analytical workflow, applied to an unfiltered compendium of MIS 5a and 5c RSL indicators, yielded peak GMSL bounds of −18±1 m and −20±1 m for MIS 5a and MIS 5c, respectively; notably, repeating this sensitivity analysis on a RSL database filtered to include only those with high-quality (predominately uranium-series) chronology widened these bounds to −22±1 m and −24±2 m, respectively (Creveling et al., 2017). 1155 Continued refinement of MIS 5a and 5c peak GMSL and regional rates of Quaternary vertical tectonic uplift remains within reach. First, numerical models for glacial isostatic adjustment that adopt '3D' solid earth models with depth-varying and laterally homogenous viscoelastic structure promise to reconcile observed spatial gradients in RSL highstands and GIA model predictions (e.g., Latychev et al., 1160Latychev et al., 2005Clark et al., 2019). Such numerical advancements offer the possibility of refining GMSL estimates in the absence of further field data collection. Second, the quality ratings conferred above motivate the strategic re-surveying of a subset of MIS 5a and 5c field observations (see Sections 5.1.19, 5.2.10, and 5.3.13) in order that each site conform to the uniform approach to establishing the elevation and uncertainty of elevation measurements ages adopted for the World Atlas of Last Interglacial Sea 1165 Level (e.g, Rovere et al., 2016). The re-sampling and/or re-analysis of geochronological material may also refine the numerical ages adopted for the World Atlas of Last Interglacial Sea Level (e.g, Rovere et al., 2016). Importantly, this retroactive translation of MIS 5a and 5c RSL observations to rigorous sealevel index points (sensu Hijma et al., 2015) offers the paired promise of refining predictions of contemporaneous global mean sea level and vertical tectonic motion and the standardization of efforts 1170 to complete these research foci. Third, the proliferation of airborne LiDAR data can offer geoscientists a fresh perspective on the quantity and spatial relationships of purported terrace platforms (Bowles and Cowgill, 2012) that, once ground-truthed, may confer confidence in, or contradict, chronologies developed from terrace counting methods. In practice, simultaneous efforts to enact all three practices will enrich conclusions about MIS 5a and 5c GMSL bounds and the accompanying tectonic 1175 displacement of these RSL indicators.