Introduction

The monsoon variability in the Indian subcontinent is regulated by coupled heating-cooling between the Himalaya and southern Indian Ocean (Webster 1987; Webster et al. 1998; Chauhan 2003; Rehfeld et al. 2012; Gupta et al. 2013). The climate in Indian Himalaya is characterized by three monsoon patterns (Fig. 1), e.g. (i) Indian Summer Monsoon (ISM) or Southwest (SW) monsoon which brings rain over Indian subcontinent during June to September; (ii) Western Disturbances (WDs) which bring winter rains (from October to May) and play significant role over the northwestern India (Kotlia et al. 2012, 2015, 2017; Liang et al. 2015); and (iii) northeastern (NE) monsoon winds carry winter rains to southern part of the country during October to December; thus, there are diverse precipitation regimes from east to west. From the beginning of the Holocene (~ 11.7 ka BP onwards), the Himalaya underwent several wet and dry periods (Phadtare 2000; Kotlia et al. 2010). Earlier observations provoked that the monsoon intensity was weakened during the cold/dry events, i.e. Last Glacial Maximum (LGM), Older Dryas (OD), Younger Dryas (YD), 8.2 ka BP, 4.2 ka BP and Little Ice Age (LIA) (Sinha et al. 2005; Kotlia et al. 2010, Gupta et al. 2013). The homogeneous Indian monthly rainfall data during 1970–2000 shows several pulses of strengthening and failure of ISM, causing droughts and floods (Pant 2003). Variations in monsoon intensity and droughts have widespread socio-economic impacts over Indian subcontinent (Joseph et al. 2009; Kotlia et al. 2015). On the other hand, the Westerlies/Western Disturbances (WDs), originating from Mediterranean/North Atlantic Ocean, are also the important sources of rainfall for northwest India. The WDs have mechanical effect on atmospheric circulation, particularly during the winters when the winter jet streams carry moisture and blow from west to east and split into two parts, north and south of the Tibetan Plateau (Liang et al. 2015; Fig. 1), and prevent the southward flow of cold continental air towards the Indian subcontinent (Benn and Owen 1998).

Fig. 1
figure 1

Location of Tityana cave (TC1, present study) with trend of different monsoon system (ISM/EASM) and speculation of Winter Jet stream (WDs), split into two parts north and south of the Tibetan Plateau. Location of the caves of different precipitation regimes, e.g. Dongee cave (Dykoski et al. 2005); Sainji cave (Kotlia et al. 2015); Qunf cave (Fleitmann et al. 2007); Soreq cave (Bar-Matthews et al. 2003) to compare with TC1 paleo-precipitation record

Full size image

The high-resolution short-term climatic events from the Central Himalayan domain between mid and late Holocene are either less understood or scattered. Prior to instrumental records, one needs to understand the past monsoon variability using high-resolution archives although a few palaeoclimatic records using continental archives (palaeolakes, peat bogs, modern lake sediments, etc.) from the Indian Himalaya are available (e.g. Kotlia et al. 1997, 2010; Phadtare 2000; Rühland et al. 2006; Kotlia and Joshi 2013; Demske et al. 2016), but the multi-centennial scale resolution and coarser sampling interval with age uncertainty offer only limited palaeoclimate information.

Stalagmites are one of the best continental archives which involve the deposition of calcium carbonate over long periods of time with decadal to century scale resolution and absolutely dated time series of the palaeoenvironmental variables (Fleitmann et al. 2007; Kotlia et al. 2012, 2015; Laskar et al. 2013). δ18O variations in the speleothems can reflect the changes in isotopic composition of cave water which is directly related to the rainfall amount (Bar-Matthews et al. 1999) and is known as “amount effect” (Dansgaard 1964; Rozanski et al. 1992; Fleitmann et al. 2004; Sinha et al. 2005; Kotlia et al. 2012, 2015). Therefore, δ18O and δ13C variations in the stalagmites can be used to infer the palaeoclimatic reconstruction (Bar-Matthews et al. 1999) with low values (more negative) reflect higher amount of rainfall and vice versa (Burns et al. 1998, 2001, 2003; Neff et al. 2001; Fleitmann et al. 2004).

A few stalagmite δ18O-based palaeoclimatic records have been obtained from the Lesser Himalaya, e.g. Bølling-Ållerød interstadial between 15.2 and 11.7 ka from the Timta cave (Sinha et al. 2005), the last ~ 400 years from the Chulerasim cave (Kotlia et al. 2012; Duan et al. 2013) and two millennia records from the Dharamjali cave (Sanwal et al. 2013) as well as from the Sahiya cave (Sinha et al. 2015), the last ~ 4 ka BP record from the Sainji cave (Kotlia et al. 2015) and 750 years records from the Panigarh cave (Liang et al. 2015). Most of these speleothems-based δ18O records have documented not only the ISM variability but also highlighted the role of the WDs over the northwestern Himalaya during the late Holocene. Beyond the Himalayan records, a few speleothems δ18O records are also available from the Peninsular India (Yadava and Ramesh 2005; Yadava et al. 2004; Sinha et al. 2007; Laskar et al. 2013; Lone et al. 2014; Raza et al. 2017). However, existing speleothems records between mid and late Holocene from the ISM/ASM and WDs domains such as Qunf cave, Oman (Fleitmann et al. 2007), Sainji cave, Central Himalaya (Kotlia et al. 2015), Dongge cave, China (Dykoski et al. 2005) and Soreq cave, Mediterranean/black sea regions (Bar-Matthews et al. 2003) are also discussed in text to compare our TC1 record. The present study from the central part of the Lesser Himalaya can extend our understanding of the behaviour of both source regions, e.g. ISM and WDs between ~ 4.0 and ~ 1.6 ka BP.

Study area, cave details and climatic perspectives

The Tityana cave (30° 38′ 30.7″ N, 77° 39′ 07.4″ E, altitude 1470 m; (Fig. 1) is located in the Deoban limestone of Precambrian age (Thakur and Rawat 1992), more precisely Mesoproterozoic or mid-late Riphean (Bhargava et al. 2011). The length of the cave is about 13 m with entrance diameter of 3 m × 1 m. The cave surface is undulated and has few subchambers. Presently, the cave is dry and has flowstones, stalactites and unbroken stalagmites. We collected a stalagmite sample at a distance of 12.1 m from the entrance in March, 2013, when the cave air temperature was about 22 °C with relative humidity inside the cave as 86% and 53% at cave ventilation. The surrounding vegetation is dominated by Pinus vegetation along with shrubs. The mean annual temperature of the area is between 22 and 25 °C with average annual precipitation of about 290 cm. Presently, the area receives 80% rainfall during summer season (June to September) and the remaining 20% is contributed by the WDs (October to May). The mean annual precipitation and temperature data for the past 43 years (1951–1995) at the nearest meteorological station of Dharamsala (www.imd.gov.in) are used in the present study (see Fig. 2). The station received the highest precipitation in 1958 (around 472 cm) and lowest in 1965 (161 cm) since the beginning of the instrumental records (see Fig. 2). There are also examples that the ISM clubbed with WDs and has created disastrous flood hazards in the neighbouring areas (Joshi and Kotlia 2015; Ray et al. 2016; Indian Disaster Report 2013).

Fig. 2
figure 2

Temperature and precipitation data of past 44 years (during 1951–1995) from the nearest meteorological station at Dhramsala (Source: www.imd.gov.in)

Full size image

Material and analytical procedures

The stalagmite was cut into two halves and gently polished using emery paper. Initially, the Hendy test was performed to recognize if the stalagmite was grown under the isotopic equilibrium (Hendy 1971). To test this hypothesis, we used two different layers from centre to right at 1.7 cm (layer 1) and 6.9 cm (layer 2) from the top. The Hendy test was performed at National Geophysical Research Institute, Hyderabad, India. Simultaneously, four samples (at 0–1, 47–48, 50–51 and 126–127 mm from the top) were analysed at Mass Spectrometry and Environmental Change Laboratory, National Taiwan University for 230Th/U dating. The samples were analysed using multi-collector inductively coupled plasma mass spectrometer (MC-ICP-MS), following a standard procedure in Shen et al. (2003, 2012). An additional set of five samples (from depths at 0–1, 30–31, 63–64, 93–94, 127–128 mm), were sent to Poznan Radiocarbon Laboratory for 14C AMS dating. The OxCal v.4.2 software package (after Bronk Ramsey 1995, 2001) was used to calibrate 14C determination into calendar years and to model the age-depth relationship through a Poisson process depositional model (Bronk Ramsey 2008). This model allows flexible change in sedimentary rate and proves effective in simulating growth of speleothems (Scholz et al. 2012). We set up the critical values as 68 and 95%, respectively, for agreement index and convergence index in the model (Bronk Ramsey 1995).

A total of 130 samples (in addition to the Hendy test) were milled at an interval of 1 mm along its central growth axis for δ18O measurement using a micro-driller (MANIX-180). All the measurements were done at National Geophysical Research Institute, Hyderabad, India, using a Delta Plus Advantage isotope ratio mass spectrometer (IRMS) coupled with a Kiel-IV automatic carbonate device. IAEA standard NBS-18 and NBS-19 were used to express the sample δ13C and δ18O values with respect to Vienna Pee Dee Belemnite (VPDB) standard. The analytical precision was better than 0.1‰ for δ13C and δ18O (Ahmad et al. 2008). The VPDB standards were used for all isotopic compositions with analytical precision better than 0.10‰.

For mineralogy, 1 g powder was drilled with a 1-mm drill bit from different darker and lighter layers of TC1 at various depths using the micro-driller (NICRAFT MB-130). The mineralogical composition was analysed using Philips single crystal X-ray diffractometer with CCD facility (Siemens). X-ray diffraction (XRD) pattern has recorded 2θ position ranges at 10–80°. Further, three thin sections (3 × 3 cm) and ten stubs (~ 0.5 cm) were prepared for petrography under scanning electron microscope (SEM). The composition of stalagmite identified using energy dispersive X-ray analyser (EDS, Cambridge). The XRD and SEM analysis were carried out at Jawaharlal Nehru Centre for Advance Scientific Research (JNCASR), Bangalore, India.

Results and discussion

Hendy test

Hendy (1971) suggested that δ18O and δ13C values remain constant along a single growth layers if the sample was formed under the isotopic equilibrium. In the case of TC1, a small shift is observed in δ18O and δ13C values along layer 1 (Fig. 3a) which may be due to shifting of the drill bit while milling the sample. However, the low (0.042) correlation coefficient (r 2) between δ18O and δ13C suggests sufficiently slow degassing of CO2 to maintain the isotopic equilibrium during carbonate precipitation. For layer 2, the δ18O and δ13C values remains are constant (Fig. 3a). The Hendy test was further checked through the correlation between all the values of δ18O and δ13C (Fig. 3b) and the low r 2 (0.19) indicates that the kinetic fractionation was negligible (Fig. 3); hence, the δ18O in the sample may well reflect the climate signal. Besides, more recent studies have argued over the Hendy test and advocated the reliability of δ18O signal in the stalagmite despite the failure of the test (Fleitmann et al. 2004; Dorale and Liu 2009; Zhou et al. 2011; Kotlia et al. 2012)

Fig. 3
figure 3

a Hendy test for two layers of TC1 to indicate the kinetic equilibrium status. The results show the independent relationship between both δ13C and δ18O isotopes. The right panel shows the correlation between both δ13C and δ18O isotopes. b δ13C and δ18O plot of all isotopic data of TC1 along centre growth axis for further checking the kinetic fractionation. The plot also show the poor correlation between δ13C and δ18O (r 2 = 0.19)

Full size image

Chronology and age-depth relationship

The 230Th/U dating of TC1 stalagmite was unsuccessful due to multiple sources of non-authigenic 232Th and low uranium concentration (< 10 ppb) (Table 1). The high content of thorium appears unlikely a product of the decay of 234U in the sequence but more detrital or allogenic in origin (e.g. Blyth et al. 2017). The activity of detrital 230Th seems not produced from the decay of 234U due to high contribution of 232Th from recrystallization of rock or may be allogenic transported from the nearby area through water and wind activity during leaching of the calcium carbonate and causing high uncertainty in 230Th/232Th. The precision of the 230Th/U dating strongly depends on the concentrations of uranium and detrital thorium (Goslar et al. 2000) and such a problem is not uncommon in using uranium-series method to date impure carbonates (e.g. Garnett et al. 2004; Durand et al. 2016) and in fact excludes a great amount of stalagmites, sometimes across a large geographical region, from being applicable in palaeoenvironmental reconstruction (Lechleitner et al. 2016).

Table 1 The 230Th/U date of TC1 stalagmite. The instrumental analysis on MC-ICP-MS (after Shen et al. 2003; Shen et al. 2012). Analytical errors is 2 s of the mean
Full size table

The 14C AMS dating is an alternative method to resolve the chronological issue of these stalagmites (Laskar et al. 2013; Zhao et al. 2015). Zhao et al. (2015) suggested that the younger 14C AMS dates than 230Th/U dates can reflect negligible dead carbon influence with strong influence of initial 230Th in 230Th/U ages. Therefore, the 14C AMS method has successfully been applied to build chronologies for young speleothems (Goslar et al. 2000; Mattey et al. 2008; Laskar et al. 2013; Zhao et al. 2015). In the present study, 14C AMS dates are much younger than the 230Th/U dates and possibly suitable for palaeoclimatic reconstruction (see Tables 1-2). We used five calibrated 14C AMS dates to build TC1 chronology (Table 2).

Table 2 14C activity (presented as pMC) of the 14C dates from TC1 stalagmite and their calibrated age (in calendar years) after dead carbon correction
Full size table

The five 14C AMS dates are in the stratigraphic order and thus provide a reasonable basis for establishing the age-depth relationship (Fig. 4a). However, carbonate-related sediments are often strongly affected by 14C reservoir effect because of carbon isotopic signals introduced from the 14C-depleted bedrock (Genty et al. 2001; Hercman and Goslar 2002). Correction that removes the effect was therefore necessary to estimate precise radiocarbon ages. The dead carbon fraction (DCF) correction (after Hua et al. 2012) was performed in order to infer contemporaneous atmospheric 14C content from the measured 14C radioactivity. However, estimating the DCF in the TC1 sample proved challenging because of the lack of suitable anchor points that could be reliably dated by other radiometric methods. Furthermore, according to field observations, the growth of the stalagmite might have ceased for a long time and the top of the sample was not likely modern in age (cf. Li et al. 1996). Dead carbon correction remains a major difficulty in dating cave sediments, not only because it is not always feasible to find an independent dating series to correct the reservoir effect, but also because of the high level of variability of dead carbon contents across space and through time (Lechleitner et al. 2016). A well-recognized approach is to assume a more or less averaged, constant DCF throughout studied speleothems (see Reimer et al. 2013). In the TC1, we tentatively assumed a constant DCF value of ~ 15%, although hypothetical, as has been done in other studies (e.g. Genty and Massault 1997; Genty et al. 2001) and an observation that an age difference of ~ 1000–1500 years due to dead carbon is most commonly encountered (Laskar et al. 2011). Although limitations associated with this hypothetical DCF need to be treated with caution when interpreting the data, a good correlation between the TC1 sample and other palaeoclimatic records from nearby areas appears to strengthen the reasonability of our correction.

Fig. 4
figure 4

a TC1 cave stalagmite sample and uncalibrated 14C AMS dates and age-depth relationship with regression line have r 2 = 0.93. b TC1 stalagmite sequence and reconstructed age-depth relationship based on AMS 14C dating. The wider and lighter shade represents 95% probability range, while the narrower and darker shade represents 68% probability range. The median of the probability distributions are indicated as open circles

Full size image

The radiocarbon measurements are reported in terms of years BP (reference year is 1950 AD). The linear regression lines of calibrated 14C AMS ages have r 2 value about 0.93 (Fig. 4a). The Poisson process depositional model incorporates these corrected radiocarbon dates as shown in Fig. 4b. All agreement indexes and convergence indexes in the model were higher than the default critical values, thus validated the model. The model was set with critical values 95% confidence level and median dates were used for further extrapolation. According to the age-depth relationship, the TC1 stalagmite seems to have deposited between ca. 4.0 ka BP and 1.6 cal ka BP.

Mineralogy and petrography

The powder XRD technique and EDS reveal that the stalagmite is made up of calcite (Fig. 5a, b). The SEM images indicate that it is mostly composed of laminated compact calcite with two thin porous layers at 0.8- and 1.8-cm depth (Fig. 6a). Sometimes, the crystallization is between palisade calcite (PC) and columnar calcite (CC). The CC crystals at 0.5, 3.5 and 8–8.5 cm (Fig. 6b–d) and the PC crystals at 2–2.5 and 7–7.5 cm are observed (Fig. 6e, f) and a multiple growth of small PC and equant calcite (EC) are viewed at 10.5–11 cm (Fig. 6g). The CC crystals are elongated with orientated crystallographic axis parallel to the earlier growth surface (Fig. 6b–d) and may have been formed either by direct precipitation or as a recrystallization product of precursor aragonite (e.g. Perrin et al. 2014). In subtropical climates, calcite crystallization is favoured during the wet season because of higher cave humidity (Denniston et al. 2000). The palisade calcite (PC) shows growth direction parallel to the crystal boundaries (Fig. 6e, f). Large and well-developed crystals of equant calcite (EC) with multiple growth of small palisade are also observed (Fig. 6f) and they are closely packed and oriented parallel to each other (Fig. 6g). The cavities show the growth of small equant calcite towards the voids from the wall of stalagmite (Fig. 6g). Two thin porous EC layers are observed along the central growth axis at 5–5.5 and 9–9.5 cm (Fig. 6h). The boundary between the CC and EC layers is clearly visible with two small cavities (Fig. 6i).

Fig. 5
figure 5

a The representative graphs of energy dispersive X-ray analysis (EDS) in different depths. b The X-ray diffraction analysis (XRD) of TC1 from the various depths of stalagmite indicate that all are calcite peaks

Full size image
Fig. 6
figure 6

The micrograph showing a the large elongated compact calcite crystals; b elongates columnar crystals of calcite with their orientation and termination; c orientation of large columnar calcite crystal; d parallel columnar calcite (CC) crystal radiating from the stalagmite growth direction; e straight parallel crystals of palisade calcite (PC); f Multiple growth of small palisade and equant calcite crystal; g closely packed and well-developed large equant calcite crystals showing inside growth with small cavity on stalagmite surface; h growth of numerous small crystals of equant calcite in porous layers; i the columnar and equant calcite along with the boundary between compact and porous layer. All the arrows are indicating the direction of crystal orientation

Full size image

Palaeoclimatic reconstruction using δ18O and δ13C variations

The long-term records of rainwater δ18O concentration are not available from the northwest India. The only long-term record (1940–2004) of temperature, humidity and rainwater δ18O measurements are available from Delhi Metrological station (GNIP; http://www.iaea.org/water) (Fig. 7a). The rainfall δ18O values at New Delhi show that the precipitation during the ISM months (i.e. July–August–September) is characterized by lower δ18O values, higher relative humidity and less temperature compared to restively hotter months (i.e. May–June) (Fig. 7a). Therefore, higher rainfall with higher humidity should lead to a minimum evaporation under reduced kinetic fractionation, thus maintaining the low δ18O values (Kotlia et al. 2012, Fig. 7a). Nevertheless, New Delhi (average ISM δ18O values as − 6.8‰; GNIP, http://www.iaea.org/ water) and cave site have different elevation of ~ 1250 m and thus the δ18O precipitation at TC1 is supposed to be about 3.6‰ lower compared to the New Delhi due to altitude effect as δ18O values are lower with ~ 0.3‰/100 m towards higher elevations (Hren et al. 2009; Kotlia et al. 2015), resulting in − 10.4‰ around the cave site. However, the TC1 average δ18O values are − 9.25‰, still higher than the expected values. Therefore, it appears that the TC1 was influenced not only by the ISM but also by another moisture source which could be ascribed to the WDs. It is important to note that the δ18O isotopic composition is directly related to precipitation (Hern et al. 2009; Kumar et al. 2010; Kotlia et al. 2017).

Fig. 7
figure 7

a Rainfall δ18O values and temperature data during 1960–2004 at New Delhi metrological station, indicating that the lower δ18O associated with stronger monsoon precipitation and has relatively lower temperature than hotter month in northern India (May–June). b The GNIP precipitation vs δ18O values for the year 2005 at Uttarkashi area (nearest to present cave location) also representing the higher rainfall and lighter δ18O during monsoon months. c Lower δ18O and δ2H values for August–September and November indicates the contribution of WDs. In general, the higher rainfall related to more negative δ18O suggest the amount effect

Full size image

Due to lack of rainfall δ18O data from any nearby stations, we used 1-year data from Uttarkashi (Kumar et al. 2010) to understand the rainfall isotopic composition near the cave location (Fig. 7a, b). The data set (from − 0.8 to − 20.8‰) shows lower δ18O values during ISM months (Fig. 7b). Therefore, the higher rainfall giving rise to depleted δ18O values perhaps indicate the “amount effect” in precipitation (Gonfiantini et al. 2001; Kumar et al. 2010; Liang et al. 2015; Kotlia et al. 2017). Additionally, the δ18O and δ2H values have significant variations (δ2H ranges from − 157 to + 14.1‰ and δ18O ranges from − 20.8 to − 0.8‰, respectively; Fig. 7c) due to amount effect and also possibly indicate contribution of the WDs (Kumar et al. 2010; Liang et al. 2015; Kotlia et al. 2015, 2017). Similarly, the δ18O values at New Delhi for winter months (December–March) range between 0.2 and − 1.8‰ (Fig. 7a) while the values range from − 2 to − 6‰ for Uttarkashi locality for the same period (Fig. 7b), further indicating contribution of the WDs. A large variation in stalagmite δ18O values of the northwest Himalayan sites compared to the ISM dominated caves has previously been attributed to the input of WDs (Kumar et al. 2010; Kotlia et al. 2015, 2017; Liang et al. 2015).

Stalagmite δ13C primarily reflects vegetation cover and dense cover are generally characterized by low δ13C values due to dissolving of more biogenic CO2 into the seepage water (Kotlia et al. 2017). Contrary to this, stalagmite δ13C values are higher due to less contribution of biogenic CO2 under stressed climatic conditions (Kotlia et al. 2017). For palaeoclimatic reconstruction, use of δ13C is based on widely differing 13C/12C ratios of the C3 and C4 plants. The C3 and C4 plants, separated on the basis of their photosynthesis pathways of carbon fixation have δ13C values in the range of − 26 to − 28‰ (with a mean of − 26‰) and − 11 to − 14‰ (with a mean of − 12‰), respectively (Smith 1972; Smith and Epstein 1971). Thus, the lower δ13C values under the C3 vegetation should broadly indicate warmer/wetter conditions as the δ13C values in cave speleothems depend on the CO2 in the soil, fractionation and the extent of exchange between seepage drip water and host rock (Smith 1972; Smith and Epstein 1971; Kotlia et al. 2015). In the TC1, the δ13C values are increased (from − 10.57 to − 0.19‰) with increasing aridity between ~ 4.0 and ~ 3.4 ka BP. The δ13C values ere highest around ~ 3.4 ka BP, suggesting peak aridity. Thereafter, the values decrease from ~ 3.4 to 2.0 ka BP (0.19 to − 9.43‰), reflecting improved C3 vegetation under favourable climatic conditions. The δ13C values show positive relationship with δ18O during the arid events around ~ 3.4, ~ 1.9 and ~ 1.6 ka BP,

Based on the δ18O variations, the TC1 deposit is divided into three distinct climate zones: (1) reduced precipitation between ~ 4.0 and ~ 3.4 ka BP, characterized by enhancement in δ18O values by about > − 1‰ (between − 9.78 and − 8.63 ‰) with abruptly higher δ18O at ~ 3.4 ka BP (− 8.11‰) (see Fig. 8), (2) slightly improved and favourable climatic conditions between ~ 3.4 and 2.7 ka BP by depletion in δ18O values for about > − 1.5‰ (− 8.63 to − 10.18‰; Fig. 8) and (3) arid conditions from ~ 2.7 ka BP onwards prevailing until 1.5 ka BP with enhancement in δ18O values for about − 2‰ (− 10.18 to − 8.08‰; Fig. 8). Between ~ 2.7 and ~ 1.5 ka BP, two peak drought events were observed at ~ 1.9 ka BP (− 8.04‰) and ~ 1.6 ka BP (− 8.08‰) and are characterized by abrupt enhancement in δ18O values (Fig. 8). Similar results have been obtained from the Sainji cave, Kumaun Himalaya (Kotlia et al. 2015) and Soreq cave, from eastern Mediterranean (Bar-Matthews et al. 2003).

Fig. 8
figure 8

δ18O and δ13C variation of TC1 with three distinct phases of past precipitation with major drought events centred at ~ 3.4, ~ 1.9 and ~ 1.6 ka BP. The vertical yellow panels show the drought period

Full size image

Comparison with other records

The mid to late Holocene climatic records from northwest Himalaya are mainly from the modern lakes, palaeolakes and peat bogs, although some with coarse sampling interval (Fig. 9). In the TC1 data, an abrupt drop in δ18O and δ13C with peak aridity around ~ 3.5–3.4 ka BP correspond to the transition of mid to late Holocene. This event is prominent in almost all the records obtained from peat bogs and lake sediments and reflect weakening of the ISM (Phadtare 2000; Rühland et al. 2006; Kotlia and Joshi 2013; Mishra et al. 2015; Rawat et al. 2015). Thereafter, the records are sparse in NW Himalaya (Fig. 9). In general, well-dated lake records from Garhwal Himalaya (Kotlia and Joshi 2013) and peat record from Lahaul region (Himachal Pradesh) show similar climatic variability (see Fig. 9; Rawat et al. 2015). Other available continental and marine records of the past climatic variability from nearby regions (e.g. China and Tibet Plateau) can also be compared with the TC1 record and so also with the Soreq cave (Bar-Matthews et al. 2003), Sainji cave (Kotlia et al. 2015), Qunf cave (Oman; Fleitmann et al. 2007) and Dongge cave (Dykoski et al. 2005) to understand the potential of well-dated stalagmite δ18O from the climatically sensitive Himalayan sites. As mentioned above, three distinct climatic phases are recognized in the TC1 and are discussed further from the view-point of comparison.

Fig. 9
figure 9

Comparison of TC1 record with other records of NW Himalaya indicates a relative impact of WDs after transition of Mid-Late Holocene around 3.5–3.4 ka BP

Full size image

~ 4.0–3.4 ka BP (declined precipitation)

Declined precipitation during this period in the TC1 with peak aridity at ~ 3.4 ka BP seems coherent with the reduced precipitation in the Soreq cave between 4.5 and 2.5 ka BP (Bar-Matthews et al. 2003) with higher δ18O around 3.5 ka BP (Fig. 10). Reduced precipitation from ~ 4.0 to ~ 3.0 ka BP with the lowest precipitation at ~ 3.2 ka BP has also been observed in the Kumaun Himalaya (Kotlia et al. 2015). Within the period, similar fluctuations in δ18O values were recognized in the Qunf cave with higher amplitude peak around 3.4 ka BP indicating weakened ISM (Fleitmann et al. 2007) (see Fig. 10). The declined intensity of monsoon at ~ 3.6 ka BP has also been registered in the Dongge cave records (Dykoski et al. 2005). Progressive lowering of lake took place in a northwest Himalayan lake around this time (Mishra et al. 2015) and the mean annual precipitation (MAP) reconstruction from the same lake confirms enhanced aridity between ~ 3.8 and 3.6 cal ka BP with markedly low lake level and less discharge in the River Indus around 3.2 ka BP (Leipe et al. 2014). The arid conditions between ~ 5.0 and 3.0 ka BP are also recorded from the Sanai lake in the Ganga plain, northern India (Sharma et al. 2004). Further, the lakes and peat records from the NW Himalaya also register intensification of aridity between ca. 4.0 and 3.5 ka BP (Chauhan and Sharma 1996; Kotlia et al. 1997; Phadtare 2000). The historical researches have also documented de-urbanization of the Indus valley culture at ~ 4.0 ka BP (Wright 2010; Dixit et al. 2014) and the beginning of the collapse of Harappan civilization around 3.8 ka BP (Winner 2012) with its complete disappearance between ~ 3.5 and 3.0 ka BP (Leipe et al. 2014). A major dry spell from ~ 3.8 to 3.2 ka BP in the Bangong basin, (Gasse and Van Campo 1994) and significantly declined humidity in the Yunnan province around 3.4 ka BP (Xiao et al. 2014) may be positively correlated with our data set. Similarly, fall in the human civilization between 4.5 and 2.5 ka BP in the Middle East, North Africa and Egyptian kingdom (Hassan and Stucki 1987; Weiss et al. 1993; Bar-Matthews et al. 2003) confirm that this arid phase was experienced widely.

Fig. 10
figure 10

Comparison of TC1 record between 4 and 3 ka BP with stalagmite records from ISM dominated Oman cave (Fleitmann et al. 2007) and WDs influenced Soreq cave (Bar-Matthews et al. 2003), suggesting the amount effect in TC1 due to two different moisture regime, e.g. WDs and ISM. The vertical yellow panels show the peak aridity around ~ 3.4 ka BP, which is well correlated with decline of Indus valley civilization. The Soreq cave and TC1 records show close correspondence, while Qunf cave show opposite trend

Full size image

~ 3.4–2.7 ka BP (improved climatic conditions)

In general, a gradual decrease in δ18O values (− 8.63 to − 10.18‰) shows slightly favourable climatic conditions. Low δ18O values between 3.2 and 2.0 ka BP are also documented from the Central Himalaya (Kotlia et al. 2015). However, the Dongge cave records are not coherent possibly due to low-resolution chronology in the TC1, while the Soreq cave data set shows similar fluctuations (Fig. 11). Pollen records from Lahaul, NW Himalaya indicate improved climatic conditions from ~ 3.4 to 2.0 ka BP (Rawat et al. 2015) and the high monsoonal runoff was observed from the Indus River during ~ 3.5–2.2 ka BP (von Rad et al. 1999). Similar situation was prevailed in the Garhwal Himalaya from ~ 3.5 to 1.8 ka BP (Kotlia and Joshi 2013) and in the central Ganga plain (Basaha lake) from 3.3 ka BP onwards (Chauhan et al. 2004). Our results support an evidence of restoration of mixed oak forest in response to amelioration of climate in the temperate zone of Kumaun Himalaya around 2.9 ka BP (Gupta 2008). However, all the records are expanded due to temporal resolution of different proxy and perhaps withdrawal of the monsoon seasonality with increased influence of WDs in NW Himalaya.

Fig. 11
figure 11

Comparison of TC1 δ18O record with stalagmite records from ISM and WDs dominated regimes, e.g. Qunf cave (Fleitmann et al. 2007); Dongge cave (Dykoski et al. 2005); Sainji cave (Kotlia et al. 2015) and Soreq cave (Bar-Matthews et al. 2003). The vertical panels show peak drought events. See Fig. 1 for the location of all the caves.

Full size image

~ 2.7–1.5 ka BP (reduced precipitation)

The TC1 records show climatic worsening from 2.7 ka BP onwards with two drought events at ~ 1.9 ~ 1.6 ka BP (Fig. 11). Similar fluctuations in stalagmite δ18O were also observed from the Sainji cave, Kumaun Himalaya (Kotlia et al. 2015) between ~ 2.0 and 0.8 ka BP (Fig. 11). The trend of δ18O value towards positive in Soreq cave within the period was characterized by gradual decrease in rainfall (Fig. 11; Bar-Matthews et al. 2003). The relatively dry periods were also documented from eastern Mediterranean after 1.9 ka (Orland et al. 2008). In general, the Dongge cave experienced reduced precipitation after 3.5 ka BP (Dykoski et al. 2005), while the higher amplitude peaks with increased values of δ18O observed at ~ 2.1 and 1.8 ka BP can be equated with TC1 (Fig. 11) although Wang et al. (2005) noted the weakened intensity of monsoon at ~ 1.6 cal ka. A progressive lowering of Tso Moriri lake level from 2.7 ka BP onwards was also viewed by Mishra et al. (2015). Rühland et al. (2006) also pointed out the drier climatic conditions around 1.6 ka BP in the Pinder valley. Intensified glacier activity and cooler climate has been documented from Himalaya at ~ 2.0 ka BP (Chauhan 2003; Kar et al. 2002). Speleothem records from the Pokhara valley (Central Nepal) confirm the reduced monsoonal precipitation and increased aridity due to deposition of aragonite layers between 2.3 and 1.5 ka BP (Denniston 2000). Pollen data obtained from the Spiti region also specify cool/dry conditions between ~ 2.0 and 1.0 ka BP (Mazari et al. 1996) and a severe weakening of ISM activity around ~ 1.5 ka BP is evident in historical records from India (Pandey et al. 2003).

Conclusions

Variations in the TC1 δ18O values primarily reflect the past precipitation variability. A positive relationship of δ18O variations with previously established records of mid-late Holocene climatic variability shows that the 14C AMS dates used in the young speleothems are expected realistic within the age uncertainties. Three major climatic phases are documented by studying the TC1 stalagmite: (1) declined precipitation between 4.0 and 3.4 ka BP, (2) gradual improvement in the climatic condition from 3.4 to 2.7 ka BP and (3) reduced precipitation conditions from ~ 2.7 until the end of TC1 deposition. During the last ~ 4.0 ka BP, we identified three dry events at ~ 3.5, ~ 1.9 and ~ 1.6 ka BP. The ~ 3.5 ka BP event is equated with transitions between mid and late Holocene and coincides with de-urbanization of the Indus valley civilization. Diversified records and large variations in speleothem δ18O from a number of the northwest Himalayan areas indicate the relative contribution of WDs during the late Holocene, a topic which has not been much incited before.