Pliocene-Pleistocene Stratigraphy and Sea-Level Estimates, Republic of South Africa With Implications for a 400 ppmv CO₂ World

Abstract

The Mid-Pliocene Warm Period (MPWP, 2.9 to 3.3 Ma), along with older Pliocene (3.2 to 5.3 Ma) records, offers potential past analogues for our 400-ppmv world. The coastal geology of western and southern coasts of the Republic of South Africa exposes an abundance of marine deposits of Pliocene and Pleistocene age. In this study, we report differential GPS elevations, detailed stratigraphic descriptions, standardized interpretations, and dating of relative sea-level indicators measured across ~700 km from the western and southern coasts of the Cape Provinces. Wave abrasion surfaces on bedrock, intertidal sedimentary structures, and in situ marine invertebrates including oysters and barnacles provide precise indicators of past sea levels. Multiple sea-level highstands imprinted at different elevations along South African coastlines were identified. Zone I sites average +32 ± 5 m (six sites). A lower topographic Zone II of sea stands were measured at several sites around +17 ± 5 m. Middle and late Pleistocene sites are included in Zone III. Shoreline chronologies using ⁸⁷Sr/⁸⁶Sr ages on shells from these zones yield ages from Zone I at 4.6 and 3.0 Ma and Zone II at 1.04 Ma. Our results show that polar ice sheets during the Plio-Pleistocene were dynamic and subject to significant melting under modestly warmer global temperatures. These processes occurred during a period when CO₂ concentrations were comparable to our current and rapidly rising values above 400 ppmv.

Key Points

• The relative tectonic stability of South Africa provides an optimal location for study of sea-level history
• There is an abundance of excellent Pliocene and Pleistocene (5–1 Ma) marine exposures along the semiarid coastlines
• Precise measurements of these sea-level indicators provide significant implications for the historical rise in CO₂ levels above 400 ppmv

Plain Language Summary

The ancient coastal deposits of the Republic of South Africa provide an opportunity to examine past climatic effects on ice sheets and sea levels during the Pliocene. This was the last time atmospheric CO₂ was similar to our current and rising level of ~415 ppmv. Indicators of past sea levels were measured at a number of natural rock outcrops that date between about 1 and 5 million years old. Our studies have found that average sea levels ranged between 15 and 30 m higher than present. Postdepositional effects by crustal movements caused by plate tectonics or ice sheets are minor but have yet to be fully quantified. Regardless, our baseline field data have important implications for the magnitude of our present and future climate and sea-level changes.

1 Introduction

The rapid rise in atmospheric CO₂ to levels exceeding 400 ppmv (Monastersky, 2013) is driving an urgent need to better understand sea-level changes and ice sheet dynamics associated with past warmer-than-present geological intervals. The Mid-Pliocene Warm Period (MPWP; 3.3 to 2.9 Ma), and indeed most of the Pliocene, was the last time Earth experienced CO₂ levels around or above 400 ppmv (Fischer et al., 2018; Pagani et al., 2010; Stap et al., 2016). As such, the Pliocene may serve as an analogue for the Earth's near future climate changes. Understanding the potential response of Greenland and Antarctic ice sheets to Earth warming is critical, as there is strong evidence from modeling studies that the ocean-cryosphere system is dynamic (e.g., DeConto & Pollard, 2016; Pollard et al., 2015; Pollard & De Conto, 2009) and could lead to rapid ice collapse and global sea-level rise in this century (Golledge et al., 2019).

The measurement, interpretation, and dating of relative sea-level (RSL) indicators is essential to derive more accurate estimates of eustatic sea-level (ESL) highstands during the Plio-Pleistocene and the related loss of polar ice. In this paper, we present the results of extensive field mapping along different coastal sectors of the Republic of South Africa (hereafter RSA, see map in Figure 1). The primary objective of our study was to characterize high-resolution sea-level stratigraphic sections. Integral to establishing RSL and indicative meaning is a clear understanding and accurate interpretation of the sedimentary facies and stratigraphic succession (as illustrated, for example, in Figure 4) along the coastline. Once optimal field samples from key sites were collected, ages were determined using ⁸⁷Sr/⁸⁶Sr geochronological methods.

Figure 1

(a) Study area, with location of subregions discussed in the text (b–f). Upper right inset: Location of the study area within a larger geographic context. background imagery source: Esri, DigitalGlobe, GeoEye, i-cubed, USDA FSA, USGS, AEX, Getmapping, Aerogrid, IGN, IGP, swisstopo, and the GIS user community. Inset in panel (a) was drawn from global self-consistent hierarchical high-resolution geography, GSHHG. Version 2.3.7 June 15, 2017 (Wessel & Smith, 1996).

While the long-term uplift (Cretaceous to Cenozoic) history of the southern tip of the African continent is characterized by phases of uplift and quiescence (e.g., see Baby et al., 2020; Marker & Holmes, 2010; Dauteuil et al., 2015; Walford & White, 2005), the choice of RSA as study area resides in the fact that the intracontinental plate, trailing edge coastal margin of the southerly African Plate has been considered to be relatively tectonically stable throughout the Pliocene and Pleistocene (Chen et al., 2014; Kounov et al., 2015; Roberts et al., 2012). Data on linear inverse modeling of drainage networks showed that, in the Northern and Western Cape Provinces of RSA, average uplift rates for the last ca. 15 Ma were, on average, 0.011 ± 0.02 mm/a (Rudge et al., 2015).

For the RSA, glacial isostatic adjustment (GIA) models predict only slight departures from eustasy (Raymo et al., 2011; Rovere et al., 2014) and are characterized by relatively small uncertainties when different mantle viscosity profiles are used to predict GIA for Pleistocene and Pliocene time scales (Rovere et al., 2014). Also, dynamic topography (DT), which is recognized to cause significant uplift or subsidence at Pleistocene and Pliocene time scales even at passive margins (Austermann et al., 2017; Moucha et al., 2008), appears to have a minor yet, currently uncertain effects along the western and southwestern coasts of South Africa. Interpreted peak sea levels (i.e., reduced global ice volume) are indicated on the basis of δ¹⁸O around 5.1 Ma (T7/T5), the “Climatic Optimum” at 4.0 to 4.4 Ma (CN5-Gi25), the “MPWP” at 3.25 Ma (M1) to 2.94 (G17), 2.75 Ma (G7), and 2.5 Ma (MS99?; Lisiecki and Raymo 2005). More directly, Naish et al. (2009) have identified and dated glacial advances and retreats (high sea levels) from glacio-marine sediments around Antarctica and provide direct chronological evidence of an obliquity dominated sea-level cyclicity during the Pliocene, parallel to we observe on the shoreline during the MPWP. In the geologic record of RSA, we infer that the highest, longest, and latest highstand events should be discernible in the field during a generally declining sea-level succession from the Pliocene to the Pleistocene. Our inference, despite being largely unprovable (i.e., the older deposits are destroyed), is based on logic. It stands to reason that if broad terraces are formed, and sea level amplitude is only 10–20 m that geomorphic surfaces must be reoccupied multiple times—the “buzz saw” effect (see section 5). Older deposits would be reworked and degraded; replaced by more recent ones.

2 Paleo Sea Level Proxies in the RSA—Previous Work

The study of coastal marine deposits in RSA expanded through the 20th Century, driven by the desire to document the evolution of large vertebrates (including hominids) on the continent of Africa, as well as exploration and exploitation of diamonds and other resources in coastal sedimentary deposits. Early works by Rogers (1905), Krige (1927), and later Sisser and Dingle (1981) provided a geological framework for many works to follow along the west coast of RSA. Krige's (1927) project embraced the entire South African coast. Coinciding with Krige's study, placer diamond deposits were found in sediments of the West Coast marine terraces, which spurred a surge of research bearing on their geomorphology, biostratigraphy, and sedimentology (Haughton, 1931) particularly in the area of Hondeklip Bay (see location in Figure 1b), in Namaqualand. This groundwork was later strengthened by Carrington and Kensley (1969), who expanded and integrated the knowledge of the biostratigraphy and chronology of these deposits. In particular, they focused on the use of marine marker fossils as biostratigraphic indicators (mainly the bivalves Isognomon sp. the surf clam Donax sp., and the limpet Fissurella sp.) to distinguish and correlate sea-level events. Terraces of marine origin at elevations around 90 (not further addressed), 50, 30, and 18 m were previously described along the west coast of RSA (Roberts & Brink, 2002) and more recently, further north in Namibia (Dauteuil et al., 2015; Stollhofen et al., 2014; Zhu, 2016). Hereafter, we provide a brief account of the levels of marine terraces that are described in the literature.

2.1 Marine Terraces at 50, 30, and 21–18 m

Pickford (1998) interpreted what he defined as a “50 m Package” (i.e., a terrace level containing shell accumulations) to be late Miocene to earliest Pliocene (7–5 Ma) based on mammal fossils “similar” to those from Langebaanweg (Roberts et al., 2011). Pickford (1998) further correlated a series of lower terraces, which included a “30-m Package,” with the Plio-Pleistocene between 3 and 2.5 Ma. This correlation was based in part on the presence of Equus, species of which colonized the Old World including Africa during the late Pliocene after about 3.4 Ma (Forstén, 1992; Jónsson et al., 2014; Lindsay et al., 1980).

Molluscan “zone fossils” in marine deposits from these terrace levels were similarly used to distinguish events during the Plio-Pleistocene, particularly making use of mollusk bivalve shells of the genus Donax. Carrington and Kensley (1969) correlated D. haughtoni with a “45-50 m strand line” where this taxon occurs “… in extraordinary abundance.” The thicker and wider shell of D. rogersi is instead found in younger and lower marine deposits that Carrington and Kensley (1969) considered as Pleistocene in age.

Carrington and Kensley (1969) found evidence for temporally distinct marine transgressions of ~30 and ~20 m in excavations at Hondeklip Bay, and both levels contained the zone fossil D. rogersi. They considered these shorelines to be early-mid Pleistocene in age. Pether (1986a, 1986b, 1994a, 1994b) regarded the evidence for the +20-m level as spurious and recognized only the “30 m Package.”

More recently, Roberts et al. (2011) summarized the Pliocene stratigraphy in the context of the stratigraphy and paleontology of Langebaanweg, a site located in the West Coast National Park. At this site, we have further documented the major units and the differential GPS (hereafter dGPS) elevations of their stratigraphic boundaries. The phosphatic deposits of the upper Varswater Formation contain marine microfossils (Dale & McMillan, 1999) that are regarded as marine/estuarine in origin at +30.7 m (elevation measured by our dGPS) and suggest the proximity of paleo-sea level during deposition (Hendey, 1981a, 1981b, 1983). Various authors suggest the rich vertebrate fossil deposits at Langebaanweg originate from an RSL highstand peaking at 50 m the Early Pliocene around 5.5 to 4.8 Ma (vis-à-vis Pickford, 1998), while the 30-m terrace is suspected to correlate with a Plio-Pleistocene transition after 3.3 Ma (Roberts et al., 2011).

Key marine biostratigraphic fossil indicators have been used to distinguish sea-level events of the Miocene and Pliocene. During our investigation, we only documented marine deposits or terrace landforms below +50 m. In our results section, we describe key sites yielding information on past RSLs that we surveyed at elevations between +41 and +10 m, which are the basis for the following results and discussion.

2.2 MIS 5e and MIS 11 RSL Indicators

The measured elevation of younger Quaternary sea-level highstands (i.e., MIS 5e and MIS 11) in RSA provide important benchmarks on long-term vertical land motions. In the Western Cape Province, Carr et al. (2010) report deposits dated to MIS 5e from Swartvlei and Groot Brak estuaries (respectively, located at 81 and 130 km from Still Bay [see location in Figure 1f]) suggesting that the highstand was between +6 and +8.5 m above present sea level. These elevations do not change much further East. In fact, at a locality close to Nahoon River (620 km east of Still Bay), Jacobs and Roberts (2009) report a “Shelly, Pebbly beachrock” between +2.5 and +6 m above present sea level, dated to 117.3 ± 6.2 ka with optically stimulated luminescence (their sample NHN1). At Cape Agulhas, Carr et al. (2010; see location in Figure 1e) surveyed a gravel beach and an overlying sandy shoreface facies at elevations up to +7.5 m. These facies were dated to 118 ± 7.2 ka with optically stimulated luminescence. At Hoë Walle, 35 km West of Cape Agulhas, the same authors report an aeolianite at +4 m dated 104 ± 7 ka. Aeolianites few meters above sea level at Kraal Bay (130 km North of Cape Town and ca. 25 km South of Saldanha, see Figure 1d for location) were also attributed to MIS 5e (see Roberts, 2008 for details). These elevations of RSA MIS 5e sea levels are in general agreement with global averages (Hearty et al., 2007; Kopp et al., 2009; O'Leary et al., 2013).

Roberts et al. (2012) documented an MIS 11 sea stand at +13 m, subsequently corrected to account for GIA to a range of +8 – 11.5 m (Chen et al., 2014). These estimates of MIS 11 sea level are in general agreement with Bermuda, now dGPS surveyed at +18.3 m (Hearty et al., 1999; Olson & Hearty, 2009) later GIA corrected to +9 – 14 m (Raymo & Mitrovica, 2012) to take into account the location of Bermuda on the forebulge of the former Laurentide ice sheet. These precisely measured and well-dated interglacial sequences provide evidence that little tectonic displacement has occurred since the late and middle Pleistocene.

3 Methods

The definition of paleo RSL from field proxies (here defined as RSL indicators) is related to the quantification of three main properties. These were first formalized by Shennan (1982) and van de Plassche (1986) for Holocene sea level studies (see a recent review in Shennan et al., 2015 and applications in Khan et al., 2019) and successively refined and adapted also to standardize RSL databases of older time periods (Rovere et al., 2016). In the simplest definition, each indicator used to reconstruct a paleo RSL should have three main characteristics: (i) its elevation must be measured as precisely as possible and referred to a known tidal datum or a geoid model.; (ii) it should be possible to quantify its indicative meaning (i.e., its relationship to the paleo sea level, composed by Indicative Range and Reference Water Level as described in Shennan et al., 2015); (iii) it should be possible to attribute an age to the facies indicating the paleo RSL. In this study, these three aspects were addressed as described below.

3.1 Elevation Measurements

Elevations reported in this study were measured with a high-accuracy differential GPS (hereafter dGPS) receiving real-time Omnistar HP corrections (nominal 2σ or 95% accuracy of 0.1 m). In general, due to the presence of cliffs or other natural features masking the optimal satellite reception of the GPS signal, the average accuracy we obtained in the field is slightly lower than the nominal one: the measurement error associated with our points averages 0.6 m (1σ). All our elevation measurements were referred to the South Africa Geoid 2010 (“SAGEOID2010,” Chandler & Merry, 2010). The discrepancy of this geoid with respect to mean sea level was estimated, in general, to be below 0.1 m. The measurement error associated with each RSL indicator was calculated as the square root of the sum of the squares of dGPS elevation measurement error and the geoid accuracy (0.1 m).

3.2 Quantification of the Indicative Meaning

Concerning the interpretation of RSL indicators, we subdivide them in different types (Table 1). To each indicator, we assign values of reference water level (RWL) and indicative range (IR) that were calculated either from comparison with a modern analogue (where available) or using the IMCalc tool (Lorscheid & Rovere, 2019). These values (RWL and IR) together compose the indicative meaning and were used to calculate the elevation of the former RSL and the uncertainty associated with it (see formulations in Rovere et al., 2016). Landforms or deposits indicating a submerged environment or a terrestrial one have been referred to as, respectively, marine or terrestrial limiting points (i.e., the sea level was respectively above, or below, the measured elevation of the landform or deposit). Because the approach of Lorscheid and Rovere (2019) assumes that both tidal ranges and wave properties in the past were similar to modern ones, we increased the final paleo RSL uncertainties of a further ±20% to account for possible deviations from this assumption.

Table 1. RSL Indicators Surveyed This Study, Reference Water Level (RWL) and Indicative Range (IR) Associated to Each Indicator

RSL indicator	RWL	IR	Definitions
Shore Platform (marine abrasion platform of older bedrock)^a	(MHHW + Poe)/2	MHHW − Poe	MHHW, MLLW—Mean Higher High Water, Mean Lower Low Water. Obtained from 19-year predictions calculated using the OTIS software (Egbert et al., 1994; Egbert & Erofeeva, 2002).
Marine geomorphic terrace/sedimentary constructional platform	(SWSH + db)/2	SWSH − db
			Ud—Upper distribution. Upper distribution of intertidal organisms measured in the field (see Laborel & Laborel-Deguen, 2005 for a review of similar indicators).
Transition between subtidal/intertidal (or shoreface/foreshore) facies	(db + MHHW)/2	MHHW − db
			db—Breaking depth. The depth at which significant waves interact with the bottom.
Horizontal barnacle/oyster band on cobble/boulder pavement above a shore platform	(Ud + MLLW)/2	Ud − MLLW
			Poe—Platform outer edge. Seaward edge of a shore platform, defined by Kennedy (2015) as: “the point where active erosion of the bedrock ceases.”
			SWSH—Storm wave swash height. The maximum elevation reached by extreme storm waves on the beach.
			All values above were calculated using the IMCalc software (Lorscheid & Rovere, 2019).
Marine limiting	Ophiomorpha burrows, subtidal cross-bedding and coarse pebbles Contact between marine and terrestrial deposits Subtidal cross beds and boulder with Patella sp.
Terrestrial limiting	Deposits that were surely emplaced above MHHW (e.g. high-energy storm deposits)

• ^a In the case of Donkergat (ZDG), we use the elevation of the shore platform inner margin measured in the field instead of MHHW.

3.3 Field Research Protocols

Each of the potential sites was initially reconnoitered to assess the quality of the exposures based on stratigraphic completeness and detail, abundance of potential sea-level related information, and presence of marine fossils for Sr dating. Once the optimal sections were identified, GPS and dGPS coordinates were taken. Key geomorphic, stratigraphic, and sedimentologic features, particularly major contacts and RSL indicators within the sections, were photographed for broad and close up features. Our team of five marine geology specialists (with a combined century of experience) followed outcrops on foot, in the field for 100 s of meters or longer in some cases, until the outcrop played out in all dimensions. While in the field, we observed, recorded, discussed, and regularly challenged our own collective observations/interpretations of the outcrops. The key stratigraphic sections were vertically scaled to ±0.1 m using dGPS elevations. Key features (RSL indicators, contacts, sedimentary facies transitions, etc. [e.g., Figure 4]) were logged electronically, photographed, and recorded in field notes. Once documentation of the sections was achieved, samples of sediment and fossils were collected from identified stratigraphic units and saved in labeled and sealed plastic zip storage bags for future analyses.

Field sites and samples ID names all carry the preface “Z” for ZA ( = RSA). In the Northern Cape Province, near the village of Hondeklip, site and sample names are taken directly from those reported on quarry site maps provided by the De Beers Consolidated Mines. In other areas, “Z” prefixes two or three-letter abbreviations of nearby geographic place names from topographic maps, followed by numbered sections from within that area.

3.4 Chronological Attribution and Strontium Isotope Dating

In order to obtain accurate Sr isotope stratigraphy (SIS) ages of the shells collected from RSA paleo-shorelines, we applied a suite of criteria to assess the degree of postdepositional alteration of the shells. For the sample ages reported here, all chemical processing occurred at Lamont Doherty Earth Observatory (LDEO) and ⁸⁷Sr/⁸⁶Sr isotopes were measured using thermal ion mass spectrometry (TIMS) at SUNY Stonybrook.

Bivalve shells were initially screened based on visual inspection of shell thickness, the presence or absence of microborings and/or Fe and Mn staining (del Río et al., 2013; McArthur et al., 1994). The most promising samples were fragmented, sonicated with quadruple distilled water, and then imaged using an optical microscope with CCD camera and an ASPEX Express scanning electron microscope (SEM). These images were assigned a preservation score ranging from “1” (no visible alteration) to “3” (significant alteration discernible) for each sample for both optical and SEM images, based on irregularities in crystal structure and above screening criteria (similar to Cochran et al., 2010). Minor and trace elements were also measured in duplicate for Sr, Mn, Mg, Al, Fe, and Ba on a Thermo iCap Q quadrupole ICP-MS at LDEO. High values of Mn, Mg, and Fe have been shown to be indicative of postdepositional alteration in marine carbonates (Gothmann et al., 2015; McArthur, 1994; Shen et al., 1991). We use these data to further eliminate samples that are likely influenced by diagenesis and recrystallization.

We used the “leach method” described by Bailey et al. (2000), which measures four or five sequential leachates on a single sample and which assumes that the more loosely bound diagenetic ⁸⁷Sr/⁸⁶Sr reservoirs will tend to be partitioned into the first leaches (Bailey et al., 2000; Li et al., 2011; Sandstrom et al., 2014). This advanced leaching protocol has the additional benefit of providing minimum or maximum SIS ages on the samples, based on trends within each leach set. If altering fluids are from a granitic source with high radiogenic ⁸⁷Sr/⁸⁶Sr ratios, SIS ages will be a minimum, as opposed to maximum ages if the sample is altered by a basaltic or older carbonate source with lower ⁸⁷Sr/⁸⁶Sr ratios. The trends in our sample leach sets always suggested a regional radiogenic (granitic) derived altering fluid, making SIS ages appear younger with more diagenesis (Sandstrom et al., 2014).

Sr was isolated and dried down using typical separation techniques with Eichon exchange resin and ⁸⁷Sr/⁸⁶Sr ratios were measured dynamically on an IsotopX Phoenix TIMS at Stonybrook University. The long-term instrument 2σ uncertainty determined by the Sr isotope standard NBS 987 (calculated annually over 3 years) varied between ±0.0000126 (2014; n = 25) and ±0.0000079 (2016; n = 40; 17–11 ppm, respectively). All Sr was corrected for mass fractionation on the ⁸⁶Sr/⁸⁸Sr ratio of 0.1194 and adjusted by the NBS 987 standard value of 0.710248. After Sr isotope analyses of leaches, sample preservation was further evaluated from variation between ⁸⁷Sr/⁸⁶Sr measurements on each shell leach set (with smaller variation between initial and inner leaches generally associated with better preservation). SIS numerical ages were derived using the LOWESS Sr marine calibration curve from McArthur et al. (2012). Sample and terrace ages are based on averaging the better-preserved inner leaches (by filaments measured), age uncertainties are calculated using 2σ standard error of the mean. Only samples with an average preservation score less than “2” were used in terrace average.

4 Results

All our RSL data (including quantification of indicative meaning) and SIS ages are available in the PANGAEA repository as Hearty et al. (2019).

4.1 Study Sites

The areas surveyed in this study are the western and southern coastlines of RSA (Figure 1). Overall, we surveyed the coastline between Hondeklip Bay in the Northern Cape Province to Groot Brak River near the town of Mossel Bay on the southern coast of the Western Cape Province. In the Western Cape Province, at the mouth of the Olifants River, our surveys include an area called “Avontuur” (same site as ZCP in this study, Roberts et al., 2011; Figure 1c) that yielded several excellent natural cliff exposures and sections. Three areas were also surveyed along the Western Cape Province coastline between Cape Agulhas and Port Elizabeth, near Stilbaai (Figure 1f). The abundance of RSL indicators found at numerous sites (Table 2) identified in this study contrast markedly with the paucity of existing published data available from coastal Pliocene elsewhere in the world (Hearty et al., 2012, 2013; James et al., 2006; Kaufman & Brigham-Grette, 1993; Raymo et al., 2011; Rovere et al., 2014, 2015). In Table 3 and Figure 2, we present an overview of the samples that were dated using Sr isotopes. In the following paragraphs, we present the description of each RSL indicator and the rationale for our RSL interpretations.

Table 2. Overview of RSL Indicators, their location (including elevation) and quantification of Indicative Meaning (Reference Water Level, RWL and Indicative Range, IR)

Site code	Section	Geographic name	RSL indicator (see Table 1 for quantification of the indicative meaning)	Lat [deg]	Lon [deg]	Elevation [m]	Elevation error [m] 1σ	RWL [m]	IR [m]	Paleo RSL [m]	Paleo RSL uncertainty [m] 1σ
Northern Cape Province—Hondeklip Bay
ZSN29	Section 1	Somnaas	Horizontal barnacle/oyster band on cobbles/boulders above a shore platform	−30.1699	17.24864	33.8	0.3	0.1	1.4	33.7	0.9
ZSN29	Section 2	Somnaas	Ophiomorpha burrows, subtidal cross-bedding and coarse pebbles	−30.1707	17.2504	32.8	0.5	Marine limiting
ZSLT23		Koingnass	Storm deposit	−30.2927	17.3049	41.1	0.3	Terrestrial limiting
ZLK_N14.3		Langklip	Horizontal barnacle/oyster band on cobbles/boulders above a shore platform	−30.3417	17.3180	28.8	0.4	0.1	1.4	28.7	1
ZLK_N6	Section South	Langklip	Contact between marine and terrestrial deposits	−30.3543	17.3004	14.8	0.3	Marine limiting
ZLK_N6	Section North	Langklip	Contact between marine and terrestrial deposits	−30.3532	17.3026	12.6	0.2	Marine limiting
ZLK_N3		Langklip	Transition between subtidal/intertidal (or shoreface/foreshore) facies	−30.3772	17.3230	30.0	0.2	−1.4	4	31.4	2.4
Western Cape Province—Olifants River mouth
ZCP	Section 1	Cliff Point	Shore Platform (marine abrasion platform of older bedrock)	−31.5874	18.1194	29.1	0.3	−0.6	2.5	29.7	1.5
ZCP	Section 2	Cliff Point	Transition between a subtidal (Upper shoreface) to intertidal facies	−31.5873	18.1196	33.9	0.3	−1.2	3.7	35.1	2.2
ZCH	Section 1/2	Channel	Transition between a subtidal (Upper shoreface) to intertidal facies	−31.5141	18.0511	15.0	0.3	−1.2	3.7	16.2	2.2
ZDG	Section 4	Donkergat	Shore Platform (marine abrasion platform of older bedrock)	−31.80061	18.2313	14.0	0.4	−0.2	3.2	14.2	2.0
Western Cape Province—West Central coast
ZPPH		Prospect Hill	Subtidal cross beds and boulder with Patella sp.	−33.0012	17.9200	11.5	0.5	Marine limiting
Western Cape Province—South coast
ZBR		Breasdorp	Subtidal sediments	−34.6748	19.9245	21.9	0.2	Marine limiting
ZDH		De Hoop	Sediments deposited in the swash zone	−34.4544	20.3977	26.3	1.0	0.0	6.7	26.3	4.2
ZST	Section1	Still Baai	Shore Platform (marine abrasion platform of older bedrock)	−34.3559	21.4183	17.8	0.4	−0.4	2.2	18.2	1.4
ZST	Section2	Still Baai	Shore Platform (marine abrasion platform of older bedrock)	−34.3600	21.4174	15.8	0.5	−0.4	2.2	16.2	1.5
ZST^a	Section3	Still Baai	Shore Platform (marine abrasion platform of older bedrock)	−34.3437	21.4084	22.3	4.1	−0.4	2.2	22.7	5.1

• Note. The paleo RSL uncertainty column includes an additional 20% uncertainty that accounts for possible difference in tidal and wave parameters since the Pliocene.
• ^a At this site, it was not possible to achieve a fix to the Omnistar satellites; hence, the accuracy of the vertical measurement is greatly reduced (4.1 m, ±1σ).

Table 3. Summary of SIS Ages Obtained From the Samples Collected in this Study

Sample metadata			Preservation index					Sr isotope stratigraphy ages
Sample name	SESAR ISGN ID	Bivalve classification	Optical score	SEM score	T.E. scores (mg/Mn/Fe)	Leach variation Score^a	Preservation score (Avg.)	n (filaments)	^87Sr/^86Sr (Corr. 987)	2σ Error^c	Shell leach variation (ppm)	Mean SIS age (ma)	Max SIS age (ma)	Min SIS age (ma)
ZBR1-A	IEMRS003Z	Glycyermis sp.	1	1	1.0	1	1.0	4	0.7090612	0.0000027	1.11	3.10	3.38	2.87
ZBR1-D	IEMRS0042	Ostrea sp.	1	1	1.3	1	1.1	6	0.7090629	0.0000043	3.62	2.95	3.38	2.66
ZDH1-C (Z3-C)	IEMRS0046	Glycyermis sp.	2	N.A.	N.A.	2	2.0^b	1	0.7090837	0.0000079	10.99	2.02	2.27	1.83
ZCP2-B	IEMRS004P	Ostrea sp.	2	2	2.0	1	1.8	4	0.7090454	0.0000033	5.86	4.63	4.78	4.44
ZCP2-E	IEMRS004S	Ostrea sp.	2	2	2.0	N.A.	2.0^b	2	0.7090997	0.0000126	N.A.	1.67	1.94	1.43
ZDG4a-A	IEMRS004A	Donex sp.	1	1	1.0	N.A.	1.0	3	0.7091390	0.0000056	N.A.	0.97	1.07	0.86
ZDG4b-C	IEMRS004I	Dosinia sp.	1	2	1.3	2	1.6	3	0.7091316	0.0000160	11.48	1.10	1.37	0.79
ZSLT23	IEMRS004W	Ostrea sp.	3	2	2.3	3	2.6^b	2	0.7090643	0.0000126	44.94	2.85	4.19	2.23
Average terrace SIS age^d
Site Code	Elevation (msl)	Location	Preservation Score (Avg.)	n samples	n filaments	^87Sr/^86Sr (Corr. 987)	2σ SEM (sample)	2σ Total Uncertainty^e	Mean SIS Age (Ma)	Max SIS Age (Ma)	Min SIS Age (Ma)
ZBR	21.90 ± 0.20	Breasdorp	1.1	2	10	0.7090622	0.0000027	0.0000039	3.01	3.40	2.73
ZDG	13.96 ± 0.43	Donkergat	1.3	2	6	0.7091353	0.0000082	0.0000084	1.04	1.18	0.88
ZCP2	33.70 ± 0.29	Cliff Point	1.8	1	4	0.7090454	0.0000033	0.0000054	4.63	4.87	4.28

• Note. For more detailed metadata regarding each sample, please refer to the SESAR IGSN ID (http://www.geosamples.org/).
• ^a Leach variation score is based on total leach variation (ppm) of initial vs. inner leaches, with <11 ppm = “1,” 11–17 ppm = “2,” and >17 ppm = “3.”
• ^b Samples with an average preservation score ≥“2” are considered unreliable and not included in terrace age calculations.
• ^c Samples with ≥3 filament measurements have age uncertainties calculated as 2σ SEM (bolded), samples with less (italics) have uncertainty based on 2σ of external standard NBS987.
• ^d Terrace ages based on averging inner leach filaments from samples that passed our screening criteria (preservation score <2).
• ^e Total uncertainty is based oncombining the 2σ SEM of sample filaments and lowess SIS curve uncertainties. This is the error used in calculating terrace ages; N.A.—Not measured.

Figure 2

Sr isotope stratigraphy (SIS) ages from the best-preserved bivalves on three different terraces in RSA. Colored circles with black error bars are based on individual shell averages (from an average of inner leach individual filaments) with errors based on 2σ standard error of the mean of those filaments. Shaded ellipses (behind circles) are terrace ages derived from averages of all screened sample filaments. The size of the ellipse is based on combined uncertainty from 2σ SEM of sample filaments and LOWESS SIS curve error. Yellow bars are exceptionally warm periods and interglacials where sea level is thought to be the same or higher than present (de Wet et al., 2016; Dutton et al., 2015; Lisiecki & Raymo, 2005).

4.2 Northern Cape Province—Hondeklip Bay

Along the Northern Cape Province (Figure 1b), during a cessation of operations at De Beers Consortium Mines, we visited a considerable number of exposures over several tens of kms in quarries and test pits. The exposures typically were cut through terrestrial colluvium and marine shoreface and foreshore deposits down to the bedrock abrasion platforms. Raw diamond deposits are excavated in gullies, moulins, and potholes in bedrock. In many cases, these sections expose the entire thickness of the marine formation in pits and quarry walls. This optimal setting allowed us to work on “clean” exposures (Figure 3d) devoid of vegetation, colluvium, slumping, and infrastructures that often obscure details and architecture of Quaternary and older shorelines.

Figure 3

Details of stratigraphic sections surveyed along the Northern Cape Province–Hondeklip Bay area. (a) Subtidal-intertidal contact at ZLK_N3. (b) Detail of the subtidal deposit in a; (c) large Ostrea shell at ZLK_N6. (d) Stratigraphy of site ZLKN_14.3, showing an imbricated Boulder pavement deposited on a wave abrasion surface, capped by beach sands and terrigenous deposits. (e, f) details of D, showing barnacles (e) and oysters (f) in living position on the boulders. (g) Modern boulder pavement provides an analogue of site ZLKN_14.3 and several other sites. (h) Boulder at site ZSN29, with fossil barnacles attached. (i) Overgrowth of barnacles on oysters at site ZSN29. (j) Ophiomorpha burrows at site ZSN29.

4.2.1 ZSN29

Two stratigraphic sections were surveyed at this site. As in ZLKN_14.3 (see below), a distinct marine abrasion surface is carved in gneiss, forming the base of the section. Decimeter-scale cobbles with pebbles and sand (Figure 3h) on the abrasion surface are covered with articulated in situ Ostrea shells. The Ostrea themselves are further encrusted with the barnacles forming a sub-horizontal band surveyed up to +33.8 ± 0.3 m (Figure 3i). A few tens of meters laterally within the same quarry pit, the cobble zone transitions to a sandy sedimentary environment exposing about 1.5 m thick, high-angle subtidal, cross-bedding with abundant Ophiomorpha burrows (Figure 3j) and coarse pebble zones, reaching a maximum elevation of +32.8 ± 0.5 m. A 0.5 m thick layer of concave-down (swash-oriented) and well-worn bivalve shells overlie the trough cross beds. The swash-oriented shells are buried by more than a meter of beach or eolian sand with terrestrial snail fossils (Trigonephrus?) and higher calcrete and terra rossa soil at +36 to +37 m, respectively. These sedimentary and biological RSL indicators combined are of the highest quality and indicative of a shallow marine environment.

4.2.2 ZSLT23

This site is exposed in an excavated trench test pit well above and landward of the several other investigated Hondeklip sites and thus represents the highest elevation fossiliferous marine site we studied in RSA. The base of the section contains largely disturbed/collapsed sandy sediment, presumably eolian capped by a deeply reddened paleosol about 0.3 m thick. Overlying is an imbricated boulder pavement with clasts ranging from 0.1 to >1.0 m and matrix of fossiliferous sand and gravel rises to +41.1 ± 0.3 m (Table 2). A supratidal or transitional eolian structureless sand 1–2 m thick and is capped by a reddish orange palaesol (0.3 – 0.5 m thick) and younger massive sands and brown soils. We interpret this facies as the uppermost reaches of the beach and/or storm deposit, therefore a terrestrial limiting (i.e., this section, at the time of formation was probably above Mean Higher High Water). An abraded and highly altered Ostrea shell from this deposit was measured using the leach method and showed older SIS ages with each leach, indicative of highly radiogenic ⁸⁷Sr/⁸⁶Sr alteration fluids. From this leach trend and degree of weathering, we can only assign a strict minimum SIS age of 2.23 Ma to the terrace with a large uncertainty (2.23 to 4.19 Ma) due to analytical uncertainty and diagenetic alteration of the shell sample (Table 3); the actual terrace age is probably significantly older than 2.23 Ma based on morphostratigraphic context.

4.2.3 ZLK_N3

A succession of high-angle subtidal trough cross beds transitioning to planar beach beds is exposed at ZLK_N3. This transition represents the most accurate paleo RSL indicator in this section measured at +30 ± 0.2 m (Figure 3a). This shallowing-upward sequence is truncated and incised by a small runnel or gully filled with highly abraded, concave downward bivalve shells characteristic of a wave swash environment (Figure 3b). Both subtidal and intertidal beds contain oyster and bivalve mollusk shells.

4.2.4 ZLK_N6

Two sections, North and South, were logged at this location. The base of the South section exposes an assemblage of multiple storm (?) beds containing coarse abraded shelly material with numerous disoriented valves of the oyster Ostrea and abundant whale bones (Figure 3c). There are no precise RSL indicators in these two sections. Although it could represent an upper-shoreface or near-foreshore environment (with whale bones resulting, for example, from strandings in shallow water or on the beach), we opt to characterize this site as a marine limiting. As elevation of this site, we use that of a sharp erosional contact between subtidal marine beds and terrestrial deposits at +14.8 ± 0.3 m. A similar complex succession 50 m from the South section occurs at North section with subtidal trough cross beds and Ophiomorpha burrows. Within the subtidal beds, an Ostrea valve measuring over 55 cm long (Figure 3c) was observed. The subtidal unit is overlain by a series of alternating sandy and coarse storm beds (possibly deposited on the shelf) similar/equivalent to the adjacent southern site. At a sharp erosional contact, subtidal cosets are truncated by terrestrial deposits at +12.6 ± 0.2 m.

4.2.5 ZLKN_14.3

At this site, a wave abrasion surface on gneissic bedrock (Figure 3d) is mantled by a submeter-scale boulder pavement with abundant attached oysters (Figure 3f). A subhorizontal band of barnacles can be traced continuously along the boulder deposit (Figure 3e) at +28.8 ± 0.4 m. About 1 m of structureless beach sand above the boulders grades upward into terrigenous deposits and a paleosol. The lack of sedimentary structures in this sandy unit may be the result of bioturbation by nearshore and land organisms. This geomorphological and palaeo environmental setting of this site is analogous to modern intertidal shore platform and boulder fields in the adjacent modern shoreline, where we measured a modern band of barnacles at +0.8 m, slightly below the high tide mark measured at +1.8 m on the same modern shoreline (Figure 3g).

4.3 Western Cape Province—Olifants River Mouth

4.3.1 ZCP1/2

The ZCP site (Figure 4) at Cliff Point is located north of the Olifants river mouth (Figure 1c). Based on similar elevation and stratigraphic components, we correlate Cliff Point with sites surveyed in the Northern Cape Province (e.g., ZLK_N14.3; ZSN29) near Hondeklip. We measured a distinct wave abrasion platform carved in the bedrock, composed of Neoproterozoic Gariep Supergroup metasediments (Pether et al., 2000; Roberts et al., 2011), gently sloping towards the sea. The modern abrasion platform is shown in Figure 5a. The highest measurable point on the platform is +29.1 ± 0.3 m (Figures 5a, 5b and 4). The platform surface is mantled with reworked pebbles and boulders up to 1 m in diameter, many encrusted on their upper surfaces by marine calcareous algae and serpulid tubes (Figure 5d). The boulder beds are overlain by a 5- and 6-m thick, shallowing-upward foreshore sedimentary sequence. Nearby, we surveyed a weakly bedded subtidal unit, composed of coarse sands and gravels, 2–2.5 m higher than the platform surface, which is then succeeded by prominent high-angle trough cross beds. Above a sharp transition at +33.9 ± 0.3 m (Figure 5c), the intertidal beds are capped by a coarse cobble and shell-rich bed about 1.5 to 2.0 m thick, interpreted as the swash zone. The highest capping shell bed, rising to near +38 m as backshore supratidal facies, is composed almost entirely of Donax rogersi or D. haughtoni, many of which are articulated, but chaotically distributed and disoriented in a terrigenous sandy matrix, unlike their living habitat in the lower surf zone. However, the field interpretation of the marine and supratidal facies appears to have caused some of the discrepancy between highstand elevations. Our dGPS elevation of the abrasion platform differs from Roberts et al.'s (2011) “Avontuur” (same site) measure by less than a meter (+30 m vs +29.1 m), and we interpreted that the highest storm deposits marking the highest reach of the sea is about +38 m, in agreement with Roberts' measurements. Nonetheless, we interpret the sharp shoreface-foreshore facies transition at +33.9 m (Figure 4) as the best RSL indicator in the sequence. The SIS age from the best preserved Ostrea shell at this deposit yielded an average age of 4.63 Ma (4.28 to 4.87 Ma) (Table 3, Figure 2). This sample had a shallow downward sloping Sr leach trend (less than 6 ppmv ⁸⁷Sr/⁸⁶Sr isotope variation across the first to last leach) indicating good preservation despite relatively elevated imagery and elemental scores. This shallow dipping leach trend is consistent with a radiogenic ⁸⁷Sr/⁸⁶Sr regional alteration fluid, causing sample SIS ages to appear younger with diagenesis.

Figure 4

Annotated photo of key ZCP1–2 with labels of various physical boundaries and representative sedimentary shoreline facies, rising from bedrock abrasion platform to soils. Refer to Table 1 for corresponding information on RSL indicators.

Figure 5

Details of stratigraphic sections surveyed in the Western Cape Province–Olifants River mouth area (a) modern and paleo abrasion platforms at site ZCP1. (b) Detail of the abrasion platform at ZCP1. (c) Transition between subtidal beds and cobble-shell deposit at ZCP2. (d) Encrusting marine calcareous algae on cobbles at ZCP2. (e) Sandy-pebbly layers at ZCH1/2. (f) Modern abrasion platform at ZDG. (g) Paleo abrasion platform at ZDG. (h) Sand with abundant rhizomorphs and fossil termite structures (dashed line) among articulated valves of D. rogersi at site ZDG1–4.

4.3.2 ZCH1/2

The sites we surveyed at ZCH, Channel Site, are stratigraphically complex and more challenging to interpret. However, some important sea-level data can be derived from two neighboring sections. At ZCH1, a horizontal marine abrasion platform (with borings of marine organisms) is formed in altered gneiss at circa +14 m and blanketed with a coarse cobble and pebble zone about 20–30 cm think (Figure 5e). The alternating sand and pebble zones above are interrupted by more indurated lenses of finer sand. Subtidal trough cross beds rise to +15.0 ± 0.3 m and, together with the abrasion platform at +14 m, show that a former RSL was close to this elevation. At this site, a 1.5 m thick bed of D. rogersi in a coarse pebble-sand matrix is exposed above a second sharp erosional contact. The shells are roughly oriented concave down in the lower part, but chaotically distributed in the upper few decimeters, up to +16.1 ± 0.3 m. Most of the lower part of the section ZCH2 exposed in a nearby gully is covered except for the upper part subtidal beds that are similar to ZCH1. The chaotic bedding with D. rogersi is repeated up to +18.6 ± 0.5 m where it is capped by terrestrial colluvium.

4.3.3 ZDG1–4

At Donkergat, four adjacent sections (ZDG1–4) can be represented as a composite. A marine abrasion platform at +14 ± 0.4 m, carved into reddish Paleozoic sandstones, is mantled by large (some >1 m) boulders, cobbles in a coarse matrix of shelly sands (Figures 5f and 5g). At the inner margin of the active marine abrasion platform adjacent to this section, we measured the platform at +1.4 m. Numerous bivalves in this marine sand, including D. rogersi at +15.5 m are articulated indicating shallow subtidal conditions. Small alternating lenses of indurated finer sand and coarser layers with surf-oriented bivalves rise up to +18.7 m. Above the densely packed intertidal, oriented shells in beach beds, lies a deeply reddened sand with abundant rhizomorphs and fossil termite mounds and 0.5 to 1.0 m thick beds of chaotically bedded and containing numerous articulated valves of D. rogersi rising to +19.3 m (Figure 5h). As in other cases, we consider these beds to be the product of a severe storm or tsunami that ripped up the shallow subtidal and beach environment, transporting the living surf clams well up into the supratidal zone; soon thereafter being recolonized by termites and land plants. Two bivalves (a Donax and Dosinia sp.) were dated from this deposit and both passed our screening tests. The average SIS age from those two shells (six measurements) is 1.04 Ma (0.88 to 1.18 Ma) (Table 3, Figure 2). This terrace age thus corresponds with the mid-Pleistocene and overlapping within uncertainty to the interglacial MIS 31, which has been shown to be an exceptionally warm period (de Wet et al., 2016).

4.4 Western Cape Province—West Central Coast

4.4.1 ZPPH

An informative shallowing-upward shoreline succession is observed in an active quarry at ZPPH (Prospect Hill), near the city of Saldanha (Figure 1d). At the base of the quarry exposure, a large granite boulder (~1.5 m diameter) is surrounded by high concentrations (compared to other parts of the exposure) of Patella/Fissurella (many in situ cemented on the boulder surface) and rock grazing gastropods (Figures 6a and 6b). This boulder in a shallow subtidal environment would function as a breakwater, “attracting” coarser sediments and shells, as well as the limpets attached to the granite higher in an intertidal zone. In a 3 and 4 m-thick sequence, subtidal trough cross beds around the boulder transition to pebbly planar beach beds at +11.5 ± 0.5 m and upward into a structureless (bioturbated), pedogenic sand with abundant pulmonate gastropods (Trigonephrus globulus). A second, higher aeolianite several meters thick, on a reddish weathering surface, contains large festoon cross-bedding and the same land snails at the upper surface.

Figure 6

Western Cape Province—west central coast area. (a) Overview of the site ZPPH with subtidal-beach transition at +11.5 ± 0.5 m. (b) Limpets including Fissurella broken from the boulder surface at ZPPH.

4.5 Western Cape Province—South Coast

4.5.1 ZDH

In the region near De Hoop Nature Preserve, a broad marine terrace at +25 – 33 m extends several tens of kilometers inland from near the modern coastline (Figure 7a). At its seaward edge, the terrace is mantled by a relatively thin “caprock” measured at +26.3 ± 1 m, composed of fossiliferous marine calcareous grainstone and biorudstone (Figure 7b) with occasional large preserved shells including abundant, ~10 cm size, concave-down oriented Glycymeris (Figure 7c), indicating a swash zone depositional environment. With no other sedimentologic constraints, we interpret this deposit as forming from the Storm Wave Swash Height to the depth of breaking of modern significant waves in this area. Using as RSL indicator the elevation of the inner margin of the large marine terrace (+28.3 ± 1.0 m), Rovere et al. (2014) calculated that this site represents a paleo-RSL at +29.8 ± 1.8 m, which we recalculate in this study (with a slightly different interpretation of the IR) to +26.3 ± 4.2 m (Table 2).

Figure 7

Western Cape Province—south coast area. (a) Long and wide Pliocene marine terrace at ZDH. The dashed line indicates the approximate inner margin. (b) “caprock” composed of fossiliferous marine calcareous grainstone and bio-rudstone (dashed line) at ZDH. (c) Large, concave-down oriented Glycymeris at site ZDH (arrows). (d) Overview of ZBR1 site. (e) Marine shells, many articulated in a matrix of fine marl and in variable in orientation at ZBR1, indicating deposition in a few meters of water.

4.5.2 ZBR1

Along the open terrace plain south of Bredasdorp (located 50 km southwestward on the same marine terrace plain as ZDH, Figure 7d), we discovered a moderately indurated marine surface at +21.9 ± 0.2 m near the roadside. However, unlike De Hoop, the shells of ZBR1 were in a matrix of fine marl and in variable in orientation, with many valves articulated in living position (Figure 7e). The random orientation of the shells and occasional articulation would likely be the result of active bioturbation on a living shallow sea floor. Rovere et al. (2014) interpreted this facies as marine limiting. We note that the same study interpreted the inner margin of the scarp at ZBR1 representing paleo SL elevations at +24.02 ± 1.51 m, while here we rely only on the stratigraphy of the site to interpret the sea level information. In fact, we interpret the roadside outcrop to indicate a deeper offshore subtidal marine sand/marl with fossils deposited in water depths generally below wave base. This interpretation better reconciles RSL at this site with that derived from the nearby site ZDH. A Glycymeris shell from ZBR1 produced a highly robust SIS age of 3.10 Ma (2.87 to 3.38 Ma), while an Ostrea shell from the same deposit recorded a SIS age of 2.95 Ma (2.66 to 3.38 Ma)  (Table 3, Figure 2). Variation between the leach sets are 1.1 and 3.6 ppm, respectively, demonstrating excellent preservation. The average SIS age of the terrace based on these two shells (10 measurements) is 3.01 Ma (2.73 to 3.40 Ma)  (Table 3, Figure 2). A large part of this uncertainty is based on the shallow slope of the SIS curve over this time period, as the actual ⁸⁷Sr/⁸⁶Sr leach measurements show little variation (Table 3).

4.5.3 ZST1 to 3

Three areas were surveyed along the southern coastline between Cape Agulhas and Port Elizabeth. Near Stilbaai, steep, and vegetated outcrops in three separate road cuttings yielded limited datable shell material, but at these sites, we could document wave abrasion surfaces in weathered and unstable bedrock of deformed Paleozoic shale averaging at +17.8 ± 0.4, +15.8 ± 0.5, and +22.3 ± 4.1 m (the larger error range at this latter site is due to a poor GPS precision at the time of measurement). On some of these terraces, we sampled small samples of weathered shell hash deposited among a coarse cobble bed generally less than 0.5 m thick. Unstructured marine sand and silt 1–3 m thick cap the exposed sections.

5 Discussion

5.1 Plio-Pleistocene RSL Indicators in South Africa

Over the last decade, there has been a growing inventory of preserved field proxies related to Pliocene sea-level highstands (Dumitru et al., 2019; Grant et al., 2019; Raymo et al., 2011; Rovere et al., 2014, 2015). The arid western and southern coasts of RSA offer nearly ideal conditions for the formation and preservation of fossil coastal deposits. Here, we document numerous Pliocene and Pleistocene sites (Tables 2 and 3). The large number of RSL indicators we surveyed confirms previously published observations of at least two major terraces of different ages between +32 and +17 m along the Northern and Western Cape Provinces, in general agreement with the classic study of Carrington and Kensley (1969) and, more recently, Roberts et al. (2011).

Some of the sites reported here (e.g., ZLK_N_14.3 and ZSN29) have preserved very precise fixed biological RSL indicators that constrain former RSL positions with submeter accuracy. Dumitru et al. (2019) document similar precision from the physical phreatic overgrowth of Pliocene speleothems in Mallorca (Spain, NW Mediterranean Sea), yet sea-level estimates involve large background uncertainties (~68%) when isostatic adjustment, DT, and uplift rates especially in an active tectonic plate zone as in the Mediterranean. While most RSL indicators we report are of high-quality and precisely measured, we also recognize that there might be factors affecting our quantification of the indicative meaning. In fact, the precision of almost all our indicative meaning estimates depend, ultimately, on measured or modeled tidal or wave data. In order to account for this possible bias, we added a further uncertainty of 20% to our field data. Of note is that there are active debates on whether paleo waves (for MIS 5e, see Hansen et al., 2016; Hearty & Tormey, 2017, 2018; Rovere et al., 2017, 2018) were stronger than modern ones. Also, tidal ranges are naturally subject to greater or lesser amplitudes when sea levels change (Hill, 2016; Lorscheid et al., 2017; Wilmes et al., 2017). The quantification of such changes with hydrodynamic models is, to some extent, possible within the Holocene, where shelf bathymetric changes are relatively well constrained (e.g., Hill et al., 2011) and where high-resolution, fine sedimentary successions are available (e.g., Roep & Beets, 1988). Expanding the concepts already employed for Holocene and Pleistocene tidal and wave energy changes also to older periods is surely a promising avenue for future research.

We also remark that, among our sites, only two (ZCP Section 2 and ZDG) qualify as valid RSL indicators following the definition of Shennan (1982) and van de Plassche (1986), mostly due to dating constraints. In fact, finding and identifying well-preserved samples for Sr isotope dating are always a challenge with Plio-Pleistocene aged carbonate shorelines. There is also an unavoidable error of 0.11 to 0.47 Ma for ⁸⁷Sr/⁸⁶Sr dating of fossil materials due to the flatness of the Sr-isotope curve during the Pliocene period (McArthur et al., 2012). Microsampling combined with a range of preservation screening protocols (this study) within single organisms adds needed confidence to the robustness of SIS age estimates. A third site dated with SIS is ZBR (elevation +21.9 ± 0.2 m), that is interpreted in this study as marine limiting, that is, deposited below mean lower low water. Correlating ZBR with ZDH, located 50 km to the East and resting upon the same extensive terrace (Rovere et al., 2014), it is possible to assign the ZBR age to the paleo RSL at ZDH (+26.3 ± 4.2 m), as shown in Figure 8b.

Figure 8

Overview of Plio-Pleistocene RSL indicators surveyed, interpreted and dated in this study. (a) Elevation and (where applicable) paleo RSL estimates for the sites described in the text (Table 2). (b) Lisiecki and Raymo (2005) delta ¹⁸O stack compared with ages and elevations obtained in this study. The graph shows the correlation between ZDH and ZBR as per Rovere et al., 2014.

In addition to the RSL information we provide in this paper, we note that a few sites contain chaotic deposits, mainly containing the surf clam Donax, often articulated but generally set in rubified, terrigenous sands often among fossil termite mounds (Figure 5h). These deposits (ZCP1, ZCH1/2; ZDG1) have elevations ranging from +38 to +18 m, and we propose they are most likely the result of rapid sea-level shifts, intense storms or tsunami events.

5.2 Topographic Zones

Based on our research, we describe three main topographic zones in which a number of sites from our study occur. These arbitrary zones do not necessarily define isolated sea-level events, but reflect simply the topographic zones by elevation in which different sites occur. We refer to the zones from highest elevation to the lowest as Zone I (+25 – 35 m), Zone II (+15 – 20 m), and Zone III (that we attribute to mid to late Pleistocene; <14 m). The RSL indicators we measured in RSA and the correlated ages and paleo RSL interpretations are summarized in Figure 8.

5.2.1 Topographic Zone I (+25–35 m)

This zone includes sites ZSN 29, ZLK_N14.3, ZLK_N3, ZCP, and ZDH (Figure 8a) and appears to be distributed along the Northern and Western Cape Provinces. One section at ZCP (Cliffs Point) yielded a Sr age of 4.63 Ma (4.28 to 4.87 Ma), which appears to closely correspond with the mid Pliocene Climatic Optimum (4.0 – 4.4 Ma, Figure 8b). A broad terrace along the Southern coast includes both DeHoop (ZDH) and Bredasdorp (ZBR) sites, which yielded marine facies from two sites between +28 and >+22 m; the latter associated with a Sr age of 3.01 Ma (2.73 to 3.40 Ma). Another site, in the Northern Cape Province (ZSLT23), yielded only a minimum age (older than 2.23 Ma) and represents a terrestrial limiting point, but is in agreement with the overall characterization of Zone I with early and late Pliocene events.

Topographic Zone I appears to correlate with a broad terrace 850 km north of our northernmost site (ZSN29), in Namibia. There, Stollhofen et al., 2014 report a marine terrace level at +33.8 m. However, these authors invoked a tectonic element to explain this high sea-level position. Along the Roe Plains of southwestern Western Australia, James et al. (2006) described the nature of the Pliocene Roe deposits and estimated the maximum sea-level position at +30 m. In the southeastern United States, Rovere et al. (2015) tracked and documented the dGPS elevation the Pliocene (MPWP?) “Orangeburg Scarp” and equivalents from Georgia to Virginia (Dowsett & Cronin, 1990). Indeed, the geomorphic sea-cliff scarp measured elevations of ~37 m in Georgia, ~60–80 m in South and North Carolina, and ~80 m in central Virginia (Rovere et al., 2015). As we anticipated, the Pliocene of the southeast United States highlights a more extreme example of postdepositional effects (±20 m on same scarp), particularly GIA and DT attributed to the proximity of ice sheets and other tectonic effects.

5.2.2 Topographic Zone II (+15–20 m)

In both Northern and Western Cape Provinces, RSL indicators were also measured at several sites around 15–20 m (Figure 8a). One site, ZDG, is associated with a Sr age of 1.04 Ma (0.88 to 1.18 Ma). In the same area, 10 km from this site, ZCH1/2 can be also attributed to Topographic Zone 2. The two sites measured at Still Baai (ZST, Figure 8a), may be correlated with ZDG, but caution is advised in such correlation as there are more than 540 km of coastline separating ZDG and ZST. In any case, Topographic Zone II would yield an apparent correlation with MIS 31, that is usually dated at 1.08 – 1.06 Ma (Lisiecki & Raymo, 2005).

5.2.3 Topographic Zone III (<15 m)

Zone III incorporates later Pleistocene sites reported in literature (Carr et al., 2010; Roberts et al., 2012), with MIS 11 and MIS 5e coastal deposits found along the coastlines we surveyed at, respectively, +13 – 14 m and +4 to +8 m.

5.3 The Pliocene Sea Level “Buzz Saw”

Pliocene shorelines were constructed under obliquity-dominated (41 kyrs) sea-level regime (Lisiecki & Raymo, 2005). Naish et al. (2009) showed that high-frequency, low-amplitude SL oscillations of ca. ~13 m over glacial–interglacial cycles persisted between 3.3 and 2.5 Ma (Grant et al., 2019). It is likely that oscillations of similar magnitude persisted throughout most of the Pliocene, for a duration of perhaps 3 million years. This resulted in the formation of vast, broad terraces tens of km wide, and in some cases, hundreds of km long (James et al., 2006; Kendrick et al., 1991). Examples such features are the De Hoop coastal plain (Figure 7a) and similar coastal features in Western Australia and the U.S. East Coast (Rovere et al., 2014, 2015).

Hearty et al. (2012, 2016) characterized this process of terrace cutting as a sea level “buzz saw.” In morphological contrast, eccentricity-dominated (100 kyrs) mid to late Pleistocene highstand events (high amplitude, low frequency) for the past 500 ka incised much narrower terraces and benches over typically short, 10–15 ka highstand intervals. Given the sheer morphological scale of 10s to 100s of km of Plio-Pleistocene terraces and deposits around the world it is postulated that the highstand events that we have identified are among some of the higher and later of the potentially large numbers of highstand deposits that occurred in this broad topographic zone. Erosional processes likely to have occurred would include repeated transgressions, reoccupation of fledgling terraces, erosion and removal of older terrace sediments, and their replacement with contemporaneous deposits.

The Pliocene sea level “buzz saw” may appear to be at odds with the terrace erosion modeling results of Trenhaile (2014), who conclude that “The large oscillations of the middle to late Quaternary were more conducive to erosion than the smaller oscillations of the Pliocene and early Quaternary …” Volumetrically, the large sea-level oscillations of the later Pleistocene may indeed have eroded more material; however, in terms of geomorphic expression, the Pliocene buzz saw processes certainly created the larger and more impressive landforms by comparison. We highlight that, in this regard, our study area and the results we present in this study may represent an ideal benchmark to test different erosion model parameters.

5.4 Long-Term Vertical Land Motions

The high-resolution data we report in this study represent RSL estimates: This means that they are still uncorrected from any quantifiable departure from eustasy that might have affected their elevation since they were deposited. Even considering tectonics as negligible, we cannot directly discount potentially relevant effects of DT in our study areas. Rovere et al. (2014) highlight that even at passive margins and with highly precise RSL indicators, the large uncertainty in DT models make assumptions on the ESL more challenging in the Mid Pliocene Warm Period (ca. 3 Ma).

We find a striking match between one of our dated records (ZCP Section 2, SIS age = 4.28 to 4.87 Ma) and the oldest phreatic overgrowth on the Mallorca speleothems recently reported by Dumitru et al. (2019; AR-02, U–Pb age = 4.39 ± 0.39 Ma, 2σ). With a full account, and given uncertainties related to postdepositional displacement of the Mallorcan site, Dumitru et al. (2019) calculated that the global mean sea level at this time (without correction for thermal expansion) was between +10.6 and +28.3 m above present (16th and 84th percentiles uncertainty bounds). For the sake of discussion, we offer this assumption: the predicted GIA reported by Rovere et al. (2014) for the MPWP (ca. 3 Ma, −3.6 ± 1.2 m) is not significantly different from GIA at the time of ZCP formation. Using these values of ESL, GIA, and RSL obtained at ZCP (35.1 ± 2.2 m), we calculate that a first-order estimate of long-term postdepositional vertical land movements in the area of interest is in the order of 4.2 m/Ma (50th percentile, with 33rd–66th percentile bounds 3.5-4.8 m/Ma, see Rovere (2020) for methods). We note that rate is slightly lower than the one proposed by Rudge et al. (2015) over the last 13 ± 5 Ma for Hondeklip Bay (ca. 160 km to the North, 8 ± 3 m/Ma) and one order of magnitude lower than the uplift rate calculated by the same authors for Saldanha Bay (ca. 160 km to the South, 20 ± 10 m/Ma).

Our calculation implies that while numerical models of DT in our study area are discrepant (e.g., see Austermann et al., 2017; Moucha et al., 2008; Spasojevic & Gurnis, 2012), there is evidence only for limited uplift, in the range of few meters per million years. If assumed linear through time, the rate calculated here would affect only minimally MIS 5e (125 ka) and MIS 11 (400 ka) (on average, less than 2 m). The MPWP, dated at ZBR (and correlated with ZDH RSL reconstructed at +26.3 ± 4.2 m, Figure 8b), would have been displaced by 12 m, and the MIS 31 RSL we dated at ZDG (RSL = 14.2 ± 2) would have been displaced, on average, by about 4 m. Refining the long-term uplift rate in the area will surely allow to better reconstruct paleo global mean sea level estimates from the RSL data presented in this study.

6 Conclusion

In this study, we reported RSL indicators from the Northern and Western Cape provinces in South Africa. While only three of 17 surveyed sites yield reliable ages (plus one minimum age), these are among the few well-dated, precisely measured and interpreted RSL indicators for the Pliocene and Early Pleistocene. In fact, we document that paleo RSL imprints were left along these shorelines at two prominent topographic zones, at +25 – 35 m and +15 – 20 m. The highest (Zone I) appears correlated to multiple Pliocene sea levels (ca. 3–5 Ma), while the second (Zone II) with a peak sea level during the early Pleistocene (ca. 1 Ma, possibly coinciding with MIS 31). A third topographic zone (Zone III, below +14 m) has conserved the imprints of late Pleistocene sea level changes (MIS 5e and MIS 11, as reported in Carr et al., 2010 and Roberts et al., 2012).

In the Western Cape Province, the morphology of the broad terrace associated with Zone I matches similar landforms at different sites (e.g., U.S. East Coast, or Australia, Rovere et al., 2014). Here, we define this as a sea level erosional terrace, that was produced by repeated small-amplitude sea-level oscillations (Grant et al., 2019). The results of this study represent an ideal real-world benchmark as the science on long-term landscape erosion models progresses (de Gelder et al., 2020).

The Plio-Pleistocene paleo RSL estimates presented here represent a robust data set against which future studies of long-term tectonic uplift or global mean sea level can be based. The overlap between our oldest RSL indicator (ZCP) and the oldest RSL indicator reported by Dumitru et al. (2019; both with average ages between 4.4 and 4.6 Ma) allowed us to speculate on the possible long-term vertical land motions in our study area, which would be relatively low (3.5–4.8 m/Ma). While caution should also be adopted in propagating long-term rates linearly through time (Stocchi et al., 2018), propagating these rates onto the two younger shorelines dated in this study (3 Ma and 1 Ma, ZBR/ZDH and ZDG, Figure 8b) sustains the recent results by Grant et al. (2019), who limited the upper bound of Pliocene sea level rise to less than +25 m. The same calculations indicate that MIS 31 may have peaked at a similar elevation to MIS 11 (+9 – 14 m, Raymo & Mitrovica, 2012).

The glaciological implications of this study are significant. The minimum sea-level elevations of Zone I of ca. +27 m (that would be only marginally affected by DT under our assumptions) implies that melting of Greenland, West Antarctica, and marine-based ice sheets of East Antarctica (about a third of all polar ice) occurred during a Pliocene 400 ppmv world. Confirmation of significantly higher sea-level elevations would involve the retreat of additional land-based Antarctica ice. During the middle Pleistocene, sea level also rose to +9 to +17 m in both MIS 11 and 31, despite the fact that CO₂ concentrations probably did not exceed 300 ppmv. Our study confirms that ice sheets are highly vulnerable under warming climates and are likely to experience dynamic changes when subject to climatic parameters of today and those predicted for decades and centuries to come.

Acknowledgments

Dr. David Lester Roberts (Dave) was instrumental in unraveling the late Cenozoic natural history of South Africa (and elsewhere) over the past decades. He made significant discoveries in ancient rock record, palaeoclimate, SL, megafauna, human footprints, and origins in South Africa, Gabon, and Madagascar. Our collaboration with Dave in the field was friendly and adventuresome but safe, inspiring and highly informative, thanks to his extensive knowledge of the region. His contributions were instrumental in the major findings of this paper. We were truly saddened to learn of his untimely death in 2015. This research investigation was among the primary objectives of the “PLIOMAX” grant National Science Foundation OCE-1202632 (PI MER and co-PI PJH). MER also appreciates support from the Vetlesen Foundation. The authors acknowledge PALSEA, a working group of the International Union for Quaternary Sciences (INQUA) and Past Global Changes (PAGES) for useful discussions and comments at a 2012 meeting (Rome). PALSEA is supported by Swiss Academy of Sciences and the Chinese Academy of Sciences. Council of Geosciences RSA provided full logistic and field support for coauthor D Roberts. The authors also thank E Bergh (Natural History Department Iziko South African Museum) for his dedicated work in the field and Iziko Museum in Cape Town for providing important materials. De Beers Consolidated Mines provided access, vehicle, and a guide and allowed several days of examination and collection of the quarries and test pits. Mr. Justin Gillis joined our expedition to tell our story to The New York Times (https://www.nytimes.com/2013/01/22/science/earth/seeking-clues-about-sea-level-from-fossil-beaches.html). This research was conducted in the Republic of South Africa under SAHRA Permits (Ref. 80/12/06/001/61, 9/2/066/0001). The background maps in Figure 1 of this article were created using ArcGIS® software by Esri. ArcGIS® and ArcMap™ are the intellectual property of Esri and are used herein under license. Copyright© Esri. All rights reserved. For more information about Esri® software, please visit www.esri.com. We appreciate the informative and constructive reviews (and rereview) provided by T. Naish, T. Törnqvist, and an anonymous reviewer that greatly improved this manuscript.

Open Research

Data Availability Statement

Data for this research are available in PANGEA at this DOI: (https://doi.pangaea.de/10.1594/PANGAEA.910120) under a Creative Commons Attribution 4.0 International license. All our samples are registered with an IGSN in SESAR (http://www.geosamples.org/igsnabout).