Central Santa Catarina coastal dunefields chronology and their relation to relative sea level and climatic changes

During the past decades, there have been contrarian explanations for the formation and stabilization of coastal dunefields: while many authors believe the dunes formation would be enhanced by falling sea level, others argue that a rising or stable sea level context would be favorable. For Brazilian coastal dunefields, the second hypothesis seems to be more consistent with the luminescence ages found so far; however, most of these data were obtained without using the SAR protocol. Another point of concern is the role of climate change in the aeolian system, which is still not very clear. The aim of this paper is to try to clarify these two questions. To this end, five coastal dunefields were selected in central Santa Catarina coast. The remote sensing and dating results allowed the discrimination and mapping of at least four aeolian generations. Their age distribution in relation to the global curve of relative sea level variation during the Late Pleistocene allows us to suggest that the formation of Aeolian dunefields in the coastal context is supported by stable relative sea level. However, relative sea level is not the only determinant for the formation and preservation of the aeolian coastal dunes. Evidences of climatic control indicate that the initiation of dunefields would be favored by periods of less humidity while their stabilization would occur preferably during the periods of rain intensification, connected to monsoon activity.


INTRODUCTION
Santa Catarina central coast, between Ouvidor and Florianópolis, has five active dunefields with 1 to 3 km length (Fig. 1).Some of these dunefields were previously studied with emphasis on the geomorphology and sedimentology of their deposits (Bigarella 1975, Bigarella et al. 2005, Miot da Silva 2006, Hesp et al. 2009, Dillenburg et al. 2014).No work has been published for this region, focussing on the chronology of dunefield evolution throughout the late Quaternary, except for attempts by Hesp et al. (2009) to distinguish aeolian Holocene pulses in Pinheira strandplain.This kind of focus is becoming more common in both national (Giannini & Suguio 1994, Barbosa 1997, Dillenburg et al. 2006, Giannini et al. 2007, Barreto et al. 2002, 2004, Martinho & Giannini 2001, Claudino-Sales & Peulvast 2002, Sawakuchi et al. 2008) and international research (Pye 1983, Endo 1986, Short 1988, Stokes et al. 1997, Loope & Arbogast 2000, Lees 2006), in particular during the last two decades.The main reasons for the renewed interest in this subject are the possible implications of the different generations of aeolian dunefields and their time-space distribution for the study of sedimentary supply, climate and relative sea level (RSL) behavior.
Similar studies were performed in the region of Jaguaruna-Imbituba (Giannini 1993, Sawakuchi 2003, Giannini et al. 2007), to the south of the present study area.The results of those studies allowed recognizing at least four dunefield generations.Morphostratigraphic interpretations, reinforced by the results of luminescence dating in this region, led Giannini et al. (2007) to suggest a model for the formation and preferential preservation of coastal aeolian deposits in terminal transgressive and high sea system tracts.The average ages obtained by luminescence in Santa Catarina centre-south (Giannini et al. 2007) and Rio Grande do Norte (Barreto et al. 2004) are in reasonable agreement.These results seem to strengthen the hypothesis of inter-regional correlation between the generations of Aeolian dunefields on the Brazilian coast (Giannini 2007), which would be sea level controlled.
However, one question that arises from this hypothesis of dunefield formation deserves some refinement or revision: the two youngest aeolian generations found in Santa Catarina centre-south, in Rio Grande do Norte and in other Brazilian coastal sectors were formed during the last millennia, after the Holocene maximum RSL (around 5.5 ky BP).Thus, factors that control the coastal dunefield dynamics are still not fully understood, since one changes from the tens of thousandsof yearstimescale (in which the influence of RSL variation of more than 100 m seems relevant) to the millennial or decadal scale (in which the RSL variation accounts only for less than 3 m).Two questions were addressed in the present study: Are the climatic oscillations in southern South America (Cruz et al. 2006, 2007, 2009, Wang et al. 2006, Haylock et al. 2006), which are connected to sub-Milankovitch events, such as the Heinrich events (Heinrich 1988) and the Little Ice Age (LIA), the main factors controlling coastal dunefield dynamics on a short timescale?Does the beginning and stabilization of dunefields mean a defined climate throughout the late Holocene?
To address these questions, the Santa Catarina central coast was chosen for two reasons.Firstly, because the area is immediately adjacent in the north to the region between Imbituba and Jaguaruna, which has been previously subject to the above mentioned studies about stratigraphy and chronology of coastal dunefields (Giannini et al. 2007).Secondly, because it covers precisely the limit between the dunefields migrating to SW (Fig. 1), that are predominant from Pinheira toward south, and the dunes and blowouts migrating to N to NW, that are found from Pinheira to north in south-southeast Brazil, especially at Santa Catarina island (Hesp et al. 2007) and São Francisco do Sul (Zular 2011), in Santa Catarina State, and at Ilha Comprida (Sawakuchi et al. 2008, Guedes et al. 2011) in São Paulo State.

Geologic and geomorphologic features
The study area is located in the second southernmost physiographic-structural sector of the Brazilian coast (Giannini 2007), which stretches with a general NNE orientation from Imbituba (28° 15' S) to Itajaí-Joinville (26°15' S).In comparison with its northern and southern neighbors, this coastal sector is more interrupted by headlands and with a narrower and steeper inner continental shelf (0.07 to 0.26°).
According to Martin et al. (1988), the study area is a coastal plain formed by four main deposit types: two marine sandy terraces, with Pleistocene and Holocene ages; coastal dunes, mainly from the Holocene; and clayey or silty sands, also from the Holocene.The Pleistocene terraces, related to the high RSL 120,000 yr BP, are preserved in the innermost portions of the coastal plain, where, according to those authors, they could reach about 9.5 m high.The Holocene sandy terraces occur along the outer edge of the Pleistocene terraces, sometimes separated from them by low lands and marshlands (Martin et al. 1988) related to partially filled lagoons and estuaries.According to the authors, they can reach 4 m high.The clayey and silty sands are mostly found at south from Pinheira in small bays and lowlands around the estuarine and lagoon systems.In Pinheira strandplain, the Holocene terrace clearly corresponds to a strandplain, as evidenced by the sets of foredune ridges (Hesp et al. 2009, Barboza et al. 2009, Hein et al. 2013, Dillenburg et al. 2014, Hein et al. 2014).
Coastal dunefields become well developed from the studied area to the south coast of Brazil (Bigarella 1975).Based on stratigraphic, geomorphologic, mineralogical and textural features, Giannini (1993Giannini ( , 2002) ) and Giannini et al. (2005) could recognize at least four aeolian dune generations in south Santa Catarina State.Giannini et al. (2007) grouped them in two sequences, indicating a major RSL influence in their formation and, therefore, a regional expression.The generation 1 (G1) was formed before the last Pleistocene highstand (120 ka).The generation 2 (G2) comprises dunefields formed before the Holocene highstand, while generations 3 and 4 (G3 and G4) were formed after this highstand.The difference between these last two generations is that G4 is still active at present.
Santa Catarina coast has more than 30 dated indicators of RSL positions during the Holocene, obtained in vermetid (encrusting marine gastropods) remains (Angulo et al. 2006).These data indicate a maximum RSL of 2.1 ± 1.0 m higher than the present RSL, reached 5916-5597 14 C cal years before present (BP) (Angulo et al. 2006).

Climate and oceanography
According to Cruz et al. (2006Cruz et al. ( , 2009) ) and Garreaud et al. (2009), there are two important sources for the moisture involved in southern South America precipitation.The first one, named South American Monsoon System (SAMS), is related to the Intertropical Convergence Zone (ITCZ) and the South Atlantic Convergence Zone (SACZ), and is intensified during austral summer.It has major importance for the rainfall in most of South America, being responsible for approximately 60% of Santa Catarina state precipitation (Cruz et al. 2007).The second source of moisture, called frontal system (cold fronts), is responsible for the winter rains and reaches the Brazilian south and southeast coast.These climatic phenomena result from migratory cyclones along the subtropical Atlantic coast (Vera et al. 2002).Both climatic systems affect the wind patterns (Fig. 2): the stronger S winds are associated with the cold front system, as the intensification of SACZ pushes the N wind offshore.
Imbituba's harbor tidal board marks the mean astronomical tide oscillation at 0.6 m within a semi-diurnal regime (Giannini 1993).Based on the records (2002)(2003) of the Arvoredo Island, located 35 km offshore from the island of Santa Catarina, Hesp et al. (2009) identified two trends in the significant wave height and direction distribution of swells arriving at the central coast of Santa Catarina.The first trend, most common and related to large-scale climatic phenomena, is represented by swells from S (17), with an average period of 12s and height of 1.0 to 1.5 m.The second one is related to air masses generated locally and translates into waves coming from ENE (70° azimuth), with a 7 s mean period and up to 1.0 m height.
Geomorphological, granulometric and mineralogical evidences, applied in different beaches in the sector between Laguna and Imbituba (Giannini 1993, 2002, 2007, Martinho et al. 2006), indicate predominance of net longshore sediment transport from S-SW to N-NE.

Remote sensing
Aerial photographs and satellite imagery were used to identify aeolian morphologic elements and to map morphostratigraphic units (possible generations of aeolian deposits).The historical series of aerial photographs includes the years 1938 (available only for the Ingleses, Joaquina and Pinheira dunefields), 1957 and 1978.These photos were georeferenced and compared with each other and with satellite images (2001 to 2009) from the Google Earth ® software.

OSL dating
The OSL method was applied to quartz grains from 13 sand samples, 11 of which are aeolian paleodunes, and 2 are marine constructive terraces.These samples were selected among 119 samples collected during the fieldwork.The sample preparation followed the standard coarse-grain quartz extraction procedure, including treatment with HCl and H 2 O 2 , etching with HF, and density separation, all under red light.
The OSL-SAR procedure described by Murray and Wintle (2000) and Wintle and Murray (2006) was applied to quartz grains of 120 -150 µm diameter, which is the most accepted procedure for the determination of the radiation dose accumulated in quartz grains.The luminescence signals were measured with a Risø DA-15 TL/OSL system at the Radiation Dosimetry Lab at Oklahoma State University -USA.The 90 Sr/ 90 Y beta source attached to this equipment provided a dose rate of 93.2 ± 4.1 mGy/s.The heating rate used was 5°C/s, and all aliquots were prepared with the same volume (~400 grains).
The annual radiation dose rate was measured with a high purity germanium detector (HPGe), at the Physics Institute of Universidade de São Paulo.The contribution of the cosmic radiation to the annual dose was calculated according to Barbouti and Rasin (1983) and Prescott and Stephan (1982), using latitude, longitude, altitude and collection depth, as well as the density of each sample.Total error of the annual dose was estimated according to the Gaussian law of error propagation.

OSL dating
Out of the 11 OSL ages obtained from aeolian samples, five belong to G3, four to G2 and two to G1.Out of the two OSL ages from marine terraces, one is Pleistocene (IN-TP-06) and the other is Holocene (PI-TH-49, Tab. 1).
The G1 samples present two ages that overlap within error limits, with a range between 350 to 240 ky.The four ages present in G2 appear distributed over three age ranges.
The oldest one comprises one sample and is located around 50 ky.The second range comprises two samples, around 21 ky, and the last range has one sample of about 10 ky.
The five dates for G3 are distributed over two temporal ranges, the first and oldest one is 5.5 ± 0.8 ky.The second temporal range comprises four samples and is almost continuous, from about 1,900 to 600 years.There is a hiatus between these ranges at around 2.5 ky.

Distribution of the Quaternary sedimentary units
The Quaternary aeolian sedimentation of Santa Catarina central coast can be divided into four generations (Figs. 3 to 7), matching the ones observed by Giannini et al. (2007).
The oldest generation G1 occurs associated with crystalline rock hills (Figs. 3, 4 and 6).It appears dissected by drainages so the aeolian features are not recognizable neither in satellite images and aerial photos, nor in the field.In the Ingleses dunefield area (Fig. 3) and at the south portion of Garopaba region (Fig. 6), the identification of G1 was mostly based on morphologic criteria (absence of depositional forms) as suggested by Giannini et al. (2007).In the Joaquina dunefield area (Fig. 4) and in the north portion of Garopaba (Fig. 6), G1 was also indicated by OSL dates.The new information found in the present study, compared to what had been described so far for this generation, is the existence of ages older than the last Pleistocene highstand (about 120 ky).These ages (274 ± 25 and 318 ± 34 ky, respectively Fig. 4 and 6) are also the oldest found for Quaternary sedimentary coastal deposits in all of the south and southeast Brazilian coast, at least for published OSL-SAR dating results.G2 occurs as softly wavy terrain (undulations about 3 m high and 150 m long), with a medium altitude around 25 m, covering the Pleistocene terraces from 10 m high, attributed to the high RSL of 120 ky BP.As far as area is concerned, it is the most representative aeolian generation for the region and represents G3 and G4 substrate in the Ingleses (Fig. 3), Joaquina (Fig. 4) and Garopaba (Fig. 6) dunefields, although aeolian features cannot be recognized in the morphology there.However, it was possible to identify sedimentary structures typical of wind deposition in outcrops, i.e., trough cross stratifications, in sets up to 3 m thick (Fig. 8).The festoon cross-stratification pattern, found in the section WNW-ESE, is compatible with the migration direction of the barchanoid aeolian chains, which is SSW_NNE in this region.Eighteen measurements of cross stratification confirm the dominance of an N 17º direction paleowind.Waved terrains with similar characteristics, and with possible tenuous parabolic shapes in aerial photos, were already described in the regions of Imbituba and Jaguaruna, center-south littoral of Santa Catarina, by Giannini (1993), who classified them as G2.
G3 has a well preserved aeolian morphology and it is found in the surroundings of active dunefields (G4) as precipitation ridges with a parallel orientation to the present dunefield (Fig. 8).An exception is in Pinheira, where different G3 dunefield pulses appear dispersed on the south portion of Holocene strandplain (Fig. 5).Stratigraphically, G3 occurs over the pre-Cenozoic crystalline basement, above the Pleistocene marine terrace and, to a greater extension, above the sandy terrain of wavy surface identified as G2.
The active dunefields (G4) show a spatial distribution related to the G3 deposits, which are sometimes partially overlapped by them.G4 appears surrounded almost exclusively by G3 and suffers a stabilization process by vegetation growth in the deflation zone, and by the consequent reduction of sedimentary supply for the more distal wind features, such as depositional frontal lobes and precipitation ridges (Mendes & Giannini 2015).

86
Central Santa Catarina coastal dunefields chronology The paleolagoon corresponds to the area where the Pleistocene terrace and its overlapping G2 paleodunes have been drowned (and subsequently silted) since the Holocene maximum inundation.It appears to have formed preferably in locations with incised drainages (incised valleys) under a lower than present base level, as exemplified by the extended occurrence of its deposits throughout the contact between the crystalline rocks and the Pleistocene terrace and G2 sedimentary deposits (Fig. 3).

Geological significance of OSL ages in aeolian deposits
In theory, the OSL age represents the time of sediment burial and, if obtained from the most distal depositional lobe of the dunefield, it provides a good estimate for the stabilization of the aeolian system.Nevertheless, depending on the uncertainty of the OSL method and the duration ("lifetime") of the dunefield active period, there is a probability that the time frame obtained with OSL does not only include the stabilization ("death") of the dunefield but also its initiation ("birth").The greater the sand volume of the dunefield, the longer is its active period.Therefore, in relatively small coastal dunefields as the ones studied here, the life-time of the aeolian system tends to be shorter than the precision of the OSL dating method (like few hundred years), which for our samples is close to 10%, so the dunefield initiation can be included in the age as well as the stabilization.This should be viewed therefore as a general premise to the two following topics of discussion.

Age distribution in relation to RSL variations
The aim was to characterize the RSL behavior during the periods that coincide with the obtained OSL ages of the aeolian dunes.Five situations were distinguished, using the available OSL results (Fig. 9): 1. strongly ascending RSL; 2. strongly declining RSL; 3. stable RSL with gentle decrease; 4. stable high RSL linked to tendency inversion (positive peak of the variation curve); and 5. stable low RSL, linked to tendency inversion (negative peak of the variation curve).
From the nine available youngest OSL ages, and considering the respective error margins, the distribution of the ages is as follows: one in the case 1, one in the case 2, five in the case 3, one in the case 4 and one in the case 5.The two oldest ages have too large error margins and therefore fit several cases at the same time, making their classification regarding RSL behavior ambiguous.As mentioned in section "Geological significance of OSL ages in Aeolian deposits", this is an inherent limitation of OSL dating, since the error grows proportionally with the determined age.
There is a clear predominance of stable RSL cases (seven cases out of nine), mainly in gentle decrease but also in tendency of inversion.Five out of these seven cases refer to dunes formed with an RSL similar to the present, in a variation margin of about 3 m, all posterior to the last transgressive maximum (G3).This high representation of G3 dunes is surely caused by a sampling bias.Stabilized dunes formed with high RSL, similar to the present, and at coast lines close to the present ones are more or better exposed at the present coastal zone than dunes formed under RSL and at coast line positions very discrepant from the present ones (Giannini 2007, Giannini et al. 2007, Guedes 2012).Their presence can be favored also by a preservation bias, where more recent deposits generally are better preserved than older ones.
Nevertheless, stable RSL conditions are as predominant among the younger ages (G3) as among the oldest ages (G2).So, stable RSL seems to be in fact an effective controlling factor for aeolian dune formation and/or preservation.A point in favor of the possible connection between stable RSL and aeolian supply, defended by Giannini et al. (2001Giannini et al. ( , 2007)), is the maintenance of the distance between the forming dunefields and their source areas at the coastline.The stability of this distance would constitute a favorable factor in the formation or maintenance of the active dunefields.The authors' argument is that the stability of the coastline, not assured but favored by a stable RSL, is a necessary condition for the activity of the aeolian system, since wind transport and deposition dilute throughout a wide coastal area,if the coastline varies quickly.
The deposits of the two oldest generations were reached by the sea during the two latest maximum floodings of the Brazilian coast (Giannini 1993).Giannini (2007) and Giannini et al. (2007) considered this as evidence that these generations were formed, at least partially, under a transgressive terminal context (shortly before the maximum inundation).They based this assumption on the premise that, because these paleodunes are close to the present coast, they were generated at similar shorelines and RSL as present.In these authors' model, the two youngest generations, attributed to the regression that followed the maximum Holocene flood, would have still formed in the same high RSL context responsible for the previous generations, however without reaching such internalization.Following the reasoning of the favorability of the shoreline stability in the formation of dunefields, proposed by Giannini (2007) and Giannini et al. (2007), the stable shoreline periods, during glacial maxima, and therefore with a low RSL, would also be favorable to the development of coastal aeolian systems (Guedes 2012).However, there is a relative higher difficulty for the formation of coastal dunefields during high sea level, when compared to ones formed during low sea level, which relates to the relative duration of the RSL maintenance.At least during the last million years of Earth history, when Milankovitch orbital cyclicity is dominated by the component of eccentricity variation, with time periods around 100 thousand years, the interglacial periods are shorter in time than the glacial (Ruddimann 2008).Therefore, the high RSL periods are also less durable than the low RSL periods.That means that the profitable time to accumulate big aeolian deposits with coastal source is much shorter under high sea conditions than under low sea level conditions.
Another exception to the model of predominance of aeolian dunes under high RSL relates to the preservation question.Due to the cyclic character of the RSL variations, connected to the succession of glacial and interglacial stages, a dunefield formed under high RSL conditions tends to be, in the passage to a new glacial stage, submitted to a substantial fall of the base level of erosion, due to the lowering of RSL and groundwater.This base level fall would hinder the preservation of its deposits for a long time, i.e., for a time compatible with an eccentricity cycle.Dunefields formed under a high RSL would be more subject to subaerial erosion, mainly because of the subsequent RSL fall (forced regression).On the other hand, the deposits formed during a low RSL could be reworked by the shoreline transgression.Meanwhile, if that transgression is a result of the RSL rise with a high rate, only the superficial aeolian deposits would be reworked, which would considerably increases their preservation potential in geological time (Guedes 2012).

Age distribution in relation to climate variations
Nine from the 11 dated aeolian dunes have their age distribution analyzed in relation to the more accepted climate variations during the last 120 ky (there are no climatic reconstructions for the period of the two oldest samples).The OSL ages were plotted inside graphics of δ 18 O variation over time, obtained in southern Brazil speleothems (Figs. 10 and 11).In these graphics, the more negative isotope ratio values, connected to the rain amount effect, result from the SAMS intensification and, therefore, they are correlated with greater atmospheric humidity values (Cruz et al. 2006(Cruz et al. , 2007(Cruz et al. , 2009)).The aeolian ages distribution analysis was made separately in two different timescales.In the first and most embracing, corresponding to only five of the older samples, the ages are all related to climate transition moments, from less to more humid and/or humidity minima followed by humidity maxima (Fig. 10).
Transition periods, with a humidity increase, and consequently with a sand moisture increase, and periods of minimum moisture followed by its quick increase would be favorable to the development and stabilization of aeolian dunes for two main reasons.The first, connected to the dunefield formation, would be the decrease of the vegetation cover, with the resulting encouragement of wind transport (Carter 1990, Hesp & Thom 1990, Hesp 2000).The second reason, connected to the stabilization of the aeolian dunes, would be a possible reduction of wind intensity with rain increase.Similar effect is observed in the recent stabilization of the central coastal dunes of Santa Catarina (Miot da Silva & Hesp 2013, Mendes & Giannini 2015) and medium north of Rio Grande do Sul (Martinho et al. 2010).Weather data from the last century points to the rise of the annual precipitation associated with a decrease of the wind intensity as the main control factor of stabilization observed on the dunefields of the Santa Catarina coast.In this context, the dune stabilization would be supported by the combined effects of: 1. the vegetation grown and/or development of wetlands on the deflation zones; 2. the increase of the transportation threshold due to the sand cohesion induced by the moisture increase; and 3. the decrease of the wind intensity (Mendes & Giannini 2015).The second timescale is comprised inside the first one as slight oscillations, representative of the secular climate change.It is observed more easily during the three last millennia, corresponding to G3.The error involved in the dates is from about hundred to tens of years.On this timescale, three of the four youngest OSL ages include falling or negative peaks of humidity, with humidity increase in the end (Fig. 11), a pattern similar to that found in the first timescale (Fig. 10).This way, the mechanism of climate control for the formation and stabilization of dunes inferred from the longer timescale would also act on this shorter timescale.This coincidence implies, as a general rule, that the dunefields would be generated during moments of monsoon weakening and stabilized during its intensification.
Summarizing, in the climate variation context, the dunefields would be preferentially formed during the less humid climate moments, connected to the weakening of the monsoons, and stabilized during the subsequent more humid moments, of monsoon strengthening.Based on the available OSL-SAR ages, this pattern is apparently valid, both on the tens of thousands of years scale, as on the last thousands of years scale, when the peaks of less humidity can be represented, for example, by MCA and CWP (Fig. 11), and on the most humidity ones, by the Bond events and the LIA.

CONCLUSIONS
Based on the results obtained and discussed in this article, the main conclusions can be summarized as follows: 1.Four aeolian generations are recognizable based on remote sensing, field work and OSL dating.They are correlated with the generations previously defined for the centersouth coast of Santa Catarina and possibly extendable to other coastal regions with aeolian dunefields in Brazil.2. Based on the OSL ages, it is possible to infer at least three stages of dune formation inside the older generation (G1), which would correspond to high relative sea levels since the Late Pleistocene highstand (120 ky).Therefore, from the chronology point of view, the G1 unit could be divided into three units.3. The relation between the aeolian generations age distribution and the global curve of RSL variation during the Late Pleistocene allows to suggest that the formation of aeolian dunefields of coastal context is favored by stable RSL, and their preservation is helped by the subsequent increase of RSL.However, RSL clearly is not the only determinant for the formation and preservation of the aeolian coastal dunes, given the evidences of climatic control.The formation of dunefields would be favored by periods of less humidity while their stabilization would occur preferably during the periods of rain intensification, connected to monsoon activity.

Figure 2 .
Figure 2. Precipitation and 925 hPa wind map (left) and surface winds (right) for South America.Squares show the study area.During austral summer, (A) the SACZ is intensified and pushes the northern winds offshore; on the other hand, during austral winter, (B) the cold fronts and the associated southern winds (left maps modified from Garreaud et al. 2009; right images modified from http://earth.nullschool.net).SACZ: South Atlantic Convergence Zone; ITCZ: Intertropical Convergence Zone

Figure 3 .48°29Figure 4 .
Figure 3. Distribution of the Quaternary sedimentary units for the Ingleses dunefield area.The dated samples are labeled with the sample name and the OSL age.

Figure 5 .48°39Figure 6 .Figure 7 .
Figure 5. Distribution of the Quaternary sedimentary units for the Pinheira strandplain region.The dated samples are labeled with the sample name and the OSL age.

Figure 8 .
Figure 8. Aerial and outcrop aspects of the main depositional lobe in the Ingleses dunefield.Left: G3 precipitation ridges parallel to the G4 (active) dunefield (image from Google Earth ® ).Right: Contact (highlighted by the dashed line) between the Pleistocene marine terrace, with external tabular shape and sub-horizontal pedogenic impregnation layers, and G2, with a wavy top and trough cross stratifications in festoon pattern.Upper right box: circular histogram of cross strata dip directions (18 measurements), indicating average azimuth of N 17º.

Figure 10 .
Figure 10.Distribution of the OSL ages for the aeolian deposit samples in relation to the climate variations registered in stalagmites from Botuverá cave(Cruz et al. 2007), sited in Santa Catarina State, about 80 km west of the study area.Apparently, the samples are placed in moments of curve decline (humidity increase) and/or in its positive peaks (humidity minima), followed by a quick decrease (abrupt humidity augment).

Figure 11 .
Figure 11.Distribution of the dated samples (ages with error bars) in relation to the high resolution δ 18 O record, from the Cristal Cave (about 300 km NNW from study site) (adapted from Vuille et al. 2012).

Table 1 .
Relation of the dated samples with the corresponding depositional facies, ages, associated error and sedimentary units