Late Quaternary Variations in the South American Monsoon System as Inferred by Speleothems—New Perspectives Using the SISAL Database

Abstract

Here we present an overview of speleothem δ¹⁸O records from South America, most of which are available in the Speleothem Isotopes Synthesis and Analysis (SISAL_v1) database. South American tropical and subtropical speleothem δ¹⁸O time series are primarily interpreted to reflect changes in precipitation amount, the amount effect, and consequently history of convection intensity variability of convergence zones such as the Intertropical Convergence Zone (ITCZ) and the South America Monsoon System (SAMS). We investigate past hydroclimate scenarios in South America related to the South American Monsoon System in three different time periods: Late Pleistocene, Holocene, and the last two millennia. Precession driven summertime insolation is the main driver of convective variability over the continent during the last 120 kyrs (from present day to 120 kyrs BP), including the Holocene. However, there is a dipole between speleothem δ¹⁸O records from western and eastern South America. Records located in the central region of Brazil are weakly affected by insolation-driven variability, and instead are more susceptible to the variability associated with the South Atlantic Convergence Zone (SACZ). Cold episodic events in the Northern Hemisphere, such as Heinrich and Bond Events, and the Little Ice Age, increase the convective activity of the SAMS, resulting in increased precipitation amount in South America.

1. Introduction

Rainfall variability in South America is primarily driven by the Intertropical Convergence Zone (ITCZ) [1], the South American Monsoon System (SAMS) [2], and to a lesser extent, other forcings such as South Atlantic cyclones (SACs), the Southern Hemisphere westerly winds (SHWW), and past Atlantic and Pacific Ocean sea surface temperatures. To gain a deeper understanding of past variations of South American climate, it is essential to reconstruct past changes of South American hydroclimate in space and time, inter regional and continental teleconnections from these reconstructions, and identify the underlying mechanisms driving climatic variability. Understanding past climate variability in South America is pertinent to forecasting future hydroclimate variability in response to global climate change. Further, climate reconstructions are essential in understanding ecological and biogeographical responses to past hydrologic change in regions such as the Amazon rainforest, Pantanal, caatinga (Brazilian savannah), Cerrado, and Atlantic forests, which are among the regions with the richest biodiversity on Earth.

Speleothems are one of the most powerful terrestrial archives of past hydroclimate globally. The Speleothem Isotopes Synthesis and Analysis (SISAL) database [3,4] (for details see [5]) is a major contribution to the scientific community in its effort to explore these scientific questions. The SISAL aims to collect all available speleothem data into a single database, allowing for spatio-temporal analyses of time periods, to compare on the one hand new reconstructions with existing speleothem records and on the other hand to test new concepts. Here, we explore the available South American records in the current SISAL database (SISAL_v1) [3] and incorporate some other speleothem records that have not been included in SISAL_v1 yet (

Table 1. Cave sites (name and id. in SISAL) and entities (name and id. in SISAL) (i.e., speleothems) in South America. Cave sites and entities in SISAL_v1 have a unique identifier (id.). This field is empty for caves and entities not in the current version of the database. Minimum (min) and maximum (max) year are the youngest and oldest age of the isotope time series in the SISAL_v1 database. Headers in italic (e.g., Site name) are searchable variables in SISAL_v1 database.

Site_name	Side_id	Country	Entity_name	Entity_id	Latitude ° N	Longitude ° E	Min Year BP	Max Year BP	Reference
Abissal Cave	18	Brazil	Abissal	79	−5.6	−37.73	24,164	25,694	[8]
Abissal Cave	18	Brazil	Ale-1	80	−5.6	−37.73	15,057	17,199	[8]
Botuverá Cave	144	Brazil	BTV21a	311	−27.22	−49.16	196	9211	[9]
Botuverá Cave	144	Brazil	BT2	312	−27.22	−49.16	−55	116,100	[10]
Botuverá Cave	144	Brazil	BTV4A		−27.22	−49.16			[11]
Botuverá Cave	144	Brazil	BTV4C		−27.22	−49.16			[11]
Botuverá Cave	144	Brazil	BT3A		−27.22	−49.16			[12]
Cascayunga Cave		Peru	CAS-A		−6.09	−77.22			[13]
Cascayunga Cave		Peru	CAS-D		−6.09	−77.22			[13]
Chiflonkhakha Cave		Bolivia	Boto1		−18.12	−65.77	234	375	[14]
Chiflonkhakha Cave		Bolivia	Boto3		−18.12	−65.77	180	1106	[14]
Chiflonkhakha Cave		Bolivia	Boto7		−18.12	−65.77	91	315	[14]
Chilibrillo Cave	78	Panama	CHIL-1	167	9.17	−79.62	690	2180	[15]
Cristal Cave		Brazil	CR1		−24.58	−48.58			[7]
Cueva Del Diamante		Peru	NAR-C		−5.73	−77.5			[16]
Cueva Del Diamante		Peru	NAR-D		−5.73	−77.5			[16]
Cueva Del Diamante		Peru	NAR-F		−5.73	−77.5			[16]
Cueva del Tigre Perdido		Peru	NC-A		−5.94	−77.31			[17]
Cueva del Tigre Perdido		Peru	NC-B		−5.94	−77.31			[17]
Curupira Cave	35	Brazil	CUR4	110	−15.02	−56.78	−21	155	[18]
Diva Cave	38	Brazil	DV2	113	−12.37	−41.57	39	2765	[19]
El Condor Cave		Peru	ELC-A		−5.93	−77.3			[16]
El Condor Cave		Peru	ELC-B		−5.93	−77.3			[16]
Huagapo Cave		Peru	P00-H1		−11.27	−75.79			[20]
Huagapo Cave		Peru	P09-H2		−11.27	−75.79			[20]
Jaraguá Cave	10	Brazil	JAR7	55	−21.08	−56.58	5489	18596	[21]
Jaraguá Cave	10	Brazil	JAR14	56	−21.08	−56.58	15,331	22,305	[21]
Jaraguá Cave	10	Brazil	JAR13	57	−21.08	−56.58	21,852	27,906	[21]
Jaraguá Cave		Brazil	JAR4		−21.08	−56.58	-50	760	[22]
Jaraguá Cave		Brazil	JAR1		−21.08	−56.58	499	1508	[22]
Lapa Doce Cave	103	Brazil	LD12	203	−12.37	−41.57	−61	39	[22]
Lapa Grande		Brazil	LG3		−14.36	−44.28			[23]
Lapa Grande		Brazil	LG11		−14.36	−44.28			[23]
Lapa Grande		Brazil	LG12B		−14.36	−44.28			[24]
Lapa Grande		Brazil	LG19		−14.36	−44.28			[24]
Lapa Grande		Brazil	LG25		−14.36	−44.28			[24]
Lapa Sem Fim		Brazil	LSF3		−16.26	−44.6			[24]
Lapa Sem Fim		Brazil	LSF9		−16.26	−44.6			[24]
Lapa Sem Fim		Brazil	LSF11		−16.26	−44.6			[24]
Lapa Sem Fim		Brazil	LSF13		−16.26	−44.6			[24]
Lapa Sem Fim		Brazil	LSF15		−16.26	−44.6			[24]
Lapa Sem Fim		Brazil	LSF16		−16.26	−44.6			[24]
Lapa Sem Fim	24	Brazil	LSF16	91	−16.15	−44.63	14,373	18,168	[25]
Lapa Sem Fim	24	Brazil	LSF3	92	−16.15	−44.63	17,296	19,304	[25]
Marcelo Arévalo Cave		Chile	MA-1		−52.68	−73.38			[26]
Marcelo Arévalo Cave		Chile	MA-2		−52.68	−73.38			[26]
Marcelo Arévalo Cave		Chile	MA-3		−52.68	−73.38			[26]
Marota Gigante Cave		Brazil	MAG		−12.58	−41.03			[24]
Pacupahuain Cave		Peru	P09		−11.24	−75.82			[27]
Pacupahuain Cave		Peru	PH2		−11.24	−75.82			[27]
Paixão Cave		Brazil	PX7		−12.61	−41.02			[24]
Paixato Cave	113	Brazil	PX7	228	−12.62	−41.02	14,983	19,144	[25]
Palestina Cave	25	Brazil	PAL3	93	−5.92	−77.35	22	851	[28]
Palestina Cave	25	Brazil	PAL4	94	−5.92	−77.35	100	1537	[28]
Paraiso Cave	3	Brazil	PAR01	20	−4.07	−55.45	714	4812	[29]
Paraiso Cave	3	Brazil	PAR03	21	−4.07	−55.45	−48	768	[29]
Paraiso Cave	3	Brazil	PAR06	22	−4.07	−55.45	40,683	44,636	[29]
Paraiso Cave	3	Brazil	PAR07	23	−4.07	−55.45	17,957	45,144	[29]
Paraiso Cave	3	Brazil	PAR08	24	−4.07	−55.45	26,490	28,444	[29]
Paraiso Cave	3	Brazil	PAR16	25	−4.07	−55.45	4440	10,927	[29]
Paraiso Cave	3	Brazil	PAR24	26	−4.07	−55.45	9476	28,286	[29]
Pau d’Alho Cave	52	Brazil	ALHO6	128	−15.21	−56.8	90	1458	[18]
Rainha Cave	111	Brazil	RN1	219	−5.6	−37.73	4045	23,890	[8]
Rainha Cave	111	Brazil	RN4	220	−5.6	−37.73	4393	17,222	[8]
Santana Cave		Brazil	St-8		−24.53	−48.73			[30]
São Bernardo Cave		Brazil	SBE3		−13.81	−46.35	−60	827	[22]
São Matheus Cave		Brazil	SMT5		−13.81	−46.35	749	1686	[22]
Shatuca Cave		Brazil	Sha-2		−5.7	−77.9			[31]
Shatuca Cave		Brazil	Sha-3		−5.7	−77.9			[31]
Tamboril Cave	27	Brazil	TM0	97	−16	−47	−32	1678	[32]
Tamboril Cave	27	Brazil	TM2	98	−16	−47	1485	3847	[32]
Torrinha Cave	62	Brazil	TR5	146	−12.37	−41.57	−57	76	[19]
Umajalanta		Bolivia	Boto10		−18.12	−65.77	604	1328	[14]

, Figure 1).

Speleothem-based records from South America are distributed over the entire continent, ranging from the tropical climate of Venezuela in northern South America to the colder climate of Patagonia at the southern tip of South America, as well, from the wet and cold climates of the Andes in western South America, the tropical climate in the Amazonas, and the dry climate of Nordeste in northeastern Brazil (Figure 1). Despite this continental coverage, almost all records included in SISAL_v1 are located in the tropical and sub-tropical climate zones of central and southeastern Brazil (between 0° and 30° S), which are influenced mainly by the SAMS. Thus, this review is focused mainly on past variations of the SAMS and associated precipitation changes that are revealed by speleothem δ¹⁸O time series, following the relationship for δ¹⁸O: δ¹⁸O = (¹⁸R_speleothem/¹⁸R_VPDB-1), wherein ¹⁸R_x is the isotope ratio between ¹⁸O and ¹⁶O in the speleothem or of the VPDB (Vienna Pee Dee Belemnite) standard. Decreasing speleothem δ¹⁸O values are primarily interpreted as a strengthening of upstream convection in the core monsoon region and increased precipitation amount (known as the amount effect) from strengthening of SAMS activity [7].

The onset of the monsoon season in tropical and subtropical South America begins in the austral spring (September to November, SON) when easterly trade winds are associated with the southward shift of the ITCZ, following maximum sea-surface temperatures (SSTs) and enhanced moisture transport from the tropical Atlantic over the continent. Southward migration of the insolation maximum drives an enhanced land-sea thermal gradient, leading to low-level convergence over the continent, in turn increasing precipitation and establishing a zone of deep convection from 10° S to 15° S over the southern Amazon Basin. This intense convection during the onset of the SAMS contributes to the formation of an upper-level atmospheric high over Bolivia (Bolivian High; 15° S, 65° W) and a low-level atmospheric low over the Chaco region (Chaco low; ~25° S). This dynamic draws humid air from the core monsoon region, which is then transported south to La Plata Basin region by the South American Low-Level Jet (Figure 2), as the easterlies at the core monsoon region are blocked and redirected by the Andes [2,33]. The monsoon reaches its mature phase in the austral summer (December to February, DJF), when the Bolivian High as well as the South Atlantic Convection Zone (SACZ) are well established. The SACZ, a dominant feature of the mature SAMS, is a NW–SE trending band of convection that extends from the core monsoon region to the western South Atlantic, (Figure 2). Upper-level convergence conditions develop over northeastern Brazil (the Nordeste Low) as a consequence of the establishment of the Bolivian High (Figure 2), leading to subsidence of air and a parallel decrease in summer precipitation over Nordeste, a well-noted antiphase relationship between precipitation in the western and eastern South America [33].

In the following section, we discuss the speleothem records available in SISAL_v1 along with speleothem records identified in the literature but not yet in SISAL_v1 (Section 2). In Section 3, we discuss past variations in the SAMS during the Late Pleistocene, the Holocene, and the last 2 kyrs, utilizing primarily speleothem δ¹⁸O time series that are in SISAL_v1. However, we also discuss critical speleothem δ¹⁸O time series that are not yet available in SISAL_v1. The Pleistocene Section 3.1 focuses on SAMS variations associated with changes in insolation due to variations in Earth’s orbit and millennial scale variations. The Holocene Section 3.2 details discrepancies and similarities between speleothem δ¹⁸O time series and potential drivers of these variations. The last Section 3.3 focuses on hydroclimate scenarios of the last two millennia, such as the Little Ice Age (LIA), the Medieval Climate Anomaly (MCA), and the Current Warm Period (CWP), and identifies the possible climate drivers of the SAMS and ITCZ during this period.

2. Assessing SISAL_v1 for South America

We have identified 75 individual speleothem isotope records from South America from the literature (Table 1). Of these, 31 are available in SISAL_v1, while the remaining 44 will be added to the SISAL database in the near future. Almost all isotope records in SISAL_v1, as well as many of the other speleothem records, are in the SAMS domain (Figure 1 and Figure 2). Records from Patagonia, Marcelo Arévalo Cave (MA Cave), and Panama (Chilibrillo Cave) are the only records available in the literature that are outside the SAMS region. SISAL_v1 is useful to investigate spatio-temporal variations of the SAMS, but highlights the need to increase coverage of speleothem records across the entirety of the continent.

SISAL_v1 δ¹⁸O records cover the last 120 kyrs, however, the distribution of speleothems δ¹⁸O records is clearly biased towards the last 30 kyrs (Figure 3). Only one δ¹⁸O record spans the entirety of the last 120 kyrs, BT2 from Botuverá Cave. For the last 30 kyrs, there are peaks in coverage for the last 2 kyrs, the early Holocene (10 to 8 kyrs BP), and the period from 20 to 15 kyrs BP. This non-uniform temporal distribution of δ¹⁸O records largely reflects the focus of the research utilizing speleothem δ¹⁸O records in South America. The temporal distribution of speleothem δ¹⁸O time series coverage (Figure 3), combined with the spatial coverage over the SAMS region (Figure 2), makes the SISAL_v1 database suitable for characterization of SAMS variability over the past 30 kyrs, but again, highlights the need to increase spatio-temporal coverage of speleothem records.

3. Results and Discussion

3.1. Pleistocene South American Monsoon System Variability as Recorded by Speleothem δ¹⁸O Time Series

3.1.1. Insolation Forced Changes of the SAMS

Thus far, only a few studies report speleothem δ¹⁸O time series on orbital time-scales in regions where the precipitation pattern is dominantly influenced by SAMS variability. These include stalagmites from southeastern Brazil, namely BT2 [10], BTV4A*, BTV4C* [12] and BTV3A* [12] from Botuverá Cave, and St8* [30] from Santana Cave. An asterisk indicates entities that are not included in SISAL_v1. Speleothem δ¹⁸O time series from El Condor (ELC*) and Cueva del Diamante (NAR*) [16] on the eastern slope of the Andes(northern Peru) in western Amazonia (WA) (Figure 2) also show pronounced orbital scale changes. Both regions are currently located at the termini of the cone of influence of the SAMS (Figure 2).

BT2 grew continuously during the last ~112 kyrs and the δ¹⁸O time series exhibits pronounced orbital scale changes that closely follow austral summer insolation (Figure 4a) [10], which is also evident in wavelet analyses (Figure 5a). The wavelet analysis reveals a strong precession signal (20 kyrs period) of the BT2 δ¹⁸O time series. However, this precession periodicity is statistically only significant from 60 to 40 kyrs BP. This is related on the one hand to the length of the time series, which prohibit assessing the significance of the precession periodicity (it is outside the cone of influence) and on the other hand because during the last 20 kyrs millennial scale climate variability (Heinrich stadial 1) and the Pleistocene-Holocene transition (termination 1), possibly diminished the insolation forced variability in the BT2 δ¹⁸O time series. Changes in BT2 δ¹⁸O values are driven by varying contribution of moisture that originates from the Amazon Basin and is associated with monsoonal rainfall (low δ¹⁸O) during summer, and the adjacent South Atlantic ocean (high δ¹⁸O) during winter [10,36]. Therefore, changes in BT2 δ¹⁸O reflect mainly the varying contribution of monsoonal rainfall to the annual rainfall in southeastern Brazil, caused by shifts of the SACZ [8]. These shifts are possibly caused by latitudinal shifts of the termini of the Hadley Cell (i.e., the width) or by the rainfall history of the moisture that is transported from the Amazon Basin towards southeastern Brazil by the South America Low-Level Jet (SALLJ) [10]. Comparison of the BT2 δ¹⁸O time series with Mg/Ca and Sr/Ca ratios in BT2, which are indicators of the recharge of water at Botuverá [37,38], show that decreasing δ¹⁸O values co-occur with decreasing Mg/Ca and Sr/Ca ratios and vice versa (Figure 4a,b). This indicates that when the influence of monsoonal precipitation at Botuverá increased, precipitation amounts increased as well.

Because of their strong covariance and proximal location, the δ¹⁸O records from El Condor (ELC-A*, ELC-B*) and Cueva del Diamante (NAR-C*, NAR-D*, NAR-F*) are referred to as the WA δ¹⁸O time series in the following. Although the WA δ¹⁸O time series covers the last 250 kyrs [16], here we compare its last 112 kyrs with the BT2 δ¹⁸O time series.

The WA δ¹⁸O time series follows, with exception of the period from 70 kyrs BP to 20 kyrs BP, austral summer insolation, and is thereby also broadly in-phase with the BT2 δ¹⁸O time series (Figure 4a,b). This is also evident by the wavelet analysis of the WA δ¹⁸O time series (Figure 5b). Marked differences between the δ¹⁸O time series from southeast Brazil and western Amazonia occur from 70 kyrs BP to 20 kyrs BP when the two time series appear to be anti-correlated and when the precession signal is absent in the WA δ¹⁸O time series (Figure 5b). That is, when the BT2 δ¹⁸O time series decreases, indicating wetter conditions in southeast Brazil, the WA speleothem δ¹⁸O values increase, inferring drier conditions in western Amazonia. This change in the phase-relationship between BT2 and the WA speleothems is also observed between the WA δ¹⁸O time series and the austral summer insolation (Figure 4b and Figure 5b). The coherence of orbital-scale changes between southeast Brazil and western Amazonia speleothem δ¹⁸O time series from ~120 kyrs BP to 60 kyrs BP and from ~20 kyrs BP to the present indicates that SAMS rainfall variability broadly co-varies between these spatially distant regions and appears to be mainly controlled by changes in austral summertime insolation, where increasing austral insolation provokes an intensification of the SAMS [10,16]. It is proposed that increasing insolation leads to a stronger continental heating, and thereby a greater land-sea heat gradient, which in turn provokes stronger convective activity over the continent. This increased convection leads to increased convective heating and vertical updraft in the core monsoon region. This mechanism also leads to an intensification of the Nordeste Low and increased subsidence in northeast Brazil, resulting in generally arid conditions in this region (higher δ¹⁸O values) [8,16].

The period when the WA δ¹⁸O time series is out-of-phase with the austral summer insolation in the period from ~70 kyrs BP to 20 kyrs BP coincides with generally wetter conditions at Botuverá (Figure 4b) and with a weakening of the Atlantic Meridional Overturning Circulation (AMOC) (Figure 4c), when it switches to its glacial (cold) mode [40]. This implies that the oceanic meridional heat transport is reduced in this period, which should, according to energetic constraints, force a southward migration of the ITCZ [43,44]. This alters rainfall pattern associated with the SAMS, particularly in the northern portion of the core monsoon region containing the Amazon Basin and western Amazonia. Rainfall in western Amazonia, then, is primarily driven by the competition between insolation and AMOC forcing, while the Botuverá δ¹⁸O record is mainly driven by insolation. This dynamic could also affect the SALLJ which brings moisture from the Amazon Basin to southeast Brazil, and this may contribute to the generally wetter climate at Botuverá in this period.

3.1.2. Millennial-Scale Climate Events

Millennial scale climate variability associated with Dansgaard-Oeschger (DO) and Heinrich stadials in the Northern Hemisphere are observed across the SAMS domain. These events in the SAMS domain are anti-correlated to precipitation variations associated with the Asian Monsoon [11]. Precipitation amounts drastically increase in Nordeste (Figure 6j) during cold events (Heinrich stadials) in the Northern Hemisphere. Semi-arid conditions prevail today in the Nordeste, preventing modern growth of speleothems in this region. However, speleothem and travertine growth periods occurred during Heinrich stadials over the last 220 kyrs due to increased rainfall amounts in this region [45]. This is further supported by changes in speleothem δ¹⁸O values across the SAMS region (Figure 6). This includes northeast Brazil (Paixão Cave, PX5*, PX7; Marota: MAG*) [24,25], eastern Brazil (Lapa Sem Fim Cave: LSF3, LSF9*, LSF11*, LSF13*, LSF15*, LSF16; Lapa Grande: LG3, LG10*, LG11, LG12B*, LG25*), and central Brazil (Jaraguá Cave: JAR7, JAR14, JAR13) [21]. Precipitation increases are also observed for eastern (Paraíso Cave: PAR01, PAR03, PAR06, PAR07, PAR08, PAR16, PAR24) [29] and western Amazonia (El Condor: ELC-A*, ELC-B*; Cueva del Diamante: NAR-C*, NAR-D*, NAR-F*, Pacupahuain Cave) [16,27] as well as southeastern Brazil (Botuverá Cave: BT2, BTV3A*, BTV4A*, BTV4C*) [10,11,12] (Figure 6). However, the Botuverá Cave δ¹⁸O record shows a much stronger orbital-scale changes than the other speleothem δ¹⁸O records from the SAMS region (Figure 6), which leads to a reduction in the amplitude of the millennial scale events there.

Comparison of the western Amazonia δ¹⁸O time series during MIS 3 (Figure 6d–f) indicates that precipitation amounts in WA were higher during Northern Hemisphere (NH) cold events compared to the following DOs. An exception from this is the precipitation response to HS2. HS2 is almost as pronounced as HS1 in the ²³¹Pa/²³⁰Th record (Figure 6b) and should have caused a distinct precipitation response in WA. However, precipitation changes are modest at the beginning of HS2 and peak at its termination, shortly before the AMOC returns from its off-mode into its cold-mode (Figure 6d–f). What is also evident in the δ¹⁸O time series from Pacupahuain Cave is the increasing precipitation associated with the evolution of DO12. This pattern is less clearly pronounced in the two δ¹⁸O time series from El Condor Cave (Figure 6d) and absent in the record from Cueva del Diamate (Figure 6e). The latter is likely caused by the low temporal resolution of this record at the time of DO12 (Figure 6e). During MIS 3 similar responses as for the WA δ¹⁸O time series are observed for eastern Amazonia (Figure 6g), eastern Brazil (Figure 6h), and northeastern Brazil (Figure 6i). The speleothem δ¹⁸O time series from Rio Grande do Norte (Figure 6j) covers only the end of HS2 in MIS 3 but is consistent with the other stalagmite δ¹⁸O time series, indicating a trend from wetter to dryer conditions. In central Brazil precipitation changes during HS2 are coherent with the other regions influenced by the SAMS, indicating that distinct precipitation changes occurred only during the end of HS2. The speleothem δ¹⁸O time series from Botuverá Cave has a very low temporal resolution and large age uncertainties during MIS 3 (Figure 6l), which makes it difficult to identify HS and DOs in this record. However, some of Heinrich and DO stadials match quite well with this record and when considering the findings from other speleothems from this cave [11,12], inferring precipitation increases during HS, the emerging figure is that precipitation amounts at Botuverá Cave increased during HS and decreased during DOs.

In MIS 2, until the development of HS1 at about 18 kyrs BP, all speleothem δ¹⁸O time series indicate that average precipitation amounts were constant (Figure 6d–l). In this period only the δ¹⁸O time series from Pacupahuain Cave shows millennial scale events (Figure 6f). It is not possible to link these events to variations of the SAMS, because no other δ¹⁸O record from the SAMS region indicates such pronounced precipitation changes during this period, possibly because the temporal resolution of the other records is too low (Figure 6d,e). Precipitation changes during HS1 in the SAMS area are different between regions (Figure 6d–l). In western Amazonia the δ¹⁸O time series from Cueva del Diamante and Pacupahuain Cave indicate decreasing precipitation amounts (Figure 6e,f) whereas the δ¹⁸O time series from El Condor shows no precipitation changes on average (Figure 6d). The precipitation decrease at Cueva del Diamante is possibly faster than shown in Figure 6d, because there are only a few U-Th ages measured during this time. In central, eastern and northeastern Brazil (Figure 6h,i,k) precipitation amounts increased at the beginning of H1. Precipitation amounts decreased again at about 16 kyrs BP; this precipitation decrease is especially pronounced in central Brazil, whereas precipitation changes were modest in eastern and northeastern Brazil. Precipitation amounts increased in Rio Grande do Norte as well as in southeastern Brazil during HS1 (Figure 6j,l), but do not indicate similar precipitation variability as observed for central, eastern and northeastern Brazil. During the Bølling-Allerød (BA) (DO1) and the Younger Dryas (YD) (Figure 6a) precipitation changes are similar again in the different regions of the SAMS. All speleothem δ¹⁸O time series indicate that precipitation amounts were reduced during the BA and higher during the YD (Figure 6).

The coherent response in central, eastern and northeastern Brazil during HS1 is termed the Mega-SACZ [25]. The response in rainfall amount during MIS 3 across the SAMS domain suggests that the development of a Mega-SACZ is not only restricted to HS1 but is present and even more pronounced during all Heinrich stadials in MIS 3. Moreover, the speleothem and travertine growth periods from northeastern Brazil, dating back 220 kyrs, suggest the development of the Mega-SACZ before the Eemian (MIS 5e, c. 130 kyrs BP to 115 kyrs BP) [45]. Testing this hypothesis requires the generation of new high-resolution (<30 years) speleothem δ¹⁸O time series covering the last 120 kyrs from the region. Strengthening of the cross-equatorial heat transport, related to changes in sea surface temperatures (SSTs) in the tropical Atlantic, are thought to drive the development of the Mega-SACZ [25]. This is consistent with modelling experiments investigating the effect of a mid-latitude Northern Hemisphere cooling through an AMOC shutdown, which would drive changes in SSTs in the tropical Atlantic [47]. Chiang and others (2012) [47] suggest dry conditions should prevail in northeast Brazil during an AMOC shutdown, however, this is not consistent with the speleothem-based rainfall reconstructions presented discussed here (Figure 6j). It is not clear why modelling and the palaeoclimate reconstructions exhibit opposite precipitation trends in response to Heinrich stadials (NH cooling); further work is necessary to resolve this apparent discrepancy between models and observations, and highlights Nordeste as a key region for data-model comparisons. Compared to the precipitation response during Heinrich and DO stadials in MIS 3, the precipitation variability during HS1 is different in the domain of the SAMS (Figure 6). This is particulary the case in western Amazonia where precipitation amounts progressively decrease (Figure 6e,f) while precipitation amounts in the other regions increase; except eastern Amazonia, where no speleothem δ¹⁸O time series is available. Kanner et al. (2012) [27] noted an influnce of Southern Hemisphere (SH) high latitude climate variability on precipitation changes at Pacupahuain Cave during the Antarctic Isotope Maxima 8 (40 to 37 kyrs BP) and 12 (48.5 to 46 kyrs BP). A similar high-latitude influence on precipitation amounts in the northern realm of the SAMS domain could also explain the apparent differences between the WA speleothem δ¹⁸O time series and the other Brazilian δ¹⁸O time series. With the beginning of HS1 in the NH, reconstructed Antarctic temperaures increase until about 14 kyrs BP [48]. Therefore, the opposite hemispheric temperature trends, a cooling NH associated with HS1 and a warming SH, strengthened the inter-hemispheric temperature contrast, which, based on energetic constrains [1,44,47], provokes a southward shift of the ITCZ (similar to the arguments in Section 3.1.1), in turn, shifting and/or tilting SAMS precipitation pattern. This would causes a precipitation decrease in western Amazonia. In summary, the precipitation variability of the SAMS from 50 kyrs BP until the end of the YD is heavily dependent on the high-latitude NH climate variablity associated with NH cold events (Heinrich stadials) and DOs, respectively. The emerging pattern during MIS 3 is that precipitation amounts were higher during NH cold events compared to DOs. Modest precipitation changes are observed during HS2 and this is unexpected when considering the observed precipitation increase to HS in MIS 3. However, one explaination for this can be, that HS2 is not followed by a DO, while all other NH cold events in MIS 3 are followed by a DO (Figure 6a–c). This may have weakened the response of the SAMS and precipitation changes to HS.

3.2. Holocene Variations of the South American Monsoon System as Recorded by Speleothem δ¹⁸O Time Series

Many published records from the SAMS region span part of the Holocene (Table 1). However, very few of these records are currently included in the SISAL_v1 database, and even fewer of these records cover the entire extent of the Holocene. Continuous records of the entire Holocene are necessary to document climate dynamics related to the Holocene climate variability of the SAMS (Figure 7). Records included in SISAL_v1 are PAR01, PAR03, and PAR16 [29] from Paraíso Cave in eastern Amazonia, RN1 and RN4 from Rainha Cave in northeast Brazil [8], and BTV21a* [9] from Botuverá Cave in southeast Brazil. Records not yet included in SISAL_v1, include speleothem δ¹⁸O time series from western Amazonia (Sha-3* and Sha-4* from Shatuca Cave [31]; ELC-A* and ELC-B* from El Condor; NAR-C*, NAR-D*, NAR-F* from Cueva del Diamante; P00-H1* and P09-H2* from Huagapo Cave) which are similar to the Holocene Botuverá record from southeast Brazil and records from central Brazil (LG3* and LG11* from Lapa Grande Cave).

Greenland and Antarctic ice core records [49,50] suggest that the Holocene is climatically more stable than previous intervals. However, there are abrupt climate changes in the Holocene, such as the 8.2 kyr event, Bond events [51], and the 4.2 kyr event [52,53]. How these events are manifested in tropical and subtropical South America remains uncertain. Austral summertime insolation increased throughout the entire Holocene period (Figure 7) and corresponds with decreasing speleothem δ¹⁸O records from southeast Brazil and western Amazonia. However, no such correspondence is present between austral summertime insolation and both eastern Amazonia and central Brazil speleothem δ¹⁸O records. The Paraíso (eastern Amazonia) δ¹⁸O record is coupled with insolation from the early- to mid-Holocene but diverges from ~8 kyr BP to the present. The mean δ¹⁸O value of the Lapa Grande (central Brazil) record remains relatively unchanged and exhibits no trend(s) through the Holocene.

3.2.1. Insolation Driven Holocene δ¹⁸O Records

The proposed mechanism driving the Botuverá δ¹⁸O record in southeast Brazil is intensification of the SAMS, as driven by increasing austral summertime insolation, suggesting a progressive increase in precipitation amount throughout the Holocene. However, this does not necessarily suggest that increased precipitation in southeast Brazil is coupled with increasing monsoon intensity, but rather that southeast Brazil received more δ¹⁸O-depleted precipitation from increased convection upstream, within the core of convection in the monsoon region (Figure 2) [7]. That is, increased convection within the core monsoon region leads to more depleted water masses, which then lead to depleted precipitation along the entire water mass trajectory, relative to periods of weakened monsoon convection. Thus, it might be expected that any regions along the monsoon trajectory will exhibit the same long-term δ¹⁸O trend, unless mechanisms beyond monsoon intensity are driving these records [7]. Such mechanisms may be increased (or decreased) local precipitation (i.e., the amount effect) or changes in relative contributions of water-mass sources (e.g., [8]). However, a speleothem Sr/Ca time series from Botuverá Cave, a proxy for the local effective rainfall [38], shows that rainfall amounts increased in this region with increasing influence of SAMS rainfall during the course of the Holocene [9]. This highlights the pertinence of complementing speleothem measurements, such as trace elements, which may provide further information on processes driving speleothem δ¹⁸O variability, past monsoon variability, and variability in local moisture conditions. Speleothem δ¹⁸O records from the western Amazonia exhibit long-term trends that follow both the Botuverá Cave δ¹⁸O record and austral summertime insolation (Figure 7), suggesting that these sites along the monsoon water-mass trajectory (Figure 2) are influenced by increased monsoon convection and/or increases in local precipitation amount as driven by increased water availability due to increased monsoon strength.

3.2.2. Dipole of Holocene δ¹⁸O Values

The South American rainfall dipole, inferred by speleothem δ¹⁸O records from eastern and western South America, is well-documented [8] and is persistent during the Holocene as evidenced by δ¹⁸O records (Figure 7). The proposed rainfall dipole between Nordeste and the other regions influenced by the SAMS is distinct in the period from the mid- to the late Holocene (Figure 7) but is not readily apparent in the early-Holocene (~12 to 8 kyr BP) as evidenced by the δ¹⁸O time series, where the Nordeste record exhibits a distinct decreasing trend, in-phase with other records across the SAMS region (Figure 7c). Notably, the RN records are in-phase with the Paraiso δ¹⁸O records (eastern Amazonia) from ~12 kyr BP to the end of the RN records at 4 kyr BP (Figure 7). However, the RN long-term trend is notably steeper than the Paraiso long-term trend from ~8 kyr BP to 4 kyr BP. This correspondence suggests a similar mechanism(s) driving precipitation δ¹⁸O values over both Nordeste and eastern Amazonia during the Holocene, although these two regions fall in different climate regimes in the present day.

3.2.3. Holocene Variations of the South Atlantic Convergence Zone

The constant mean δ¹⁸O values at Lapa Grande Cave, central Brazil, is unique among the speleothem δ¹⁸O time series available in SISAL_v1 (Figure 7). One possible driver of the variability in the Lapa Grande δ¹⁸O time series are changes in regional rainfall amounts associated with latitudinal fluctuations of the SACZ (Figure 2). Latitudinal fluctuations and shifts of the SACZ affect rainfall amounts in boundary regions, while precipitation amounts in the centre of the SACZ are inferred to be constant [22]. Thus, Lapa Grande’s position in the centre of the SACZ, and steady mean δ¹⁸O values of Lapa Grande speleothems during the Holocene, imply little changes in precipitation amount at this site and suggest it was under the influence of the SACZ during the entire Holocene. This would suggest that speleothem δ¹⁸O time series from Lapa Grande Cave are not dominantly influenced by upstream rainout in the core monsoon region, but rather local rainfall amount associated with the strength of the SACZ. If so, this suggests the SACZ has not increased or decreased in intensity during the Holocene, despite increasing insolation and increasing intensity in the core of monsoon convection in the Amazon Basin. Stríkis et al. (2011) [23] identify depleted δ¹⁸O episodes in the Lapa Grande record that coincide with the cold Bond events in the Northern Hemisphere. They suggest the low δ¹⁸O values were the result of a southward movement of the ITCZ, following the same mechanism that occurred during Heinrichs stadials, but with smaller amplitude. Notably, abrupt climate events that occurred at 2.7 kyrs and 8.1 kyrs are also documented in speleothems in Northeast Brazil [19,54].

3.3. Variations of the South American Monsoon System during the Last 2 kyrs as Recorded by Speleothem δ¹⁸O Time Series

There are 10 stalagmite records from 6 different regions covering the last two millennia in the SISAL_v1 database. Most of the records only have δ¹⁸O data, their resolution is approximately annual, and, with exception of the ALHO6 stalagmite that is annually laminated, all speleothems’ age models are based on U/Th ages. The records included are DV2 (Diva Cave), TR5 (Torrinha Cave), LD12 (Lapa Doce Cave) from northeast Brazil [19]; CR1* (Cristal Cave) from southeast Brazil [7]; SBE3 (São Bernardo Cave), SMT5 (São Matheus Cave), TM0 (Tamboril Cave) from central Brazil [22,32]; ALHO6 (Pau d’Alho Cave), CUR4 (Curupira Cave) from middle-western Brazil [18,22]; PAR1 (Paraiso Cave) from eastern Amazonia [29]; CAS-A (Cascayunga), CAS-D (Cascayunga), PAL3 (Palestina Cave), PAL4 (Palestina Cave) from the Peruvian Andes [13,28,31] and Boto1 (Chiflonkhakh Cave), Boto3 (Chiflonkhakh Cave), Boto7, (Chiflonkhakh Cave), Boto 10 (Umajalanta Cave) from the Bolivian Andes [14].

The main climate features over the last two millennia in South America were forced by NH warm and cold periods during the Medieval Climate Anomaly (MCA) and the Little Ice Age (LIA), respectively [7,55]. Warm conditions in the NH force a northward positioning of the ITCZ that weakened the convective activity of the SAMS. This weakening was exhibited as an increase in speleothem δ¹⁸O values during the MCA in the records: Diva Cave from Northeast Brazil [19], Pau d’Alho and Jaraguá caves from mid-western Brazil [18,22], Palestina [28] and Cascayunga caves [13] from the Peruvian Andes. However, speleothem records from Tamboril [32] and São Bernardo [22] caves located in central Brazil, and Cristal Cave [7] from southeast Brazil do not show a significant anomaly during the MCA. Conversely, the speleothem records from the Umajalanta–Chiflonkhakha Cave system in the eastern Bolivian Andes [14] suggest increased convective activity during the MCA period, with decreasing speleothem δ¹⁸O values, which was interpreted as a result of a different mechanism (not ITCZ/SAMS system), implying that moisture was sourced mainly from the southern tropical Atlantic [14].

The ITCZ likely shifted southward during the LIA, leading to wetter conditions, as evidenced by several speleothem records in South America [22], such as Pau d’Alho [18], Jaraguá [22], Palestina [28], Cascayunga [13], and Cristal [7] caves. However, the SACZ is thought to have also shifted southward during this period, driving an antiphase of convection between the southwest and northeast portions of the SAMS domain [22]. Thus, while records from the southwest region of the SAMS domain indicate wet conditions [7,13,14,18,22,28], speleothem records from the northeast region of the SAMS, represented mainly by the speleothems from Diva Cave [19], suggest dry conditions. Further, records located near to the node of this dipole in central Brazil (São Matheus and Tamboril caves) probably experienced a continuous influence of the SACZ, indicated by the stable mean δ¹⁸O values during the LIA period, as well over the entire last millennium (Figure 8) [22]. During the CWA, the convective conditions over South America appear to be similar to the MCA, with warming in the NH leading to a northward positioning of ITCZ and a progressive weakening of the SAMS.

A centennial periodicity of the SAMS (around 210 years) was reported from different sites [18,19,28] and is associated with the de Vriess–Suess solar cycle, exhibiting depleted δ¹⁸O during periods of high solar irradiance. It is proposed that this relationship is due to several feedbacks involving amplification of solar forcing by coupled air-sea dynamics, cloud formation and stratospheric warming due to enhanced UV absorption through increased stratospheric ozone concentration. A resulting adjustment in the Hadley cell modulates then the position of the ITCZ and SACZ, respectively [18]. Multidecadal variability (60–90 years) associated with the Atlantic Multidecadal Oscillation (AMO) also affects the SAMS, with a more southerly position of the ITCZ during its negative phase (warm in southern tropical Atlantic) [18,19,28]. Although the El Niño–Southern Oscillation (ENSO) climate mode may be a predominant forcing of the present SAMS, there is no evidence of the influence of this mode during the past two millennia. The argument against an ENSO influence during the last two millennia is related to the often synchronous climate conditions observed between the Peruvian Andes and southeast Brazil during this period, even though the canonical ENSO impact in these two regions is the exact opposite [7,22].

4. Conclusions and Outlook

4.1. Summary on South American Monsoon System Variability

Speleothem δ¹⁸O records from South America are interpreted mainly as proxies of amount effect and convective activity of the monsoon and are mostly used to infer past variations of the ITCZ, the SAMS and the SACZ. Precession driven insolation is the main driver of convective variability over the continent during the last 250 kyrs, including the Holocene. A climate dipole is observed between the west and east portions of the continent, which is most prominent on orbital time scales, and apparent during the Holocene. Records located in the central region of Brazil appear to be weakly affected by insolation and are more susceptible to variations of either the positioning or intensity of the SACZ. Cold episodic events in the Northern Hemisphere force a southward displacement of the ITCZ, which in turn increases convective activity of the SAMS domain on all times scales, as documented during Heinrich events during the last glacial period, Bond events during the Holocene, and during the LIA. Variations in Northern Hemisphere temperatures were the main forcing of changes in SAMS during the last two millennia, which resulted in predominantly dry conditions during the MCA and wet conditions during the LIA, followed by progressively more arid conditions during the CWP.

4.2. Outlook and Potentials for Future Speleothem-Based Empirical Research

Speleothem records from South America are located predominantly in the tropical and subtropical region of the Southern Hemisphere. No South American speleothem records are available from north of the equator and only one record from south of the subtropics (Figure 1 and Figure 2). New speleothem records from outside the SAMS domain, and the generation of longer records from the whole of South America, are required to improve our understanding of South American hydroclimate. Most existing records consist of composite records shorter than 50 kyrs, and only few time series are longer than 100 kyrs (Figure 3). Other types of measurements on speleothems, such as trace element concentrations or elemental isotope ratios (e.g., ⁸⁷Sr/⁸⁶Sr), could be used to explore the relationship between local precipitation and regional monsoon strength [9,32]. Speleothem δ¹³C records could also be exploited for reconstruction of the palaeovegetation or soil dynamics [56,57].

4.3. Next Steps and Potentials for SISAL-Based Research

Speleothem isotope time-series not included in the SISAL_v1 database (Table 1) would improve the spatio-temporal coverage for the SAMS domain (Figure 1) and allow detailed characterization of past variations of the SAMS. Such analysis should not only be focused on speleothem δ¹⁸O time series but also on the available δ¹³C time series. This is of particular interest, because δ¹³C yields insights in the response of the vegetation to variations of the SAMS. Better understanding of vegetation feedbacks are important for a better understanding of the contribution of moisture recycling in δ¹⁸O variability [29] and also for predicting the response of the tropical rain forest (an important carbon sink and source) to precipitation changes in the Amazon Basin in response to global warming.

Speleothem δ¹⁸O time series in the SISAL database are very appropriate for data-model comparisons. This does not necessarily include only isotope enabled climate models, because trends in speleothem δ¹⁸O time series—and patterns—also yield information on whether specific regions became wetter or drier in the past, which is in many cases already sufficient to test the response of a climate model. The Heinrich stadials or the transition from the early to the late Holocene could be an ideal opportunity for data-model comparisons, since there are many speleothem δ¹⁸O records available. A strategic region for such tests is northeast Brazil (Nordeste), which is at present dominated by semi-arid conditions, but was much wetter during Heinrich stadials or in the mid-Holocene [8,45]. Since variations of precipitation in northeast Brazil are linked to the low-level circulation of the SAMS, which appears to be linked to the tropical atmospheric circulation in Africa [8], accurate modelling of the precipitation in Nordeste is a crucial benchmark test for the tropical atmospheric circulation in climate models.

A deeper understanding of past spatio-temporal changes of the SAMS could be gained using Monte Carlo Principal Component Analyses (MC-PCA) techniques [58]. This would help to constrain the different forcing and coherence of precipitation changes in the core monsoon region and regions that are mainly influenced by shifts of the SACZ and to understand the drivers of speleothem δ¹⁸O variations, in particular the role of local effects and downstream processes associated with precipitation history and moisture recycling [7,16,29]. In summary, the SISAL database makes it possible to conduct a variety of different analysis of the SAMS, which will ultimately yield a better understanding of its past changes.