Sediment sources and accumulation mechanisms

Unit A exposed in the outcrops represents the interstadial MIS 5c sea-level high stand that reached ∼4 m above the contemporaneous mean sea level (a msl) in the southern CPRS around 104–101 ka (Lopes et al., Reference Lopes, Pereira, Caron, Dillenburg, Rosa, Barboza, Saviani, Sawakuchi, Tatumi and Yee2024) and is followed by a continental succession, the base of which is represented by Unit B. The strong reddish-to-white color and plant root traces of the upper Unit A2 (Fig. 4C) indicate subaerial weathering after sea-level fall. Root traces and plant debris are present between units A2 and B1 (Fig. 4A and B) at AC-01A but are scarce at AC-01B and AC-02, which could indicate differences in preservation across the outcrops or that plant cover was sparse and thus probably did not contribute much to retention and stabilization of sand (Carter and Wilson, Reference Carter, Wilson and Pye1993). The gradual change from Unit A to B without any evident depositional hiatus or erosive surface indicates that sediment supply and accumulation were constant during the marine–continental transition.

Transgressive aeolian deposits are characteristic of coasts subject to high-energy waves, abundant sediment supply, and strong winds (Rowe and Bristow, Reference Rowe and Bristow2015; Hesp, Reference Hesp, Wolanski and McLusky2024). As the transgressive sediment accretion depends on the sediment budget, that is, the equilibrium between input, loss, and retention of sediment (Nickling and Davidson-Arnott, Reference Nickling, Davidson-Arnott and Davidson-Arnott1990), the continuous accumulation between Units A and B indicates a positive budget (high sediment supply) during the marine–continental transition. This high availability of sand could have been favored by the orientation of Barrier III (N33°E, Fig. 1F) relative to the incoming southerly waves that eroded the shoreface–foreshore and promoted longshore redistribution of sediments, which were then carried landward by SE winds to feed transgressive aeolian systems, as observed today on Barrier IV (Fig. 1G) along the southern CPRS (Tomazelli et al., Reference Tomazelli, Villwock, Dillenburg, Bachi and Dehnhardt1998; Dillenburg et al., Reference Dillenburg, Roy, Cowell and Tomazelli2000; Seeliger et al., Reference Seeliger, Cordazzo, Oliveira and Seeliger2000; Martinho et al., Reference Martinho, Hesp and Dillenburg2010; Cecilio and Dillenburg, Reference Cecilio and Dillenburg2019; Hesp et al., Reference Hesp, Da Silva, Miot da Silva, Bruce and Keane2022; Watanabe et al., Reference Watanabe, Barboza-Pinzon, Hesp, Schlosser and Salgado2025a).

The coastal aeolian systems are built from sediments supplied by waves and wind during transgressive high-stand phases of sea-level change (Cooper, Reference Cooper1958; Cattaneo and Steel, Reference Cattaneo and Steel2003; Lees, Reference Lees2006; Gianinni et al., Reference Giannini, Sawakuchi, Martinho and Tatumi2007; Tamura et al., Reference Tamura, Kodama, Bateman, Saitoh, Watanabe, Matsumoto and Yamaguchi2011) or by aeolian deflation along the large fetch provided by the subaerially exposed continental shelf during regressive/low-stand phases (Sayles, Reference Sayles1931; Schofield, Reference Schofield1970, Reference Schofield1975; Coetzee, 1975–Reference Coetzee1976; Clemmensen et al., Reference Clemmensen, Andresen, Nielsen and Sten1996; Bateman et al., Reference Bateman, Holmes, Carr, Horton and Jaiswal2004; Lees, Reference Lees2006; Peterson et al., Reference Peterson, Murillo-Jiménez, Stock, Price, Hostetler and Percy2017). Deflation can be more intense if the stabilizing coastal vegetation is reduced due to arid/semiarid climate (as during the last glacial; see “Climate Influence” section), and/or lowered water table (Seeliger et al., Reference Seeliger, Cordazzo, Oliveira and Seeliger2000; Watanabe et al., Reference Watanabe, Barboza-Pinzon and Hesp2025b) driven by sea-level fall. The coastal barriers of the CPRS (Fig. 1C) were built with sediments reworked from the subaerially exposed continental shelf by the Quaternary marine transgressions, thus implying a major allogenic (i.e., eustatic) control on sediment supply and accumulation during transgressive high-stand phases. Reworking and redeposition of sediments from older units are suggested by the unexpectedly older ages of ∼128 and ∼141 ka obtained in Unit A and ∼131 ka in Unit B (Fig. 5A) at AC-02 (Lopes et al., Reference Lopes, Pereira, Caron, Dillenburg, Rosa, Barboza, Saviani, Sawakuchi, Tatumi and Yee2024).

The building of transgressive aeolian systems can be initiated by several mechanisms such as drier climate, reduced plant cover, lowering of the water table, shoreline erosion, and/or changes in wind and wave regimes, but the role of each can be difficult to distinguish (Hesp et al., Reference Hesp, Da Silva, Miot da Silva, Bruce and Keane2022; Watanabe et al., Reference Watanabe, Barboza-Pinzon, Hesp, Schlosser and Salgado2025a, Reference Watanabe, Barboza-Pinzon and Hesp2025b). In the study area, sea-level fall was probably a major influence on sediment supply and availability. The absence of a clear depositional/erosive hiatus between Units A and B indicates that accumulation of the latter could have started early during the eustatic fall (forced regression) of MIS 5b. As the sea level fell, sediments were eroded from the shoreface and deposited by waves on the paleo-coastline, or accumulated on the subaerially exposed continental shelf surface, where they would have been deflated and carried landward by onshore winds. The planar and trough cross bedding of Unit B observed at AC-03 (Fig. 4I) likely represent small incipient dunes, probably formed of beach-supplied sand, that may be coeval with the aeolian deposits dated as ∼98 ka (MIS 5c) at Cassino Beach (Dillenburg et al., Reference Dillenburg, Barboza, Rosa, Caron and Sawakuchi2017), and thus would represent the regression after the high stand of ∼101 ka (Lopes et al., Reference Lopes, Pereira, Caron, Dillenburg, Rosa, Barboza, Saviani, Sawakuchi, Tatumi and Yee2024). This positive sediment budget, however, would have been available only during the early phase of the regression when the sources were still relatively close to the barrier (Illenberger, Reference Illenberger1996; Bateman et al., Reference Bateman, Holmes, Carr, Horton and Jaiswal2004, Reference Bateman, Carr, Dunajko, Holmes, Roberts, McLaren, Bryant, Marker and Murray-Wallace2011), whereas sediment supply (and accumulation) would have decreased as the coastline shifted basinward.

Although the magnitude of the regression in the study area is unknown, dated corals from Barbados and Papua New Guinea point to a sea-level fall at a rate of 5.6 m/ka during the MIS 5c–5b transition (Cutler et al., Reference Cutler, Edwards, Taylor, Cheng, Adkins, Gallup, Cutler, Burr and Bloom2003). The low slope of the continental shelf adjacent to the studied outcrops (0.033°) implies that a 2 m sea-level fall in this area would result in a basinward coastline shift of ∼820 m (Dillenburg et al., Reference Dillenburg, Roy, Cowell and Tomazelli2000). Assuming rates of sea-level fall similar to those of Cutler et al. (Reference Cutler, Edwards, Taylor, Cheng, Adkins, Gallup, Cutler, Burr and Bloom2003) as a reference, the paleo-coastline could have shifted some 2296 m in 1000 years, and considering that coastal aeolian deposits usually do not extend more than 5–8 km landward from the shoreline (e.g., Illenberger, Reference Illenberger1996; Tomazelli, Reference Tomazelli1994; Bateman et al., Reference Bateman, Holmes, Carr, Horton and Jaiswal2004; Giannini et al., Reference Giannini, Sawakuchi, Martinho and Tatumi2007; Dillenburg and Hesp, Reference Dillenburg and Hesp2009; Sawakuchi et al., Reference Sawakuchi, Giannini, Martinho and Tanaka2009), sediment supply for Unit B would have lasted only a few thousand years. The surface S1 (Fig. 3A) could represent the end of the accumulation of Unit B as a result of diminished sediment input (depositional hiatus) or aeolian deflation greater than accumulation (erosive hiatus) of uppermost Unit B2, controlled by the MIS 5b sea-level fall that reached a minimum around 88 ka (Spratt and Lisiecki, Reference Spratt and Lisiecki2016). In either case, the accumulation could have ended or was highly reduced at the time that the regressive coastline reached a distance >8 km from the outcrops, when the supply of marine sediments reaching inland areas was no longer substantial.

The accumulation of Unit C above S1 indicates another episode of increased landward sand transfer, which could have been initiated by a positive sea-level oscillation during MIS 5a (84.7–71 ka), as suggested by the age of ∼82.3 ka obtained at the base of the unit (Fig. 3A). Similar ages of ∼86.4 and 85 ka from aeolian sediments overlying marine deposits in the northern CPRS (Poupeau et al., Reference Poupeau, Soliani, Rivera, Loss and Vasconcellos1988) suggest that this aeolian episode was coeval along the entire coastal plain. Coastal marine deposits of similar age are not observed directly in the study area, but one electron spin resonance (ESR) age of ∼87 ka from a shell obtained in subsurface deposits seaward of Barrier III suggests it was reworked from a marine MIS 5b deposit by the MIS 5a transgression that did not reach an elevation comparable to current sea level (Lopes et al., Reference Lopes, Pereira, Kinoshita, Mollemberg, Barbosa and Baffa2020, 2024a). The building of coastal aeolian systems by a marine transgression is initiated while sea level is still below than the present, at least −50 m or less in some coasts (Bateman et al., Reference Bateman, Holmes, Carr, Horton and Jaiswal2004; Tamura et al., Reference Tamura, Kodama, Bateman, Saitoh, Watanabe, Matsumoto and Yamaguchi2011). Considering the landward reach of coastal aeolian systems, the accumulation of Unit C could have started when the transgressing coastline was at a distance of at least 5 to 8 km from the outcrops during the MIS 5b–5a marine transgression, corresponding to a depth of ∼19.5 m below the present according to the estimates of Dillenburg et al. (Reference Dillenburg, Roy, Cowell and Tomazelli2000).

Aeolian sediment accumulation was apparently continuous from MIS 5a onward, as indicated by the lack of distinct hiatuses between the levels dated as ∼82.3 and ∼41.6 ka (Fig. 5E), although some periods of relative stabilization are suggested by the presence of protosols within Unit C. Accumulation between ∼82.3 and ∼71.8 ka could have been driven by the two MIS 5a high-stand peaks at 85–82 and 77 ka (Muhs et al., Reference Muhs, Wehmiller, Simmons, York, Gillespie, Porter and Atwater2003; Potter et al., Reference Potter, Esat, Schellmann, Radtke, Lambeck and McCulloch2004; Schellmann et al., Reference Schellmann, Radtke, Potter, Esat and McCulloch2004; Dumas et al., Reference Dumas, Hoang and Raffy2006; Spratt and Lisiecki, Reference Spratt and Lisiecki2016; Wainer et al., Reference Wainer, Rowe, Thomas, Mason, Williams, Tamisiea, Williams, Düsterhaus and Henderson2017), at an average rate of ∼7.5 cm/ka estimated from the distances between dated layers at AC-01A (Figs. 3 and 5E). On the other hand, the lower estimated accumulation rates of ∼3.5 cm/ka between 71.8 and 41.6 ka could have been driven in part from plant growth and sediment stabilization, but also by lower sediment supply due to the post-MIS 5a sea-level fall. Aeolian accumulation at that time is also indicated in the northern CPRS by one TL age of 74.8 ka in the northern CPRS (Poupeau et al., Reference Poupeau, Soliani, Rivera, Loss and Vasconcellos1988).

Sea level fell and remained tens of meters lower than the present (up to −80 to −120 m) from MIS 4 up to the Early Holocene (MIS 1) (Cutler et al., Reference Cutler, Edwards, Taylor, Cheng, Adkins, Gallup, Cutler, Burr and Bloom2003; Spratt and Lisiecki, Reference Spratt and Lisiecki2016). This implies that accumulation during low stands may have been mostly from autogenic processes, such as migration of bedforms, cannibalization of dunes, and/or deflation from other areas, possibly favored by climate conditions (see “Climate Influence” section), instead of direct inputs of sea-supplied sand. Nevertheless, coastal sand inputs could have increased episodically as a result of several high stands during MIS 3 (at 60, 52, 49, 44, 36, and 32 ka) that reached from −70 to −40 m relative to the present sea level (Lambeck and Chappell, Reference Lambeck and Chappell2001). Accumulation around ∼41.6 ka was coeval with the marine high stand during MIS 3 recorded in the CPRS when sea level reached about −22 to −5 m relative to the present between ∼47 and 36 ka BP (Dillenburg et al., Reference Dillenburg, Barboza, Rosa, Caron, Cancelli, Santos Fischer, Lopes and Ritter2020).

The few discernible primary sedimentary structures at the base of Unit D are similar to those of Unit C. The absence of an erosive/non-depositional contact and the same unimodal (mostly fine) sand in the two units point to continuous aeolian accumulation that seemingly lasted up to 10.5 ka (MIS 1) based on the OSL age at the top of Unit D (Fig. 3), with an estimated average accumulation rate of ∼8 cm/ka between ∼41.6 and 10.5 ka (Fig. 3). As the shelf-sourced aeolian sand supply probably was reduced during MIS 2 when sea level reached its lowest (about –120 to –130 m) levels of the last glacial period (Lambeck et al., Reference Lambeck, Rouby, Purcell and Sambridge2014), accumulation of Unit D could have occurred through autogenic mechanisms facilitated by sparse vegetation and increased aeolian processes at that time.

Although the Holocene postglacial marine transgression that started around 18 ka (Lambeck et al., Reference Lambeck, Rouby, Purcell and Sambridge2014) could, in theory, have contributed to the accumulation of the upper Unit D and Unit E, this seems less likely in the study area, because most sediments transferred landward would have built Barrier IV or became trapped in the Lagoon IV (Mangueira Lagoon, Fig. 1G) as the barrier migrated landward (Caron, Reference Caron2014). Moreover, Unit E was probably built by aeolian deflation of Unit D, as suggested by its light brown color (Fig. 6H), whereas the Holocene aeolian deposits are light gray to white-colored.

Depositional environments of the continental units

The sedimentary succession overlying marine deposits (Unit A) accumulated essentially in continental environments dominated by aeolian processes. Nevertheless, the facies changes recognized through the succession, represented by the depositional units, indicate sediment accumulation under variable environmental conditions.

The dominant sedimentary structures in Units B and C across AC-01 and 02 are thin, subhorizontal, planebed to wavy rippleform laminations without internal features (see Figs. 4D and 5B and C), although cross bedding is present locally (Fig. 4G). Those styles of laminations are analogous to the horizontal or undulated laminae and strata with very low dip angles described from ergs (sand seas), characteristic of tabular sand sheet deposits accumulated in interdune basins, where aeolian sediments deposited from suspension are trapped (McKee and Tibbitts, Reference McKee and Tibbitts1964; Bigarella, Reference Bigarella, Rigny and Hamblin1972; McKee and Moiola, Reference McKee and Moiola1975; Fryberger et al., Reference Fryberger, Ahlbrandt and Andrews1979; Kocurek, Reference Kocurek1981a, Reference Kocurek1981b; Hummel and Kocurek, Reference Hummel and Kocurek1984). Interdune deposits tend to occupy large areas of dunefields and have greater preservation potential than the dunes. In coastal ergs, or areas of interior ergs located close to the shore, interdunes are thicker than deposits found in other areas of sand seas (McKee and Tibbitts Reference McKee and Tibbitts1964; Kocurek, Reference Kocurek1981a).

Despite the similar laminated structure, the differences between Units B and C at AC-01 represent changes in depositional environments. The gray hues and redox depletion mottling of Unit B (Fig. 9C) point to accumulation in a wet/damp, vegetated aeolian system (Kocurek, Reference Kocurek1981b; Kocurek and Havholm, Reference Kocurek, Havholm, Weimer and Posamentier1993; Crabaugh and Kocurek, Reference Crabaugh, Kocurek and Pye1993; Carr-Crabaugh and Kocurek, Reference Carr-Crabaugh, Kocurek, Shanley and McCabe1998). This type of a system can be observed today in the dunefields developed in Barrier IV in the southernmost CPRS. This system includes ephemeral interdune lakes formed between the mobile dunes during rainy periods (Fig. 9A), and vegetated ponds (Fig. 9B) developed on aeolian deflation basins located between the frontal and the mobile dunes (Tomazelli, Reference Tomazelli1993; Martinho et al., Reference Martinho, Hesp and Dillenburg2010).

Figure 9.

(A) Flooded interdune areas of Barrier IV dunefield in the southern coastal plain of Rio Grande do Sul (CPRS). (B) Vegetated slacks developed on wet/damp interdune areas of the Barrier IV dunefield (a free dune is visible in the background). (C) Comparison between Units B1 and B2 showing physical features interpreted to record changes in depositional conditions, from wet–damp (gray hues, depletion mottles) to dry (reddish hues, laminated white sand). (D) Damp interdune (darker areas) of Barrier IV partially covered by dry, rippled sand sheets. (E) Google Earth image of the Barrier IV dunefield showing ephemeral lakes (dark areas) and creeks (black arrows) between barchanoid ridges. (F) Interdune area of the Barrier IV dunefield covered with low dunes and sand sheets, forming a possible analog of Unit C.

The physical features of subunits B1 and B2 (Fig. 9C) suggest alternate periods of wet–dry conditions. The light gray sand between the top of subunit A2 and bottom of subunit B1 indicates subaerial weathering and leaching of Fe by percolating water, as observed today on the uppermost Unit C exposed in gullies (Fig. 5G), followed by relatively dry, oxidizing conditions. The bluish-gray color and root mottling in both subunits point to vegetated, wet environments subject to reducing conditions. Together with the absence of well-defined sedimentary structures, which become indistinct as wet sediments homogenize (Bigarella Reference Bigarella, Rigny and Hamblin1972; Mountney and Jagger, Reference Mountney and Jagger2004), those features indicate saturation with stagnant water for prolonged periods, similar to the vegetated ponds of Barrier IV (Fig. 9B). On the other hand, the Fe oxide mottlings, sharp upper contacts, and light gray laminated sand layers with ripples at the top of both B1 and B2 suggest change from wet, to damp, and finally dry conditions. Damp interdune deposits favor the retention of dry windblown sand through adhesion, resulting in laminated and rippled sand deposits (Kocurek, Reference Kocurek1981b; Hummel and Kocurek, Reference Hummel and Kocurek1984). The light gray laminated sand layers on top of both B1 and B2 may be analogs to dry sand sheets accumulated on top of damp interdune areas of Barrier IV (Fig. 9D).

The development of wet aeolian systems at basin scale can be controlled by tectonic subsidence or uplift (not regarded as influencing factors in the tectonically stable CPRS), climate, or by a rise in the water table, which in coastal areas is mostly induced by relatively high sea level (Carr-Crabaugh and Kocurek, Reference Carr-Crabaugh, Kocurek, Shanley and McCabe1998; Kocureck et al., Reference Kocurek, Robinson and Sharp2001). As discussed in the previous section, sea level still could have been relatively high at the time when the accumulation of Unit B started, but rainy conditions after the MIS 5c high stand could also have contributed to a high water table (see “Climate Influence” section). When the water table is high, the capillary fringe is at or close to the surface of interdune areas, and if the rate of water table rise is in equilibrium with sediment input, the moisture contributes to trapping windblown sand and prevents aeolian deflation, thus favoring sediment aggradation (Mountney and Thompson, Reference Mountney and Thompson2002). When the water table falls, the dry sediment is redistributed across the interdune flats, generating bedforms such as wind ripple laminations (Kocurek, Reference Kocurek1981b; Crabaugh and Kocurek, Reference Crabaugh, Kocurek and Pye1993). The apparent dry–wet–damp–dry succession in both B1 and B2 could have resulted from oscillations of the water table driven by variations in rainfall or by small-amplitude positive sea-level oscillations within a regressive trend. The differences between subunits B1 and B2, such as stronger bluish-gray color in B1 and more reddish iron mottlings in B2 (Fig. 9C), suggest that the latter accumulated under relatively drier conditions than the former, which could be explained by the progressive lowering of the water table in response to drier climate and/or sea-level fall. In either case, the B1–B2 succession seems to represent the progressive drying and infilling of a lake/pond developed within an interdune depression.

The rain-fed ephemeral lakes/ponds developed within the dunefields of Barrier IV in the southern CPRS (Fig. 9A and B) can last for months, and under prolonged/intense rainfall conditions, such as during El Niño events, the lakes occupy large portions of the interdune basins (Fig. 9E). The cut-and-fill structures at AC-01B exhibiting mudclasts, lags of heavy minerals, and coarser grains, and inclined to subhorizontal stratification (Fig. 4F), indicate channels excavated by water flowing along ephemeral creeks connecting interdune lakes/ponds (Picard and High, Reference Picard and High1973; Hummel and Kocurek, Reference Hummel and Kocurek1984; Granja et al., Reference Granja, De Groot and Costa2008), as observed today on Barrier IV (Fig. 9E). As those channels cut into Unit A below (Fig. 4F), they are probably correlated to the wetter conditions represented by Unit B at AC-01A, whereas the channel infill (Fig. 4F) and overlying cross strata (Fig. 4G) at AC-01B point to cessation of flow and infill of the channel through aeolian accumulation under drier conditions. The intercalated subhorizontal and cross-strata dipping at 10–30° (Fig. 4G) indicate bedform migration, possibly under changing wind direction/intensity. Variations of wind regime or intensity are also suggested by the heavy mineral lag concentrations along the bedding planes (Sawakuchi et al., Reference Sawakuchi, Giannini, Martinho and Tanaka2009; Fig. 4F).

The wet/damp environments of Unit B observed at AC-01 are seemingly absent in the other outcrops. Unit B at AC-02 exhibits subhorizontal laminations (Fig. 4A) virtually indistinguishable from those of Unit C, whereas at AC-03, the laterally adjacent subhorizontal laminations (Fig. 4H) and trough cross bedding (Fig. 4I) suggest small, incipient dunes, possibly developed on foreshore–backshore settings. The variation of Unit B across the outcrops seem to reflect the zonation observed across the Holocene Barrier IV, characterized by incipient/foredunes seaward, foredunes and/or aeolian transgressive sand sheets (TSSs) in intermediate areas (Caron, Reference Caron2014; Watanabe et al., Reference Watanabe, Barboza-Pinzon, Hesp, Schlosser and Salgado2025a), and wet environments landward (Fig. 9E).

The surface S1 between Units B and C (Fig. 5B) represents a change from wet/damp to dry deposits and thus can be characterized as a supersurface (Crabaugh and Kocurek, Reference Crabaugh, Kocurek and Pye1993; Kocurek and Havholm, Reference Kocurek, Havholm, Weimer and Posamentier1993; Jones et al., Reference Jones, Scherer and Kuchle2016). This surface originated via non-deposition or aeolian deflation, probably facilitated by the sea-level fall during MIS 5b that reached a minimum around 88 ka Spratt and Lisiecki, Reference Spratt and Lisiecki2016), consistent with its position between the levels dated as ∼104 and 82.3 ka at AC-01A (Fig. 3). In the Wilderness coast of South Africa, a depositional hiatus formed between 90 ± 4 and 73 ± 3 ka (Bateman et al., Reference Bateman, Holmes, Carr, Horton and Jaiswal2004) on a sandy cordon suggests a similar eustatic control on its origin.

The tabular subhorizontal (planebed) and low-angle lamination and undulating rippleform laminations without internal structures or foresets in Unit C are characteristic of dry interdune areas that develop when the water table is well below the surface (Bigarella, Reference Bigarella, Rigny and Hamblin1972; Kocurek Reference Kocurek1981a; Hummel and Kocurek, Reference Hummel and Kocurek1984; Mescolotti et al., Reference Mescolotti, Varejão, Warren, Ladeira, Giannini and Assine2019). The structure of Unit C (Fig. 5C) suggests subhorizontal tabular deposits formed by laterally translatent strata and by translatent strata climbing at low (subcritical) angles relative to the underlying surface, produced by wind ripple migration parallel to the depositional surface (Hunter, Reference Hunter1977a, Reference Hunter1977b; Kocurek and Dott, Reference Kocurek and Dott1981; Mountney and Jagger, Reference Mountney and Jagger2004). The subhorizontal and low-angle laminations indicate low net deposition and high rates of ripple migration (Hunter, Reference Hunter1977a; Allen, Reference Allen1982). The accumulation of Unit C likely occurred under conditions similar to those of the interdune areas of Barrier IV during dry periods, when dry aeolian sand sheets accumulate on top of damp deposits (Fig. 9D).

The lowering of the water table increases the deflation of interdune sediments, which are cannibalized by dunes that grow at the expense of the reduction of interdune areas (Mountney and Jagger, Reference Mountney and Jagger2004). Dry sand tends to be easily deflated by wind unless there is some mechanism that prevents deflation, such as plant cover and/or wet/damp sediments (Bigarella, Reference Bigarella1975; Illenberger, Reference Illenberger1996; Kocurek et al., Reference Kocurek, Townsley, Yeh, Havholm and Sweet1992). The aggradation of Unit C indicates constant accumulation of sediments deflated from other areas, probably forming TSSs similar to those on the Albardão dunefield (Watanabe et al., Reference Watanabe, Barboza-Pinzon, Hesp, Schlosser and Salgado2025a, Reference Watanabe, Barboza-Pinzon and Hesp2025b; Fig. 9F), facilitated by sparse vegetation and lowered water table due to lower sea level. Nevertheless, the lower estimated accumulation rates between ∼71.8 and 41.6 ka described in the previous section could have resulted from periods of relative stabilization of the aeolian system due to the expansion of vegetation, as indicated by the brownish protosols. These organic concentrations of plant-derived humus indicate relatively humid conditions that favor plant development and stabilization of sediments (Bigarella, Reference Bigarella, Rigny and Hamblin1972; Clemmensen et al., Reference Clemmensen, Andresen, Nielsen and Sten1996) and in coastal deposits indicate short-term stabilization of aeolian deposits related to diminished sediment deflation/accumulation (Murray-Wallace, Reference Murray-Wallace2018). The intercalated protosols and sand layers (Fig. 5E and F) indicate episodic accumulation of the TSSs in response to reduced vegetation cover, lower water table, and/or other factors, as observed in the Albardão dunefield and other sectors of Barrier IV (Martinho et al., Reference Martinho, Dillenburg and Hesp2008; Dillenburg et al., Reference Dillenburg, Barboza, Hesp, Rosa, Caron and Guadagnin2024; Watanabe et al., Reference Watanabe, Barboza-Pinzon, Hesp, Schlosser and Salgado2025a, Reference Watanabe, Barboza-Pinzon and Hesp2025b).

Together with the protosols, the absence of cross strata characteristic of dunes within Unit C suggests flat interdune environments, which are wetter and, in general, occupy larger areas than dunes in aeolian systems (McKee and Tibbitts, Reference McKee and Tibbitts1964). Another possibility is that the study area was subject to constant changes in wind direction and/or speed, which hindered the development of dunes but favored the development of tabular sand sheets. Those changes can also generate heavy mineral lags along bedding planes, such as those observed at Unit C (Fryberger et al., Reference Fryberger, Ahlbrandt and Andrews1979; Kocurek et al., Reference Kocurek, Townsley, Yeh, Havholm and Sweet1992; Sawakuchi et al., Reference Sawakuchi, Giannini, Martinho and Tanaka2009; Fig. 5C).

Aeolian accumulation on dry interdune areas continued during the deposition of units C and D, but the latter exhibits poorly sorted sediments with bimodal grain-size distribution (Fig. 3, Table 2) due to the presence of silt and clay. The silt is probably of aeolian origin, considering the presence of loess deposits in the southern CPRS that accumulated between MIS 4 and 2 (Lopes et al., Reference Lopes, Dillenburg and Schultz2016), whereas the clay was probably formed by the weathering of less stable minerals in the sediments (Villwock and Tomazelli, Reference Villwock and Tomazelli1995).

The uppermost Unit E exhibits similarities to Unit C, being the very pale brown, laminated sand with intercalated light brown protosols (Fig. 6H). These can be interpreted as episodes of increased aeolian reworking–redeposition of Pleistocene sediments intercalated with periods of stabilization/plant growth in the last <1000 years, based on the OSL age obtained at its base.

Pedogenic modifications

Although the composition and texture of the coarse sediment fraction (mainly fine- and medium-grained quartz sand) is essentially homogeneous through the stratigraphic succession, the differences in macroscopic physical features between Facies Associations 2 and 3 point to postdepositional modification of the sediment that reflect mainly climate-driven pedogenic (soil-forming) processes. The most evident physical modifications are color variation and the development of Fe-Mn coatings and nodules, but also include variation in MS (Fig. 3).

Color

The color variation through the succession can be ascribed to redoximorphic processes that involved the reduction, translocation, and oxidation of Fe and Mn compounds subject to hydromorphic (alternate wet and dry) conditions (Schwertmann and Fanning, Reference Schwertmann and Fanning1976; Bouma et al., Reference Bouma, Fox, Miedma and Douglas1990; Vepraskas et al., Reference Vepraskas, Lindbo, Stolt, Stoops, Marcelino and Mees2018). The bluish-gray and brown to reddish colors through the succession indicate weathering and precipitation of Fe released from minerals such as ilmenite, magnetite, pyroxenes, and hornblende (Silva, Reference Silva1979; Table 1) through hydrolysis, followed by reduction and oxidation (Schwertmann, Reference Schwertmann, Stucki, Goodman and Schwertmann1988). The postdepositional mobilization of iron would have resulted from the anaerobic reduction of structural ferrous iron (Fe²⁺) released from minerals in water-saturated sediment via microbial oxidation of organic matter, followed by movement through the sediment by convection or diffusion until it reached aerobic zones, where the iron would have undergone reoxidation to ferric (Fe³⁺) oxides, a process called ferrolysis (Brinkman, Reference Brinkman1970; Cornell and Schwertmann, Reference Cornell and Schwertmann2003; Levin et al., Reference Levin, Tsoar, Maia, Sales and Hermann2007; Vepraskas, Reference Vepraskas2015).

The pale brown to brownish-yellow colors of Unit A probably resulted from subaerial oxidation during sea-level fall and lowering of the water table. The low chroma (light gray to white) color at the gradual contact between A2 and B1 (Fig. 4C) represents the predominance of sand quartz and indicates complete removal of Fe and/or Mn (redox depletion) resulting from leaching of soluble Fe and Mn driven by chemical reduction due to prolonged saturation with water (Barrell, Reference Barrell1908; Bouma et al., Reference Bouma, Fox, Miedma and Douglas1990), followed by subaerial (aerobic) dry conditions indicated by the root traces. The loss of color (whitening) in the upper A2/lower B1 could also be explained by seasonal wetting–drying cycles, as observed in interdune areas of dunefields along the northeastern Brazilian coast, which are seasonally occupied by ponds (Levin et al., Reference Levin, Tsoar, Maia, Sales and Hermann2007). Similarly, the whitening of the uppermost Unit C exposed in gullies (Fig. 5G) today could be explained by Fe leaching driven by seasonal changes in moisture (rainy–dry seasons).

The bluish-gray color in Unit B indicates Fe-reducing conditions under stagnant, anaerobic water (Barrell, Reference Barrell1908; Bouma, Reference Bouma, Fox, Miedma and Douglas1990; Schwertmann and Fitzpatrick, Reference Schwertmann and Fitzpatrick1992; D’Amore et al., Reference D’Amore, Stewart and Huddleston2004). The low chroma (dark gray) in reduced layers could indicate partial substitution of the Fe oxide (probably goethite) by manganese, which can occur in hydromorphic soils (Stiers and Schwertmann, Reference Stiers and Schwertmann1985; Gilkes and McKenzie, Reference Gilkes, McKenzie, Graham, Hannam and N.C1988; Ebinger and Schulze, Reference Ebinger and Schulze1990). The invertebrate bioturbations (Fig. 4E) and depletion mottling (Fig. 9C) indicate biological activity, implying the presence of organic matter that probably contributed to the dark gray chroma (Chukhrov, Reference Chukhrov1973), the bacterial decomposition of which could have favored anaerobic conditions for the reduction of Fe and Mn in the sediment. The permanence of the gray color together with Fe oxide mottles in B1 and B2 (Fig. 9C) suggests that dissolved Fe²⁺ was not totally leached (despite being mobile in solution in soil water) and/or that oxidation to Fe³⁺ was only partial, possibly due to damp instead of fully dry/aerobic conditions caused by lower water/oxygen diffusion, favored by lower permeability due to the presence of 10% mud in Unit B1 (Fig. 3). On the other hand, the light gray color of the root mottling and sediment overlying the reduced layers of B1 and B2 (Fig. 9C) indicate Fe leaching driven by periodic oscillations of the water table. This would be similar to the process responsible for groundwater soils (gleys) characterized by a reduced horizon overlain by an oxidized (weathered) horizon (Bouma, Reference Bouma, Fox, Miedma and Douglas1990; Schwertmann and Fitzpatrick, Reference Schwertmann and Fitzpatrick1992).

The predominantly yellow-brownish yellow (10YR hue) and pale yellow (5Y hue) grain coatings in units A, B, and C (Fig.7A and B) suggest the presence of fine (∼0.2 μm) crystals of goethite (αFeOOH), the most abundant and stable Fe oxide in soils (van Houten, Reference van Houten1968; Kämpf and Schwertmann, Reference Kämpf and Schwertmann1983). The goethite found in soils can precipitate directly from microbially reduced iron in solution, released from solid Fe²⁺ compounds or from Fe³⁺ oxides, or indirectly by alteration of other minerals such as the metastable ferrihydrite (5Fe₂O₃·9H₂O) which is usually the first product of weathering of Fe-bearing minerals (Chukrov, Reference Chukhrov1973; Senkayi et al., Reference Senkayi, Dixon and Hossner1986; Torrent et al., Reference Torrent, Qingsong, Bloemendal and Barrón2007). Ferrihydrite is more abundant in young (latest Pleistocene–Holocene) soils rather than in older units (McFadden and Hendricks, Reference McFadden and Hendricks1985).

The stronger (high chroma) yellowish color directly above S1 at the base of Unit C (Fig. 5B) is derived from higher concentration of Fe oxides that points to increased weathering under conditions favorable to the formation of goethite, that is, cool to temperate climate with high excess moisture (rainfall > evapotranspiration) and the presence of organic matter (van Houten, Reference van Houten1968; Kämpf and Schwertmann, Reference Kämpf and Schwertmann1983; Schwertmann, Reference Schwertmann, Stucki, Goodman and Schwertmann1988). On the other hand, the paler color in the overlying sediments suggests less intense weathering of Fe-bearing minerals due to cold and dry glacial conditions (see “Climate Influence” section) and/or a greater stability of goethite due to partial substitution of Fe by Al (Bigham et al., Reference Bigham, Fitzpatrick, Schulze, Dixon and Schulze2002). If the aerobic oxidation occurs in situ, the distribution of FeO will reflect the distribution of Fe in the parent material (Cornell and Schwertmann, Reference Cornell and Schwertmann2003). The relatively homogeneous pale brownish color observed in units A, B, and C suggests oxidation of dissolved Fe²⁺ that was dispersed through waterlogged sand during an anaerobic/reduced stage.

The most distinct physical features of Unit D are its greater mechanical resistance to erosion and reddish color (Fig. 10A). The resistance is related to the presence of silt and clay (Fig. 3) and Fe oxides, which besides acting as cements, are also responsible for the brown (7.5YR hue) to reddish-brown (5YR hue) colors that largely obscure the primary sedimentary structures. The red colors (5YR–10R hues) in soils are positively correlated with the amount of hematite (αFe₂O₃; Torrent et al., Reference Torrent, Schwertmann and Fechter1983; McFadden and Hendricks, Reference McFadden and Hendricks1985), whereas the reddish/brown colors of Unit D also suggest also the presence of the hematite precursor ferrihydrite, goethite, and maghemite (γFe₂O₃; Schwertmann, Reference Schwertmann, Bigham, Ciolkosz and Schwertmann1993; Bigham et al., Reference Bigham, Fitzpatrick, Schulze, Dixon and Schulze2002). Although subunit D1 exhibits yellow mottles (probably goethite/ferrihydrite) associated with Fe-leached light gray mottles (Fig. 10B), the prevalent reddish-brown color of D1 and D2 indicates the presence of fine (∼0.1 μm) crystals of hematite, the second most common oxide in soils. This suggests that Unit D was subject to weathering under similar conditions to those that produce red beds (Walker, Reference Walker1967a; van Houten, Reference van Houten1973; Kämpf and Schwertmann, Reference Kämpf and Schwertmann1983; Torrent and Schwertmann, Reference Torrent and Schwertmann1987). Continental red beds are rocks or detrital sedimentary deposits with reddish-brown color derived from the weathering of iron-titanium oxides, which form fine pigment particles and coarser opaque particles of Fe-Ti minerals (van Houten, Reference van Houten1973; Turner, Reference Turner1980). The characteristic color of red beds is due to the fine-grained hematite coating on larger detrital grains and infilling pores, and although other oxides can be present, the hematite pigment masks their colors (Cornell and Schwertmann, Reference Cornell and Schwertmann2003).

Figure 10.

(A) Differential erosion between Unit D and the weathered Unit C below; faint subhorizontal lamination is visible at the base of Unit D (indicated by the arrow). (B) Detail of Unit D1 showing redoximorphic Fe-depletion mottles (indicated by arrows), absent at Unit D2 above. (C) Profile of Unit D in an excavation for a wind farm, with Fe-depletion mottling (gray hue) at the base and reddish Fe-concentrations around plant roots above shown in detail on the right. (D) Eolian deposits of Barrier II on the shore of Mirim Lagoon, with iron oxide crusts (E) at the base.

The diffuse lower boundary of Unit D indicates the reddish color is the product of in situ weathering instead of exogenous input of Fe-rich sediment (Catt, Reference Catt1990). Red beds develop under oxidizing conditions in both arid and tropical/subtropical warm climates, where hematite is formed by redox processes in soils subject to seasonally short wet periods intercalated with longer dry/warmer intervals under conditions with low excess moisture (rainfall < evapotranspiration) and less organic matter than required to form goethite (Walker, Reference Walker1967b; Norris Reference Norris1969; van Houten, Reference van Houten1973; Kämpf and Schwertmann, Reference Kämpf and Schwertmann1983; He et al., Reference He, Yang, Du, Hu, Zhang, Hou and Li2022). Formation of hematite is favored over goethite under moderate (200–600 mm/yr) seasonal rainfall (Balsam et al., Reference Balsam, Ellwood, Ji, Williams, Long and A.E2011). The presumed presence of goethite in Unit D1 (Fig. 10B) and hematite in overlying D2 would be consistent with the zonation observed in weathered soils developed under a warm and wet climate (van Houten, Reference van Houten1968). It could also reflect an early stage of weathering under wetter/cooler conditions that favored goethite/ferrihydrite formation, followed by weathering under drier/warmer conditions that favored hematite.

The in situ formation of hematite and maghemite can occur in the vadose or phreatic zones through weathering of detrital magnetite, ilmenite, and Fe-Mn silicates such as hornblende, pyroxene and biotite, or transformation of ferrihydrite (Walker, Reference Walker1967a; van Houten, Reference van Houten1968; Hubert and Reed, Reference Hubert and Reed1978; Schwertmann, Reference Schwertmann, Stucki, Goodman and Schwertmann1988; Tianhu et al., Reference Tianhu, Quiaoquin, Huifang, Jun, Junfeng, Huayu and Balsam2010; Sandler et al., Reference Sandler, Taitl-Goldman and Ezersky2023). In the studied deposits, either (or both) processes may have formed the yellow mottling and coatings of goethite/ferrihydrite in Unit D1 (Fig. 7C) and the hematite in Unit D2. The lower relative proportion of detrital magnetite (13%) compared to ilmenite (58%) in the assemblage could indicate that most of the former was converted into hematite and possibly maghemite, whereas ilmenite is more stable and resistant to reducing conditions (van Houten, Reference van Houten1968; Turner, Reference Turner1980). Nevertheless, the weathering of ilmenite could have produced (at least in part) the rutile and leucoxene in the assemblage (Silva, Reference Silva1979; Wust et al., Reference Wust, Rizzi, Takehara, Quairoga, Girelli and Chemale2024), besides Mn oxides.

The reddening, or rubification, of sedimentary deposits that originate red beds through formation of hematite has led different authors (e.g., Norris and Norris, Reference Norris and Norris1961; Price, Reference Price1962; Walker, Reference Walker1967b; Norris, Reference Norris1969; Bigarella, Reference Bigarella, Rigny and Hamblin1972) to regard the concentration of hematite as proportional to the age of the deposits. The reddish color of Unit D overlying older brownish-yellow sediments of Units B and C, however, indicates that this is not the case in Barrier III. Instead, the brownish-red color can be attributed to more intense climate-driven weathering (see “Climate Influence” section) under a warmer climate with marked seasonal variations of moisture (van Houten, Reference van Houten1968; Kämpf and Schwertmann, Reference Kämpf and Schwertmann1983).

The subaerial portion of Barrier III is up to about 4 to 5 km wide in the study area (Fig. 1E), and variations in redoximorphic features in Unit D across it can be related to the “edge effect” (Daniels and Gamble, Reference Daniels and Gamble1967; Richardson and Daniels, Reference Richardson, Daniels, Bigham and Ciolkosz1993), that is, differences in drainage related to topographic variations across the landscape. The brownish-red sediments occur only on the well-drained, dissected seaward edge of the barrier at the transition to the topographically lower Holocene System IV (Fig. 1F), well above the water table. On poorly drained, flatter areas landward of the edge, where the water table is higher and its surface is occupied by wetlands, the color is gray (reduced, Fe-leached sediments) with centimetric Mn masses and higher clay content (Fig. 6D and E). Farther landward, the low-gradient-slope aeolian deposits of Barrier III that partially cover moderately drained areas of Lagoon III are also subject to a relatively high (but seasonally variable) water table. Unit D developed on these areas also contains more clay, but exhibits redox features in the form of Fe concentration along plant root channels and Fe depletion (Figs. 6C and 10C). The yellowish Fe oxides and absence of red hues in these poorly/moderately drained areas of Barrier III indicate preferential reduction and solution of hematite, thus increasing the relative proportion of goethite (Richardson and Daniels, Reference Richardson, Daniels, Bigham and Ciolkosz1993).

The in situ origin of the physical modifications of Barrier III implies that other Pleistocene deposits were likely affected by weathering as well. The subaerial portion of Barrier II exhibits vertical stability and brownish color (Fig. 10D) besides hard crusts of Fe oxide (Fig. 10E). The predominant pale brown (2.5Y hue) to yellowish-brown (10YR hue) colors suggest the absence of hematite, possibly due to longer exposure to weathering processes compared with Barrier III, considering that it was likely formed by the previous interglacial sea-level high stand of MIS 7 some 100 ka earlier. A comparison with Barrier III allows us to assume that Barrier II underwent a similar evolution, being subject to aeolian processes and weak weathering during the next glacial period (MIS 6, 190–130 ka) after its formation, and then underwent stronger weathering and pedogenic processes during the next warmer and wetter interglacial (MIS 5e). If hematite was formed, it was probably lost either by leaching or transformation into goethite through weathering under cooler and wetter conditions during the last glacial period (Schwertmann, Reference Schwertmann1971; Melo et al., Reference Melo, de Oliveira, Batista, Cherobim and Favaretto2020).

Fe-Mn coatings, nodules, and concretions

The aggregates, masses, nodules, and concretions of sand grains with black to reddish coatings (Figs. 5C and H and 6F and G) within the succession are products of postdepositional redox processes. The elemental analyses of the small aggregates (Figures 7E and 8A) and irregular nodules (Fig. 8B) from units A, B, and C and the concretions from Unit D (Fig. 8C) show that the black coatings are mostly Mn oxide, containing also Fe and other trace elements (Table 4). The differences in hardness and relative proportions of constituent elements of those features point to successive stages of development driven by variations in weathering mechanisms through time, including abiotic redox processes driven by wet–dry cycles or biotic redox processes mediated by microorganisms (Bromfield, Reference Bromfield1956; Gilkes and McKenzie, Reference Gilkes, McKenzie, Graham, Hannam and N.C1988; White and Dixon, Reference White and Dixon1996; Tan et al., Reference Tan, Liu, Li, Hu and Huang2006; Yu et al., Reference Yu, Fu, Brookes and Lu2015). The potential sources of Mn in the studied sediments include ferromagnesian silicates (amphibole, pyroxene) and such non-silicates as ilmenite and magnetite (Silva, Reference Silva1979; Wust et al., Reference Wust, Rizzi, Takehara, Quairoga, Girelli and Chemale2024).

Coatings of Mn oxides usually contain iron, because Fe²⁺ is oxidized in the presence of MnO (Lee et al., Reference Lee, Phillips, Ammos, Lietzke and Douglas1990; Jing et al., Reference Jing, Qingze, Yixuan, Hongyan, Minwang, Runliang, Hongping, Michael and M.F2022). In soils subject to reducing conditions and temperatures > 5°C, bacterial decomposition consumes oxygen, and Mn is reduced before Fe. Precipitation of Mn is favored by higher soil pH or redox potential (Eh); therefore, Fe is precipitated first under lower Eh (Veneman et al., Reference Veneman, Vepraskas and Bouma1976; Gilkes and McKenzie, Reference Gilkes, McKenzie, Graham, Hannam and N.C1988; Bouma et al., Reference Bouma, Fox, Miedma and Douglas1990; Dixon and White, Reference Dixon, White, Dixon and Schulze2002; Yu et al., Reference Yu, Fu, Brookes and Lu2015). If the periods of saturation in water are short, only Mn coatings/nodules are formed, but if the saturation time is longer, Fe becomes more abundant than Mn (Veneman et al., Reference Veneman, Vepraskas and Bouma1976). The dark brown aggregates with Mn > Fe found within units A, B, and C (Fig. 8A, Table 4) could have developed from bacterially precipitated Mn²⁺ concentrations, which acted as nuclei over which more manganese was precipitated during the wet period, followed by oxidation under dry conditions (Taylor et al., Reference Taylor, McKenzie and Norrish1964; Burns and Burns, Reference Burns and Burns1975). The small size of the aggregates suggests short wet periods with ideal conditions for precipitation and bacterial crystallization of Mn oxides. Moreover, microbial reduction of Mn oxides is less effective at lower T°C (prevalent during glacial periods), and microbial oxidation of Mn is less effective in sandy soils (van Veen Reference van Veen1973; Sparrow and Uren Reference Sparrow and Uren2014).

Mn nodules and concretions are formed by precipitation of secondary oxides as cements infilling pores and intergranular spaces in the soil, forming aggregates with little or no Fe that act as nucleation structures and can develop into larger nodules as more Mn and Fe in solution are incorporated (Winters Reference Winters1938; Taylor et al., Reference Taylor, McKenzie and Norrish1964; Uzochukwu and Dixon Reference Uzochukwu and Dixon1986; Sanz et al., Reference Sanz, Garcia-González, Vizcayno and Rodriguez1996). The nodules found in the weathered upper Unit C (Fig. 5H) are larger and harder and exhibit stronger black-reddish color and higher Fe content than the aggregates (Fig. 8B), possibly due to leaching of Mn (Dixon and White, Reference Dixon, White, Dixon and Schulze2002). These nodules occur only on the leached sediment (Fig. 5G), thus indicating that they probably developed around the aggregates by incorporation of more Fe due to contemporaneous weathering of the detrital ilmenite and magnetite that would release more Fe compounds (Temple, Reference Temple1966).

The larger, black Fe-Mn masses in Unit D probably formed by growth and coalescence of smaller nodules/aggregates in response to longer periods of saturation with water, considering the presence of similar (albeit smaller) masses in waterlogged parts of Barrier III located landward (Fig. 6D and E). The concentration and depletion mottlings in Unit D1 (Fig. 10B) point to wetter conditions than on the underlying Unit C, which would have favored the development of larger Mn masses associated with yellowish Fe concentrations, probably of goethite and ferrihydrite (Cornell and Schwertmann, Reference Cornell and Schwertmann2003). On the other hand, the absence of Fe mottlings within Unit D2 points to relatively dry conditions, possibly due to shorter wet periods and/or a long-term climate change, which favored precipitation of hematite (Vepraskas et al., Reference Vepraskas, Lindbo, Stolt, Stoops, Marcelino and Mees2018).

The soft Fe-Mn aggregates and masses with irregular borders indicate formation under poor drainage conditions, whereas hard nodules and concretions with sharp boundaries, such as the ones found at the top of Unit D2 (Fig. 6F), form in well-drained soils (Zhang and Karathanasis, Reference Zhang and Karathanasis1997). The Fe-Mn concretions form in soils where the horizontal remobilization surpasses the vertical translocation or leaching of oxides; thus, they are concentrated in Horizons A and B (Drosdoff and Nikkiforoff, Reference Drosdoff and Nikkiforoff1940). The formation of hard concretions in the uppermost Unit D was probably favored by the decreased the permeability of the sediment and vertical mobility of water due to the presence of aeolian silt and pedogenic clay.

The black nucleus of the concretions, containing similar proportions of Fe and Mn (Table 4) and surrounded by a Fe-rich layer with less Mn (Fig. 8C), points to a change in redox conditions. Concretions exhibiting concentric, intercalated Fe-Mn rings around the nucleus result from repeated cycles of reduction–oxidation driven by multiple (usually seasonal) wet–dry episodes or oscillations of the water table (White and Dixon, Reference White and Dixon1996; Yu et al., Reference Yu, Fu, Brookes and Lu2015). During short, wet periods of high redox, only Mn precipitates, whereas the following drier and low-redox conditions favor precipitation of Fe. The absence of concentric Fe-Mn rings in the concretions, however, does not indicate seasonal wet–dry cycles, but the outer Fe-rich layer suggests a change from wet to dry environmental conditions instead. Spherical concretions with sharp boundaries form in well-drained soils subject to periodic oscillation of the water table and have black nuclei surrounded by Fe coatings, because Mn is concentrated first, and in later stages, when Mn is exhausted, Fe is precipitated (Schwertmann and Fanning, Reference Schwertmann and Fanning1976; Zhang and Karathanasis, Reference Zhang and Karathanasis1997; D’Amore et al., Reference D’Amore, Stewart and Huddleston2004). The concretions in Unit D could have started to form during a wet period that favored nucleation and growth of Mn aggregates and proceeded under relatively dry and/or better drained conditions that caused Fe precipitation around the nucleus.

The changes in the relative proportions of major elements in the masses, nodules and concretions also suggest that differences between these features represent distinct stages of development under variable weathering intensity. Compared with the surrounding sediments, Fe-Mn concretions contain less Al and Si, which become less abundant with an increase in size of the concretions (Winters, Reference Winters1938; Childs, Reference Childs1975; Zhang and Karathanasis, Reference Zhang and Karathanasis1997), a pattern consistent with the decrease in those two elements recorded by the elemental analyses (Fig. 11A). The smaller proportion of these elements in the concretions relative to the aggregates and nodules could indicate that Si and Al were not involved in the formation of concretions or that the aluminosilicates in the concretions were decomposed (Childs, Reference Childs1975). Although the FeO and MnO content increases with the size of the nodules/concretions, the amount of Mn is negatively correlated to the amount of Fe (Tan et al., Reference Tan, Liu, Li, Hu and Huang2006), and this relationship is clearly demonstrated in the analyzed samples (Fig. 11A).

Figure 11.

(A) Changes in the relative proportion of elements in the Fe-Mn materials from Units C and D shown in Fig. 8. (B) Average and standard deviation values of magnetic susceptibility (MS) in sediments of the southern coastal plain of Rio Grande do Sul (CPRS).

Figure 11 Long description

Line plot labeled A with the vertical axis labeled percent, ranging from 0 to 90. The horizontal axis shows categories a, b, c, b a (inn.), b (out.) and d, with group labels below: C (aggregate) under a and b; C (nodule) under c; D (concretion) under b a (inn.), b (utt.) and d. Five labeled lines are present: Si, Mn, Al and Fe, plus one additional unlabeled line. Si line values: a about 30, b about 40, c about 20, b a (inn.) about 35, b (utt.) about 10, d about 5. Mn line values: a about 25, b about 30, c about 25, b a (inn.) about 40, b (out.) about 20, d about 10. Al line values: a about 10, b about 15, c about 15, b a (inn.) about 12, b (out.) about 5, d about 0. Fe line values: a about 5, b about 10, c about 30, b a (inn.) about 0, b (out.) about 35, d about 80. Unlabeled line values: a about 15, b about 25, c about 20, b a (inn.) about 45, b (out.) about 35, d about 5. Horizontal error bar plot labeled B with the horizontal axis labeled MS times 10 superscript 3 SI, ranging from 0 to 0.35 with tick labels at 0, 0.05, 0.1, 0.15, 0.2, 0.25, 0.3 and 0.35. The vertical axis lists categories: Unit E, Unit D, ps4, ps1, Unit C, Unit B, AC-01B, AC-01A, Unit A, Barrier IV (topsoil), Barrier IV (subsoil), Barrier IV (dune), Barrier IV (beach), wetlands, CF (loess), SVA (fluvial) and Barrier III (marine). Each category has a point with a horizontal error bar. Unit E point about 0.30 with error bar about 0.25 to 0.35. Unit D point about 0.25 with error bar about 0.20 to 0.30. ps4 point about 0.20 with error bar about 0.15 to 0.25. ps1 point about 0.18 with error bar about 0.12 to 0.22. Unit C point about 0.22 with error bar about 0.18 to 0.26. Unit B point about 0.24 with error bar about 0.20 to 0.28. AC-01B point about 0.08 with error bar about 0.05 to 0.10. AC-01A point about 0.20 with error bar about 0.18 to 0.22. Unit A point about 0.10 with error bar about 0.07 to 0.12. Barrier IV (topsoil) point about 0.12 with error bar about 0.10 to 0.14. Barrier IV (subsoil) point about 0.12 with error bar about 0.10 to 0.14. Barrier IV (dune) point about 0.18 with error bar about 0.15 to 0.20. Barrier IV (beach) point about 0.20 with error bar about 0.18 to 0.22. wetlands point about 0.25 with error bar about 0.22 to 0.28. CF (loess) point about 0.27 with error bar about 0.24 to 0.30. SVA (fluvial) point about 0.10 with error bar about 0.07 to 0.12. Barrier III (marine) point about 0.06 with error bar about 0.05 to 0.07.

Magnetic susceptibility

The composition of the parent material (quartz sand with heavy minerals) through the stratigraphic succession of Barrier III is essentially homogeneous, and considering the processes that can enhance MS in soils, the observed upward-increasing trend and variations in MS values are probably explained by the concentration of detrital ferrimagnetic minerals (especially magnetite), inorganic neoformation of fine magnetite and maghemite through oxidation of Fe²⁺ solutions, and the activity of magnetotactic or reducing bacteria (Bigham et al., Reference Bigham, Fitzpatrick, Schulze, Dixon and Schulze2002) driven by changes in environmental conditions.

Magnetite is the most magnetic mineral, occurring in soils as detrital grains derived from mafic volcanic rocks or in the form of ultrafine (<100 μm) grains produced by magnetotactic and reducing bacteria or by abiotic processes such as redox cycles or transformation (Maher and Taylor, Reference Maher and Taylor1988; Fine et al., Reference Fine, Singer, La Ven, Verosub and Southard1989; Schwertmann and Fitzpatrick, Reference Schwertmann and Fitzpatrick1992; Fassbinder et al., Reference Fassbinder, Stanjek and Vali1993; Heller and Evans, Reference Heller and Evans1995; Spassov et al., Reference Spassov, Heller, Kretzschmar, Evans, Yue and Nourgaliev2003). The contribution of pedogenic magnetite to soil magnetism increases with the degree of pedogenesis. In waterlogged, anoxic soils, bacteria can assimilate the Fe³⁺ in complexes and transform it into magnetite through reduction and partial reoxidation (Maher, Reference Maher, Frankel and Blakemore1991; Singer et al., Reference Singer, Verosub, Fine and TenPas1996; van Velzen and Dekkers, Reference van Velzen and Dekkers1999; Bigham et al., Reference Bigham, Fitzpatrick, Schulze, Dixon and Schulze2002). On intermittently wet soils with abundant Fe supply, bacterial reduction can transform ferrihydrite into magnetite during wet periods, and it can subsequently oxidize to maghemite through removal of Fe or addition of O₂ during subsequent dry intervals (Schwertmann and Fitzpatrick, Reference Schwertmann and Fitzpatrick1992; Heller and Evans, Reference Heller and Evans1995; Dearing et al., Reference Dearing, Hay, Baban, Huddleston, Wellington and Loveland1996; Maher, Reference Maher1998; van Velzen and Dekkers, Reference van Velzen and Dekkers1999).

Maghemite is regarded as the dominant ferromagnetic phase in aerobic soils under oxidizing conditions (Torrent et al., Reference Torrent, Barrón and Qingsong2006; Balsam et al., Reference Balsam, Ellwood, Ji, Williams, Long and A.E2011). Besides oxidation of magnetite, maghemite can also form in the soil through thermal transformation (e.g., by fire), dehydration of lepidocrocite, or authigenically through redox cycles, the latter being regarded as the main formative process of clay- and sand-sized maghemite grains (Fine et al., Reference Fine, Singer, La Ven, Verosub and Southard1989; Singer et al., Reference Singer, Verosub, Fine and TenPas1996). Maghemite is an intermediate product in the transformation of ferrihydrite to hematite through dehydration (Torrent et al., Reference Torrent, Barrón and Qingsong2006, Reference Torrent, Qingsong, Bloemendal and Barrón2007). Hematite can form through transformation of ferrihydrite under neutral pH, high T°C, and low water activity, and being 200 times weaker than magnetite, its relative contribution to soil magnetism is low (Heller and Evans, Reference Heller and Evans1995; Bigham et al Reference Bigham, Fitzpatrick, Schulze, Dixon and Schulze2002; Spassov et al., Reference Spassov, Heller, Kretzschmar, Evans, Yue and Nourgaliev2003). Formation of hematite in soils is favored by short wet periods intercalated with longer warm and dry intervals (Carter-Stiglitz et al., Reference Carter-Stiglitz, Banerjee, Gourlan and Oches2006).

The lowest values measured in the marine Unit A1 (Fig. 3, Table 4) could indicate that its MS is controlled by the weakly magnetic oxyhydroxides such as ferrihydrite and goethite produced by weathering. On the other hand, the similar values measured in Pleistocene marine deposits of Barrier II, on the swash zone of Cassino Beach, and in subsoil sediments of the Holocene regressive beach ridges of Barrier IV (Fig. 11B), suggest that weathering of Unit A1 had a minor influence on the production of ferrimagnetic minerals. This implies that magnetism in the sea-supplied coastal sediments is primarily controlled by the concentration of detrital magnetite, and thus the lower MS on Unit A could be explained by dispersed magnetite grains in the sediment. The slightly higher MS measured at the base of Unit A2 probably results from the higher concentration of Fe oxides leached from the light gray weathered layer above (Fig. 4C).

The low MS values measured within one cut-and-fill structure of Unit B1 at AC-01B (Table 4) are similar to those of Unit A1, indicating transport and redeposition with minimal modification of magnetic properties. Nevertheless, the slightly higher MS could have resulted from higher proportion of heavy minerals, as indicated by the lags/concentrations along bedding planes (Fig. 4F and G). On the other hand, the substantially higher values measured in B1 and B2 at AC-01A (Table 4) could be explained by bacterial activity. Large quantities of ultrafine magnetite are produced by the activity of dissimilatory Fe³⁺-reducing bacteria in the presence of organic matter under waterlogged/anoxic conditions (Bell et al., Reference Bell, Mills and Herman1987; Lovley et al., Reference Lovley, Stolz, Nord and Phillips1987; Lovley, Reference Lovley, Frankel and Blakemore1991; Maher, Reference Maher, Frankel and Blakemore1991; Evans and Heller, Reference Evans and Heller2003), consistent with the gray hues in Unit B (Fig. 9C). Another possibility is that bacterial activity produced strongly magnetic Fe sulfides. Whereas the magnetic properties of well-drained sols are controlled by Fe oxides, in waterlogged soils they are dominated by Fe sulfides such as the ferrimagnetic greigite (Fe₃S₄) produced by bacterial sulfate reduction (Maher, Reference Maher1998; Snowball and Torii, Reference Snowball, Torii, Maher and Thompson1999).

The average MS of Unit C is similar to that measured in Holocene coastal dunes at Cassino Beach (Fig. 11B). The lowest values measured in the strongly weathered layer at the base of Unit C (Fig. 5C, Table 4) could be explained by the presence of weakly magnetic goethite and ferrihydrite, as suggested by its yellow color. The higher values in the less weathered sediments above (Fig. 11B) suggest abundant magnetite grains in wind-supplied sand as the source of MS enhancement, although comparatively higher activity of magnetotactic bacteria cannot be excluded. The MS values measured in protosols ps1 and ps4 (Figs. 3 and 5E, Table 4) are even higher and probably indicate ultrafine magnetite produced by magnetotactic bacteria, which could be responsible for the comparable values measured on Late Pleistocene loess deposits of the Cordão Formation (CF) and overlying Holocene organic-rich wetland deposits (Fig. 11B). The higher values of the protosols compared with the Holocene topsoil (Fig. 11B) probably result from the younger age (and lower degree of development) of the latter.

There is a positive correlation between the degree of rubification and MS (derived mostly from neoformed ferrimagnets) in soils derived from acid igneous rocks, whereas in soils derived from basic igneous rocks, the MS is a function of detrital ferrimagnets and there is weak correlation with the reddish color (Preetz et al., Reference Preetz, Igel, Hannam and Stadler2017). Because the ilmenite and magnetite in the heavy mineral assemblage indicate provenance from basic rocks, whereas the other minerals derive from acidic rocks (Silva, Reference Silva1979), the higher MS in Unit D points to increased in situ neoformation of ferrimagnetic minerals by stronger weathering processes, that is, the production of ultrafine magnetite by magnetotactic bacteria, and the oxidation of detrital magnetite into maghemite, part of which was likely modified into hematite, as indicated by the reddish color of the sediment.

Unit E above is probably formed of sediments deflated from Unit D, and its lower MS and very pale brown color (both comparable to Unit C) could have resulted from transformation of hematite to goethite (Schwertmann, Reference Schwertmann1971; Melo et al., Reference Melo, de Oliveira, Batista, Cherobim and Favaretto2020) or to illuviation/leaching of the magnetic minerals concentrated in the finer size classes, as suggested by the smaller silt and clay content (Fig. 3). However, the increasing upward trend of MS values (Fig. 3) suggests higher production of ultrafine magnetic minerals by magnetotactic bacteria associated with increased rainfall and expansion of vegetation (better developed than in the Holocene topsoil) during the Late Holocene.

Climate influence

The depositional processes and postdepositional modifications identified within the continental succession of Barrier III are directly or indirectly related to rainfall, temperature, and wind regime, which in turn are controlled by climate patterns on different timescales. The substantial time interval of some 90 ka represented by the succession implies that the depositional units accumulated and were modified under variable climate conditions that characterized the last glacial–interglacial cycle (Jouzel et al., Reference Jouzel, Masson-Delmotte, Cattani, Dreyfus, Falourd, Hoffmann and Minster2007). It is therefore possible to try to establish connections between the changes observed in the sediment layers and large-scale climate fluctuations during the last glacial period, which were controlled by orbital processes.

The marine–continental transition after the sea-level high stand of ∼101 ka inferred from the OSL ages at AC-01 and AC-03 (Fig. 12A and B) is marked by the subaerial accumulation of sediment, followed by wet conditions represented by the redoximorphic features and cut-and fill-structures in Unit B and by a subsurface layer of organic-rich sand and peat positioned at a similar altitude on top of lagoon margin deposits of Barrier III to the north of the study area (Lopes et al., Reference Lopes, Pereira, Caron, Dillenburg, Rosa, Barboza, Saviani, Sawakuchi, Tatumi and Yee2024; Fig. 1F). The estimated period of accumulation of Unit B, between <101 and >82 ka coincides with a period of decreased aeolian processes in southern South America (SSA; Fig. 12C), inferred from variations in the volume of dust deflated from the SSA and accumulated on Antarctic ice, as recorded in the EPICA Dome C (EDC) ice core (Lambert et al., Reference Lambert, Bigler, Steffensen, Hutterli and Fischer2012; Fig. 12C). The inferred higher rainfall during deposition of this unit could be related to strengthening of the South America Monsoon System (SAMS) driven by a cooling episode (Heinrich stadial 9, Fig. 12D) in the subtropical North Atlantic, as evidenced in speleothems from central-eastern and northeastern Brazil (Strikis et al., Reference Strikis, Cruz, Barreto, Naughton, Vuille, Cheng and Voelker2018). However, the high summer insolation and obliquity (Fig. 12E) at that time suggest an orbital control on rainfall. Speleothem records show that high summer insolation leads to increased summer rainfall over subtropical South America due to stronger SAMS (Cruz et al., Reference Cruz, Burns, Karmann, Sharp, Vuille, Cardoso, Ferrari, Dias and Viana2005; Wang et al., Reference Wang, Auler, Edwards, Cheng, Ito, Wang, Kong and Solheid2007), whereas higher obliquity enhances seasonal contrasts (van den Heuvel, Reference van den Heuvel1966; Williams, Reference Williams1993). This combination could have increased annual precipitation, leading to the development of wet environments on interdune areas, dominated by subaqueous accumulation and reduced aeolian remobilization and transport due to sediment saturation with water.

Figure 12.

(A) Timing of accumulation inferred from optically stimulated luminescence (OSL) ages of terrestrial and marine sediments obtained at outcrops AC-01 and 03 (Lopes et al., Reference Lopes, Pereira, Caron, Dillenburg, Rosa, Barboza, Saviani, Sawakuchi, Tatumi and Yee2024), electron spin resonance (ESR) ages of marine shells (Lopes et al., Reference Lopes, Pereira, Kinoshita, Mollemberg, Barbosa and Baffa2020), and thermoluminescence (TL; Poupeau et al., Reference Poupeau, Soliani, Rivera, Loss and Vasconcellos1988) and optically-stimulated luminescence (OSL; Dillenburg et al., Reference Dillenburg, Barboza, Rosa, Caron and Sawakuchi2017) ages from Barrier III. (B) Sea-surface temperatures (SSTs; Shakun et al., Reference Shakun, Lea, Lisiecki and Raymo2015), relative sea level (Spratt and Lisiecki Reference Spratt and Lisiecki2016), and sea-level high stands (HS) from (a) Lambeck and Chappell (Reference Lambeck and Chappell2001) and (b) Dillenburg et al. (Reference Dillenburg, Barboza, Rosa, Caron, Cancelli, Santos Fischer, Lopes and Ritter2020). (C) Dust concentrations in Antarctica from the Epica Dome C (EDC) ice core (Lambert et al., Reference Lambert, Bigler, Steffensen, Hutterli and Fischer2012). (D) Air temperature variations in Antarctica (Jouzel et al., Reference Jouzel, Masson-Delmotte, Cattani, Dreyfus, Falourd, Hoffmann and Minster2007) with Heinrich stadials (H1–12) indicated. (E) Orbital parameters (obliquity, summer insolation at 32°S, and eccentricity; Laskar et al. Reference Laskar, Robutel, Joutel, Gastineau, Correia and Levrard2004) and wet periods (WP) in Brazil (Wang et al. Reference Wang, Auler, Edwards, Cheng, Ito, Wang, Kong and Solheid2007); the dashed line indicates current average insolation and obliquity values.

Figure 12 Long description

This multi-panel scientific figure compares sediment chronologies with sea level, Antarctic dust and temperature and orbital forcing over the last 130 thousand years. Panel A shows sediment ages using different methods: optically stimulated luminescence (OSL), electron spin resonance (ESR) and thermoluminescence (TL) for terrestrial, marine and shell samples. Panel B is a line plot of relative sea level (RSL) in meters and sea surface temperature (SST) in degrees Celsius, showing fluctuations with notable peaks and troughs. Panel C presents dust concentration in Antarctica in micrograms per kilogram, with peaks corresponding to cold periods. Panel D displays temperature anomalies in degrees Celsius, highlighting Heinrich stadials H1 to H12 with shaded bands, indicating cooler periods. Panel E shows orbital parameters: obliquity in degrees, insolation in watts per square meter and eccentricity (unitless), with curves indicating cyclical changes. Key trends include low sea levels aligning with high dust concentrations and temperature anomalies during stadials, suggesting climatic interactions. Each panel′s x-axis represents time in thousand years ago (ka).

The sediments overlying the supersurface S1 indicate renewed aeolian accumulation represented by Unit C, possibly favored by the two sea-level high stands during MIS 5a (Fig. 12B). The strongly weathered layer at the base of Unit C (Fig. 5B) may be related to a wet period recorded in speleothems from northern and southern Brazil, attributed to the strengthening of the SAMS, possibly related to Heinrich stadial 8 (Fig. 12D), at ∼87 ka (Wang et al., Reference Wang, Auler, Edwards, Cheng, Ito, Wang, Kong and Solheid2007). Summer insolation at that time was similar to the contemporaneous values (Fig. 12E), resulting in higher summer rainfall, and seasonal contrasts were increased due to higher obliquity. In combination with cooler temperatures, higher rainfall may have increased hydrolysis of Fe-bearing minerals and, consequently, the formation of goethite and ferrihydrite at the base of Unit C. The pale color of the sediments overlying the basal layer suggests that pedogenic processes (hydrolysis of Fe-bearing minerals and soil bacterial activity) became less intense, probably due to lower temperatures and a relatively drier climate, favored by smaller insolation amplitude (modulated by low eccentricity), especially during periods of low summer insolation and/or decreased obliquity, which caused equatorward expansion of temperate climate belts (Williams, Reference Williams1993). These conditions could have prevented the extensive production of Fe oxides and the development of the Fe-Mn aggregates into larger masses.

As sea level fell and conditions became cooler and drier from MIS 5a onward, sediment transport and accumulation became controlled mainly by aeolian processes. Different geomorphological records (Clapperton, Reference Clapperton1993; Iriondo, Reference Iriondo1999; Zárate and Tripaldi, Reference Zárate and Tripaldi2012; Peltier et al., Reference Peltier, Kaplan, Birkel, Soteres, Sagredo, Aravena, Araos, Moreno, Schwartz and Schaefer2021) and dust from SSA transported to Antarctica (Fig. 12C) show that aeolian processes in SSA were intensified, especially during the stadials MIS 4 and 2 (Fig. 12D), represented in the southern CPRS by deposition of windblown silt (Lopes et al., Reference Lopes, Dillenburg and Schultz2016). Drier climate during the last glacial would have been favored by decreased ocean-sourced moisture due to lower sea levels (Fig. 12B) that shifted the coastline tens of kilometers to the east and by lower sea-surface (SST; Fig. 12B) and atmospheric (Fig. 12D) temperatures, which reduced evaporation (Jouzel et al., Reference Jouzel, Masson-Delmotte, Cattani, Dreyfus, Falourd, Hoffmann and Minster2007; Shakun et al., Reference Shakun, Lea, Lisiecki and Raymo2015; Spratt and Lisiecki, Reference Spratt and Lisiecki2016). The lower estimated accumulation rates between ∼71.8 and ∼41.6 ka, in addition to being influenced by a diminished supply of coastal sand due to lower sea level, relatively weaker aeolian processes during MIS 3 (Fig. 12D), and development of protosols that stabilized the sediments, could have been affected by relatively moist conditions during MIS 4 suggested by the greater extent of Andean glaciers at that time (Peltier et al., Reference Peltier, Kaplan, Birkel, Soteres, Sagredo, Aravena, Araos, Moreno, Schwartz and Schaefer2021).

The expansion of vegetation above the level dated as ∼71.8 ka indicated by the more common brownish laminations and development of thicker protosols (Fig. 5E and F) could have been favored by increased soil moisture during MIS 4 and especially during the wetter and warmer MIS 3 (Gasparini et al., Reference Gasparini, Rabassa, Deschamps and Tonni2016), episodes of strengthened SASM (Wang et al., Reference Wang, Auler, Edwards, Cheng, Ito, Wang, Kong and Solheid2007), and/or fluctuations of the water table driven by sea-level oscillations (Lambeck and Chappell, Reference Lambeck and Chappell2001; Dillenburg et al., Reference Dillenburg, Barboza, Rosa, Caron, Cancelli, Santos Fischer, Lopes and Ritter2020) (Fig. 12B and C). The cyclicity inferred from ps4 (Fig. 5E and F) suggests enhanced seasonality, driven by increasing summer insolation/rainfall and obliquity between ∼60 and 41 ka (Fig. 12D) and episodes of strengthened SASM at 47.6, 43.5, and 42.1 ka (Wang et al., Reference Wang, Auler, Edwards, Cheng, Ito, Wang, Kong and Solheid2007; Fig. 12C). Moreover, the higher MS of ps4 compared to ps1 (Fig. 11B) also points to warmer/wetter conditions with higher soil bacterial activity. Except for the protosols, plant cover does not seem to have been extensive, which would have allowed sand mobility and accumulation predominantly in the form of TSSs, similar to the unvegetated interdune flats of the Albardão dunefield (Figs. 9D and F).

Although the sea level in southern Brazil rose to −23 to −5 m relative to the present between 47.7 and 36.2 ka BP (Dillenburg et al., Reference Dillenburg, Barboza, Rosa, Caron, Cancelli, Santos Fischer, Lopes and Ritter2020), it is not clear whether this high stand contributed to onshore sand accumulation. Nevertheless, aeolian processes increased during MIS 2 (Fig. 12D), especially during the driest and coldest last glacial maximum (LGM), and progressively decreased, although accumulation seems to have lasted up to ∼10.5 ka based on the OSL age obtained at the top of Unit D (Fig. 3). The predominance of TSSs from the LGM up to ∼6 ka in the coast of Santa Catarina indicate dry conditions, whereas the development of younger dunefields indicates expansion of vegetation and increased moisture (Rodrigues et al., Reference Rodrigues, Giannini, Fornari and Sawakuchi2020).

The change from brownish-yellow colors of Unit C to the reddish-brown color of Unit D is also observed in coeval deposits of Barrier III in the northern CPRS (Poupeau et al., Reference Poupeau, Soliani, Rivera, Loss and Vasconcellos1988; Tomazelli and Dillenburg, Reference Tomazelli and Dillenburg2007), indicating that rubification was a regional-scale phenomenon. Although reddening is regarded as a function of time (chronofunction sensu Catt, Reference Catt1990), the rubification of the younger Unit D indicates this process was instead controlled by climate (climofunction). The rubification and coeval increased MS were probably driven by a change from a cold glacial climate to a warmer climate around 10 ka. The mottles in D1, larger Fe-Mn masses/concretions nuclei, and clay formation-iluviation in D2 point to increased moisture, whereas rubification, MS increase, and development of a Fe ring around the Mn nuclei are consistent with a shift to drier conditions. This change could be related to a period of increased SASM around 12 ka (Wang et al., Reference Wang, Auler, Edwards, Cheng, Ito, Wang, Kong and Solheid2007) followed by progressively drier conditions due to enhanced seasonality, driven by increasing obliquity that reached a maximum at 10–9 ka (Laskar et al., Reference Laskar, Robutel, Joutel, Gastineau, Correia and Levrard2004; Fig. 12C) and less summer rainfall due to decreased insolation (minimum at 11–10 ka), which could have resulted in contrasting wet (winter–spring) and dry (summer–autumn) seasons.

The modification of Unit D may also have been favored by higher temperatures that characterize the Hypsithermal, or Holocene Climatic Optimum (HCO), identified in terrestrial and marine records and ice cores (Deevey and Flint, Reference Deevey and Flint1957; Ciais et al., Reference Ciais, Petit, Jouzel, Lorius, Barkov, Lipenkov and Nicolaïev1992; Iriondo, Reference Iriondo1999; Masson et al., Reference Masson, Vimeux, Jouzel, Morgan, Delmotte, Ciais and Hammer2000). Ice cores show that the HCO occurred between ∼10 and 6 ka, when global temperatures reached a maximum of ∼1°C above the average values of the 1990s. In the Southern Hemisphere, it was characterized by oscillations of temperature, with two warming intervals at ∼11–9 ka and 8–6 ka intercalated with minima at ∼10.5 and 8 ka, with a peak warming (“Altithermal”) at 7.4 ± 3.7 ka BP (Ciais et al., Reference Ciais, Petit, Jouzel, Lorius, Barkov, Lipenkov and Nicolaïev1992; Masson et al., Reference Masson, Vimeux, Jouzel, Morgan, Delmotte, Ciais and Hammer2000; Shakun and Carlson, Reference Shakun and Carlson2010). Diatom assemblages from the Southern Ocean indicate a 1–2°C increase in SSTs between ∼12.5 and 6.2 cal ka BP intercalated with cold episodes centered at 10.3 and 9.3 cal ka BP (Nielsen et al., Reference Nielsen, Koç and Crosta2004). In the Argentinean Pampas, increased pedogenesis and mobilization of Fe in the soils between 8.5 and 3.5 ka BP also point to a warmer (MAT > 20°C) and wetter (MAP >2000 mm/yr) climate, followed by a cooling trend from 3.5 to 1.5 ka BP (Iriondo et al., Reference Iriondo, Brunetto and Kröhling2009).

Aeolian activity was resumed in the last 1000 years, as indicated by the OSL age (Fig. 3), alternated with (probably wet) intervals when vegetation cover expanded, as indicated by the protosols within Unit E (Fig. 6H). Aeolian processes were active up to the last century (Fig. 13A) but the surface of Barrier III is now mostly vegetated (Fig. 13B). The expansion of vegetation and consequent stabilization of aeolian sediments during the last century are also recorded in coastal dunefields along the CPRS (Martinho et al., Reference Martinho, Dillenburg and Hesp2008, Reference Martinho, Hesp and Dillenburg2010; Silva et al., Reference Silva, Hesp, Keim, Martinho and Ferligoj2013) and in Santa Catarina (Silva and Hesp, Reference Silva and Hesp2013; Mendes and Giannini, Reference Mendes and Giannini2015). These may be related to decreased seasonal contrasts due to lower obliquity and an intensification of the SASM driven by high insolation (Fig. 12C), resulting in higher summer rainfall over southern Brazil.

Figure 13.

Aerial photograph of the study area in 1967 (A) and Google Earth image today (B) showing the expansion of vegetation over the dunes of Barrier III in the last decades.

Figure 13 Long description

Aerial images of a coastal region near the Atlantic Ocean. A) Shows Lagoon III, Barrier III and active dunes with Chuy Creek flowing through. Dashed lines outline the barriers and creek. B) Displays the same area with marked points AC-01, AC-02 and AC-03. A scale bar indicates 1 km and a north arrow is present. The images highlight the spatial arrangement of natural features and human-modified landscapes.