Abrupt changes in Indian summer monsoon strength during 33,800 to 5500 years B.P.

1 Introduction

The sudden arrival of the rainy season after a period of hot and dry summer marks the onset of the summer monsoon in India [Chao, 2000]. About 80% of India's total annual rainfall occurs during summer or southwest (SW) monsoon season (June–September) with the exception of a few areas in the northwest and extreme south [Parthasarathy, 1960]. Historically, Indian summer monsoon (ISM) failures led to the demise and/or displacement of numerous cultures and civilizations in South Asia [Staubwasser et al., 2003; Buckley et al., 2010; Dixit et al., 2014]. Extreme future changes in Indian summer monsoon (ISM) rainfall may have adverse impacts as well [Intergovernmental Panel on Climate Change (IPCC), 2013].

Climate models predict increases in extreme and heavy precipitation events in a warm world [Trenberth et al., 2003; Allan and Soden, 2008; IPCC, 2013]. These studies indicate a direct relation between rainfall extremes and temperature, with intense rain events increasing during warm intervals and decreasing during cold intervals. It is also anticipated that as temperatures rise, the hydrological cycle will experience a shift in precipitation form from snow to rain and early snowpack melting [Pavelsky et al., 2012]. Understanding changes in the intensity and frequency of these precipitation extremes is important to the welfare of human societies and the ecosystems on which they depend.

Model and observational studies indicate that the Asian summer monsoon weakened and the westerlies strengthened significantly during the Younger Dryas (YD) and the Last Glacial Maximum (LGM) [Porter and Zhisheng, 1995; Sinha et al., 2005]. The East Asian summer monsoon precipitation also decreased significantly during the YD [Wang et al., 2001; Yuan et al., 2004]. Similarly, ISM strength decreased during Holocene North Atlantic cold intervals; the so-called Bond events [Gupta et al., 2003]. These studies demonstrate a link between North Atlantic temperature changes and Asian monsoon strength. Model experiments show that the westerlies transmitted these North Atlantic anomalies to the Asian region [Sun et al., 2011].

Late Pleistocene speleothem records from the Hulu and Dongge Caves of China indicate centennial- to millennial-scale changes in East and South Asian monsoon precipitation [Wang et al., 2001; Yuan et al., 2004]. However, such high-resolution, long-ranging ISM precipitation records are scarce from the Indian landmass especially from the Indian Himalaya [Sinha et al., 2005]. Oxygen isotope records in speleothems (cave carbonates) across Asia are commonly interpreted as a proxy for the strength of the Asian monsoons [Wang et al., 2001; Yuan et al., 2004; Sinha et al., 2005]. Recent studies, however, suggest that the cave records from Asia also reflect changes in moisture source, changes in fractionation at the source, and changes in transport pathways as well as the isotopic composition of Bay of Bengal (BoB) surface water [LeGrande and Schmidt, 2009; Breitenbach et al., 2010; Dayem et al., 2010; Pausata et al., 2011; Berkelhammer et al., 2012].

The stable oxygen isotope (δ¹⁸O) record of speleothems from Mawmluh Cave, NE India, provides information about changes in ISM strength on decadal to centennial timescales during the past few millennia. Mawmluh Cave (25°15′44″N, 91°52′54″E altitude 1290 m) is in a region of NE India near Cherrapunji which is swept by strong SW monsoon winds (Figure 1 and supporting information Figures S1–S3). Cherrapunji is among the wettest locations on Earth with an annual average precipitation of >11 m, 70% of which is received during the summer monsoon months from June to September, during the active SW monsoon season [Murata et al., 2007].

The weighted mean δ¹⁸O versus precipitation and δ¹⁸O versus temperature relationships for Shillong GNIP site near the cave indicate that precipitation amount accounts for ~30% of the variance in modern δ¹⁸O of precipitation while temperature shows no relation to the δ¹⁸O of precipitation (supporting information Figure S4) (source: http://www.univie.ac.at/cartography/project/wiser/gui/gnip_all_index.php). These modern relationships suggest that changes along the transport path and in the isotopic composition of source surface water account for the majority of the modern variance in the local rainfall δ¹⁸O [Breitenbach et al., 2010; Liu et al., 2014] (Figure S4). Therefore, we relate changes in speleothem δ¹⁸O to changes in isotopic composition of source waters and processes (evaporation and precipitation) occurring along the transport path; all of which, in turn, reflect changes in ISM strength [Yuan et al., 2004; Pausata et al., 2011].

To assess long-term changes in ISM strength and to understand whether LGM, YD, and Heinrich cold events were accompanied by decreased summer monsoon strength in the Indian subcontinent, we studied cave carbonate δ¹⁸O from the Mawmluh Cave, near Cherrapunji, Meghalaya (NE India) spanning in age from 33,800 to 5500 years B.P. (Figure 1). Our main objective is to evaluate the climatological significance of the abrupt changes recorded in this δ¹⁸O time series.

2 Methods

Samples were taken at 0.5 mm interval along the central growth axis of the speleothem sample with average sampling intervals varying from 15 to 77 years. The δ¹⁸O profile is based on 367 isotopic measurements carried out at Brown University, USA (supporting information Table S1). Replicate analysis of Carrara marble, treated in the same as the speleothem samples, indicates reproducibility of ±0.07 (1σ, N = 26). Hendy test [Hendy, 1971] of one layer suggests no significant correlation between δ¹⁸O and δ¹³C (supporting information Figure S5). Comparison of δ¹⁸O values with the earlier Mawmluh Cave record [Berkelhammer et al., 2012] from 12,000 to 3650 years eliminates the chances of kinetic fractionation and validates our data (Figure 2).

The chronology of Mawmluh Stalagmite-1 (MWS-1) was constrained by 11 absolute U-Th series dates determined by multiple collector–inductively coupled plasma–mass spectrometry at the University of Minnesota using the technique as described by Cheng et al. [2013] (Figure 4 and Table 1). All ages fall close to the regression line, suggesting that the growth of the stalagmite occurred without any major break during 33,800 to 5500 years B.P. (Figures 4 and 5). The relationship between age and the distance of the proxy measurement along the growth axis of the speleothem is established using the StalAge Model [Scholz and Hoffmann, 2011]. The StalAge program is run in the open source statistical software R [R Development Core Team, 2010]. The StalAge program at first step identifies the major outliers. In next step, the age data are screened for minor outliers and age inversions, and the uncertainty of potential outliers is increased using an iterative procedure. Finally, the age model and corresponding 95% confidence limits are calculated by a Monte Carlo simulation fitting ensembles of straight lines to subsets of the age data (Figure 5). Calculated uncertainties are in the range of those calculated by combined application of Bayesian chronological ordering and a spline, providing the stratigraphic information in order to reduce age model uncertainty.

Sample ages were interpolated linearly between adjacent age measurements. The typical age uncertainty is 269 years from 33,800 to 24,000 years B.P. and less than 100 years from 24,000 to 5500 year B.P. (Table 1). Spectral analysis of the oxygen isotope time series shows statistically most significant (strongest) periodicity (95% confidence level) centered at 1700 year using PAST Red Fit method [Hammer et al., 2001] (supporting information Figure S6) which resembles 1667 year suborbital cycle reported in Hulu Cave record from China [Clemens, 2005]. The other significant cycles lie at 1118, 802, and 602 years indicating millennial-scale changes in ISM strength/precipitation (Figure S6).

3 Results and Discussion

This δ¹⁸O data set from the Mawmluh Cave and the earlier published record from the same cave [Berkelhammer et al., 2012] show fairly a good match; however, small divergence in the two records may have arisen due to different karst processes controlling different drips which fed the two cave samples (Figure 2). A Hendy test [Hendy, 1971] of one layer suggests no significant correlation between δ¹⁸O and δ¹³C (Figure S5). Comparison of δ¹⁸O values with the earlier Mawmluh Cave record [Berkelhammer et al., 2012] from 12,000 to 3650 years using a different speleothem sample argues against kinetic fractionation and validates our data (Figure 2).

Speleothem δ¹⁸O values are influenced by orbital variations that bring changes in local precipitation processes including transport pathways, moisture source, and seasonal precipitation balances [Wang et al., 2001; Cheng et al., 2009; LeGrande and Schmidt, 2009; Pausata et al., 2011]. Breitenbach et al. [2010] proposed that the BoB surface waters serve as the primary source of moisture to Mawmluh Cave and therefore are the dominant influence on Mawmluh Cave δ¹⁸O. Such a linkage can be explained by the fact that strengthened monsoons increase river output to the BoB, therefore lowering the surface seawater δ¹⁸O.

A comparison of MWS-1 cave δ¹⁸O with that of Hulu/Sanbao [Wang et al., 2001] and Dongge Caves [Yuan et al., 2004] from China suggests reasonable similarity which might be due to similar isotopic composition of source of air parcels in two regions. The eastern China caves receive moisture both from the Indian Ocean and from the Pacific Ocean, whereas moisture sources for the Mawmluh Cave is the Indian Ocean including the Bay of Bengal Arabian Sea [Breitenbach et al., 2010; Liu et al., 2014]. As such, MWS-1 speleothem δ¹⁸O mainly reflects the Indian monsoon signal.

Mawmluh Cave δ¹⁸O are plotted with maximum insolation at 25°N [Huybers, 2006], orbital precession [Berger and Loutre, 1991] and the Greenland Ice Sheet Project 2 (GISP 2) record to better assess the role of orbital insolation and high-latitude glaciation in driving changes in ISM strength (Figure 3). Mawmluh Cave δ¹⁸O is exceptionally well aligned with GISP δ¹⁸O from 25,000 years ago to the present. The most enriched Holocene values (−7‰ at 8 ka) lag maximum local solar insolation by 3500 years, consistent with the lag (relative to July 21 insolation) measured for the late Pleistocene Hulu/Sanbao record [Wang et al., 2001, 2008]. These relationships indicate links to both solar insolation and North Atlantic climate signals transmitted via a westerlies teleconnection and sea ice. Increased North Atlantic sea ice extent causes cool temperatures throughout major parts of the Northern Hemisphere and the northern Indian Ocean. These cool anomalies, transported in the westerlies, delay the arrival of the Indian monsoon as well as reduce precipitation over the Indian subcontinent [Gupta et al., 2003; Pausata et al., 2011]. Such changes are likely to be recorded in heavier δ¹⁸O values of cave carbonates.

Our record also shows good correlation with Northern Hemisphere tropical and subtropical climate proxy records from the BoB and Cariaco Basin, indicating both the tropics and subtropics were swept by similar climatic regimes during the studied interval (Figure 2). Gupta et al. [2003] linked such contemporaneous changes in the tropics and Northern Hemisphere to solar variability.

The Mawmluh Cave δ¹⁸O record indicates a strengthened ISM circulation from 33,500 to 32,500 years B.P. and weakened circulation in the later part of marine oxygen isotope stage (MIS) 3 and early part of MIS 2 with a weak ISM phase during 26,000 to 23,500 years B.P. coinciding with a cold phase in the North Atlantic (Figure 3). During the latter interval, monsoon runoff to the Bay of Bengal weakened owing to a reduction in ISM rainfall [Govil and Naidu, 2011]. During the LGM the ISM circulation strengthened similar to that during the late MIS 3 followed by its weakest phase lasting from 17,000 to 15,000 years B.P. (Heinrich event H1). The Mawmluh Cave δ¹⁸O record shows an abrupt increase in ISM strength during the Bølling-Allerød from 15,000 to 12,900 years B.P. and early Holocene (~10,000 to 6500 years B.P.) warm intervals suggesting increased ISM strength, which corresponds to increased river discharge to the BoB [Govil and Naidu, 2011]. The Bølling-Allerød increase is consistent with the ISM maximum in northern India as documented in the speleothem δ¹⁸O record from the Timta Cave [Sinha et al., 2005]. ISM strength decreased significantly during the Heinrich events H1–H3 and YD cold event (Figure 2).

The strengthened Holocene ISM (10,000–7000 years B.P.) is followed by a sudden decrease at 6500 years B.P. indicating a weak phase of ISM [Gupta et al., 2003]. Similarities among Mawmluh Cave δ¹⁸O, BoB planktic foraminfer δ¹⁸O, Chinese cave records and solar insolation suggest that changes in δ¹⁸O of precipitation along the transport pathway and source effects were perhaps linked to solar variability [Wang et al., 2001; Gupta et al., 2003]. We agree with Yuan et al. [2004] that rainfall isotopic composition integrated from source to cave site increases with δ¹⁸O depletion and vice versa. The same mechanism worked on centennial to millennial timescales [Pausata et al., 2011]. Mawmluh Cave δ¹⁸O terminations align well with the GISP 2 ice core record, implicating a westerlies mechanism for transmitting the millennial-scale ice volume and sea ice signal from the North Atlantic to moisture δ¹⁸O composition in Indian and Asian caves during cold intervals.

The largest amplitude excursions recorded in Mawmluh Cave δ¹⁸O are associated with cold events including the YD (~13,000 years B.P.), Heinrich #1 (~17,000 years B.P.), and Heinrich #2 (~25,000 years B.P.). Variance within the warm Holocene interval (~10,000 years B.P. to the present) is significantly reduced compared to that in the colder MIS3 interval (35,000 to 23,000 years B.P.). If the underlying physics are similar, these data suggest that increased warmth will be accompanied by reduced amplitude monsoon variability.

4 Conclusions

Our results demonstrate that abrupt weak/strong phases in Indian summer monsoon (ISM) strength correspond to cold/warm intervals driven by weakening/strengthening of thermal contrasts between land and sea since MIS3. The novelty of our work lies in the fact that it is the southernmost longest record that is uniquely an Indian signal, offering the opportunity to assessisotopic contrasts relative to the Chinese records and the millennial-scale differences with other climate records. Our study suggests that δ¹⁸O depletion along the transport pathway and isotopic composition of Bay of Bengal surface water mainly controlled δ¹⁸O of Mawmluh Cave carbonate. We have also demonstrated explicitly for the first time the presence of YD cold event of Greenland with sharp boundaries in a cave carbonate record from the Indian landmass.