INTRODUCTION
The impact of future climatic change in the eastern Mediterranean (EM), a region already exposed to severe agricultural and environmental water stress, must be better constrained and may be assessed by investigating past climatic variability (Masson-Delmotte et al., Reference Masson-Delmotte, Schulz, Abe-Ouchi, Beer, Ganopolski, Gonzalez Rouco and Jansen2013). The climate of the EM is influenced by weather systems originating from the North Atlantic Ocean, passing from Europe to the Mediterranean Sea. In the EM region, many records from lake and marine cores (Rohling et al., Reference Rohling, Cane, Cooke, Sprovieri, Bouloubassi, Emeis and Schiebel2002; Emeis et al., Reference Emeis, Schulz, Struck, Rossignol-Strick, Erlenkeuser, Howell and Kroon2003; Jones and Roberts, Reference Jones and Roberts2008; Almogi-Labin et al., Reference Almogi-Labin, Bar-Matthews, Shriki, Kolosovsky, Paterne, Schilman, Ayalon and Matthews2009; Develle et al., Reference Develle, Herreros, Vidal, Sursock and Gasse2010) and speleothems (Bar-Matthews et al., Reference Bar-Matthews, Ayalon, Gilmour, Matthews and Hawkesworth2003; Verheyden et al., Reference Verheyden, Nader, Cheng, Edwards and Swennen2008; Cheng et al., Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015) have been used to define past climate conditions. Regional paleoclimate records suggest that the region is sensitive to large-amplitude glacial–interglacial changes and climatic fluctuations on millennial to decadal timescales (Bar-Matthews et al., Reference Bar-Matthews, Ayalon, Gilmour, Matthews and Hawkesworth2003; Almogi-Labin et al., Reference Almogi-Labin, Bar-Matthews, Shriki, Kolosovsky, Paterne, Schilman, Ayalon and Matthews2009; Bar-Matthews and Ayalon, Reference Bar-Matthews and Ayalon2011). However, this sensitivity is unequal among regions because of the heterogeneity of the EM climate (Fig. 1) over short distances (Ulbrich et al., Reference Ulbrich, Lionello, Belusic, Jacobeit, Knippertz, Kuglitsch, Leckebusch and Lionello2012). The precipitation distribution shows high spatiotemporal variability, with most of the effective moisture occurring during winter–spring seasons and being concentrated in mountainous regions. Confidently reconstructing this variability requires a dense network of precisely dated and highly resolved paleoclimate records. Past spatiotemporal climate variability in the EM is still poorly documented due to unevenly distributed records (Burstyn et al., Reference Burstyn, Martrat, Lopez, Iriarte, Jacobson, Lone and Deininger2019).
(A) Location of Qadisha Cave (this study) and other paleoclimatic records spanning the Holocene: Lisan lake (Torfstein et al., Reference Torfstein, Goldstein, Stein and Enzel2013a, Reference Torfstein, Goldstein, Kagan and Stein2013b), Soreq Cave (Bar-Matthews et al., Reference Bar-Matthews, Ayalon, Gilmour, Matthews and Hawkesworth2003; Burstyn et al., Reference Burstyn, Shaar, Keinan, Ebert, Ayalon, Bar-Matthews and Feinberg2022), Zalmon Cave (Keinan et al., Reference Keinan, Bar-Matthews, Ayalon, Zilberman, Agnon and Frumkin2019), Peqiin Cave (Bar-Matthews et al., Reference Bar-Matthews, Ayalon, Gilmour, Matthews and Hawkesworth2003), Mizpe Shelagim (MS) Cave (Ayalon et al., Reference Ayalon, Bar-Matthews, Frumkin and Matthews2013), Ammiq peat record (Hajjar et al., Reference Hajjar, Haïdar-Boustani, Khater and Cheddadi2010), Jeita Cave (Verheyden et al., Reference Verheyden, Nader, Cheng, Edwards and Swennen2008; Cheng et al., Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015), Yammouneh basin (Develle et al., Reference Develle, Herreros, Vidal, Sursock and Gasse2010), El-Jurd peat record (Chedaddi and Khater, Reference Cheddadi and Khater2016), Ghab core (Van Zeist and Woldring, Reference Van Zeist and Woldring1980; Yasuda et al., Reference Yasuda, Kitagawa and Nakagawa2000), Sağlik peat (Sekeryapan et al., Reference Şekeryapan, Streuman, van der Plicht, Woldring, van der Veen and Boomer2020), Incesu Cave (Erkan et al., Reference Erkan, Bayari, Fleitmann, Cheng, Edwards and Özbakir2022), Dim Cave (Ünal-Imer et al., Reference Ünal-İmer, Shulmeister, Zhao, Uysal, Feng, Nguyen and Yüce2015), LC21 (Grant et al., Reference Grant, Rohling, Bar-Matthews, Ayalon, Medina-Elizalde, Bronk Ramsey, Satow and Roberts2012), and ODP 967 (Emeis et al., Reference Emeis, Schulz, Struck, Rossignol-Strick, Erlenkeuser, Howell and Kroon2003; Scrivner et al., Reference Scrivner, Vance and Rohling2004). (B) Maps showing seasonal precipitation amounts in the wider eastern Mediterranean region. Panels at the right illustrate seasonal precipitation amounts from June to September and from December to March, respectively. Gray areas indicate regions where the daily precipitation amount is below 0.5 mm. Data retrieved from the ERA-Interim reanalysis data set (1979 to 2015 CE) (Berrisford et al., Reference Berrisford, Kållberg, Kobayashi, Dee, Uppala, Simmons and Poli2011).
In the Levantine region stretching from southern Taurus mountains to southern Negev desert and the island of Cyprus, many studies cover the Holocene period, but few are well dated (Bar-Matthews et al., Reference Bar-Matthews, Ayalon, Gilmour, Matthews and Hawkesworth2003; Cheng et al., Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015) or record rapid climate changes (RCC). Some are located along the coast at low altitudes like Jeita Cave or Zalmon Cave, other records such as Soreq and West Jerusalem Caves are located at mid-altitudes (western flanks of the Judean plateaus). Some lake or marsh records are located in the rain shadow of the Levantine mountains (e.g., the Dead Sea, the Ammiq marsh in the Beqaa inner plain, the Ghab depression). Among the records covering the Holocene, a few are located in high mountainous areas, such as Mizpe Shelagim Cave in Mount Hermon and Incesu Cave in the Taurus Mountains (Turkey). In the central Levant, Mt. Lebanon (3088 m) is an imposing range facing the moisture coming from the Mediterranean and creates therefore a peculiar local climate system in Lebanon. None of the records spanning the Holocene were retrieved in areas located at high altitude, except for the El-Jurd marsh (Cheddadi and Khater, Reference Cheddadi and Khater2016) and Yammouneh lake records, which both provide low-resolution (millennial-scale) data. To analyze the effect of the mountainous climate and its variability during the Holocene, a reconstruction of past climate variations from highly resolved archives such as speleothems is needed. These secondary cave deposits (e.g., stalagmites) are currently considered to be the most suitable terrestrial archives for establishing high-resolution proxy time series in paleoclimate research (Genty et al., Reference Genty, Blamart, Ouahdi and Gilmour2003; Cheng et al., Reference Cheng, Zhang, Spötl, Edwards, Cai, Zhang, Sang, Tan and An2012; Fairchild and Baker, Reference Fairchild and Baker2012).
We report new stalagmite stable isotope data (δ13C, δ18O) from Qadisha Cave (Lebanon) located at 1720 m above sea level (m asl), which provides a high-resolution and well-dated record for northern Mt. Lebanon covering the time period from 9 to 5 ka. The combined measurements of calcite δ13C and δ18O and trace elements enable us to characterize the regional climate trend versus the local mountainous effect on high-altitude records. Isotope measurements of fluid inclusion water (δD, δ18Ow) provide insights into the hydrologic cycle and allow an estimation of mineral formation temperatures for the Holocene optimum at the high-altitude Qadisha Cave.
STATE OF THE ART
During the past decades, several synthesis reports were prepared on paleoclimate studies in the EM (Robinson et al., Reference Robinson, Black, Sellwood and Valdes2006; Finné et al., Reference Finné, Holmgren, Sundqvist, Weiberg and Lindblom2011; Burstyn et al., Reference Burstyn, Martrat, Lopez, Iriarte, Jacobson, Lone and Deininger2019), compiling many marine (Rossignol-Strick and Paterne, Reference Rossignol-Strick and Paterne1999; Kallel et al. Reference Kallel, Duplessy, Labeyrie, Fontugne, Paterne and Montacer2000; Emeis et al. Reference Emeis, Schulz, Struck, Rossignol-Strick, Erlenkeuser, Howell and Kroon2003; Almogi-Labin et al., Reference Almogi-Labin, Bar-Matthews, Shriki, Kolosovsky, Paterne, Schilman, Ayalon and Matthews2009) and terrestrial records spanning the Holocene (Frumkin et al. Reference Frumkin, Ford and Schwarcz2000; Bar-Matthews et al. Reference Bar-Matthews, Ayalon, Gilmour, Matthews and Hawkesworth2003; Verheyden et al., Reference Verheyden, Nader, Cheng, Edwards and Swennen2008; Develle et al. Reference Develle, Herreros, Vidal, Sursock and Gasse2010; Rowe et al., Reference Rowe, Mason, Andrews, Marca, Thomas, van Calsteren, Jex, Vonhof and Al-Omari2012; Ayalon et al., Reference Ayalon, Bar-Matthews, Frumkin and Matthews2013; Cheng et al., Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015; Gasse et al., Reference Gasse, Vidal, Van Campo, Demory, Develle, Tachikawa and Elias2015; Ünal-İmer et al., Reference Ünal-İmer, Shulmeister, Zhao, Uysal, Feng, Nguyen and Yüce2015; Cheddadi and Khater, Reference Cheddadi and Khater2016; Flohr et al., Reference Flohr, Fleitmann, Zorita, Sadekov, Cheng, Bosomworth, Edwards, Matthews and Matthews2017; Carolin et al., Reference Carolin, Walker, Day, Ersek, Sloan, Dee, Talebian and Henderson2019; Sinha et al., Reference Sinha, Kathayat, Weiss, Li, Cheng, Reuter and Edwards2019; Jacobson et al., Reference Jacobson, Flohr, Gascoigne, Leng, Sadekov, Cheng and Fleitmann2021; Burstyn et al., Reference Burstyn, Shaar, Keinan, Ebert, Ayalon, Bar-Matthews and Feinberg2022; Erkan et al., Reference Erkan, Bayari, Fleitmann, Cheng, Edwards and Özbakir2022). Focusing more on the Levantine coast, from the Sinai and Negev Deserts to the southern flanks of the Taurus Mountains in Turkey, and including Cyprus, studies have revealed a general climatic state of a wet and warm Early Holocene from ~10 to ~6 ka during maximum summer insolation, coeval with the deposition of Sapropel S1. Sapropels are organic-rich layers deposited during periods of increased discharge of the river Nile (Rossignol-Strick and Paterne, Reference Rossignol-Strick and Paterne1999; Kallel et al. Reference Kallel, Duplessy, Labeyrie, Fontugne, Paterne and Montacer2000). This processional timescale effect results in isotopically lighter sea-surface water due to the contribution of the δ18O-depleted Nile influx into the EM sea basin. This isotopic source effect is expressed in the Soreq, Peqiin (Bar-Mathews et al., Reference Bar-Matthews, Ayalon, Gilmour, Matthews and Hawkesworth2003; Bar-Mathews and Ayalon, Reference Bar-Matthews and Ayalon2011; Burstyn et al., Reference Burstyn, Shaar, Keinan, Ebert, Ayalon, Bar-Matthews and Feinberg2022), and Jeita Cave records (Verheyden et al., Reference Verheyden, Nader, Cheng, Edwards and Swennen2008; Cheng et al., Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015), with more negative oxygen isotope values following the depletion of sea-surface δ18O as resolved from planktonic foraminifera (Grant et al., Reference Grant, Grimm, Mikolajewicz, Marino, Ziegler and Rohling2016). In the southern Taurus Mountains, both Dim (Unal-Imer et al., Reference Ünal-İmer, Shulmeister, Zhao, Uysal, Feng, Nguyen and Yüce2015) and Incesu Cave records (Erkan et al., Reference Erkan, Bayari, Fleitmann, Cheng, Edwards and Özbakir2022) show a trend toward more depleted δ18O from ~10 to ~8 ka. Although the isotopic composition of the EM source is considered to have a primary effect on the δ18O signal of the terrestrial records, the isotopic δ18O depletion is amplified by increased rainfall (Emeis et al., Reference Emeis, Schulz, Struck, Rossignol-Strick, Erlenkeuser, Howell and Kroon2003; Almogi-Labin et al., Reference Almogi-Labin, Bar-Matthews, Shriki, Kolosovsky, Paterne, Schilman, Ayalon and Matthews2009), also leading to more negative δ18O values. This effect has been invoked for the Early Holocene in some speleothem records (Burstyn et al., Reference Burstyn, Martrat, Lopez, Iriarte, Jacobson, Lone and Deininger2019) and the Yammouneh polje record (Develle et al., Reference Develle, Herreros, Vidal, Sursock and Gasse2010).
The distribution of rainfall and how it changed during the Holocene were not coeval along the Levantine coast. Regional variations in the rainfall amount cause isotopic variations, as attested by cave climate monitoring studies (Bar-Matthews et al., Reference Bar-Matthews, Ayalon, Matthews, Sass and Halicz1996; Ayalon et al., Reference Ayalon, Bar-Matthews and Sass1998, Reference Ayalon, Bar-Matthews and Schilman2004; Nehme et al., Reference Nehme, Verheyden, Nader, Adjizian-Gerard, Genty, De Bont and Clayes2019). Cave water isotopic values are biased toward the main infiltration period, which is the winter–spring season (Nehme et al., Reference Nehme, Verheyden, Nader, Adjizian-Gerard, Genty, De Bont and Clayes2019, Reference Nehme, Kluge, Verheyden, Nader, Charalambidou, Weissbach and Gucel2020). The isotopic signal of the infiltration water is transferred into the cave calcite. Rapid seasonal, decadal, or centennial isotopic variations in speleothems are therefore more likely related to rainfall amount changes (Bar-Matthews et al., Reference Bar-Matthews, Ayalon, Matthews, Sass and Halicz1996; Orland et al., Reference Orland, Burstyn, Bar-Matthews, Kozdon, Ayalon, Matthews and Valley2014; Nehme et al., Reference Nehme, Verheyden, Nader, Adjizian-Gerard, Genty, De Bont and Clayes2019), as cave temperature would not vary strongly on seasonal, annual, or, in deep, less-ventilated caves, decadal scales. The impact of temperature on the calcite δ18O values is about 0.2‰ per 1°C change (Demeny et al., Reference Demeny, Kele and Siklosy2010; Tremaine et al., Reference Tremaine, Froelich and Wang2011; Daëron et al., Reference Daëron, Drysdale, Peral, Huyghe, Blamart, Coplen and Zanchetta2019), whereas a 200 mm change in the annual rainfall amount could cause a change in the δ18O of rainfall, and therefore of the infiltrating water by 1‰, as measured at the Soreq Cave site by Bar Matthews et al. (Reference Bar-Matthews, Ayalon, Gilmour, Matthews and Hawkesworth2003).
Develle et al. (Reference Develle, Herreros, Vidal, Sursock and Gasse2010) and Cheng et al. (Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015) found contrasting patterns of climatic variability between records from the northern and southern Levant. Cheng et al. (Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015) emphasize an out-of-phase pattern between records from the northern Levant relative to the Dead Sea Basin (DSB). The contrasting pattern of precipitation is likely due to enhanced warm southerly–southwesterly flow, which intensifies winter–spring precipitation over the northern Levant (Brayshaw et al., Reference Brayshaw, Rambeau and Smith2011). Furthermore, the contrasting north–south precipitation pattern persists on millennial to centennial timescales, linked potentially to an enhanced (weakened) meridional circulation, which results in a wet (dry) northern (southern) Levant (Xoplaki et al., Reference Xoplaki, González-Rouco, Luterbacher and Wanner2004), expressed by a contrasting RCC between Jeita isotopic variations and DSB levels (Cheng et al., Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015).
Cheng et al. (Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015) emphasized the relevance of distinct effective infiltration (precipitation–evaporation [P-E]) effects, enhanced by local factors (topography, vegetation) between the northern and southern Levant, in parallel with changes in meridional circulation patterns. This interpretation, previously proposed by Develle et al. (Reference Develle, Herreros, Vidal, Sursock and Gasse2010), focuses on effective rainfall, related to P-E in both soil and epikarst that affects water supply to caves (infiltration) and lakes (drainage) and thus explains differences in water balances between sites. Indeed, Develle et al. (Reference Develle, Herreros, Vidal, Sursock and Gasse2010) stressed the different timing of RCC in the northern and southern Levant.
According to the Jeita Cave and Aammiq marsh records (Hajjar et al., Reference Hajjar, Haïdar-Boustani, Khater and Cheddadi2010), the major shift from humid to general dry conditions occurred around 6 ka, which is different than in the Dead Sea, the level of which was extremely low during the entire Holocene compared with the last glacial maximum (LGM). Migowski et al. (Reference Migowski, Stein, Prasad, Negendank and Agno2006) do not exclude the possibility that those humid conditions relative to modern times prevailed in the DSB during the Early Holocene, but the water balance contrast between the Yammouneh and DSB records relies on different P-E conditions in both regions. For example, the Yammouneh record (Develle et al., Reference Develle, Herreros, Vidal, Sursock and Gasse2010) does not reveal proxy evidence for a drought around 8 ka as observed in the Jeita record (Cheng et al., Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015) at lower altitude. More to the south, the 8.2 ka event was identified in the Soreq record (Bar-Matthews et al., Reference Bar-Matthews, Ayalon, Gilmour, Matthews and Hawkesworth2003), but was not interpreted as a drought.
Effective infiltration (P-E) also influences proxies related to local soil and epikarst conditions, such as δ13C in carbonates. Such proxies, conditioned by vegetation cover and soil microbial productivity, are additionally influenced by CO2 degassing and prior calcite precipitation (PCP) related to changing cave drip rates (Fohlmeister et al., Reference Fohlmeister, Voarintsoa, Lechleitner, Boyd, Brandtstätter, Jacobson and Oster2020). In the EM region, a contrasting pattern of hydroclimate variability between the northern and southern Levant is exemplified by comparison of the δ13C profile of Jeita with Peqiin and Soreq records. Although the Jeita and Peqiin records show similar δ13C and δ18O isotopic trends, a notable heavy excursion in the Soreq δ13C record from ~10 to ~7 ka presents an opposite trend relative to the Jeita δ13C curve. This positive Soreq δ13C event is interpreted to reflect an extremely wet period in the southern Levant, partially because of its association with lighter δ18O values (Bar-Matthews and Ayalon, Reference Bar-Matthews and Ayalon2011). Cheng et al. (Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015) challenge this interpretation and suggest that the Soreq record could reflect drier conditions during the Early Holocene, consistent with low Dead Sea levels. The recent publication of Burstyn et al. (Reference Burstyn, Shaar, Keinan, Ebert, Ayalon, Bar-Matthews and Feinberg2022) follows the interpretation of Bar-Matthews and Ayalon (Reference Bar-Matthews and Ayalon2011) and involves a complex opposite response system of magnetic particle influx to rainfall to demonstrate the coupling between the inflow of magnetic particles (IRMflux) and δ13C in Soreq speleothems. The contrasting climates between the northern and southern Levant are therefore still under debate, and factors in the local P-E response to the synoptic climate systems (e.g., topography) are still not well constrained, mainly due to different interpretations of the δ13C response to water availability.
CURRENT CLIMATIC SETTINGS OF THE LEVANT
Today, the Levant region is mainly influenced by the midlatitude westerlies, which originate from the Atlantic Ocean, forming a series of subsynoptic low-pressure systems across the Mediterranean Sea (Gat et al., Reference Gat, Klein, Kushnir, Roether, Wernli, Yam and Shemesh2003; Ziv et al., Reference Ziv, Saaroni, Romem, Heifetz, Harnik and Baharad2010). In winter, cold air plunging south over the relatively warm Mediterranean enhances cyclogenesis, creating the Cyprus Low (Alpert et al., Reference Alpert, Price, Krichak, Ziv, Saaroni, Osetinsky and Kishcha2005). Moist air is then driven onshore, generating intense orographic rainfall across Mt. Lebanon, Mt. Hermon, and the Syrian mountains in the northern Levant. In summer, the westerly belt is shifted to the north, following the northern shift of the North African subtropical high pressures, and the region experiences hot and dry conditions with more southerly winds. In Lebanon, the annual rainfall varies between 700 and 1000 mm along the coastline and more than 1400 mm in the mountains. Average snow coverage is 5 months from December to April (Shabaan and Houhou, Reference Shabaan and Houhou2015) in basins located at mid-altitudes (1200–2000 m) and up to 7 months from December to June (Fayad and Gascoin, Reference Fayad and Gascoin2020) in high-altitude basins (>2500 m). As a consequence of this circulation system, the climate is seasonal with wet winters (November to February) and dry, hot summers (May to October), with a significant influence of snow coverage (up to 75%) on the water supply budget in karst networks and springs at high altitudes (Koeninger et al., Reference Koeniger, Margane, Abi-Rizk and Himmelsbach2017).
A general west–east gradient in rainfall (amount, isotopic composition) from the Lebanese coastline to the inner Beqaa plain is evident in the published local meteoric water line (Aouad-Rizk et al., Reference Aouad-Rizk, Job, Khalil, Touma, Bitar, Bocquillon and Najem2005) as a consequence of the altitudinal effects related to Rayleigh distillation processes (Dansgaard, Reference Dansgaard1964; Rozanski et al., Reference Rozanski, Araguas, Gonfiantini, Swart, Lohmann, McKenzie and Savin1993). This gradient is mirrored in the cave stream and drip-water isotopic composition (Nehme et al., Reference Nehme, Verheyden, Nader, Adjizian-Gerard, Genty, De Bont and Clayes2019), with a clear altitudinal gradient from the coastline to the highest Mt. Lebanon peaks. The north–south gradient at Mt. Lebanon and to the south is also expressed in the isotopic composition and amount of rainfall. The studies of Aouad-Rizk et al. (Reference Aouad-Rizk, Job, Khalil, Touma, Bitar, Bocquillon and Najem2005), Gat et al. (Reference Gat, Klein, Kushnir, Roether, Wernli, Yam and Shemesh2003), and Saad et al. (Reference Saad, Kazpard, El Samrani and Slim2005) showed more positive isotopic rainfall values and a lower rainfall amount toward the south, impacting the water balance (P-E) and effective infiltration in karstic systems along the Levantine coast and mountain chains.
CAVE SITE AND SAMPLE DESCRIPTION
Qadisha Cave (34°14′38″N, 36°02′11″E) is located 1720 m asl in the northern part of Mt. Lebanon (Fig. 1), in the vicinity of the highest peak of the mountain chain (Mt. Makmel), which reaches 3088 m asl. The latter, with its northeast–southwest direction, faces the EM Basin. The cave develops in Quaternary deposits derived from dolomitized Cretaceous limestone, located in the vicinity of the Qadisha catchment basin (Dubertret, Reference Dubertret1975; Nader et al., Reference Nader, Abdel-Rahman and Haidar2006). The basin is fed by ample snowmelt and has a mean elevation of 2244 m. The duration of snow cover is variable across years and reaches 6 to 7 months in the Qadisha water catchment basin (Telesca et al., Reference Telesca, Shaban, Gascoin, Darwich, Drapeau, El Hage and Faour2014; Fayad and Gascoin, Reference Fayad and Gascoin2020). The cave is horizontal, with more than 1076 m of explored galleries, comprising an upper relict part and lower active part with a permanent spring (discharge rate up to 1 m3/s) (Edgell, Reference Edgell1997). Qadisha Cave was partially transformed into a tourist cave in 1934 (first show cave in Lebanon). It is located near the cedar forest of Bsharreh, which is believed to have covered a wider area in the past. Today, the Quaternary deposits above the cave, consisting mainly of scree and cemented rock debris, are covered with shrub vegetation (Dubertret, Reference Dubertret1975). A previous monitoring study in the cave (Nehme et al., Reference Nehme, Verheyden, Nader, Adjizian-Gerard, Genty, De Bont and Clayes2019) showed a mean cave air temperature of 9.0 ± 0.5°C and a pCO2 concentration of 600 ppmv. Water percolation through the cave persists generally throughout the year. Two speleothems, Qad-1 and Qad-2, were retrieved from the upper gallery of the cave, which hosts many active stalagmites.
METHODS
230Th dating
The chronologies of Qad-1 and Qad-2 were established using 10 and 9 230Th dates, respectively (Table 1). Exploratory 230Th dating was performed first at Xi'an Jiaotong University (China) in 2011 and 2014, and the rest was completed at the University of Minnesota (USA) in 2018, by using Thermo-Finnigan Neptune/Neptune Plus multi-collector inductively coupled plasma mass spectrometers. The methods in both laboratories were identical (Cheng et al., Reference Cheng, Edwards, Shen, Polyak, Asmerom, Woodhead and Hellstrom2013). Standard chemistry procedures (Edwards et al., Reference Edwards, Chen, Ku and Wasserburg1987) were used to separate uranium and thorium. A triple-spike (229Th-233U-236U) isotope dilution method was used to correct instrumental fractionation and to determine U/Th isotopic ratios and concentrations (Cheng et al., Reference Cheng, Edwards, Shen, Polyak, Asmerom, Woodhead and Hellstrom2013). U and Th isotopes were measured on a MassCom multiplier behind the retarding potential quadrupole in the peak-jumping mode using standard procedures (Cheng et al., Reference Cheng, Edwards, Shen, Polyak, Asmerom, Woodhead and Hellstrom2013). Uncertainties in U/Th isotopic measurements were calculated offline at 2σ, including corrections for blanks, multiplier dark noise, abundance sensitivity, and contents of the same nuclides in spike solution. Corrected 230Th ages assume an initial 230Th/232Th atomic ratio of 4.4 ± 2.2 × 10−6, the values for a material at secular equilibrium with a bulk earth 232Th/238U value of 3.8 (Cheng et al., Reference Cheng, Edwards, Shen, Polyak, Asmerom, Woodhead and Hellstrom2013).
230Th dating results of stalagmites Qad-stm1 and Qad-stm2 (error is given as 2 SE).a
a U decay constants: l238 = 1.55125 × 10−10 (Jaffey et al., Reference Jaffey, Flynn, Glendenin, Bentley and Essling1971) and l234 = 2.82206 × 10−6 (Cheng et al., Reference Cheng, Edwards, Shen, Polyak, Asmerom, Woodhead and Hellstrom2013). Th decay constant: l230 = 9.1705 × 10−6 (Cheng et al., Reference Cheng, Edwards, Shen, Polyak, Asmerom, Woodhead and Hellstrom2013).
b d234U = ([234U/238U] activity − 1) × 1000.
c d234U initial was calculated based on 230Th age (T), i.e., d234Uinitial = d234Umeasured × eλ234 × T.
d Corrected 230Th ages assume the initial 230Th/232Th atomic ratio of 4.4 ± 2.2 × 10−6. Those are the values for a material at secular equilibrium, with the bulk earth 232Th/238U value of 3.8.
e The errors are arbitrarily assumed to be 50%. Ages are defined as the year before 1950 CE.
fAge considered as outlier.
Calcite and water stable isotopes measurements
Samples for stable isotopic analyses were taken along the growth axes of Qad-1 and Qad-2 stalagmites (Fig. 2) for δ13C and δ18O measurements. Overall, 230 samples were measured in both stalagmites. Samples were drilled along the stalagmite growth axis at 2.5 mm resolution using a Merchantek Micromill mounted on a Leica microscope, with a 0.3 mm resolution for specific parts. Between every sample, the drill bit and sampling surface were cleaned with compressed air. The samples were analyzed using a Nu Carb carbonate device coupled to a Nu Perspective mass spectrometer (MS) at Vrije Universiteit Brussel (Belgium). Parts of the samples were measured at the Laboratory for Environmental and Raw Material Analysis at the Karlsruhe Institute of Technology (Germany), using a Thermo Gasbench II connected to a DELTA V IRMS in continuous-flow mode. All δ18O and δ13C values are calibrated against Vienna Pee Dee Belemnite (VPDB) and are reported in per mil (‰). Analytical uncertainties were better than 0.1‰ (1σ) for oxygen and 0.05‰ (1σ) for carbon on both instruments. Percolation and stream waters as well as recent calcite samples underneath active drip water were collected previously from Qadisha Cave in 2011 and 2014 (Nehme et al., Reference Nehme, Verheyden, Nader, Adjizian-Gerard, Genty, De Bont and Clayes2019).
Age model of both Qadisha-1 (Qad-1) and Qadisha-2 (Qad-2) stalagmites using StalAge (Scholz and Hoffmann, Reference Scholz and Hoffmann2011). Sampling track for stable isotopes, trace elements, and locations of fluid inclusions (squares) along the growth axis are shown on each stalagmite. Red rectangle in the upper part of the Qad-2 stalagmite shows a high-resolution image of discontinuities D3 and D4 along the growth axis. F1, fluid inclusions; SI, stable isotope; TE, trace elements.
Trace element data
Elemental abundances were determined by laser ablation-inductive coupling plasma–MS (LA-ICP-MS), at the Institute for Geosciences, Johannes Gutenberg University Mainz (Germany), using an ESI NWR193 ArF excimer LA system equipped with the TwoVol2 ablation cell, operating at 193 nm wavelength, coupled to an Agilent 7500ce quadrupole ICP-MS. Ablation was performed in line scan mode and surfaces were pre-ablated before each line scan to prevent potential surface contamination. Line scans were performed at a scan speed of 10 μm/s, using a spot size of 110 μm and a laser repetition rate of 10 Hz. Laser energy on the samples was about 3 J/cm2. Measured ion intensities were monitored in time-resolved mode, and background intensities were measured for 15 s. Synthetic glass NIST SRM 612 was used to calibrate element concentrations, with the preferred values given in the GeoReM database being applied (Jochum et al., Reference Jochum, Pfänder, Woodhead, Willbold, Stoll, Herwig and Hofmann2005, Reference Jochum, Scholz, Stoll, Weis, Wilson, Yang and Andreae2012). Quality-control materials (QCMs) (USGS MACS-3 and USGS BCR-2G) were used to monitor the accuracy and precision of the LA-ICP-MS analysis and calibration strategy. Raw data were processed using TERMITE (Mischel et al., Reference Mischel, Scholz, Spötl, Jochum, Schröder-Ritzrau and Fiedler2017), an R script for data reduction. The internal standard was 43Ca, applied as an internal standard, at a Ca concentration of 390,000 μg/g. The values for trace element results are reported in the GeoReM database for the QCMs. Element concentrations determined for the QCMs had a precision of <0.02% (1σ).
Fluid inclusion stable isotope (H-O) analyses
Five calcite samples, from three levels in the Qad-2 stalagmite were measured using a custom-built extraction line connected to a Picarro L2130i analyzer using cavity ring down spectroscopy. This technique allows simultaneous measurement of hydrogen and oxygen isotopes for minute water amounts released from calcite. The extraction line follows the design of Affolter et al. (Reference Affolter, Fleitmann and Leuenberger2014) and is described in detail in Weissbach et al. (Reference Weissbach, Kluge, Affolter, Leuenberger, Vonhof, Riechelmann and Fohlmeister2023). In brief, the calcite samples were hydraulically crushed, and the released fluid inclusion water instantly vaporized in the heated extraction system and transferred to the analyzer. Reference water injections with known δ18O and δD values were used for calibration and quality control. Glass capillaries (microliter size) were used for high-precision water amount calibration and isotopic control. The precision of fluid inclusion water analyses is dependent on the released water amount and is 0.5‰ for δ18O and 1.2‰ for δ2H, if fluid water amounts are >0.2 μl, and reaches 0.1–0.3‰ for δ18O and 0.2–0.7‰ for δ2H for fluid water amounts >1 μl (Weissbach et al., Reference Weissbach, Kluge, Affolter, Leuenberger, Vonhof, Riechelmann and Fohlmeister2023).
RESULTS
Petrography and chronology
Over most sections, both samples of 11 cm for Qad-1 and 13 cm for Qad-2 display a translucent to whitish calcite with a columnar fabric along their growth axis. Both stalagmites are highly laminated. With the aid of a camera mounted on a Leica Microscope, images of laminae were taken at several depths, and the thickness of each was measured. Layer thickness of laminae varied between 60 and 280 μm. The laminar growth is irregular and alternates between thin (60–70 μm) to thick laminae (200–280 μm).
A total of 19 U/Th ages were obtained from stalagmites Qad-1 and Qad-2 (Table 1). The age distribution based on the StalAge model (Scholz and Hoffmann, Reference Scholz and Hoffmann2011) indicates that stalagmite Qad-1 grew from ca. 6.643 ± 0.038 to 3.247 ± 0.127 ka, including two discontinuities (Fig. 2). The first discontinuity, D1, ranges from 5.787 to 3.247 ka (extrapolated ages), and D2 covers a shorter period from 6.408 to 6.108 ka (extrapolated ages). Qad-2 grew from 9.145 ± 0.021 to 0.005 ± 0.007 ka, including two discontinuities: D3 (7.016 to 0.005 ka) and D4 (7.844 to 8.326 ka). The age model of Qad-1 was constructed with eight ages, with one other age (Qad-stm1-00) considered to be an outlier. The latter is sampled on the edge of the growth axis and is not in stratigraphic order with the other dating points. At the base of Qad-1, the growth rate is around 60 μm/yr, evolving to a higher rate around 150 μm/yr and up to 500 μm/yr in the upper part of the stalagmite. For the Qad-2 stalagmite, all the ages are in stratigraphic order and were used in the calculation of the age model. The growth rate at the base of Qad-2 is very high, reaching 650 μm/yr and is reduced to 60 and 20 μm/yr in the middle and upper part of the stalagmite.
Stable isotopic composition of calcite, modern water, and fluid inclusions
Calcite δ18O and δ13C values were analyzed at a multi-annual to decadal resolution. The δ18O values for the Qadisha record (Fig. 3) range from −5.7‰ to −7.5‰, with a mean of −6.8‰. The δ13C values range from −6.2‰ to −8.6‰, around a mean of −7.5‰. Both δ18O and δ13C values generally covary during the entire period from ~9.1 to ~5.7 ka but are distinct in the medium- and long-term trends. For example, the δ18O values show an overall trend toward more positive values from the Early to the Middle Holocene, unlike the δ13C values, which show a trend toward more negative values when reaching the mid-Holocene. Significant variations are noticeable in the stable isotope curves, with clear δ18O and δ13C excursions around ~9.1, ~8.9, ~7.7, ~7.4-7.2, ~6.5, and ~5.9 ka.
δ18O and δ13C profiles of both Qad-1 and Qad-2 stalagmites with the trace elements curves (Mg/Ca, Sr/Sa, P/Ca, U/Ca) and a moving average curve (in black) for some of the trace elements. Growth rate is displayed in logarithmic scale. Both stable isotope and trace element data are plotted against time (yr 1950 CE), modeled using StalAge. Black dots refer to the dating points and the gray shading to the identified discontinuities (D2, D4). Blue rectangles highlight significant variations toward wetter conditions in both stable isotopes and trace element data, and orange rectangles highlight drier conditions. Dashed lines crossing the δ18O and δ13C profiles indicate present-day values. Rapid dry and wet events are numbered from 1 to 10.
Stable isotopes of Qad-2 calcite dated at 0.005 ± 0.007 ka (calcite age between 1938 and 1952 CE) show average δ13C and δ18O values of −7.2‰ and −6.6‰, respectively. On a comparative basis, present cave and spring water sampled in 2011 and 2014 in Qadisha Cave (Nehme et al., Reference Nehme, Verheyden, Nader, Adjizian-Gerard, Genty, De Bont and Clayes2019) show average values of −8.6‰ for δ18Ow and −46.9‰ for δ2Hw.
Average fluid inclusion δ18Ow and δ2Hw values for Qad-2 are −6.5‰ and −41.8‰, respectively. Samples taken at approximately the same level B in the stalagmite (~9.0–9.1 ka) show a certain variability of 3.6‰ for δ2H and 2.6‰ for δ18O, between samples. The sample with the highest water yield shows the most negative δ18O value. At ~7.2–7.3 ka, δ18Ow is −7.0‰ and δ2Hw is −42.8‰ (Table 2). Level E, which spans the time period of the last century, yielded a very low amount of water and was therefore rejected.
Qadisha fluid inclusion samples in chronological order (old to young).a
a Stable isotope measurements are shown for 1σ error.
b Measurements with an asterisk (*) are considered outliers due to a low water amount of less than 0.5 μl/g.
c Apparent formation temperature was determined using the calcite–water fractionation factors after Kim and O'Neil (Reference Kim and O'Neil1997).
High-resolution trace element data
Main trace element ratios presented in Figure 3 are all above-background levels. The overall sampling resolution is annual to multi-annual. The average elementary ratios show a slight shift between Qad-1 and Qad-2, with a higher variability for Qad-1 in general. Mg/Ca ratios show a few significant shifts that are mirrored in most other investigated ratios (Fig. 3). Particularly strong trace elemental signals are visible around ~9.0–8.9 and ~7.7 ka, whereas weaker signals are noticeable at ~8.5, ~7.3 and ~7.1, ~6.6, and ~5.8 ka.
DISCUSSION
Age-growth behavior of the studied speleothems
Both stalagmites grew during the Holocene, roughly between 9.2 and 5.7 ka, one succeeding the other, suggesting favorable (humid and warm) conditions for stalagmite deposition at the high altitude of Qadisha Cave. Although growth is relatively constant over most parts of the stalagmites, several hiatuses are observed. Former studies on stalagmites from Mt. Lebanon (Verheyden et al., Reference Verheyden, Nader, Cheng, Edwards and Swennen2008; Nehme et al., Reference Nehme, Verheyden, Noble, Farrant, Sahy, Hellstrom, Delannoy and Claeys2015, Reference Nehme, Verheyden, Breitenbach, Gillikin, Verheyden, Cheng and Noble2018) suggest that cave calcite precipitation, and thus growth in the central Levant, is conditioned by effective infiltration (Cheng et al., Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015). This growth cessation may be site-specific but could also reflect unfavorable conditions for calcite deposition, such as dry conditions with low effective infiltration within the epikarst or scarcer vegetation with less bio-vegetation activity. Based on the number of currently analyzed speleothems, one cannot determine which option is correct.
Growth rates of both stalagmites are of the same order, reflecting similar drip rates and cave pCO2. Similarities in the petrography of both speleothems corroborate consistent growth conditions. High growth rates of 0.15–0.6 mm/yr between 6.1 and 5.8 ka and of 0.3–0.9 mm/yr between 9.15 and 9.0 ka allow for annual and higher-resolution climate reconstruction. Growth rates of 15–50 μm/yr for most other parts of the record allow at least for multi-annual resolution using a traditional micro-milling technique for stable isotope analysis. Two growth pauses at 8.3–7.8 ka and 6.4–6.1 ka are supported by two clear petrographic discontinuities. The growth pauses between 8.3 and 7.8 ka in the Qad-2 stalagmite, although supported petrographically by a thin dust layer, are not supported by changes in other proxies before or after the hiatus. If regional aridification characterized the 8.2 ka event, as inferred from the lowland Jeita Cave record (Cheng et al., Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015), then a combination of low infiltration and scarcer vegetation within the Qadisha karst basin would make speleothem growth highly unfavorable, plausibly leading to a centennial-scale hiatus. Such a hiatus is observable in Qad-2, but cannot be attributed confidently to climatic change.
Unlike the growth stop dated between 8.3 and 7.8 ka in Qad-2, the one dated between 6.5 and 6.1 ka in Qad-1 seems to be clearly framed by changes in stable isotope values and trace elements. More positive δ18O and δ13C values suggest drier conditions before and directly after the growth stop and may record unfavorable growth conditions, although prior calcite precipitation related to local factors at the drip rate may occur, and other speleothems need to be studied to confirm our hypothesis. This seems also in agreement with a change toward increased Mg/Ca values and decreased Sr/Ca values just before and after the hiatus.
Interpreting the changes in geochemical proxies from Qadisha Cave
Speleothem δ18O and δ13C
In the Levant, it is now widely established that the δ18O signal is interpreted as related to effective recharge in the epikarst (precipitation amount vs. evapotranspiration), and therefore indicates water balance (P-E) in the epikarst (Bar-Matthews et al., Reference Bar-Matthews, Ayalon, Gilmour, Matthews and Hawkesworth2003; Verheyden et al., Reference Verheyden, Nader, Cheng, Edwards and Swennen2008; Cheng et al., Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015; Nehme et al., Reference Nehme, Kluge, Verheyden, Nader, Charalambidou, Weissbach and Gucel2020). Higher infiltration (more positive P-E) is related to pronounced low-pressure systems with significant rainfall. Higher rainfall amounts are known to cause more negative rainfall δ18O values (Bar-Matthews et al., Reference Bar-Matthews, Ayalon, Gilmour, Matthews and Hawkesworth2003; Aouad-Rizk et al., Reference Aouad-Rizk, Job, Khalil, Touma, Bitar, Bocquillon and Najem2005), and reduced annual rainfall (with concomitant reduced P-E) produces higher rainfall δ18O values.
At mid-latitudes, δ13C and growth rate in speleothems are related to biological CO2 production, which is dependent on soil and vegetation conditions in the catchment (Genty et al., Reference Genty, Baker, Massault, Proctor, Gilmour, Pons-Branchu and Hamelin2001a, Reference Genty, Baker and Vokal2001b). Warmer and wetter periods usually enhance the production of biogenic, δ13C-depleted CO2 and increase the growth rate (Genty et al., Reference Genty, Baker, Massault, Proctor, Gilmour, Pons-Branchu and Hamelin2001a, Reference Genty, Baker and Vokal2001b). Cold/dry conditions reduce the vegetation cover and the biogenic CO2 supply. Lower/higher calcite δ13C values and faster/slower growth rates are therefore usually indicative of soil development/disruptions and can be related to warmer-wetter/colder-drier conditions. An additional control on δ13C is exerted by the hydrologic state of the aquifer. Partial dewatering of the drip-feeding system induces longer residence times and prior calcite precipitation, leading to higher δ13C during drier periods (Fairchild and Baker, Reference Fairchild and Baker2012).
The stable isotopic δ18O values of both Qadisha stalagmites display a general trend toward more positive δ18O values from 8.5 to 5 ka. This general δ18O trend comprises three independent segments that include several rapid shifts toward negative values at ~9.1 ka (10), ~8.9 ka (8), ~8.5 ka (7), ~7.7 ka (6), and 5.8 ka (2) (see Fig. 3 for numbers in parentheses). Such negative “peaks” are interpreted as wet intervals with high effective recharge (high growth intervals of the stalagmites) and a positive water balance (P-E) (Cheng et al., Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015). Rapid changes toward more positive δ18O at ~9.0 ka (9), ~7.2–7.1 ka (5), ~6.6 ka (4), ~5.9 ka (3), and ~5.7 ka (1), are particularly well expressed and chronologically constrained. Some of these periods interpreted here as “drying” events, particularly at ~9.0 (9), are characterized by unusually high growth rates. Such peculiar periods seem to indicate, at least locally, a wet period, as confirmed by negative (lower) δ13C values, before the short dry period of ~80 yr, but still with high growth rates. Albeit with low resolved growth rates compared with the resolution of geochemical proxies (annual for trace elements and decadal for the δ13C/δ18O), such peculiar drying events preceded by short wet events reflect a rapid decadal variability, inferring a shift in the distribution of effective infiltration.
Trace elements
In general, the (trace) element concentrations of Mg, Sr, P, and U in calcite are controlled by hydrologic processes and less by cave temperature (Roberts et al., Reference Roberts, Smart and Baker1998). Speleothem Sr/Ca ratios constitute a sensitive proxy for infiltration changes in the epikarst, with higher values being recorded during times of reduced effective moisture availability (Fairchild and Treble, Reference Fairchild and Treble2009). Other mechanisms can explain Sr mobility within the epikarst and include changes in weathering rates, aerosol input, and soil activity that can induce high-frequency variability on short timescales (decadal to annual) (Verheyden et al., Reference Verheyden, Keppens, Fairchild, McDermott and Weis2000; Fairchild et al., Reference Fairchild, Smith, Baker, Fuller, Spötl, Mattey and McDermott2006; Sinclair et al., Reference Sinclair, Banner, Taylor, Partin, Jenson, Mylroie and Miklavič2012; Baker et al., Reference Baker, Mariethoz, Comas-Bru, Hartmann, Frisia, Borsato and Asrat2021). For Mg/Ca ratios in speleothems, it is common to attribute an increase to a response to PCP during times periods of low effective infiltration (Verheyden et al., Reference Verheyden, Keppens, Fairchild, McDermott and Weis2000; Tooth and Fairchild, Reference Tooth and Fairchild2003; McDermott, Reference McDermott2004; Fairchild et al., Reference Fairchild, Smith, Baker, Fuller, Spötl, Mattey and McDermott2006). Because Sr/Ca ratios are similarly influenced by PCP, Mg and Sr ratios are often positively correlated.
An exception to this behavior occurs in the presence of dolomitized limestone in the karst aquifer, as is the case in the Qadisha basin, where processes of incongruent calcite dissolution and/or differentiated soil input may play a role (Roberts et al., Reference Roberts, Smart and Baker1998; Hellstrom and McCulloch, Reference Hellstrom and McCulloch2000; Huang et al., Reference Huang, Fairchild, Borsato, Frisia, Cassidy, McDermott and Hawkesworth2001; Sinclair et al., Reference Sinclair, Banner, Taylor, Partin, Jenson, Mylroie and Miklavič2012, Jamieson et al., Reference Jamieson, Baldini, Brett, Taylor, Ridley, Ottley and Prufer2016).
Phosphorous and U are less commonly studied but increasingly used as proxies in the speleothem archive. This element is generally mobilized from soils during early autumnal/winter storms. High P/Ca in speleothems reflects maximum soil infiltration during heavy rainfalls at the start of wet seasons (Borsato et al., Reference Borsato, Frisia, Fairchild, Somogyi and Susini2007). The P/Ca ratio can be complemented by other paleo-hydrologic proxies such as Ba/Ca or U/Ca. Uranium is generally leached from the bedrock during high-rainfall events and results in high U/Ca in the calcite. (Ayalon et al., Reference Ayalon, Bar-Matthews and Kaufman1999; Treble et al., Reference Treble, Shelley and Chappell2003; Fairchild and Treble, Reference Fairchild and Treble2009)
In the Qadisha speleothems, the trace element concentrations vary at maximum by two orders of magnitude. Wet and dry episodes identified by stable isotopes and growth rate changes (section 7.2.a, Fig. 3) can overall be linked to changes in the trace element concentrations, with generally high Mg/Ca, and low Sr/Ca, P/Ca, and U/Ca corresponding to higher stable isotope values, suggesting an overall common driver, although with different sensitivities between proxies. Therefore, in the Qadisha speleothems, we assign the periods with high δ13C/δ18O and Mg (and generally low Sr, P, and U) to dry periods, with a lower water recharge in the epikarst, and the opposite changes to wet periods. At 9.0 ka and 7.7 ka, significant changes in δ13C and δ18O values, as well as in U and P concentrations, may suggest the occurrence of important flushing episodes.
Infiltration water isotopes based on fluid inclusion analysis
Isotope measurements of fluid inclusion water (δD; δ18Ow) can provide insights into the hydrologic cycle and allow estimation of mineral formation temperatures. Three successful fluid inclusion analyses from the Qad-2 stalagmite with water amounts >0.5 μl/g were retained here (Table 2). The δ18O and δD values of water sampled in pools, cave streams, and drip water inside Qadisha Cave vary between −8.4 and −9.0 and between −45.7‰ and −49.5‰, respectively. The d-excess of the cave water is between 21.4‰ and 22.6‰ (Nehme et al., Reference Nehme, Verheyden, Nader, Adjizian-Gerard, Genty, De Bont and Clayes2019). As the cave water represents an average of the annual precipitation, rainfall values in contrast show a higher variability from −12.6‰ to +1.2‰ and −86‰ to +11.6‰ for δ18O and δD values, respectively (Aouad-Rizk et al., Reference Aouad-Rizk, Job, Khalil, Touma, Bitar, Bocquillon and Najem2005). The d-excess values are between 2‰ and 27‰, with an average close to that of the cave water (Aouad-Rizk et al., Reference Aouad-Rizk, Job, Khalil, Touma, Bitar, Bocquillon and Najem2005), indicating a generally higher d-excess for Lebanon compared with the global meteoric water line. Although fluid inclusions (FI) values fall within the modern-day range of rain water, compared with modern-day drip water from Qadisha, they show important variability. Especially, the samples from the B-level (9.0–9.1 ka) raise questions about the lateral coherency of the isotopic composition of FI water. The variability between different levels (B vs. E) may be related to differences in isotopic composition of the rain and therefore the fluid inclusions. Overall, the results are more positive than today's drip water isotopic composition. Following our interpretation, these values dated within the intervals of 7.3–7.2 ka and 9.1–9.0 ka represent isotopic values of water precipitated close to dry events (5) and (9) identified based on their more positive calcite δ13C and δ18O values (Fig. 4) and may suggest short-term drier conditions in the generally wetter Early to Middle Holocene. A higher fraction of summer rainfall contribution to the water budget with a higher isotope value compared with winter snow with its lower isotopic composition (Aouad-Rizk et al., Reference Aouad-Rizk, Job, Khalil, Touma, Bitar, Bocquillon and Najem2005) may be another possible explanation for this isotopic observation. In Soreq Cave, Early to Middle Holocene fluid inclusion isotope values are comparable or slightly more negative compared with modern-day values (Matthews et al., Reference Matthews, Affek, Ayalon, Vonhof and Bar-Matthews2021). This contrasting effect could be related to Sapropel S1, which has a significant effect on the isotope values in the southern Levant (Bar-Matthews et al., Reference Bar-Matthews, Ayalon, Gilmour, Matthews and Hawkesworth2003) but is not particularly expressed in the northern Levant (e.g., the Jeita record; Cheng et al., Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015), where rainfall amounts are generally higher. In addition, snowfall is of minor importance for the infiltration budget at Soreq Cave, whereas it is today the dominant water source in the region around Qadisha Cave.
Graph showing the fluid inclusion stable isotopes of the Qadisha stalagmite (this study) in comparison with fluid inclusions (FI) from the Mizpe Shelagim (Ayalon et al., Reference Ayalon, Bar-Matthews, Frumkin and Matthews2013) and Soreq records (Matthews et al., Reference Matthews, Affek, Ayalon, Vonhof and Bar-Matthews2021), all plotted against the global (GMWL; Rozanski et al., Reference Rozanski, Araguas, Gonfiantini, Swart, Lohmann, McKenzie and Savin1993) and Mediterranean meteoric water lines (MMWL; Gat et al., Reference Gat, Klein, Kushnir, Roether, Wernli, Yam and Shemesh2003). Modern rain water in Lebanon (Aouad-Rizk et al., Reference Aouad-Rizk, Job, Khalil, Touma, Bitar, Bocquillon and Najem2005; Saad et al., Reference Saad, Kazpard, El Samrani and Slim2005) is plotted in blue circles; modern cave water of Qadisha Cave is plotted in hatched circles.
The fluid inclusion isotope values of various Early to Middle Holocene speleothems along the Levant become more negative from south (Soreq Cave) to north (Qadisha Cave), in agreement with a dominating altitudinal effect and, in addition, an increase in average rainfall toward the Lebanon mountains (Nehme et al., Reference Nehme, Verheyden, Nader, Adjizian-Gerard, Genty, De Bont and Clayes2019). The difference in drip-water δ18O values between Soreq and Qadisha Caves is ~4‰, which is close to the expected difference from altitude alone. The elevation difference is about 1400 m, which explains an isotopic shift of 2.8–4.2‰ at a lapse rate of 0.2–0.3‰/100 m (Clark and Fritz, Reference Clark and Fritz1997). The difference between measured fluid inclusion isotope values of Holocene samples from the Soreq and Qadisha Caves is similar. Although hardly resolvable with the current set of fluid inclusion data, rainfall amount could also play a role. For example, Mizpe Shelagim is situated at almost the same elevation as Qadisha Cave, but about 120 km to the south, and shows a more positive δ2H value than Qad-2.
The climatic conditions during the Holocene optimum
Global and regional drivers of climatic variability during the Holocene optimum
The Qadisha stalagmites grew from 9.2 to 5.7 ka, albeit with a discontinuous trend, and show a general tendency from more negative δ18O values during the Early Holocene toward more positive δ18O values at the end of the Holocene optimum. This trend reflects a change from a wetter (more positive P-E) toward a drier climate (reduced P-E) in this high-altitude area. This trend is also noticeable at low altitude; for example, in the Jeita δ18O record and in other speleothem records from the southern Levant (e.g., Mizpe Shelagim and Soreq Caves) (Fig. 5A). The carbon signal in Qadisha indicates values reaching −8‰ during wet peaks and shifts to −6‰ during dry events (Fig. 5B), reflecting higher soil and vegetation activity during humid periods. The δ13C curve overprints the short-term changes with a slow long-term increasing trend (Fig. 5B) toward more negative values when reaching the mid-Holocene, suggesting overall improved bio-pedological soil activity in the Makmel mountains from 9 to 6 ka.
Plots of the Qadisha stable isotope curve in comparison with other speleothem records in the Levant. (A) Carbon isotope curves; (B) oxygen isotope curves. Black dots represent U-Th dating points in the Qadisha record.
The two investigated Qadisha stalagmites started to grow during the period of maximum summer insolation (Berger and Loutre, Reference Berger and Loutre1991) and a high sea-surface temperature (SST) in the EM, reaching ~18°C during the Holocene optimum (Fig. 6) (Emeis et al., Reference Emeis, Schulz, Struck, Rossignol-Strick, Erlenkeuser, Howell and Kroon2003; Scrivner et al., Reference Scrivner, Vance and Rohling2004). From 10 to 6.5 ka, the deposition of Sapropel 1 in the EM occurred as a result of increased discharge of the river Nile, leading to isotopically lighter sea-surface water. The contribution of the Mediterranean source effect as the only factor in the δ18O-depleted signals in terrestrial records has been debated in the last decades. Recent studies on Yammouneh lake (Develle et al., Reference Develle, Herreros, Vidal, Sursock and Gasse2010) and the Soreq (Grant et al., Reference Grant, Grimm, Mikolajewicz, Marino, Ziegler and Rohling2016) and Zalmon Caves (Keinan et al., Reference Keinan, Bar-Matthews, Ayalon, Zilberman, Agnon and Frumkin2019) show a contribution of rainfall amount in changes of the δ18O signal throughout glacial–interglacial cycles and even during the sapropel events, by extracting the source signal (the δ18Og.ruber) from the δ18Ocalcite of speleothems or δ18Oostrac. signal of lake ostracods. An increased precipitation over Mt. Lebanon during the Sapropel 1 event is noticeable with depleted δ18Olake (corrected from the source) signal reaching −11‰ in Yammouneh lake, from 9.2 to 8.3 ka. Likewise, the uncorrected δ18Oostrac. signal of the lowland Sağlik II peat record (Sekeryapan et al., Reference Şekeryapan, Streuman, van der Plicht, Woldring, van der Veen and Boomer2020) in the northern part of the Levant, albeit a low-resolved curve, reach the most depleted ostracod δ18O values between 9 and 7.7 ka. The well-dated lowland Jeita Cave record in central Lebanon shows the most-depleted values for the whole record around 8.5 ka, followed by a sharp δ18O drop (less negative values) from 8.5 to 8 ka and then a general decrease with less negative δ18O values until 5 ka.
Comparison of the Qadisha record with regional records in the eastern Mediterranean (EM) region. From top to bottom: Sağlik II δ18Oostracods, Turkey (Sekeryapan et al., Reference Şekeryapan, Streuman, van der Plicht, Woldring, van der Veen and Boomer2020), Qadisha δ18Oc record, Lebanon (this study) Jeita δ18Oc record, Lebanon (Cheng et al., Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015), the Summer Insolation curve (Berger and Loutre, Reference Berger and Loutre1991), Quercus pollen taxa percentage from El-Jurd core (Cheddadi and Khater, Reference Cheddadi and Khater2016), δ18OL Yammouneh record (Develle et al. Reference Develle, Herreros, Vidal, Sursock and Gasse2010), EM sea-surface temperature (SST), ODP 967 (Emeis et al., Reference Emeis, Schulz, Struck, Rossignol-Strick, Erlenkeuser, Howell and Kroon2003; Scrivner et al., Reference Scrivner, Vance and Rohling2004), Soreq δ18Oc record (Bar-Matthews et al., Reference Bar-Matthews, Ayalon, Gilmour, Matthews and Hawkesworth2003)
More locally, in the Makmel Mountains, the Quercus pollen taxa in the Al-Jurd peat core (Cheddadi and Khater, Reference Cheddadi and Khater2016), located 23 km north of Qadisha Cave, show a relatively high pollen percentage from 10 to 8 ka, followed by a reduced Quercus pollen percentage from 8 to 6.5 ka, and marked later by an increase of Quercus pollen percentage at 6.5, 5.5, 4.5, and 3 ka (Fig. 6). Albeit a low-resolved curve, the pollen record of the Al-Jurd peat core seems to show a higher percentage of Quercus pollen during the mid-Holocene (6.5 to 3 ka) rather than in the Early Holocene period from 9 to 6 ka, as indicated by several speleothems. The Quercus forest extension after 6.5 ka is rather puzzling in the local paleoclimate scheme (speleothems) of the Makmel mountains, but could suggest wetter/warmer conditions with improved bio-pedogenic activity in soils coeval with a reduced snow season and/or decreased snow cover enabling the extension of the forest. During the Early Holocene, slightly colder/drier conditions (than during the mid-Holocene) might have prevailed, with the persistence of a longer snow cover period impacting the length of the plant growing season but enhancing speleothem growth with an important recharge from the during summer. Soil erosion is enhanced during short melt seasons on the long-term trend in the Makmel Mountains and would explain the reduced Quercus forest in the area. In a similar alpine context such as in the Rio Martino Cave, Piedmont Alps, Italy (Regattieri et al., Reference Regattieri, Zanchetta, Isola, Zanella, Drysdale, Hellstrom and Zerboni2019), it was suggested that an increase in the persistence of snow cover may also impact the δ13C by reducing the length of the plant growing season, leading to reduced biogenic CO2 supply and therefore to a less negative δ13C signal in speleothems. Above the Qadisha site, the cedar forest stretching over same altitude as the El-Jurd site may have had similar reduction/extension dynamics during the Early/Middle Holocene, thus leading in general from δ13C signal (avg: −7‰) in Qadisha speleothems in the Early Holocene to a slightly more negative δ13C signal (avg: −8‰) in the mid-Holocene.
Centennial-scale climate variability
Rapid climatic changes at the centennial scale are recorded in some of the highly resolved stalagmites along the Levantine coast. Cheng et al. (Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015) detected nine cold/dry events in the Jeita record during the Holocene coeval with those in the North Atlantic sediments (Bond et al., Reference Bond, Kromer, Beer, Muscheler, Evans, Showers, Hoffmann, Lotti-Bond, Hajdas and Bonani2001). The Bond events in the Jeita record are interpreted as dry events inferred from both δ18O and δ13C variations (Cheng et al., Reference Cheng, Sinha, Verheyden, Nader, Li, Zhang and Yin2015) (Figs. 5 and 6). Some of the events, such as the ~5.1 ka event, noticeable in the Jeita record, is also recorded as a dry event in Anatolian lake records (Roberts et al., Reference Roberts, Eastwood, Kuzucuoğlu, Fiorentino and Caracuta2011). The 8.2 ka event recorded in the Jeita stalagmite corresponds to a δ18O and δ13C excursion (Fig. 4) in Soreq Cave (Bar-Matthews et al., Reference Bar-Matthews, Ayalon, Gilmour, Matthews and Hawkesworth2003). Such an event is, however, not noticeable in other speleothem records along the Levantine coast or in Yammouneh lake (Develle et al., Reference Develle, Herreros, Vidal, Sursock and Gasse2010). In Qadisha Cave, the 8.2 ka cold event is not expressed or may be related to the growth stop.
Other events, at ~8.9–9.0 ka, ~7.1 ka, and ~5.9 ka (Fig. 4) in the Jeita record are also visible in the Qadisha record, with more positive stable isotope values (dry events) but with less-clear changes in the related trace element pattern. In contrast, RCC interpreted as wet events in Qadisha Cave, such as the most prominent 7.7 ka wet period, are not noticeable in the Jeita record but correspond to the major peak in the Soreq record (Fig. 5A). The inconsistency in recording rapid climate change events between Levantine records may be due to different proxy resolution (e.g., smoothing in case of too low resolution) or differences in the local archive sensitivity to RCC.
Altitudinal and snow cover effect on the significance of climate signals
In the Levant, most of the published speleothem records are located at low altitudes facing the Mediterranean Sea, except for the Incesu and Mizpe Shelagim records located at 1615 m and 2180 m, respectively. Only at Mizpe Shelagim a plausible link between the effect of snow cover at high altitudes and speleothem growth was established on glacial–interglacial timescales. During the Holocene interglaciation, the snow cover in mountain areas participated actively in the epikarst water budget, fed drip sites, and favored speleothem growth in caves located at high altitudes, as well as affecting the vegetation dynamics. Whereas temperature remains close to 0°C during winter over large areas of Mt. Lebanon (Fayad and Gascoin, Reference Fayad and Gascoin2020), the soil beneath the snow does not freeze under present winter conditions, except during exceptional cold events. Comparison of the Qadisha records with those located at low altitudes helps in deciphering altitudinal trend and local particularities.
The Qadisha stalagmites show more negative δ18O values (Fig. 5A) than those recorded by the Jeita stalagmite related to the altitudinal effect expressed in the rain and cave water isotopic signals (Nehme et al., Reference Nehme, Verheyden, Nader, Adjizian-Gerard, Genty, De Bont and Clayes2019). Also, the contribution of snow to the infiltration and the percolation water in caves (Aouad-Rizk et al., Reference Aouad-Rizk, Job, Khalil, Touma, Bitar, Bocquillon and Najem2005) may lead to more negative drip-water and calcite δ18O values. It is noteworthy that the Qadisha stalagmites display a larger amplitude in the δ18O changes and therefore have a more dynamic response to changes than the Jeita record. This behavior may be related to the higher altitude and the enhanced convective rainout at mountain ranges as well as to the generally lower δ18O values of snow. An increase in low-pressure systems can disproportionally increase the rainfall/snow amount in high-altitude mountain regions. Qadisha records can therefore better, or at least more clearly, indicate RCC than the low-elevation Jeita speleothems. In contrast, short-term dry conditions may be better recorded by the lower-elevation archive sites, as they are more sensitive to droughts and PCP, which enhances the isotope and element signal. An example is the drought event recorded in Soreq, Mizpe Shelagim, and Zalmon Caves around 6.5 ka, which is well expressed in these caves, but less well expressed in Qadisha Cave.
More to the south, the Zalmon and Soreq records indicate changes in the δ13C values of more than 4‰ throughout the Holocene (Fig. 5B). This drastic shift in the δ13C signal of Zalmon Cave is mirrored by the δ13C signals in other records, with a trend toward less negative δ13C values by the end of the Holocene optimum. Although marked by a high δ13C shift, the vegetation in northern Galilee indicates a reduced bio-pedological activity after 8.5 ka, whereas the soil activity seems to be less variable throughout the Holocene optimum at the Jeita coastal site until 6 ka. In the case of Qadisha Cave, the carbon isotope signal reflects a higher sensitivity of the vegetation to humidity changes with step changes (< 2‰), similar to Mizpe Shelagim (Mt. Hermon). In Soreq Cave located to the south, the shift to less negative carbon values that are closer to the bedrock signal throughout the Holocene optimum, is interpreted as reflecting a soil denudation by intensive storms. The vegetation cover appears to be more stable at Jeita and changes drastically only during larger climatic shifts (e.g., 6 ka), whereas the soil system at high-altitude sites (Qadisha and Mizpe Shelagim) is more sensitive to rapid changes in the water budget and the snow cover dynamics and indicates short-term erosional phases between 7 and 6 ka.
CONCLUSIONS
New stalagmites from Qadisha Cave (Lebanon) located in the high Mt. Makmel, north of Mt. Lebanon, provide a well-dated record for the northern Levant with high-resolution geochemical proxies (δ18O and δ13C values, trace element concentrations). The Qadisha stalagmites grew from 9.2 to 5.7 ka, albeit with discontinuities, and show a general tendency from more negative δ18O values during the Early Holocene toward more positive δ18O values at the end of the Holocene optimum. This trend is in agreement with a change from a wetter to a drier climate in this high-altitude area. The δ13C values show rapid shifts along the record and a decreasing trend toward more negative values when reaching the mid-Holocene. The slight trend toward more negative values suggests overall improved bio-pedological soil activity in the Makmel Mountains during the mid-Holocene. Discrepancies between pollen data showing a wetter period at 7–6 ka than at 7–9 ka, unlike speleothems showing an overall wet Early Holocene, are interpreted as related to a reduced snow cover season and thickness since 6.5 ka, explaining the extension of Quercus at 6.5 ka. On the short-term climate trend, Qadisha stalagmites record rapid dry and wet changes on a centennial scale from 9 to 5 ka, with a tendency for drier conditions toward the mid-Holocene.
The Qadisha record is in good agreement with other Levantine records, such as the one from Jeita Cave, showing overall more humid climatic conditions from 9 to 7 ka in the region. After 7 ka, a drier climate seems to affect sites in both low- and high-altitude areas. The Qadisha record reflects particularities of a mountainous climate compared with other records: (1) more negative stable isotope values than the Jeita record due to the altitudinal rainout effect; (2) a larger amplitude in the oxygen isotopes, reflecting a more dynamic response to RCC than records located at lower altitudes; and (3) the effect of snow cover and duration regulating the effective infiltration in the Makmel area.
Acknowledgments
This study and field mission was funded by the 2014 mobility fellowship program of the Belgian Federal Scientific Policy (BELSPO). We acknowledge the assistance of Saint-Joseph University of Beirut for facilitating the access to caves with the help of ALES (Association Libanaise d'Etudes Spéléologiques) and SCL (Spéléo-Club du Liban) and the support of members of both caving clubs who collected water and calcite samples during field campaigns. Fluid inclusion measurements were supported by grant DFG KL2391/2-1 and uranium/thorium dating was funded by NSF grant 2202913 to RLE.
