Introduction

The study of climate change from the Pleistocene to the present can contribute to a better understanding of the magnitude and rates of change of natural climate variability in the context of current global warming, especially in key study areas. In this context, climate and hydrological conditions in the tropical and subtropical Atlantic play a crucial role in the worldwide climate system, either as a response to changes in the Northern Hemisphere or as a driving mechanism for global climate conditions. The Caribbean is a highly interesting study region—located in the subtropics, influenced by climate processes of hemispheric interconnection including Pacific and Atlantic oceanic and atmospheric processes1,2. The competing influence of these modes of climate variability on tropical Atlantic climate is called the Tropical Atlantic Variability (TAV)1,2. The Caribbean region represents an important area to obtain information about past natural climate variability in the tropics and its relationship to climate forcing mechanisms. Climate change has also had a significant impact on the human history of the Caribbean region. For instance, it is believed that catastrophic droughts on the Yucatan peninsula contributed to the demise of the Maya civilisation3,4,5, and paleoclimatic research using speleothems has been important in helping to define the timing, spatial extent, and severity of the droughts that occurred at this time6. Therefore, it is essential to understand the factors influencing climatic variability in the Caribbean.

There are a small number of paleoclimate reconstructions for Cuba where most are limited to the Holocene or the Pleistocene/Holocene transition, using either sediment cores7,8,9,10 or speleothems2,6,11. Recently, Schielein et al.12 took fossil corals, from the terraces on the northern coast of Cuba, for dating using Electron Spin Resonance (ESR) and Uranium-series (U-Th) to determine the chronology of the Marine Isotope Stage (MIS) 5e coral reef formation. The age data shows one MIS 5e sea-level highstand around 126 ka. Furthermore, Warken et al.13 extended previous paleoclimate evidence for Cuba back to 96 ka using a speleothem record. This study combined stable isotopes of oxygen and carbon with trace element analysis. It was found that North Atlantic cold events were indeed recorded in the Caribbean, with abrupt cooling events (Henrich stadials), in particular, associated with relatively cold and/or dry conditions in Cuba13. This record showed that H1 was the driest and coolest interval of the whole record. During Greenland warm interstadials it was found that these produced relatively warm and wet conditions—especially during MIS 5. During the transition into the Holocene, this record shows a gradual change to generally warmer and wetter conditions. However, the speleothem record by Warken et al.13 is not continuous as this record shows a hiatus during the period ~ 93 to 81 ka. Thus, additional records are needed that cover longer time periods with high temporal resolution.

In this study, we present a new high-resolution paleoclimate record from Cuba, in the Caribbean region, developed from a stalagmite (hereafter “MCS-01”), collected in the Salón de la Permanencia of the Majaguas Upper Cave that is a part of the Majaguas-Cantera Cave System in western Cuba (Fig. 1). The MCS-01 speleothem record covers the period 99 ± 3 to 84.9 ± 4.5 ka within MIS 5b, which has not been documented before in previous paleoclimate studies in the region. Analyzing the oxygen and carbon stable isotopes of MCS-01 will help fill these data gaps, and provide a more in-depth explanation for North Atlantic climatic events and their effects on sub-tropical regions. Due to the fact that speleothem archives are able to record continuous episodes of growth, their resolution can range on timescales from days to hundreds of thousands of years14. Speleothems can also provide excellent chronologies by virtue of their ability to be U-series dated14.

Figure 1
figure 1

Map indicating the approximate locations of the most relevant sites discussed in this manuscript: (1) western Cuban caves, including the Majaguas Upper cave (this study), Torch Cave (Santo Tomás Cave System)13, and Dos Anas Cave (Majaguas-Cantera Cave System)6; (2) Terciopelo Cave, Costa Rica15; (3) Sediment core ODP 999A, Caribbean Sea16; (4) Sediment core MDO3-2621, Cariaco Basin17; (5) Sediment cores ODP 1063 and CDH19, Bermuda Rise18,19,20,21; (6) Botuvera cave, SE Brazil22; (7) NGRIP Ice core, Greenland23,24,25,26,27,28. The Atlantic Ocean circulation patterns and typical positions of winter and summer ITCZ are shown as well. The map was made using Adobe Illustrator.

Full size image

Material and methods

Stalagmite sample and regional setting

The stalagmite MCS-01 was collected in the “Salon de la Permanencia” (bedroom hall) of the Majaguas Upper Cave, which is part of the Majaguas-Cantera Cave System (22° 23′ N, 83° 58′ W) (Fig. 1). The Majaguas-Cantera Cave system is found in Pinar del Rio Province, which is located on the western side of the island. Pinar del Rio Province is characterized by one of Cuba’s main mountain ranges—the Cordillera de Guaniguanico. This mountain range is divided into the Eastern Sierra del Rosario and the Western Sierra de Los Órganos. The Pinar del Rio Province has a moderate subtropical climate with two seasons: the dry season from November to April and the humid season from May to October. Pinar del Rio is classified as “Am or tropical monsoon climate” according to the Koppen-Geiger climate classification29, with the average annual temperature being 24.9 degrees Celsius and the average rainfall accumulation is 1353 cm according to the ECMWF Data over the period 1991–2021.

The Majaguas-Cantera cave system is located in the Sierra de San Carlos karst area, which is part of the Sierra de los Órganos mountain range. The system has a total length of approximately 35 km with 10 cave levels of caverns ranging from 50 to 290 m elevation30. The ground surface over the cave is covered by thick vegetation and soil.

MCS-01 developed at 110 m above sea level, 200 m under the ground surface. The MCS-01 sample (Fig. 2) was cut down its central growth axis using a Target telematic 14″ diamond blade with a continuous rim and polished to show the internal structural details of the stalagmite. X-ray Diffraction (XRD) analyses were performed on two samples from MCS-01 using a Rigaku Ultima IV Diffractometer in the X-ray Core Facility at the University of Ottawa to confirm the nature of the carbonate material (i.e. calcite or aragonite).

Figure 2
figure 2

Stalagmite MCS-01 indicating the location of U–Th dating samples, with ages reported in B.P., and the approximate sample positions for the X-Ray Diffraction (green shading).

Full size image

U-Th chronology

Calcite samples were taken from one-half of the speleothem along visible growth layers using a hand-held Dremel Rotary tool with a 0.3 mm diamond tip drill bit for U-Th radiometric dating. The positions of the preliminary samples were based on their proximity to potential hiatuses. The preliminary U-series measurements for age determination were performed at the Radiochronology Laboratory of the GEOTOP-UQAM-McGill Research Center, Quebec, Canada, whereas further U-dating was carried out at the Isotope Laboratory of the University of Minnesota, Minneapolis, MN, USA.

Regarding the analytical procedure followed at the Radiochronology laboratory of the GEOTOP-UQAM-McGill Research Center, 1 g of subsample powder was dissolved using nitric acid in a Teflon™ beaker into which a weighed amount of calibrated mixed spike 233U–236U229Th had been placed and evaporated slowly to dryness. After the dissolution, around 10 mg of the iron carrier was added and the solution was then left overnight for spike-sample equilibrium. U and Th were co-precipitated with Fe(OH)3 by adding ammonium hydroxide drop by drop until reaching pH 8–9. The precipitate was recovered by centrifugation and washed twice with deionized water, then dissolved in 6 N HCl. U–Th separation was performed on a 2 mL AG1X8 anionic resin volume. The thorium fraction was recovered through elution with 6 N HCl, and the U and Fe fraction, with water. The U fraction was purified on a 0.2 mL U-Teva™ (Elchrom industry™) resin volume. Fe was eluted with 3 N HNO3 and the U fraction with 0.02 N HNO3. After drying, thorium purification was carried out on a 2 mL AG1X8 resin in 7 N HNO3 and eluted with 6 N HCl. U–Th measurements were performed using a multi-collector inductively coupled plasma mass spectrometry Nu instrument™. 236U–235U–234U–233U and 232Th–230Th–229Th were measured on the ion counter (IC0) in peak switching mode for uranium and thorium isotopes, respectively. Knowing 236U/233U of the spike, mass bias corrections in the atomic mass unit (amu-1) were calculated and used to correct measured ratios between U isotopes and between Th isotopes31.

At the Isotope Laboratory of the University of Minnesota, uranium and thorium U/Th analyses were performed using a Neptune Multi‐Collector Inductively Coupled Plasma Mass Spectrometer (MC-ICPMS) separation procedure described by Edwards et al.32. For this, 0.2 g of carbonate subsamples were drilled. All errors are reported as two standard deviations (2σ). Standard chemistry procedures were used to separate U and Th. A triple-spike (229Th–233U–236U) isotope dilution method was employed to correct instrumental fractionation and determine U–Th isotopic ratios and concentrations. The instrumentation, standardization, and half-lives are reported in Cheng et al.33,34. All U–Th isotopes were measured in peak-jumping mode on a MasCom multiplier placed behind the retarding potential quadrupole. We followed procedures to characterize the multiplier similar to those described in Cheng et al.33. Uncertainties in U/Th isotopic data were calculated offline, including corrections for blanks, multiplier dark noise, abundance sensitivity, and spike composition. 230Th ages were corrected using an initial 230Th/232Th atomic ratio of 4.4 ± 2.2 × 10−6. The values are for material at secular equilibrium with respect to the bulk earth 232Th/238U value of 3.8. The U and Th decay constants are reported in Cheng et al.34.

Stable isotope analyses

Samples for stable isotope analysis (oxygen and carbon) were taken using the hand-held Dremel Rotary tool fitted into a Dremel Rotary Tool Workstation Drill Press. The sample was placed on a velmex tree ring counter, in order to accurately sample the speleothem along the growth axis, where ~ 700 to 800 μg of the sample was collected. Stable isotope samples were processed at the University of Ottawa’s Ján Veizer Stable Isotope Laboratory.

Oxygen and carbon isotope values (ẟ18O and ẟ13C) were measured for 363 calcite samples drilled at discrete 1 mm intervals down the growth axis of the stalagmite. Approximately 500–700 μg of calcite was weighed for each sample and transferred to the exetainer for analysis. Stable isotopic analyses were conducted at the Ján Veizer Stable Isotope Laboratory, where 0.1 mL of degassed anhydrous phosphoric acid was added to the exetainer, while on its side. Once all vials had acid added, the exetainers were flushed and filled with UHP helium off-line for 4 min at a rate of 60–70 mL/min. Following this, the vials were then immediately placed upright and in the heating block of the GasBench, and left to react for 24 h. Regular analysis of the CO2 headspace was followed, using a Thermo Finnigan Delta Plus XP IRMS. The isotopic composition of the CO2 (g) for each sample was measured relative to the international reference, the Vienna Pee Dee Belemnite (VPDB), using the conventional “δ-per mil” notation. The analytical precision of this instrument is + /− 0.15‰.

Results and discussion

Age model

A total of 44 U/Th dating analyses were performed on calcite samples from MCS-01, among which 5 preliminary dating analyses were carried out at the Radiochronology laboratory of the GEOTOP-UQAM-McGill Research Center (Table 1), and then 39 additional analyses were added in the Isotope Laboratory of the University of Minnesota (Table 2). Indeed, age reversals due to potential aragonite-calcite age discordance35 have already been shown in Western Cuba stalagmites6.The XRD results from the basal part of the stalagmite show that MCS-01 is composed of calcite. However, 11 major outliers were detected by the StalAge algorithm, and therefore excluded from the age model since their age and uncertainties are significantly far from the stratigraphic order (e.g. by several thousands of years). Another verification method that was adopted in the age dating of samples below and above each major outlier that could not be well dated to confirm them.

Table 1 U/Th dating results from the Radiochronology laboratory of the GEOTOP-UQAM-McGill Research Center.
Full size table
Table 2 U/Th dating results from the Isotope Laboratory of the University of Minnesota.
Full size table

The final age‐depth model was constructed with 33 U/Th ages (including 4 U/Th ages for the Holocene period which is not the main focus of this manuscript), using the StalAge algorithm36 (Fig. 3). Ages are reported in years BP with 1950 AD as the year of reference. The growth of the stalagmite was active when it was collected. The U/Th dating results show that the stalagmite had three main growth phases, from 100.5 ± 2.5 to 85 ± 4.2 ka BP, which corresponds to MIS 5b, followed by a growth hiatus of several tens of thousands of years. The growth of the stalagmite was restored from 11 ± 0.25 to 10.5 ± 0.25 ka BP and was interrupted until 0.5 ± 0.1 ka BP.

Figure 3
figure 3

Chronological age-depth model developed for the stalagmite MCS-01 during MIS 5b based on StalAge Algorithm36. The grey lines represent the age model’s 95% confidence intervals. Age uncertainties of dated samples used in the age model are shown as well. Minor outliers shown in the age model we still considered as their uncertainty is increased using an iterative procedure by the StalAge Algorithm.

Full size image

Description of δ18O and δ13C records

The stable isotope results show considerable variability between the growth periods (Holocene vs. MIS 5b) with respect to both the δ18O and δ13C values (Fig. 4). During MIS 5b, δ18O values ranged from − 1.1 to − 4.3‰ and the δ13C values ranged from − 2.4‰ to − 11.9‰. Interestingly, the δ18O values are consistently low from 90.2 ± 0.4 to 87.6 ± 0.5 ka BP and then undergo an abrupt increase to near − 1‰. The δ13C values in this section of the stalagmite, however, respond in an inverse manner, becoming more negative and reaching values of approximately − 10‰. Prior to this, δ18O and δ13C values show similar peaks and troughs. In the early Holocene section of the stalagmite, δ18O values range from − 0.5 to − 5.2‰ and − 1.5 to − 12.9‰ for δ13C. This time period has a much greater range and variability compared to MIS5b, and both the δ18O and δ13C data reveal a consistent increase throughout this period, before undergoing abrupt decreases. For the late Holocene section, the δ18O and δ13C data are the least variable; these values are − 3.2 to − 5.7‰ and − 7.4 to − 13.1‰ respectively. Previous results in western Cuba revealed that the Caribbean region had a substantial change in temperature37,38, confirmed by a more recent speleothem paleoclimate evidence from Puerto Rico, which indicates a cooler last glacial maximum compared to the Holocene39.

Figure 4
figure 4

The MCS-01 speleothem isotope (δ18O and δ13C) records from Majaguas Upper cave in Western Cuba (a); the early Holocene record of stalagmite MCS-01 is shown as well (b).

Full size image

The analyses of drip water and speleothems from western Cuban caves showed that the δ18O values in this area are mainly dominated by the summer rainfall1. The δ18O data from western Cuba speleothems has been interpreted in recent studies as representing a primarily climatic signal, with more positive δ18O values indicative of drier climate conditions, and more negative δ18O values in wetter conditions6,13,40. The carbon isotopic results from speleothems are more complicated and can be indicative of a number of other processes in addition to climate change, such as vegetation productivity2, changes in atmospheric pCO2, and hydrological or biological processes in the soil and/or the epikarst. Hence, the remainder of the discussion focuses mostly on the δ18O data.

Local response to North Atlantic abrupt cooling events H8 and H9

The MCS-01 record covers the period between 98.7 ± 3.1 and 84.9 ± 4.5 ka BP within MIS 5b and two relatively short periods from 11.1 ± 0.25 to 10.5 ± 0.25 ka BP and from 0.4 ± 0.1 to present 0.05 ± 0.7 ka BP. Hence, our discussion in this paragraph will focus on the MIS 5b. The δ18O record was corrected for global ice volume changes by subtracting 0.008‰ per meter of sea-level change41, using the most detailed sea level data with an independent age model by Grant et al.42. The corrected record presents δ18O values ranging between - 1 and − 4.3‰, with three distinctive periods of high δ18O corresponding to dry periods 87.6 ± 0.5 to 85 ± 4.2, 93.1 ± 0.5 to 90.2 ± 0.4, and 98.6 ± 2.9 to 95.9 ± 0.3 ka BP. We compare the MCS-01 δ18O data (Fig. 5c) to the ITCZ record from the Cariaco Basin (Fig. 5b)17 and the NGRIP ice core (Fig. 5a)23,24,25,26,27,28 to assess a tropical response to high northern latitude climatic changes.

Figure 5
figure 5

Comparison of the MCS-01 speleothem δ18O record from Majaguas Upper cave in western Cuba (c) with the North Greenland Ice Core Project (a)23,24,25,26,27,28 and the reflectance record from the sediment core MDO3-2621 from the Cariaco Basin (b) representing ITCZ movements17.

Full size image

The dry and/or cold periods of 87.6 ± 0.5 to 85 ± 4.2 and 93.1 ± 0.5 to 90.2 ± 0.4 ka BP seem to be synchronous with Heinrich events 8 and 9 (H8 and H9) and corresponding southward shifts of the ITCZ at those times. Interestingly, the H8 event seems more pronounced in western Cuba compared to H9, with persistent positive δ18O values which are the highest in the MCS-01 record. This can be explained by North Atlantic temperatures, which were colder during H8 compared to H9 as shown by the NGRIP record, and the stronger southward shift of the ITCZ. These relationships represent the bipolar seesaw in which cold conditions in the North Atlantic during H8 and H9 might have been linked via the thermohaline circulation to warm conditions in the southern hemisphere, such that cold conditions in the North Atlantic are associated with a southward shift of ITCZ, and hence dry and/or colder conditions in Cuba. This is a new result showing for the first time a local climate response in Western Cuba to H8 and H9. However, caution is still needed given the large age model uncertainties around 85 ka BP.

The North Atlantic cold events H8 and H9 were characterized by cooler and/or drier conditions not only in Cuba, but also in Central America (CA), as shown by a speleothem record from Costa Rica15 (Fig. 6a). The close connection between Cuba and Costa Rica derives from the easterly trade winds associated with the North Atlantic Subtropical High, which strengthen the Caribbean Low-Level Jet (CLLJ), and drive moisture towards both the western Gulf of Mexico and the CA isthmus to the Pacific13. Speleothem records from Western Cuba and Central America show considerable similarities during North Atlantic abrupt climate events13,15,17. Furthermore, the comparison of our data (Fig. 6b) with the Botuvera speleothem record from SE Brazil22 (Fig. 6c) reinforces the idea of anti-phased climate conditions between the Caribbean and South America caused by ITCZ shift at the millennial timescale. For instance, a southern position of the ITCZ observed during H8 leads to significant dry conditions in the Caribbean while it corresponds to a stronger SAMS regime to the south. Indeed, the highest δ18O values in the MCS-01 record are found during H8 between (~ 1‰). The Greenland Interstadial (Dansgaard-Oeschger; DO) event 22 is also associated with low δ18O values which suggest a northward shift of the ITCZ and the prevalence of wet conditions in the Caribbean, whereas dry conditions are established in the SASM domain, consistent with Strikis et al.43.

Figure 6
figure 6

Comparison of the MCS-01 speleothem δ18O record from Majaguas Upper cave in western Cuba (b) with the Terciopelo speleothem record from Central America (a)15 and the Botuvera speleothem record from SE Brazil (c)22.

Full size image

Ocean and atmospheric mechanisms during the Last Glacial period

In terms of comparisons to other local studies, the Cuba Medio (CM) speleothem record from western Cuba dates to MIS 513, but CM is mostly discontinuous though the time period covered by stalagmite MCS-01. The comparison with the CM record shows that the absolute δ18O values are not the same during the contemporary growth period of both stalagmites from 96.2 ± 0.25 to 92.8 ± 0.3 ka BP, whereas the CM record presents higher absolute δ18O values. This can be explained by drip water and site-specific conditions15. For instance, the CM stalagmite was collected in Torch Cave (western Cuba) at 170 m above sea level, which is overlain by approximately 60 m of rock13. This is in contrast to the ~ 100 m of rock overlying stalagmite MCS-01. However, both speleothem records show a similar trend with a transition from low to high δ18O values between 96.2 ± 0.25 and 92.8 ± 0.3 ka BP indicating a resumption of drier conditions during the onset of H9.

Other than the CM record of Warken et al.13, we also compare our MCS-01 results to multi-speleothem records of δ18O values from the Cuba Pequeño (CP) data from the Dos Anas cave system6 (Fig. 7e). The western Cuban multi-speleothem record (composite of MCS-01, CP and CM speleothem records) covers the last 100 ka at high resolution and allows us to assess the mechanisms that modulated rainfall variability in the region during the last glacial period through comparison with regional paleoclimate data and variability modes. On orbital timescales, the multi-speleothem δ18O record shares similarities with the ITCZ reconstruction (Fig. 7f) and the SST records in the Caribbean basin (Fig. 7d)16 with higher rainfall in western Cuba during periods of warm tropical North Atlantic Ocean. The multi-speleothem record is consistent with winter insolation on glacial-interglacial timescales13. Indeed, enhanced winter insolation may induce a warming effect on surface waters in the Caribbean, which would result in an extended humid season15,44.

Figure 7
figure 7

Comparison of the Western Cuban speleothem records (normalized data) (e) by Fensterer et al.6 (dark blue), Warken et al.13 (light blue) and this study (blue), with the North Greenland Ice Core Project (a)23,24,25,26,27,28, the Bermuda Rise 231 Pa/230Th as a proxy of AMOC strength (b)18,20,21, the North Atlantic SST from core SU90-03 (c)48, the Caribbean Sea Surface Temperature (SST) and δ18O records (d) inferred from planktonic foraminifera from the sediment core ODP 999A16, and the reflectance record from the sediment core MDO3-2621 (f) from the Cariaco Basin reflecting ITCZ movements17. Gray bars indicate the timing of the North Atlantic abrupt cooling episodes (Heinrich events).

Full size image

Previous work suggests a close relationship between the Caribbean and the North Atlantic climate during the last glacial period13. Positive peaks in Western Cuban speleothem δ18O records are synchronous with the Heinrich events (Fig. 7a). Furthermore, a remarkable consistency was observed with the Bermuda rise 231 Pa/230Th record of AMOC strength18,20,21, whereas the positive peaks in the speleothem record correspond to dry events, which coincide with cooler temperatures in the North Atlantic and periods of reduced AMOC during stadials13 (Fig. 7b). Indeed, the Heinrich events are associated with strong freshwater flow in the North Atlantic Ocean45,46, which resulted in the weakening of the AMOC due to reduced deep water formation18,20,47. Interestingly, the climate response in Western Cuba to H8 seems relatively as severe as during the mega-drought period of H1, as shown by the range of variability in the speleothem δ18O values. However, this is not particularly observed in the AMOC strength records nor in the North Atlantic SST48 (Fig. 7c) and the NGRIP records, whereas this phenomenon is prominent the ITCZ record which indicates a substantial southward shift of the ITCZ during both H8 and H1 events (Fig. 7f). Hence, the abrupt changes observed during both events are limited to the Western Cuban region and might have been amplified by the ITCZ changes.

Conclusion

The Caribbean is a key region to obtain information about past natural climate variability in the tropics and its relationship to climate forcing mechanisms. However, paleoclimate evidence from this region is still scattered and limited. In this study, we presented a new speleothem record from Majaguas Upper cave in Western Cuba. The MCS-01 record covers the period between 98.7 ± 3.1 and 84.9 ± 4.5 ka BP within MIS 5b, which was not documented before in Cuba.

The comparison of our data with the NGRIP ice core record showed that two distinct dry and/or cold periods of 87.6 ± 0.5 to 85 ± 4.2 and 93.1 ± 0.5 to 90.2 ± 0.4 ka BP are synchronous with the Heinrich events 8 and 9. The H8 event is more pronounced in western Cuba compared to H9, associated with persistent positive δ18O values which are the highest in our record. This is associated with cooler sea surface temperatures in the North Atlantic Ocean during H8 as compared to H9. Our results are consistent with the stronger southward shift of the ITCZ during H8 from the Cariaco Basin record, providing crucial information to fill the gaps of previous speleothem records from the region. Hence, a multi-speleothem record from Western Cuba spanning the last 100 ka with a high resolution reveals a consistent response to Heinrich events, as positive peaks in speleothem δ18O records indicate dry and/or cooler conditions in the Caribbean as a response to the weakening of the AMOC due to reduced deep water formation resulting from cooler sea surface temperatures in the North Atlantic high latitudes.

In summary, as compared with regional paleoclimate records, we show that the North Atlantic cold events H8 and H9 were characterized by cooler and/or drier conditions not only in Cuba but also in Central America, which resulted from easterly trade winds associated with the North Atlantic Subtropical High, that strengthen the Caribbean Low-Level Jet and drive the moisture towards both the western Gulf of Mexico and the CA isthmus to the Pacific. Moreover, our data, as compared with South American paleoclimate evidence, reinforce the idea of anti-phased climate conditions between the Caribbean and South America caused by the ITCZ shift during millennial-scale events. Therefore, we suggest that a more southward position of the ITCZ observed during H8 and H9 leads to significant dry conditions in the Caribbean whereas it corresponds to a stronger SAMS regime to the south.