Mechanisms controlling gypsum speleothem formation and the isotopic composition of GHW
Previous studies on subterranean microclimate parameters in the gypsum karst of Sorbas Basin (i.e. Covadura Cave, 2 km apart from Sima Blanca Cave) concluded that the precipitation of gypsum speleothems is favoured during the winter dry season (January–March), when relative humidity in the upper cave passages is less than 95% and occasionally below 60%5. Low winter cave relative humidity results from an intense exchange of air masses with the external atmosphere when the outside air temperature falls below the cave’s air temperature5. From April to December, relative humidity in the cave’s atmosphere is closer to water saturation (normally above 95%), decreasing evaporation in the cave and thus suppressing gypsum precipitation. Consequently, as evaporative conditions are more likely to occur during winter, we interpret the isotopic record of the SBL stalactite as reflecting winter evaporative conditions in the cave.
In addition to infiltration water, condensation of atmospheric water vapour has been found to be an important source of water for gypsum speleothem formation during summer and autumn5,22. Water condenses on the cave ceiling, dissolving Messinian gypsum host-rock and the water becomes saturated with respect to gypsum. The saturated water then migrates by gravity along the surface of the stalactite, mixing along its path with seepage water, and gypsum deposition occurs because of evaporation. The relative contribution of the condensation water source to the dripwater was estimated to be ~ 40% in the nearby Covadura Cave system. The intensity of this condensation process depends on the difference between external and internal cave temperatures and on the degree of cave ventilation5,22. Sima Blanca cave is a shallow subterranean system (~ 10 m deep) and its temperature is expected to equilibrate rapidly with the external atmospheric temperature. This has been observed in the passages of Covadura Cave system (also ~ 10 m deep), where the equilibration time is < 2 months23. As a result, we do not expect the condensation to seepage water ratio to change significantly either on an inter-annual basis or during the growth of the SBL stalactite. In addition, the modern mean isotopic composition of dripwater and condensation water in these caves differ by only ~ 1‰ for δ18O and ~ 5.5‰ for δD, whereas their d-excess and 17Oexcess are indistinguishable within error5. This reinforces the idea that long-term changes in the condensation to seepage water ratio did not significantly alter the isotopic composition of the SBL stalactite, but rather isotopic changes mainly reflect the degree of evaporation within the cave.
Once meteoric water reaches the cave and mixes with condensation water, the most important process affecting δ17O, δ18O and δD and derived parameters d-excess and 17Oexcess is evaporation5. The absence of stalagmites in Sima Blanca suggests that most dripwater is evaporated from the speleothems on the ceiling. We observe that the values of d-excess and 17Oexcess versus δ18O describe an evaporation trend (i.e. lower d-excess and 17Oexcess values when δ18O increases) (Fig. 4). In general, the SBL speleothem-forming waters are more evaporated than modern cave dripwater (2011–2013 period)5 (Fig. 4). Higher infiltration/evaporation ratios are expected during wetter periods because of enhanced rainfall and higher relative humidity in the cave. This results in a lowering of δ17O, δ18O and δD values and higher d-excess and 17Oexcess of the evaporated water14,15,24. During drier periods, the δ17O, δ18O and δD values of evaporated dripwater increase, whereas d-excess and 17Oexcess decrease when aridity increases because of lower infiltration/evaporation.
Changes in the isotopic composition of rainfall and, therefore, infiltration water to the cave may also influence the SBL stalactite record. There are two main sources of humidity that may influence the isotopic composition of rainfall in the Mediterranean part of the Iberian Peninsula25,26. Convective rainfall events from the Mediterranean are isotopically enriched, and prevail during summer and autumn and during warm climatic periods. In contrast, rainfall with lower isotopic values are transported along Atlantic storm tracks, which dominate during winter and during colder climatic periods25,26 The mean δ18O values of modern rainfall from the Mediterranean have been found to be up to 3‰ higher in this region than rainfall events from Atlantic fronts27. Also, the d-excess values in rainfall from Mediterranean convective events are normally higher than in precipitation from the Atlantic moisture source28. Changes in the contributions of the Mediterranean and Atlantic moisture sources to the total hydrological budget in SE Iberia have been proposed previously to explain the δD variability of n-alkanes in sediment profiles of alpine lake of the Sierra Nevada26. This mechanism could have partially influenced the isotopic composition dripwater in Sima Blanca Cave; however, the fact that the δ18O–δD values of the speleothem-forming water are linearly correlated, with a slope much lower than that of the Local Meteoric Water Line (Fig. 3), strongly suggests that evaporation played a dominant role in controlling the isotopic variations of the dripwater recorded by the speleothem. Thus, we interpreted the long-term δ18O, δD and d-excess changes in the SBL stalactite dominantly as a record of evaporation in the cave that could have been affected by variations in the isotopic composition of rainfall in this region.
No specific studies have been conducted to measure the 17Oexcess of past rainwater in this region. However, monitoring of triple oxygen isotopes of rainfall in southern Japan reported modern annual 17Oexcess cyclicity (values ranging from ~ 10 to ~ 40 per meg) that is positively correlated to normalized relative humidity at the oceanic moisture source29. Water evaporation under low relative humidity conditions at the moisture source generates water vapour (and rainwaters) with greater 17Oexcess values30. This contrasts with the behaviour of 17Oexcess in evaporated waters that tends to be lower when evaporation occurs under low relative humidity conditions14,15. Therefore, the 17Oexcess record of the SBL stalactite may be partially controlled by the long-term variability in the triple oxygen isotope composition of rainwater/infiltration water in the Sorbas area and the effects on 17Oexcess of dripwater evaporation in the cave varying relative humidity conditions. Other potential mechanism may include changes in the main moisture source region (i.e. Mediterranean vs Atlantic source) or changes in the contributions of a non-rainfall water source (e.g. fogs) to cave dripwater.
In summary, the SBL stalactite isotope record is interpreted here as a function of infiltration/evaporation ratio in the cave, which is affected by rainfall amount in SE Iberia and humidity in Sima Blanca Cave. Also, varying contributions of the Atlantic and Mediterranean moisture sources to rainfall in this region could have influenced the isotopic signal of dripwater in the cave. Higher δ17O, δ18O and δD values derived from drier conditions and potentially from higher contributions of rainfall from the Mediterranean moisture source; lower δ17O, δ18O and δD, indicate wetter periods and may also suggest a greater influence of the Atlantic moisture source to the precipitation in this region.
Lower d-excess values in the SBL record indicate drier climate, although the signal may have been attenuated by higher d-excess of rainfall in response to greater contributions of Mediterranean convective precipitation events during dry periods. Higher d-excess values of GHW suggest wetter climatic conditions, with possible attenuation of the signal because of greater influence of the Atlantic rain events, which have lower d-excess values. The mechanisms controlling 17Oexcess in GHW of gypsum speleothems in Sima Blanca Cave is more complicated, and there is very little information available for 17Oexcess variability of modern rainfall in this region. Therefore, interpretation of the 17Oexcess record of SBL remains speculative.
The Sima Blanca paleoclimate record
The climatic changes inferred from the oxygen and hydrogen isotope ratios in GHW of the SBL stalactite are consistent with other nearby paleoclimate records from the region (Fig. 1), including Lake Zoñar in Andalucía31,32,33,34, several alpine lakes in Sierra Nevada35,36,37,38,39 and sediment records in the Alboran Sea33 (Fig. 5). Several lines of evidence suggest that the geochemical signal of the gypsum stalactite is primary and has not been altered by post-depositional processes (i.e. isotope exchange or gypsum dissolution/re-precipitation): (1) the SBL isotope record agrees with other regional paleoclimate proxies; (2) the reconstructed δ18O and δD of the speleothem-forming water describe an evaporation line with changing distance along the line indicating different degrees of evaporation; and, (3) there are no age reversals in the U/Th dates.
The SBL stalactite started growing at ~ 800 BCE. Dry conditions in southern Iberia prior to 800 BCE, as evidenced by a lower-than-average level of Lake Zoñar31,32, may have prevented stalactite formation. Likewise, low concentration of Quercus pollen in sediments of alpine lakes of Sierra Nevada are indicative of dry conditions during the Late Iron Age37,38 (Fig. 5). The relatively high δ17O, δ18O and δD values (1.3, 2.2 and − 5‰, respectively) of speleothem-forming water and relatively low d-excess (− 22‰) at 800 BCE compared to the mean value of the entire record (δ17O = 0.7‰, δ18O = 1.4‰, δD = − 7.2‰ and d-excess = − 18.1‰) indicates generally drier climate prevailed at this time. Also, greater contribution of the Mediterranean vapor source to rainfall in this region could result in higher isotopic values of dripwater in the cave during this period.
Consistent with a trend towards wetter conditions, δ17O, δ18O and δD values of evaporated paleo-dripwater decreased from 800 until ~ 500 BCE by about 0.5, 1.1 and 2.9‰, respectively, while d-excess increased by 5.9‰. This period includes the onset of the Iberian Roman Humid Period (IRHP), as inferred from the sediments of Lake Zoñar (650–510 BCE)33. A sharp decrease in δ17O, δ18O and δD values between 400 and 300 BCE result in the lowest isotopic values observed in the entire record. This abrupt change is accompanied by an increase in d-excess by ~ 8‰, until − 14‰, which is the highest d-excess value recorded in the SBL stalactite, indicating a trend toward wetter conditions. This shift in the isotope records and inferred wetter conditions coincide with an increase in Rb/Al recorded in the Lake Zoñar sediments at 400 BCE, as a result of enhanced runoff because of increased rainfall33. In addition, the concentration of Quercus pollen in Laguna de la Mula in the Sierra Nevada started to increase from 350 BCE as a result of enhanced water availability36,38.
The δ17O, δ18O, δD and d-excess values stabilized from 300 to 100 BCE at ~ 0.3‰, 0.6‰, − 8‰ and − 13‰, respectively, which are among the lowest values of the record (Fig. 5), suggesting wetter conditions that may have been accompanied by greater contributions of rainfall events from an Atlantic moisture source26. This wetter-than-average climate occurred around the time of the Roman occupation of Iberia (197 BCE) and is consistent with increased input of fluvial sediment to the Alboran Sea at that time33. Also, a large range of paleoclimatic records support these wet conditions in southern Iberia at ~ 300–200 BCE (Fig. 5) including enhanced runoff to Lake Zoñar32,33, an increase in deciduous Quercus pollen in Laguna de la Mula36,38; a decrease in Sahara dust input and increase in detrital sedimentation in Laguna de Hondera35; and enhanced aquatic production in the Borreguil de la Caldera and the Padul peat bog records37,39. Previous studies have identified this period as the wettest stage of the IRHP that began gradually from 650 BCE and reached the wettest conditions during the Iberian–Early Roman Epoch (510–190 BCE)32. The SBL record and other lake proxies in the region suggest this change may have been more abrupt.
Following this wet period, the SBL data support an aridification trend from ~ 100 BCE to 600 CE. This is indicated by the decrease in d-excess (− 10‰) during this period. The drying inferred from Sima Blanca cave agrees well with other Iberian terrestrial and marine records (Fig. 5) that document a general late-Holocene aridification from 5,000 years BP to present. This trend is also observed in alpine lakes of the Sierra Nevada37,38,39 and recurrent desiccation of other lakes in south Iberia40,41. The driest stage of the SBL record occurred at ~ 500 CE, when δ17O, δ18O, δD and d-excess of the evaporated paleo-dripwater reach their highest values (1.2‰, 2,4, − 3.4‰ and − 22.9‰). An increase in the frequency and intensity of convective rainfall events from the Mediterranean could have contributed to the relatively high δ17O, δ18O, δD values recorded by the speleothem during this period.
The latest part of the SBL record (from ~ 600 to ~ 800 CE) indicates conditions slightly wetter than during the previous 300 years, but drier than the period from 200 to 100 BCE. Dry conditions during the early stages of the Migration Period are indicated by a prominent decrease in Mediterranean forest in SE Iberian mountains, by an increase in Cichorioideae herbs and by the decline in the water levels in alpine lakes in the area39. Additionally, a progressive drying around 700 CE has been observed in Alboran Sea sediments. The SBL stalactite stopped growing at ~ 800 CE probably because of a reduction in water infiltration to the cave. This coincided with the last stages of the Visigoth occupation of Hispania (~ 800 CE), when generally dry conditions prevailed in Iberia33,39,42.
Previous studies in Andalusia have suggested a possible link between varve thickness and geochemical proxies in several lakes of lowlands and alpine areas and the North Atlantic Oscillation (NAO) and solar variability32,38,43. The SBL stalactite record between ~ 800 BCE and ~ 800 CE also generally correlates with changes in the NAO (North Atlantic Oscillation) index44 and atmospheric radiocarbon production45. It is therefore possible that the aridification trend in SE Iberia may have been related to increased solar activity and NAO+ conditions (Fig. 6). While the 17Oexcess record of SBL tracks the variability of solar activity and NAO throughout the record, variation in δ18O, δD and d-excess is anti-correlated during the Iron Age (800–500 BCE). Additional monitoring of modern conditions, including measurement of 17Oexcess in dripwater and rainfall in the region may help to better understand the climate signal recorded by the SBL stalactite.
Socio-cultural changes in south Iberia during the formation of the SBL stalactite
The wettest conditions inferred from the SBL stalactite (Fig. 6) occurred around the time of the Roman occupation of Iberia (197 BCE). Specifically, the wettest period starting at ~ 300–200 BCE agrees (within errors of the age model) with the time of the Second Punic Roman War (218–201 BCE) and the defeat of the Carthaginian by the Romans (206 BCE). The Roman Republic slowly expanded its control over Hispania over the next 200 years, which coincided with relatively wet climate in southern Spain that was probably favorable to cropping and farming.
As a consequence of the Roman activity, the IRHP was characterized by an important increase in environmental pollution (Fig. 7). For example, there was an increase in heavy metal input to aquatic systems in both south and north Iberia33,43,46,47,48, as well as significant landscape modification45 during this period. Lead pollution from Roman mining in Iberia is recorded by lake sediment cores in Europe49 and ice cores in Greenland50, the Russian Arctic51 and the Alps52. Increased charcoal abundance in several lake sediment cores suggest intense deforestation and biomass burning during the early and middle-IRHP36,38,48,53 (Fig. 7).
Wetter conditions also prevailed in the western Mediterranean region at the time of the IRHP, as evidenced by lake records and a carbonate speleothem20,54,55,56,57. Moreover, increased rainfall during this period has been inferred from tree-ring growth in central Europe58 and analysis of carbon isotopes in ancient crop grains59, which suggest wetter climate conditions during the conquest and apogee of the Roman Empire in Iberia.
The SBL record suggests that from ~ 100 BCE until ~ 500–600 CE, a general aridification trend occurred that was punctuated by some brief wetter periods. Roman mining and intense land-use change declined in Iberia during this period36,38,53 until the Roman Empire eventually abandoned the Iberian Peninsula at ~ 400 CE. Indeed, a gradual decrease in mining and lead emissions occurred from ~ 100 to 250 CE, just before the Imperial Crisis (234–284 CE)51 (Fig. 7).
The Imperial Crisis has been attributed to a combination of migrations into the Roman territory, Barbarian invasions and civil wars, peasant rebellions, political instability, debasement of currency, and economic depression. Natural disasters, such as plague (i.e. the Plague of Cyprian, 249–262 CE) and/or increased climate variability, may have contributed to the fall of the Roman Empire60. Indeed, exceptional climate variability in Europe from ~ 250 to 550 CE is evidenced by tree-ring records, including periods of severe and prolonged droughts58,61. This was a time of increased aridification of SE Iberia. Drier winters, as suggested by the SBL record, probably resulted in crop failure and more extreme weather events that led to agricultural instability during the abandonment of Iberia by the Romans17.
The driest conditions inferred from the SBL stalactite occur at the transition between the IRHP and the Migration Period, at 500–600 CE. Cold and dry conditions occurred in the northern hemisphere during the Late Antique Little Ice Age (between 536 and 660 CE) and had an important impact on the transformation of ancient civilizations. Written sources pointing to significant veiling of solar radiation in 536 and 537 CE have been attributed to a powerful volcanic eruption that resulted in crop failures in different areas of the Roman Empire61. These events coincided with decreasing total solar irradiance in the Northern Hemisphere62 and an event of ice-rafting in the subpolar North Atlantic (cold event 1, 800 CE)63.
In summary, we demonstrate for the first time that stable isotopes of hydration water in gypsum speleothems can be used as a proxy for paleo-humidity/aridity. This novel approach offers the possibility of estimating all the oxygen (16O, 17O, 18O) and hydrogen (1H, 2H) stable isotopes of cave dripwaters as a proxy for evaporation. We suggest that gypsum speleothems in arid or semiarid regions constitute a valuable and unexploited archive of paleoclimate information.