Holocene climatic changes in the Westerly-Indian Monsoon realm and its anthropogenic impact

The Indian Summer Monsoon (ISM) with its rainfall is the lifeline for people living on the Indian subcontinent today and possibly was the driver of the rise and fall of early agricultural societies in the past. Intensity and position of the 10 ISM have shifted in response to orbitally forced thermal land-ocean contrasts. At the northwestern monsoon margins, interactions between the subtropical westerly jet (STWJ) and the ISM constitute a tipping element in the Earth’s climate system, because their non-linear interaction may be a first-order influence on rainfall. We reconstructed marine sea surface temperature (SST), supply of terrestrial material and vegetation changes from a very well-dated sediment core from the northern Arabian Sea to reconstruct the STWJ-ISM interaction. The Holocene record (from 11,000 years) shows a distinct, 15 but gradual, southward displacement of the ISM in the Early to Mid-Holocene, increasingly punctuated by phases of intensified STWJ events that are coeval with interruptions of North Atlantic overturning circulation (Bond events). Effects of the nonlinear interactions culminate between 4.6-3 ka BP, marking a climatic transition period during which the ISM shifted southwards and the influence of SWTJ became prominent. The lithogenic input shows an up to 4-fold increase after this time period signaling the strengthened influence of agricultural activities of the Indus civilization with enhanced erosion of soils 20 amplifying the impact of Bond events and adding to the marine sedimentation rates adjacent to the continent.


Introduction
Changes in latitude and strength of the ISM were gradual over the course of the Holocene, but were punctuated by distinct climatic events (Herzschuh, 2006). The ISM was strong during the Early Holocene until approximately 6.5 ka ago, and influenced regions farther north than today (Herzschuh, 2006;Prasad and Enzel, 2006). A weakening of the ISM is reflected 25 in many records after the Holocene Optimum Period, commonly attributed to declining summer insolation (Banerji et al., 2020;Herzschuh, 2006), latitudinal insolation gradients (Mohtadi et al., 2016;Ramisch et al., 2016) and feedbacks of climate anomalies such as the Arctic Oscillation (Zhang et al., 2018), the North Atlantic Oscillation (NAO) (Band et al., 2018;Banerji et al., 2020;Kotlia et al., 2015;Lauterbach et al., 2014) or the El Niño Southern Oscillation (ENSO) (Banerji et al., 2020; Changes of the ISM system not only affected hydrological conditions and vegetation in this region, but were also an important constraint on the expansion and dispersal of the Indus civilization, also known as Harappan civilization, emerging around 5 ka ago (Durcan et al., 2019;Giosan et al., 2012Giosan et al., , 2018Possehl, 1997). Growing agricultural activities adjacent the northwestern Arabian Sea increased soil erosion and, therefore, terrigenous sediment input (Gourlan et al., 2020). Further, the decrease of ISM precipitation favored the development of combined summer and winter cropping as well as drought-tolerant crops (Petrie 35 and Bates, 2017). Winter precipitation intensity and interannual variability in the present-day northwestern region of the Indian Monsoon (IM) domain is governed by STWJ-induced events of strong precipitation, so called Western Disturbances (WD) (Anoop et al., 2013;Dimri et al., 2016;Hunt et al., 2018;Leipe et al., 2014;Munz et al., 2017;Zhang et al., 2018).
Climatological data show a link between the NAO, the position of the STWJ and the strength of the WD during the winter over northwest India (Yadav et al., 2009). A positive NAO is associated with an intensified STWJ and stronger WD and, thus, 40 above normal winter precipitation in this region (Yadav et al., 2009). The interaction, position and strength of the IM and the STWJ on millennial to Holocene timescales are yet poorly understood, because the pronounced signal of the ISM dominates sedimentary records, and high-quality records in the region where IM and STWJ interact are sparse.
The northeastern (NE) Arabian Sea (AS) is located in the region of interaction of IM and STWJ. Whereas the primary productivity during the warm ISM season is driven by lateral transport of nutrient-rich waters from the Oman upwelling area 45 (Schulz et al., 1996), the NE winds deepen the mixed layer during the colder IWM season due to convective mixing leading to enhanced primary productivity and carbon exports to the deep sea (Banse and McClain, 1986;Madhupratap et al., 1996;Rixen et al., 2019). The NE AS holds high-quality archives due to excellently preserved and highly resolved sediment records deposited on the shelf and slope impinged by a strong oxygen minimum zone (OMZ) between 200 m and 1200 m water depth (von Rad et al., 1995;Schott et al., 1970;Schulz et al., 1996). Most existing records have poor temporal resolution or do not 50 bracket the entire Holocene (Böll et al., 2014Giosan et al., 2018;Lückge et al., 2001;Munz et al., 2015), but they outline the interplay of the Indian Winter Monsoon (IWM) with the westerlies in the NE AS realm during the Holocene and suggest that winter monsoon activity intensified between 3.9 and 3.0 ka BP (Giesche et al., 2019;Giosan et al., 2018;Lückge et al., 2001) and intermittently during the last ca. 1.5 ka (Böll et al., 2014;Lückge et al., 2001;Munz et al., 2015).
For a detailed and high-resolution reconstruction, we analyzed lithogenic mass accumulation rates (LIT MAR), alkenone-55 based SST and n-alkane-based land vegetation from the NE AS ( Figure 1) over the last 10.7 ka. They document the influence of the IWM and the westerlies on the marine and terrestrial environment in the NE AS region and show that the advent of agriculture in step with climate change significantly enhanced soil erosion.
https://doi.org/10.5194/cp-2020-141 Preprint. Discussion started: 17 November 2020 c Author(s) 2020. CC BY 4.0 License. Records presented here were obtained from the box core SO90-63KA (697 cm long), which was retrieved from the Arabian Sea off Pakistan (24°36.6'N, 65°59.0'E, 315 m water depth) during the RV SONNE cruise SO90 in 1993. With the upper 18 cm missing, SO90-63KA covers the last ca. 10.7 ka and encompasses the development of early Bronze Age agricultural 70 societies (Staubwasser et al., , 2003. The chronology is based on 80 14 C dates of planktic foraminifers Globigerinoides sacculifer (Staubwasser et al., , 2003 and belongs to the most-well-dated cores from the Arabian Sea. The bulk components, grain sizes and major and trace elements of SO90-63KA were published previously Staubwasser and Sirocko, 2001). We used an extended data set for the grain size analyses of SO90-63KA sediments for the endmember modelling analyses. The procedure is described in the earlier published coarser resolution data set (Burdanowitz 75 et al., 2019).
For the lipid analyses, 2.0 to 13.4 g of freeze-dried and grounded sediment were extracted with a DIONEX Accelerated Solvent Extractor (ASE 200) at 75°C, 1000 psi for 5 min using dichloromethane (DCM) as solvent. The procedure was repeated three times. A known amount of an internal standard (14-heptacosanone, squalane) was added prior to extraction. An additional working sediment standard was extracted for each ASE running sequence. The total lipid extracts (TLEs) were concentrated 80 using rotary evaporation. The TLEs were separated into a hexane-soluble and a hexane-insoluble fraction by Na2SO4 column chromatography. The hexane-soluble fractions were saponified at 85°C for 2h using a solution of 5% potassium hydroxide in methanol (MeOH). Afterwards, the neutral fraction was extracted with hexane. Column chromatography using deactivated silica gel (5% H2O, 60 mesh) was carried out to separate the neutral fraction into an apolar-(containing n-alkanes), ketone-(containing alkenones) and polar-fraction by using hexane, DCM and DCM:MeOH (1:1), respectively. The apolar fraction 85 was further cleaned by AgNO3-Si column chromatography using hexane as solvent.
Quantification of the n-alkanes and the alkenones was carried out using an Agilent 6850 gas chromatograph (GC) equipped with an Optima1 column (30 m, 0.32 mm, 0.1 µm), split/splitless injector operating at 280°C and a flame ionization detector (FID, 310°C). H2 was used as carrier gas with a flow of 1.5 ml/min. Samples were injected in hexane and duplicate measurements of each sample were carried out for the alkenones. For the n-alkanes, the GC temperature was programmed 90 from 50°C (held 1 min) ramped at 8°C/min to 320°C (held 15 min). An external standard was used for quantification, containing n-C8 to n-C40 alkanes in known concentration. Repeated analyses of the external standard resulted in a quantification precision of 6 % (1SD).
The average chain length (ACL) of the homologues C27-C33 was calculated using following equation: where Cx is the concentration of the n-alkane with x carbon atoms.
For the alkenones, the GC temperature was programmed to increase from 50°C (held 1 min) to 230°C at 20°C/min, then at 4.5°C/min to 260°C and at 1.5°C/min to 320°C (held 15 min). The C37:2 and C37:3 alkenones were identified by comparing the https://doi.org/10.5194/cp-2020-141 Preprint. Discussion started: 17 November 2020 c Author(s) 2020. CC BY 4.0 License. retention time peaks of the samples and the known working sediment standard. Quantification was carried out by using a known amount of an external standard (14-heptacosanone and hexatriacontane). 100 We calculated the alkenone unsaturation index using the following equation (Prahl et al., 1988): 37 ′ = 37:2 37:2 + 37:3 and using the core top calibration of Indian Ocean sediments (Sonzogni et al., 1997) to convert the UK'37 index to SSTs: For each sample, at least one duplicate measurement was obtained with an average precision of 0.1°C (1SD). The precision of 105 the replicate extractions of the working standard sediment (n=13) and duplicate measurements of each replicate was 0.5°C (1SD).
The total mass accumulation rates (MAR) were calculated by multiplying the dry bulk density (DBD) with the linear sedimentation rate (LSR). The LSR was ascertain by using age model depth and age interval values. DBD was calculated with following equation (Avnimelech et al., 2001): 110 For calculating the sediment particle density, we assumed 2.65 g/cm³ for inorganic sediment particles (Blake and Hartge, 1986;115 Boyd, 1995) and 1 g/cm³ for water density. We performed a density correction for organic matter contents assuming a density of 1.25 g/cm³ for organic particles (Boyd, 1995): It has to be noted that we did not calculate the LIT MAR between 9.4 and 8.5 ka BP because the core was dried out in this section. 120 The wavelet power spectrum analyses of LIT MAR were carried out using a Matlab Wavelet script (Torrence and Compo, 1998).

Holocene climate change in the NE Arabian Sea realm
Changes in the alkenone producing coccolithopores community Emiliania huxleyi and Gephyrocapsa oceanica in the NE AS 125 are controlled by changes in the nutrient availability and the mean mixed layer depth (Andruleit et al., 2004). Although, the https://doi.org/10.5194/cp-2020-141 Preprint. Discussion started: 17 November 2020 c Author(s) 2020. CC BY 4.0 License. coccolithophores community in the NE AS varies over the course of the year with higher alkenone fluxes during winter and spring, Böll et al. (2014) found that the alkenone-based SST reconstruction is representative for annual mean SST. The lithogenic material at the core site is supplied by fluvial input from the Makran rivers (Forke et al., 2019;Lückge et al., 2002;von Rad et al., 2002;Staubwasser and Sirocko, 2001;Stow et al., 2002) and by aeolian input probably from the Sistan Basin 130 region in the border region of Afghanistan and Iran Forke et al., 2019;Kaskaoutis et al., 2014;Rashki et al., 2012). At present dust from the Sistan region is transported to the NE AS by northerly Levar winds during summer and northeasterly and westerly wind during winter (Hussain et al., 2005;Kaskaoutis et al., 2015;Pease et al., 1998;Rashki et al., 2012;Sirocko et al., 1991;Tindale and Pease, 1999).
The SSTs reconstructed from our record range between 25.9 and 27.9°C throughout the Holocene. Waning glacial conditions 135 in the region during the Early Holocene resulted in consistently lower SSTs until ca. 8.7 ka BP in the NE AS Gaye et al., 2018;Giosan et al., 2018) (Figure 2). Between 8.7 and 8.1 ka BP SST increased from 26.0°C to 27.2°C and was accompanied by rising LIT MAR around 8.4 to 8.3 ka BP. Coeval rapid changes have been identified across the entire IM realm (Dixit et al., 2014;Rawat et al., 2015) and mark the so-called 8.2 ka event, which was associated with a weakened ISM (Cheng et al., 2009;Dixit et al., 2014). The North Atlantic hematite stained grain (HSG) records indicate eight pronounced 140 cold events, so called Bond events, during the Holocene (Bond et al., 2001). Recent studies suggested that changes in the NAO weakened ISM of the 8.2 ka event has been linked to one of these North Atlantic cool phases (Bond event 5) and is interpreted to signal a decrease in ocean heat transport that induced a marked southward shift of the Intertropical Convergence Zone (ITCZ) (Cheng et al., 2009). Although our SST record shows no evidence of the 8.2 ka cold event in the NE AS, δ 18 O data of Globigerinoides ruber of the same core signal abruptly declining Indus river discharge around 8.4 ka BP , accompanied by rising LIT MAR indicative for strong soil erosion. It is plausible that this cold event, observed in both 150 the continental records and marine proxies, led to a short southward migration of the ITCZ and drying on the adjacent continent, resulting in a decoupling of the oceanic vs. land proxies in our core. High SSTs prevailed after the Early-Mid Holocene transition at ca. 8.1 ka BP until ca. 5.2 ka BP and are consistent with other Arabian Sea records Gaye et al., 2018;Giosan et al., 2018). This warm period encompasses the Mid-Holocene climate optimum period and is characterized by low LIT MAR and increasing fluvial input ( Figure 2) due to high precipitation rates in the North Himalayan region 155 Dallmeyer et al., 2013;Herzschuh, 2006) and a dense vegetation cover which lowered soil erosion.  Thereafter, at about 4.6 ka BP a cooling period occurred which was punctuated by moderate initial cooling event at 4.4 ka BP 165 and a rapid SST decrease of about 1°C between 4.2 and 3.8 ka BP. Foraminiferal analyses of the same core imply lower Indus river discharge and increased IWM mixing between 4.5 -4.25 ka BP (Giesche et al., 2019;Staubwasser et al., 2003) which coincides with enhanced eolian input (EM3, Figure 2). Giesche et al. (2019) found a weakening of the IWM strength from 4.1 to 3.9 ka BP and a period of cooling between 3.7 and 3.3 ka BP by using salinity sensitive foraminera as proxies. This partly contradicts our SST record. However, the δ 18 O record of G. ruber, mainly interpreted as a salinity signal, is also affected by 170 water temperature (Giesche et al., 2019;Staubwasser et al., 2003). If the δ 18 O signal were interpreted as a temperature rather than a salinity signal, Giesche et al. (2019) argued that the observed increase of the δ 18 O around 4.1 ka BP would be consistent with a surface water cooling of about 1°C. This interpretation matches our alkenone-based SST reconstruction at that time.
Our record show lower SSTs along with decreasing fluvial input prevailing until 3.0 ka BP. The marked cooling between 4.6 and 3.0 is widespread in the Arabian Sea and on the Indian subcontinent (Gaye et al., 2018;Zhao et al., 2017) and coincides 175 with the rise and fall of the Indus civilization (Wright et al., 2008).
The last ca. 3.0 ka in the record are characterized by high and variable SSTs, LIT MAR, and increasing aeolian inputs, while fluvial inputs of the Makran rivers decrease . In contrast to the smoothly decreasing river discharges, the higher SST and especially LIT MAR and aeolian inputs, reveal pronounced variations (Figure 2). These changes are mainly associated with decreasing ISM activity as well as increasing wind in the source areas of the aeolian material and variable 180 IWM strength (Böll et al., 2014Burdanowitz et al., 2019;Ivory and Lézine, 2009;Lückge et al., 2001). The main forcing mechanisms have been suggested to be the southward shift of the ITCZ due to decreasing summer solar insolation, thermal land-ocean contrast and/or teleconnection to mid-high-latitude NH climate probably via the STWJ (Böll et al., 2014Burdanowitz et al., 2019;Fleitmann et al., 2007;Giosan et al., 2018;Lückge et al., 2001;Mohtadi et al., 2016;Munz et al., , 2017. A stronger impact of NH climate is indicated by high accumulation rated during Bond events 0-2 ( Figure 2). 185 The up to 4-fold increase of LIT MAR during the Late Holocene Bond events marks a system shift towards a strong response to mid-high latitude NH climate via the STWJ. This increased precipitation in winter, but decreased precipitation during the summer. Furthermore, this change in the seasonal precipitation pattern enhanced soil erosion due to stronger erosive forces of rivers in largely derelict farmlands exposed to desertification due to an overall increased aridification trend.

Transition period from low latitudinal to increased mid-high latitudinal influence on the Makran coast 190
The smooth and long-term decrease of the summer insolation at 30°N after ca. 11 ka BP is an important driver of climate and vegetation changes in the IM and westerlies realm (Fleitmann et al., 2007;Herzschuh, 2006).It is, however, not consistent with the marked cooling event between about 4.6 and 3 ka BP and the subsequent unstable warmer period found in our record. The decreasing summer insolation cannot be the sole driver of these environmental changes even if one considers that the maximum ISM intensity lags the maximum summer insolation by about 3 ka (Ansari and Vink, 2007;Fleitmann et al., 2007;Leipe et al., 195 2014;Reichart et al., 1998;Zhang et al., 2018;Zhao et al., 2017). A decisive influence instead may be the latitudinal insolation gradient (LIG) that triggers changes of the atmospheric pressure gradient, heat transport and determines the strength and position of regional atmospheric circulation (Bosmans et al., 2012;Lee and Wang, 2014;Mohtadi et al., 2016;Wang et al., 2017) which has received little attention in paleostudies (Clemens and Prell, 2003). Winter and summer LIG are equally important drivers for the strengths, positions and areal extents of the STWJ and the ISM (Mohtadi et al., 2016;Ramisch et al., 200 2016). During the winter months (here defined as December -February), the LIG (here defined as the gradient between 60°N-30°N) determines the southernmost position of the STWJ and the frequency of the WDs (Fallah et al., 2017). The LIG (here defined as the gradient between 30°Nequator) during the summer months (here defined as June-August) affects the strength and extension of the ISM over larger timescales (Ramisch et al., 2016). The antagonism and relative magnitudes of the winter and summer LIGs influence the duration of the ISM, as well as STWJ and the frequency of the WDs. Therefore, changes in 205 LIG have probably modulated seasonal length of the IM in the course of the Holocene.
The summer LIG reached its maximum between 8 and 7 ka BP, whereas the winter LIG peaked during the last 1 ka ( Figure   2). The net effect of ISM weakening and STWJ strengthening, in combination with dry periods on the adjacent continent to the north, resulted in more frequent dust storms that increased aeolian input to the Arabian Sea during the Late Holocene . We suggest that the time period between 4.6 and 3 ka BP marks a transition from an ISM-dominated 210 climate system towards one which is more influenced by the STWJ. This strengthened the teleconnection between the Makran coast and climate variability in the North Atlantic, which is most visible by the link between the LIT MAR record and the Bond events since the end of the time period and associated fall the Indus civilization (Figure 2). A wavelet analysis of LIT MAR data reveals further evidence for the transition period with a change of periodicities between ca 4.2 and 3.5 ka BP ( Figure   3). These findings corroborate the hypothesis of a teleconnection between the mid-high-latitude NH climate via the SWTJ and 215 IWM in the NE Arabian Sea region Munz et al., 2015Munz et al., , 2017. On a more regional scale, increasing influence of the STWJ coinciding with the transition phase is documented in marine and terrestrial records from the Arabian Sea and Indian subcontinent (Giosan et al., 2018;Hou et al., 2016). Climate simulations for the last 6 ka over Iran have shown a southward shift of the STWJ during the Late Holocene due to increasing winter insolation (Fallah et al., 2017), suggesting a "tipping point" around 3-4 ka BP in the interaction of ISM and STWJ with waxing influence of the STWJ towards the present 220 day. The "tipping point" corresponds to so-called neoglacial anomalies in the Makran region, when a strong winter monsoon and weak interhemispheric temperature contrast (30°N-30°S) were suggested to have been accompanied by a decrease in ISM precipitation (Giosan et al., 2018). Significant temperature oscillations over the last 3 ka were also recorded in the continental record of Lake Qinghai, but are commonly interpreted as an amplified response to volcanic and/or solar forcing (Hou et al., 2016). The manifestation of a meridional shift in STWJ is not restricted to the Arabian Sea region. A southward displacement 225 of the STWJ during the Late Holocene has also been reported from the southern Alps in Europe, the Japan Sea and climate simulations for East Asia (Kong et al., 2017;Nagashima et al., 2013;Wirth et al., 2013), indicating teleconnections on a hemispheric spatial scale.  Several studies have suggested ENSO as an important driver modulating the climate in the IM realm during the Holocene (Banerji et al., 2020;Munz et al., 2017;Prasad et al., 2014Prasad et al., , 2020Srivastava et al., 2017). In general, an ENSO event is 235 associated with reduced ISM precipitation (Gadgil, 2003). Even though the ENSO events have apparently increased since the last ca. 5000 years (Moy et al., 2002), roughly matching the timing of our observed transition period, we cannot find a correlation of our LIT MAR record with periods of strong ENSO events (not shown). Other studies have shown that the relationship between IM precipitation and ENSO is not linear. Observations (1880 -2005) suggest that less than half of the severe droughts occurred during El Niño years, while the amount of precipitation was normal or above normal during other El 240 Niño years (Rajeevan and Pai, 2007). A possible reason for this mismatch is the strong effect of the Indian Ocean Dipole (IOD). It can act as an amplifier or suppressor of the ENSO influence in the ISM realm (Ashok et al., 2001(Ashok et al., , 2004Behera and Ratnam, 2018;Krishnaswamy et al., 2015;Ummenhofer et al., 2011). A positive IOD can counteract even a strong El Niño forcing leading to "normal" precipitation amounts (Ashok et al., 2004;Krishnaswamy et al., 2015;Ummenhofer et al., 2011).
A recent study suggested that a positive IOD shows a tripolar pattern over the IM realm, with above normal precipitation in 245 central India and below normal precipitation in northern and southern India (Behera and Ratnam, 2018). In contrast, a negative IOD creates a zonal pattern with above normal precipitation in western India and below normal precipitation in eastern India (Behera and Ratnam, 2018). In addition, these authors have linked a positive IOD to warmer SST in the northern AS and vice versa. However, we can only speculate how strong its impact was during the transition period due to the lack of paleorecords reconstructing the IOD. 250

The environmental-anthropogenic interaction during the mid-Late Holocene
Widespread anthropogenic land use, such as deforestation or rice irrigation, and associated environmental change is recorded during the Early-Mid Holocene transition period (Petrie et al., 2017;Petrie and Bates, 2017). Especially rice irrigation, starting around 6-4 ka ago, has been suggested to have increased the atmospheric CH4 concentration since about 5 ka ( Figure 4) and thus had a small, but emblematic impact on climate (Ruddiman and Thomson, 2001). The rise of the Indus civilization began 255 around 5.2 ka BP when general aridification decreased river discharges (Giosan et al., 2012). Since ca. 4 ka BP, increasing winter precipitation enabled the development of agriculture along the Indus valley due to less intense floods favoring urbanization (Giosan et al., 2012;Wright et al., 2008). The urban Harappan civilization collapsed towards the end of the climatic transition period when less frequent annual flooding occurred, which was the basis of agriculture. Most of the people migrated from the Indus valley to the Himalaya plains, whereas the remaining society turned into a post-urban society (Clift 260 and Plumb, 2008;Giosan et al., 2012Giosan et al., , 2018Possehl, 1997).
For instance, African savanna plants produce, on average, longer chain n-alkanes than rainforest plants (Rommerskirchen et al., 2006;Vogts et al., 2009). In the region of Indus civilization, natural dominated until agriculture proliferated (Giosan et al.,265 2012) driven by the changing precipitation pattern due to the interplay of the position and strength of the STWJ and ISM (Ansari and Vink, 2007) (Figure 4). During the Mid-Holocene warm period, high SSTs are thought to have fueled the moisture transport to the adjacent continent, which is in line with our SST record (Figure 2). High average summer precipitation and/or more uniformly distributed precipitation during the different seasons favored the development of grasslands in southern Pakistan (Ansari and Vink, 2007;Ivory and Lézine, 2009). This may reflected in a decrease in the ACL of n-alkane homologues 270 C27 -C33 (ACL27-33) between ca. 7 to 5 ka BP (Figure 4). The ACL27-33 increased after the transition period, although the amplitude of ACL27-33 variation is not very high, and decreasing river discharges and increasing aeolian input (Figure 2) indicate prevailing dry conditions after 3 ka BP. It is impossible to separate effects of anthropogenic modified vegetation or climatic driven vegetation changes on ACL27-33 fluctuations s as the ACL27-33 reflects to whole vegetation community.
However, but the ACL27-33 increase suggests that humans started to influence the vegetation in the sediment source region and 275 led to the decoupling of the climate signal and the ACL27-33. This is in line with the coeval decoupling of the atmospheric CH4 concentration and orbitally driven climate changes (Ruddiman and Thomson, 2001). These authors suggested that inefficient early rice farming caused atmospheric CH4 concentration to increase at around 5 ka BP. Rice agriculture began around 4 ka https://doi.org/10.5194/cp-2020-141 Preprint. Discussion started: 17 November 2020 c Author(s) 2020. CC BY 4.0 License. BP on the Indian subcontinent and near the Indus River region, adding to the atmospheric CH4 concentration (Fuller et al., 2011). Further, climatic changes during the Mid-Holocene were suggested to have changed the agricultural system in the 280 region to a "multi-cropping" (summer/winter crops) system (Petrie and Bates, 2017;Wright et al., 2008). This clearly shows that human activities not only modified the vegetation, but also started to impact climate during this time. In a modelling study land cover change associated with prehistoric cultures modified not only local climate, but climate on a broader hemispheric scale via teleconnections (Dallmeyer and Claussen, 2011). Since the transformation of natural into cultivated landscapes favors soil erosion, it is likely that this early human land-use change intensified the impact of the NAO and Bond events on the 285 sedimentation in the NE AS during the late Holocene. The latter in turn documents a clear shift of the climatic system that was associated with the collapse of and deep crises of Late Bronze Age societies in the Mediterranean, Middle East and East Asia.  The climate transition period between 4.6 and 3 ka BP impacted not only the Indus civilization but also civilizations in Mesopotamia, the Eastern Mediterranean and China (Giosan et al., 2012(Giosan et al., , 2018Kaniewski et al., 2013;Liu and Feng, 2012;295 Wang et al., 2016;Weiss et al., 1993). Today these regions are located slightly north of the northern border of the region influenced by the Asian monsoon , but where probably within the monsoon realm prior the tranisition period. The series of climatically triggered collapses started at about 4.4 ka in China and reached Mesopotamia and the Eastern Mediterranean between 3.1 and 3.2 ka BP (Giosan et al., 2012(Giosan et al., , 2018Kaniewski et al., 2013;Liu and Feng, 2012;Wang et al., 2016). This indicates that this transition period was characterized by a southward shift of the ISM realm, which in combination 300 with socio-economic responses turned into catastrophic changes for Late Bronze Age societies that thrived in the ISM/ STWJ transition zone.

Conclusions
Our study of reconstructed SST, LIT MAR and vegetation changes at the Makran coastal region suggests a climatic transition period between 4.6 -3 ka BP from a low latitude to increased mid-high latitude influence via the STWJ. Prior to this transition 305 period the region was strongly influenced by the ISM and the ITCZ. Our LIT MAR record shows an up to 4-fold increase after the transition period coincident with Bond events in the North Atlantic. We argue that after this period the North Atlantic signals of NAO and Bond events are transmitted to the Makran coast region via the southward shifted STWJ. This supports an earlier modelling study (Fallah et al., 2017) suggesting a southward shift of the STWJ and more winter precipitation near our study area in Iran between 4 and 3 ka BP, which the authors mark as a "tipping point" in that region. Besides a long term 310 drying trend since the Mid-Holocene, this transition period and the associated change in the precipitation pattern (winter vs. summer precipitation) affected the settlements and agriculture of the Indus civilization (Giosan et al., 2012(Giosan et al., , 2018Petrie and Bates, 2017). The temporal coincidence gives rise to the hypothesis that humans themselves became environmental drivers prior to the onset of the transition period (Fuller et al., 2011;Ruddiman and Thomson, 2001), instead of only being passive victims of climate change. Whether human impacts suffice to influence the global climate is an ongoing debate, but it is very 315 likely that land cover changes associated with prehistoric cultures modified at least local climate (Dallmeyer and Claussen, 2011). Consequences included a weakened local hydrological cycle that favored desertification, especially in semi-arid regions.

Data availability
The alkenone-based SST, the LIT MAR, the extended grain size used for the EM3 and n-alkane data sets of core SO90-63KA 320 will be available at PANGAEA data repository. The previous grain size and elemental data of SO-90KA are available at PANGAEA data repository (https://doi.pangaea.de/10.1594/PANGAEA.900973).

Author contribution
BG and KE designed the study. NB analyzed LIT MAR, alkenones and n-alkanes. TR analyzed grain sizes. NB, BG and TR interpreted the data. NB prepared the manuscript with input from all co-authors. 325

Competing interests.
The authors declare that they have no conflict of interest.