Circum-Indian Holocene sea-level database:A repository of distinct relative sea-level plots across the Indian Peninsula

Shrdh Menon , Pnkj Khnn ,＊, Sntnu Bnerjee

a REEFS Lab, Department of Earth Sciences, IIT Gandhinagar, Palaj, Gandhinagar, Gujarat, 382355, India

b Department of Earth Sciences, IIT Bombay, Powai, Mumbai, 400076, India

Abstract The Indian Peninsula is one of the most well-studied regions for Holocene sea-level fluctuations in the world,however,standardized relative sea-level datasets are missing.This study provides an archive of sealevel indicators (n = 162, 20 locations) along the western and the eastern sides of the peninsula, that have been used to develop Relative Sea Level (RSL) plots.Each dated sea-level indicator is recalibrated for its elevation based on tidal and tectonic correction, as well as age with reservoir correction, and have been separated into six zones based on coastal geomorphology and number of datasets.The database spans throughout the Holocene and covers sea-level depth/elevations from -45 m to +5 m from mean sea-level(MSL).Approximately 90 % of the dataset range from 8 ka to the present day.The first transgression is highly variable and identified between 8.5-8 ka BP in Gujarat(Zone 1),～5.5 ka BP in Maharashtra(Zone 2),between 8 and 7 ka BP in Tamil Nadu (Zone 4) and between 8 and 7.5 ka BP in the Bengal coasts (Zone 6).No transgression above present sea-level is observed along Andhra Pradesh (Zone 5) (no data for Kerala - Zone 3).Further, Zones 1, 2, 4 and 6 show a strong uplift component (tectonic), whereas Zone 5 exhibits subsidence during the Holocene (Zone 3-insufficient data).Based on these findings, and given the region's coastal topography and tidal components,Zones 6 and 1 will likely undergo the largest coastal inundation,followed by Zones 5, 4, 2, and 3.These insights are critical in planning future coastal inundation measures across the Indian Peninsula.

Keywords Indian ocean, Terraces, Relative sea level curve, Holocene, Arabian sea, Bay of Bengal

1.Introduction

Several studies from around the world from satellite altimetry and tidal gauge datasets have indicated a rise in global mean sea-level (GMSL) by about 3.4±0.4 mm/yr(since 1992),with the rates predicted to continue to rise in the near future(Woodworth and Player, 2003; Church and White, 2011).This rise is partly characteristic of climate change and the consequent melting of ice sheets and glaciers(Michael et al., 2019).This consequence poses a threatening scenario especially in coastal areas; with the frequency of coastal inundation and coastal overtopping predicted to surge in the coming decades(Almar et al.,2021).In particular, the coastal zones of SE Asia are vulnerable to flooding as part of sea-level rise, which will likely lead to crucial socio-economic loss as a majority of the megacities are located in these coastal margins, thereby creating greater risk (Kraas, 2008;Lincke and Hinkel, 2021).The 7500-km-long Indian coastline hosts about 14 % of the total Indian population (‘Coastal States of India’), which would be the first area affected by the consequences of climateinduced sea-level changes.Hence, emphasis on understanding the nature and amplitude of past sea-level fluctuations, specifically in decadal time periods(Khanna et al., 2017), is pertinent for informing modelers to accurately predict future scenarios and subsequently develop effective mitigation policies.

Relative sea-level (RSL) varies temporally and spatially based on eustatic, isostasy, tectonics, and geographic variability(Khan et al.,2019).Recognition of both global and regional factors in determining the sea-level change is imperative to properly gain insights into the factors affecting the change and the rates of sea-level variability.The relative sea-level change in an area is a consequence of various mechanisms that is elucidated as follows(Shennan and Horton,2002;Khan et al.,2019)-

Here, the ESL (Eustatic Sea level) refers to the global fluctuations in sea-level by changes in the ocean volume (Lambeck et al., 2014; Rovere et al., 2016;Hieronymus, 2019).It is attributed to the melting of land-based ice and variation in ocean salinity and temperature(Khan et al.,2019;Khanna et al.,2021).The mass exchange between ice sheets and the ocean and the resultant response in the solid earth drive the Glacial Isostatic Adjustment (GIA) (Clark et al., 1978;Milne and Mitrovica, 2008).The static equilibrium is associated with gravitational, rotational, and deformational processes in response to mass exchange between the ocean and the cryosphere (Milne et al.,2009; Khan et al., 2019).Together with GIA, these processes lead to spatial variations in sea-level, leading to near-field and far-field regions.The near-field regions experience a relative sea-level fall due to isostatic readjustment (unloading).In the far-field regions (tropical) the dominant mechanism is determined by the flux of water between ice and ocean reservoirs (ocean unloading) (Mitrovica and Milne,2002; Milne and Mitrovica, 2008).Another important factor contributing to the RSL is tectonics(Baranskaya et al.,2018).The rates of tectonism vary in the active and passive margins.Passive margins are generally stable with negligible rates of tectonism.However,neotectonic events could possibly reactivate previously stable rifts and they must be considered for the preparation of RSL curves.Local factors include subsidence due to compaction of sediments, or groundwater extraction, leading to sea-level fluctuations(T¨ornqvist et al.,2008;Shennan,2015;Karegar et al.,2016).

Numerous studies have been conducted regarding sea-level changes in the Indian coastline during the Holocene period (Hashimi et al., 1999; Banerjee,2000; Loveson and Nigam, 2019; Rao et al., 2020;Sharma et al., 2021).The earliest attempts on understanding the sea-level variations during the Holocene were conducted along the western coast of India(Agrawal and Guzder, 1972; Hashimi et al., 1999).Radiocarbon dating of the sediment samples from the western continental shelf provided evidence for a lower sea-level, as compared to the present (Agrawal and Guzder, 1972; Hashimi et al., 1999).These studies indicated a rise in sea-level to about 7 ka BP after which it reached the present-day level.Sealevel curves have been plotted for the regions of the eastern coast of India,primarily in the Gulf of Mannar(Banerjee, 2000), Cauvery delta (Achyuthan and Baker, 2006; Hameed et al., 2006; Thomas, 2009;Srivastava and Farooqui,2017;Goswami et al.,2019),the Godaveri delta (Rao et al., 2012, 2015, 2020; Rao et al., 2020) and the Bengal delta (Islam and Tooley,1999; Rashid et al., 2013; Sen and Banerjee, 2016).The presence of paleo-deltas indicated a rise in sealevel during the Holocene (before 7 ka BP) (Loveson and Nigam, 2019).The Godaveri delta exhibited progradation that has been attributed to fluctuations in sea level during the Holocene (Rao et al., 2012).In the Bengal basin, the sea-level was high during 7—6.5 ka BP (Sen and Banerjee, 2016).Although several studies have plotted sea-level curves for the western and eastern coasts, yet, a comprehensive, standardized study incorporating eustatic, tectonic and local effects has not been carried out in detail.Global,regional, and local factors need to be considered to produce a robust reconstruction of Holocene relative sea-level across the Indian coastline.

In this study, we review Indian relative sea-level fluctuations during the Holocene by compiling and standardizing relative sea-level index points(n=65)and limiting data (n = 97) from the eastern to the western coasts of the Indian Peninsula.This has been done as follows.

1) Critical evaluation of the literature with published sea level indicator datasets

2) Standardization of datasets: Calibration of all14C dates and standardization of the elevation as defined by standard conventions (Shennan and Horton, 2002; Khan et al., 2019)

In doing so, we attempt to achieve the following objectives-

1) Compiling and adding to the existing global database of Holocene Sea Level fluctuations around the world.

2) Identification of the variation in amplitude and trends of the Holocene relative sea-level fluctuations along the Indian Peninsula.This would enable development of accurate models to predict future sea-level trends and adopt mitigation strategies.

2.Regional setting

The Indian Plate,which encompasses most of South Asia,the Indian subcontinent,and portions of the basin under the Indian Ocean(Jain et al.,2020),broke apart from Gondwana at about 180 Ma (Ramakrishnan and Vaidyanathan, 2010) and collided with the Eurasian Plate at about 55 Ma (Kearey et al., 2009).The Indian subcontinent extends from latitudes 8°4′N and 37°6′N and longitudes 68°7′E and 97°25′E; with a coastline of approximately 7500 km length (‘Coastal States of India’).The mainland coastline extends to about 6100 km, while the islands comprise about 1200 km in length(‘Coastal States of India’).The coastline can be broadly divided into the western and eastern margins,with varying morphology, genesis, and sedimentation patterns(Ramakrishnan and Vaidyanathan,2010;Kale,2014) (Fig.1).Ramakrishnan and Vaidyanathan (2010)and Kale(2014)have elucidated the various geological and geomorphological features that have been summarized below.

2.1.Western coast (Fig.1, B, C, D)

The western coast,from the Kutch(Gujarat)in the north to Kanyakumari in the south, is sandwiched between the Western Ghats and the Arabian Sea and is entirely devoid of deltas(Kale,2014).The tidal range is approximately a meter at the southern tip of India,however that increases to several meters northwards towards Gujarat.

2.1.1.Gujarat basin (Fig.1, B)

Four interlinked basins exist at the northern tips of the western coastlines: Narmada, Kutch, Gulf of Khambhat and Saurashtra basins.These basins were postulated to have formed due to rifting along the three orogenic trends during the different stages of evolution of the subcontinent (Gombos et al., 1995;Faruque and Ramachandran, 2014).The rifting in the basin was initiated during the Late Triassic to Late Jurassic.The Cenozoic succession consists dominantly of limestone, shale, and sandstone.The Saurashtra coast is characterized by occurrences of lime mud,corals and limestone beds.The upper slope region is characterized by phosphorite deposits within a calcium carbonate matrix.The basin formation along the Saurashtra displays carbonate facies from the Paleocene to Miocene.The Gulf of Cambay and Narmada basins are dominated by estuaries and tidal flats.Studies by Hashimi et al.(1999) attest to neotectonic events in the Kutch basin; however, these have not been quantified accurately.

2.1.2.Bombay basin (Fig.1, C)

The basin is characterized by a thick succession of limestone and calcareous shale horizons, amidst arenaceous and argillaceous sediments.The Bombay shelf is constituted by tidal flats, strand plains, and carbonate ramps.The majority of the coast is covered in Deccan volcanics and overlying Cenozoic laterites and littoral deposits.The unique feature is the carbonate platform, which is covered in relict sandy carbonate sediments and siliciclastic sediments(Hashimi et al.,1999).The other prominent feature on the shelf is the Fifty Fathom Flat;covered in carbonate oolite and sand formed during the Eocene and Holocene (Rao and Wagle, 1997; Hashimi et al., 1999).

The coast is predominantly subsiding (Hashimi et al.,1999);though quantitative estimates are lacking.

Fig.1 A) Overview map of the Indian Peninsula with the representative sites depicted; the images on the right depict each of the geographical sub-zones and the sites reviewed for the Holocene sea-level change in this study;B)Gujarat(Zone 1);C)Maharashtra(Zone 2);D) Kerala (Zone 3); E) Tamil Nadu (Zone 4) F) Andhra Pradesh (Zone 5); G) Bengal (Zone 6).

2.1.3.Konkan basin (Fig.1, B)

The predominant sediments consist of an almost continuous sequence of Cenozoic sediments over the Deccan trap sediments.The sediments are predominantly limestone beds along with carbonate bioherms,oolites and relict sands.Only vestiges of Neogene sediments are found on the shelf in the landward direction.Other features include coral-algal ridges,submarine terraces and paleo-strandlines (Kale,2014).

The basin is not active and there is no evidence of neotectonic activities in the basin.

2.1.4.Kerala basin (Fig.1, C)

The bulk of the sediments in the basin have been postulated to have formed during the Cenozoic(Campanile et al.,2008).The Cochin Formation in the basin is a clastic sequence consisting of sandstones.Siltstones, shales and claystones in this basin are observed in outer neritic environments.Other formations are composed of algal wackestones, packstones and dolomites of the Late Oligocene to Middle Miocene age (NDR (National Data Repository)).The southern part of the basin is characterized by peat/lignite beds(NDR (National Data Repository)).

The basin lacks any evidence of neotectonic events and hence is considered passive.

2.2.Eastern coast (Fig.1, E, F, G)

The eastern coastal plain extends from Rameshwaram, Tamil Nadu in the south to Bangladesh in the north, encompassing a vast region lying between the Eastern Ghats and the Bay of Bengal.The tidal range varies from as low as 0—1 m in the southernmost tip of the peninsula and increases to several meters at the northern tip.

2.2.1.Cauvery basin (Fig.1, E)

The basin was formed during the Late Jurassic to Early Cretaceous period and can be attributed to the fragmentation of the Precambrian basement into several horsts and grabens (Ramakrishnan and Vaidyanathan, 2010).Initially, sedimentation was essentially siliciclastic and non-marine.Marine incursions during the Cretaceous have been inferred from the planktonic foraminifera and ammonite zones,when the rift phase terminated.The Quaternary sediments are fluvial and the basin is also characterized by mid-Holocene beach ridges and late Holocene dune sands (Alappat et al., 2010).

2.2.2.Krishna-Godavari basin (Fig.1, F)

The NE—SW-trending Krishna-Godavari basin is superposed on the Pranhita-Godavari basin, which has been considered the failed arm of the triple junction(Ramakrishnan and Vaidyanathan, 2010).The dominant lithology includes siltstones and thin limestone bands.The progradation of the delta due south-east has been inferred during the Cenozoic (Ramakrishnan and Vaidyanathan, 2010).The delta is constituted by beach ridges, mudflats and mangrove swamps.The deposition of beach ridges indicates Holocene transgression limit (Kakani et al., 2020).Evidence of postsediment compaction of 1.1 mm/year has been evidenced by archaeological records (Kakani et al.,2020).

2.2.3.Bengal basin (Fig.1, G)

The polycyclic basin is flanked by the Indian shield in the west and the Shillong massif in the north (Ramakrishnan and Vaidyanathan, 2010).It is characterized by tectonic subsidence and refilling by the rivers (Akter et al., 2015).The basin has evolved as a result of various tectonic activities.The basin is characterized by shale-sandstone and siltstone formation over the Rajmahal traps.The south-eastern part of the basin is covered by older alluvium while the southern part has been covered by younger alluvium and deltaic sediments (Sen and Banerjee, 2016).

Various tectonic and seismological studies have inferred subsidence, while sediment compaction has also contributed to the overall subsidence of the deltaic basin (Stanley and Hait, 2000).Anthropogenic changes, particularly in the form of groundwater extraction, have exacerbated the subsidence in the delta, with an average rate of up to 5.2 ± 1.2 mm/yr(Becker et al., 2020).

3.Material and methods

3.1.Archived data and database template

A sea-level indicator database has been developed based on standardized methodologies(Shennan,2015;Khan et al., 2019) (See Table 1 for reference locations).The database consists of fields that are used to archive in detail the age, location and elevation(including tectonic and tidal corrections).This database is the first repository of the published sea-level indicators, that could later be modified (for example,with respect to a certain city/state,or type of sediment, parts of the data could be easily extracted,or additional data could be added in future).The RSL indicators utilized include corals, shells, lithified carbonates, carbonate sediments from beach ridges,peat, soil, plant material and charred material from mangrove swamps.

3.2.Compilation of data

The compilation of the data and necessary correction has been carried out as defined and formalized during the Geological Correction Program(IGCP) Project 61 (Shennan and Horton, 2002;Engelhart and Horton, 2012; Mann et al., 2019; Khan et al., 2019).Corrections and uncertainties are taken into consideration for the accurate measurement of RSL.The sea-level indicators have been classified as index points, marine limiting and terrestrial limiting points based on their elevation with respect to the tidal datum (Hijma and Cohen, 2015 and references therein;Shennan and Horton,2002;Khan et al.,2019).The steps followed to standardize the database are as follows.

●Separating original data from interpretations from the published literature.Only the elevation/depth of each sample along with the uncorrected14C age, make up the original dataset.

●All the samples are categorized into index or terrestrial/marine limiting points (Table 2).

●The radiocarbon dates (14C) have been calibrated with CALIB 8.2 (Stuiver et al., 2021)using the Marine20 calibration curve to correct for the reservoir age.The terrestrial data points have been calibrated using IntCal20(Stuiver et al., 2021).

●To standardize and correct for the elevation,the procedure of Hijma and Cohen, 2015 is followed.Here, IR represents the elevation range occupied by the index and sea level limiting points relative to tidal datums.Further, tectonic corrections are applied to each sample (if available).

3.3.Reconstruction of RSL

The standardization and identification of vertical(elevation) and horizontal (age) errors are critical to plotting a relative sea-level curve.The samples were categorized and the Indicative Meaning (IM) was derived based on the location of the sample withrespect to the present tidal datum.For instance,shells from lagoonal sands were designated as index points.Likewise, sediments derived from floodplains were designated as terrestrial limiting points.In the case of mangroves, literature yielded no particular information on the species or the tidal data range of the modern analogs of mangroves.Hence, in this study, these were designated as TL as most of the mangroves have been derived from peat and mangrove swamps.The IR of the index points is calculated by taking into account the HAT and LAT levels; while the terrestrial and marine limiting points are calculated solely based on HAT and LAT,respectively.Further, based on uplift or subsidence,further addition or subtraction is conducted to standardize the elevation of a sample/data point.The elevation of each sample is an aggregate of all the elevations calculated, viz, after tectonic correction.The associated uncertainties are calculated as the root mean square of all the uncertainties of a location(Khan et al.,2019).Data points that did not have elevation data have been rejected.The age corrections were applied to all the data points(Stuiver et al., 2021), and calibrated 2σ (95 %probability) values and associated uncertainties are utilized.

Table 1 Summary of reviewed sea level studies and respective locations across the Indian Peninsula.

Table 2 Sea-level indicators(Index points,marine limiting points and terrestrial limiting points)and their indicative meanings.Highest Astronomical Tide(HAT),Lowest Astronomical Tide(LAT).

4.Results

A total of 162 sea-level indicators have been standardized, which are characterized by 80 terrestrial limiting points, 65 index points and 17 marine limiting points (Fig.2).Shells, mostly gastropods and bivalves, comprise the majority of the index points,while carbonate sediments are the other dominant indicators classified as index points (IP).Peat, plant material, carbonized wood and charred material constitute the terrestrial points (TL) while the corals are classified as marine limiting points (ML).

The data spans throughout the Holocene and covers signals in the range of -45 m to +6 m MSL.Approximately 90 % of the sea level indicator dataset range from 8 ka to the present day.It is mostly because the samples collected in this study are currently above the present sea-level.The datasets older than 8 ka are provided in Supplementary Material (SF 1).

4.1.Characterizing Indian Peninsula into zones/segments

Fig.2 A)Age-elevation plot of the sea-level indicators database;B)Frequency distribution of the index,terrestrial and marine limiting data points reviewed in the study (IP-Index point, ML-Marine limiting point, TL-terrestrial limiting point).

Based on clustering of the available datasets(published literature), as well as the salient features of the coastal regions, the Peninsula has been subdivided into 6 zones(Zones 1,2,3,4,5,and 6;Fig.1,Table 1).A total of 20 clusters of dated sea-level indicators could be identified from these zones cumulatively.Some of the zones have a higher number of data points and some very few, that is owing to the number of studies published for a certain region(Fig.1).The western margin is divided into three zones — Zone 1: the Gujarat coast, Zone 2: the Maharashtra coast and Zone 3: the Kerala coast.The eastern margin is divided into three zones — Zone 4:Tamil Nadu coast; Zone 5: Andhra Pradesh coast, and Zone 6: Bengal coast.There are no datasets present along Orissa coastlines.

4.2.Western margin

4.2.1.Zone 1 - Gujarat coast

The Gujarat coast zone consists of the Gulf of Khambhat and Saurashtra basins.Eleven data points representing five locations have been identified from the published literature (Agrawal and Guzder, 1972 Hashimi et al., 1995; Banerji et al., 2015; Table 1)along the Gujarat coast.Amongst the dataset, five index points were identified, which include shells and carbonate sediments (Agrawal and Guzder, 1972;Banerji et al., 2015), two marine limiting points from corals(Agrawal and Guzder,1972),and four terrestrial limiting points from beach strandlines(Hashimi et al.,1995).The datasets range in age from 8.5 ka to about 2 ka,with the RSL varying from 0 to 5 m MSL(Fig.3).The first transgression is observed before 8 ka.Following the transgression,the Holocene highstand is observed from 8—2 ka.The peak of the Holocene highstand is identified at approximately 6 ka BP, ～4 meters above sea level, after which the sea-level started to fall(note—Index points are used to approximate the peak timing rather than the terrestrial limiting points).No data is available from 0-2 ka.These observations from the entire Gujarat coast have provided a generalized signal that might or might not be true for all the segments of the Gujarat coast,as it is still active and has four distinct basins.Even within the same segments along the Gujarat coast, e.g., Saurashtra, the RSL during the mid-Holocene, has been interpreted to be 2—3 m high (Juyal et al., 1995; Sharma et al., 2021).However,Banerji et al.(2015)suggested that the Mid-Holocene sea-level in the Saurashtra Peninsula was～1 m higher than the present after the corrections related to tectonic upliftment.Along another segment of the Gujarat coast-The Gulf of Kutch,maximum sea levels have been interpreted to be～2-m-high sea-level between 6 and 3 ka(Das et al., 2017).

4.2.2.Zone 2 - Maharashtra coast

The western coast of India, south of Gujarat, is represented by the Maharashtra,Goa,Karnataka,and Kerala coastlines that have been split into two zones(Zone 2 and 3).Based on the available datasets and their clusters, Zone 2 is categorized by Maharashtra,which comprises nineteen data points,represented by four locations (6,7,8,9 in Fig.1).This dataset comprises seventeen index points and two terrestrial limiting datapoints (Hashimi et al., 1995).The index points are prominently lithified carbonate sediments and shells, while the terrestrial limiting points are wood samples (Hashimi et al., 1995).No marine limiting points have been identified and dated from this zone.The dataset ranges in age from 5.5—1 ka,and vertically lies from -3 to +4 m MSL.

Fig.3 Age-elevation plots of the RSL indicators from each geographical sub-region.A)Zone 1;B)Zone 2;C)Zone 3;D)Zone 4;E)Zone 5;E)Zone 6.The green bands correspond to the transgression of the corresponding regions while the blue bands represent lack of data or datapoints below present MSL.

Even though the plot for Zone 2 (Fig.3) does not depict the exact timing of the first sea-level transgression,it is proposed to be～5.5 ka based on the available datasets.Following the transgression, a mixed signal is observed for Zone 2 Holocene highstand, wherein a high degree of vertical variation is observed between 3 and 1 ka.Since the signal is highly variable spatially,further separation of the dataset in Zone 2 is required.

The zone has been further subdivided into N Maharashtra(Zone 2a;Fig.4),S Maharashtra(Zone 2b;Fig.4)and Karwar basin(Zone 2c;Fig.4).Zone 2a consists of the majority of the points,ranging in age from 1 to 5.5 ka,followed by Zone 2b ranging in age from 1 to 2.5 ka,and Zone 2c from 1 to 2.5 ka.The RSL plot for Zone 2a represents a clearer signal of Holocene highstand from approx.6—1 ka,with some internal variability,whereas,Zone 2b is still complex with a few index points below and a few higher above MSL (paucity of datasets).In Zone 2c, a cleaner signal is observed from 2.5—1 ka,demonstrating a fall in sea level.However, this clean signal in Zone 2c might be due to the paucity of dataset.

4.2.3.Zone 3 —Kerala coast

Three sea-level indicators have been identified from this zone on the coast of Kerala that is Holocene in age(Agrawal and Guzder, 1972.Two are index points,that are based on shell dating, and one is a terrestrial limiting point i.e., wood (Agrawal and Guzder, 1972).The RSL database ranges from 7 to 5.5 ka,and varies in depth from-7 to-2 m below MSL.Due to the lack of a dataset, it is not possible to derive any conclusion for the first transgression or the Holocene highstand.

4.3.Eastern margin

4.3.1.Zone 4 — Tamil Nadu coast

The dataset in this zone primarily comes from the Cauvery delta and the coast of Rameshwaram(Vaz and Banerjee, 1997; Banerjee,2000;Achyuthan and Baker,2006; Hameed et al., 2006; Thomas, 2009; Srivastava and Farooqui, 2017; Goswami et al., 2019).A total of forty-five sea-level indicators have been identified and dated,where seven are index points(Vaz and Banerjee,1997;Achyuthan and Baker,2006;Hameed et al.,2006),thirteen are marine limiting points (Banerjee, 2000;Achyuthan and Baker, 2006), and twenty five are terrestrial limiting points (Vaz and Banerjee, 1997;Achyuthan and Baker, 2006; Hameed et al., 2006;Thomas,2009;Srivastava and Farooqui,2017;Goswami et al., 2019).The dataset ranges in age from 8 to 0.5 ka,and vertically from-5 to+6 m from MSL.Based on the plots (Fig.3), the first transgression is identified between 8 and 7 ka,following which a Holocene highstand is observed until 3 ka.The RSL appears to have dropped from 2.5 ka below present-day sea-level upto a few meters with a renewed rise～1 ka to arrive at present day sea-level.In order to identify cleaner signals for the dataset in this zone,it is further divided into Zone 4a,4b and 4c(Fig.4).

Fig.4 Age—elevation plots of the RSL indicators from the sub-regions within A)Zone 2 with 2a,b and c);B)Zone 4 with subzones a,b,and c.The green bands correspond to the transgression of the corresponding regions while the blue bands represent a lack of data or datapoints below present MSL.

In Zone 4a, the first transgression above the present sea-level is observed at about 7 ka,following which the Holocene highstand is observed until a few hundred years ago.This seems to be the zone with the longest duration of Holocene highstand across the Indian Peninsula, where the sea levels were up to 4 m above present sea-level.Within Zone 4b too, a prominent highstand is observed from 6 ka till about 3 ka.Within Zone 4c,the dataset ranges in age from 8 ka to recent, with RSL ranging from -4 to +4 m from mean sea-level.The first transgression (just one data point, not indicative of a definite transgresion) is observed from 2 ka, with no prominent Holocene highstand.

4.3.2.Zone 5 — Andhra Pradesh coast

This zone primarily consists of sea-level indicator data from Krishna-Godavari delta and consists of 64 data points:34 index points and 30 terrestrial limiting points(Rao et al.,2012,2015,2020;Rao et al.,2020).Shells of gastropods and bivalves have been identified as index points, and plant material, wood, peat and charred material as terrestrial limiting points,respectively (Rao et al., 2012, 2015, 2020).The samples range in age from 11 ka to the present, that lie from -45 m to +2 m.Even though there is ample sealevel data from this zone, it displays a high degree of variation both spatially and temporally, especially from 7.5 ka to the present(Fig.3).We suggest that the datasets are of low quality, and due to differential deltaic subsidence do not represent any clear trends.

4.3.3.Zone 6 —Bengal coast

The eighteen sea-level indicators dataset within this zone is from Bengal coastline and are exclusively terrestrial limiting points,represented by wood,peat,plant material and charred material(Islam and Tooley,1999; Rashid et al., 2013; Sen and Banerjee, 2016).The sea-level indicators range in age from 8 to 1 ka,with the RSL fluctuating from -1 m to +5 m MSL.The first sea-level transgression is observed approx.between 8-7.5 ka,following which a Holocene highstand is identified from 7.5—1.5 ka.Since no datasets are available younger than 1.5 ka,it is not conclusive as to when the Holocene highstand ended within this region.

4.4.RSL variability and common trends across the Indian Peninsula

Even though the Indian Peninsula is currently a passive margin, a high degree of spatial and temporal variability in Holocene sea-level positions has been observed.The spatial and temporal variability and common trends have been summarized below.

• In Zone 1 and Zone 6,the first transgression occurred at approx.8 ka,which is the earliest amongst other Indian Peninsula zones.These zones also represent the location of the longest recorded Holocene highstands across the Indian Peninsula.In Zone 1, the signal is a product of eustatic sea-level rise and tectonics (uplift), whereas in Zone 6, the zone has always been predicted to be undergoing high subsidence due to the Ganga Brahmaputra Delta.In essence, due to coastal subsidence, the sea-level signal should have been lower than the present sea-level during the Holocene for Zone 6,however,possibly due to the high sediment supply(Allison and Kepple, 2001), and a potential local uplift, the RSL signal seems to be higher.It should be noted that all the dataset for Zone 6 is terrestrial limiting point only, thus, it is highly likely that the sea level was lower than what is observed in the plot.

• Zone 2 showcases a high degree of lateral variability along the Indian coastline.Zones 2a and 2c represent mostly higher than present sea level during the Holocene,whereas Zone 2b represents a mixed signal, with sea-level position mostly below the present level.Zone 4 demonstrates a similar spatial variability as in Zone 2, where Zone 4a and 4b, represent mostly higher than present sea level during the Holocene, and 4c represent lower sea level.One way to explain these plots is that there has to be a tectonic component associated with the eustatic signal.However, a contrast between Zone 2 and Zone 4 is that parts of Zone 4c are along river deltas.Thus, the signal for Zone 4 (in particular Zone 4c) is a product of eustatic signal as well as delta-related sea-level variability.

• Zone 3 does not have sufficient dataset to provide any conclusions, except that the sea-level transgression did not occur until 6 ka.

• Zone 5 is similar to Zones 3 and 4 in terms of spatial variability,i.e.,5a and 5c represent relatively higher sea-levels in the Holocene than 5b.However, sealevel never appears to have exceeded present day levels.There seems to be a clear signal of local subsidence(related to the Krishna+Godavari delta).

5.Discussion

This study has substantially expanded the compiled sea-level indicator datasets for the circum-Indian Peninsula to develop RSL curves.These plots demonstrate distinct sea-level signatures of the different segments of the Indian Peninsula over the past 8 ka.The variable sea-level trends along the eastern and western coasts are a consequence of the complex interplay of global and local factors,including eustasy,GIA, coastal topography, local tectonics (uplift vs subsidence) and sediment compaction.Based on the observations from these plots, it is clear that the transgression during the Holocene was complex spatially and temporally.The Holocene highstand also appears to be highly variable in its duration along the Indian Peninsula.It should be noted that there are still gaps along the Orissa coast, as well as more datasets are required in Zones 2b, 2c and 3.

In the near future, sea-levels around the world have been predicted to rise(Church and White, 2011;Michael et al.,2019)and the Indian coast is especially vulnerable to the risks associated with sea-level rise due to the large population and infrastructure near the coastline.The coasts that are currently undergoing tectonic uplift will observe a relatively slower sealevel rise, while the coasts currently subsiding will expect a relatively higher sea-level rise as compared with the global mean sea-level rise.This will be further influenced by the coastal topography and tides, with areas having lower coastal gradients and higher tidal ranges experiencing larger coastal inundation.The Holocene sea-level fluctuations across the Indian Peninsula as well as the potential future rise are discussed in the section below.

5.1.Western margin

The western Indian Peninsula RSL plots demonstrate that Zone 1 was inundated by a marine transgression in the early Holocene approx.around 8.5—8 ka and in Zone 2 approx.by 5.5 ka (Zone 3 - inconclusive based on available datasets).Here, successively, the northern parts of the western coasts were inundated first, followed by the central and southern(most likely) parts of the western coastline.The eustatic signal would have been similar along the entire western margin of India.However, tectonics(uplift/subsidence),subsidence by sediment loading of the shelf,and GIA would have influenced the RSL in the three zones.

Tectonics and GIA: Except in Zone 1 where neotectonics events have been recorded and is considered tectonically active (Rao et al., 1996; Hashimi et al.,1999; Faruque and Ramachandran, 2014, both Zone 2 and Zone 3 coasts are considered to be a passive margin,that are 1200—1500 km away from the closest plate boundary.West of Zone 1 is a strike-slip fault between the Indian and Eurasian plates named as Chaman Fault(Barnhart, 2017).Zone 1 has recorded neotectonic eventsand theRSLfrom theregionindicates sealevel up to 4 m higher than present at～6 Ka.Further,modelled GIA from Milne and Mitrovica(2008),demonstrates that the western Indian coast could have experienced up to 5 m of uplift due to glacio-hydro-isostasy since past 6 ka.This would support higher sea-level up to 5 m above MSL in Zones 1 and 2 due to GIA and tectonic uplift.However,resolving these components individually is out of the scope of this study.

Coastal Geomorphology: The coastal carbonate sediments (excluding the Gulf of Cambay) are found along Zone 1,followed by rocky beaches with estuaries,headlands in Zone 2,and the brackish lagoons,lakes and backwaters towards south of Zone 2 and in Zone 3(Kale,2014;Nair et al.,2018).Thus,the coastal areas and the adjacent basin fills would be relatively different in these zones, leading to differential compaction and variable subsidence rates(if any),however,these have not been quantified yet.Thus,the sea-level fluctuation in Zone 2 could be explained in terms of local factors,including sediment compaction.Zone 3 has insufficient data to draw any conclusions.

5.2.Eastern margin

The eastern Indian Peninsula RSL plots demonstrate that Zone 6 was inundated by coastal flooding first,approx.around 8—7.5 ka, followed by Zone 4 approx.between 8—7 ka, whereas it might have not risen higher than the present sea-level in Zone 5.Thus,here, the northern parts of the eastern coasts were inundated first,followed by the southern part,and no coastal inundation is observed in the central part of the eastern coastline.The eustatic signal would have been similar along the entire eastern coastal margin,however,tectonics,subsidence by sediment loading of the shelf, and GIA would have influenced the RSL datapoints in the three zones.

Tectonics and GIA: Zones 4 and 5 are sandwiched between the Eastern Ghats and the Bay of Bengal,which are at least 1000 km away from a plate boundary.However,due to the ongoing collision of the Indian and Eurasian plates,Zone 6 is still tectonically active;though the rates haven't been adequately constrained.Furthermore, unlike the western peninsula, the Modelled GIA values from Milne and Mitrovica (2008),demonstrate that the eastern Indian coast could have experienced up to 2 m of uplift over the past 6 ka;although any uplift could be a consequence of both GIA and tectonics.Resolving these components individually is out of the scope of this study.

Coastal Geomorphology: Along the eastern peninsula, the southern part has abundant coral reefs and sea grass and northward moving deltas(e.g.,Cauvery,Godavari deltas),mudflats,lagoons(Rao et al.,2020),and mangrove swamps.Zone 4a exhibits relatively clean signals and would likely follow the global mean sea-level.However, Zone 4c, which forms part of the Cauvery Delta is constituted by mangrove swamps and mudflats.The subsequent sediment compaction might have played a significant role in the variations in RSL.Moreover,Zone 5,i.e.,the Andhra coast,has exhibited subsidence rates of 1.1 mm/year which have been evidenced by archaeological records (Kakani et al.,2020), suggesting the variable trend of the RSL.This could also exacerbate coastal inundation in this zone in the near future.The northernmost margin of the eastern peninsula is represented by the tidedominated Sundarbans deltas, encompassing the Bengal lowland and Bangladesh, which consists of several tidal creeks and large mangrove swamps, dune complexes, estuaries and islands (Goodbred and Kuehl,2000; Kale, 2014).Thus, the coastal areas and the adjacent basin fills would be relatively different in these zones, leading to differential compaction and variable subsidence rates.Specifically, along the Ganga-Brahmaputra delta, where highly variable subsidence rates from 1 to 10 mm/yr (Steckler et al.,2022) have been observed.The subsidence rates could substantially influence future sea-level rise rates along deltas.Thus, based on the coastal geomorphology it is clear that Zone 5 (Andhra coast) will foresee coastal inundation, followed by Zone 4 (Tamil Nadu coast).

5.3.Indian islands (Lakshadweep; Andaman and Nicobar)

The Holocene RSL fluctuations for the Lakshadweep and the Andaman and Nicobar Islands have not been studied as extensively as along the Indian Peninsula.Coral terraces and beach rocks have been observed in both these locations (Lakshadweep;Fieldwork,2022-Khanna et al.;the Maldives—Rovere et al.,2018)however dating of the sea-level indicators has not been extensively conducted.The Lakshadweep Archipelago forms part of the passive Laccadive -Chagos ridge (Siddiquie, 1975), and hence the sealevel indicators would yield an accurate estimation of the eustatic sea-level change during the Holocene.Echo sounding studies conducted by Siddiquie (1975)have yielded submarine terraces at depths of 10—15 m and 43—47 m on Kadmat Island, although these have not been dated.In Siddiquie(1980),storm beaches in the eastern shores of islands have been dated to 3—2 ka BP.However, these have been attributed to periodic cyclone events, rather than changes in sea level.Beach rocks at elevations of 2 m have also been observed in the southern part of Kavaratti Island(field observations)athough these are yet to be dated.These could potentially help infer the RSL changes in the Lakshadweep Islands and in the future add to the global dataset.

The Andaman and Nicobar Islands are considered to be a part of the outer arc ridge of the Andaman-Sunda subduction arc(Rajendran et al.,2008).Coral terraces have been dated;however,these have been attributed to the tectonic uplift and evidence of earthquakes in the area.The youngest terrace at Hut Bay, dated 620 ± 160 cal yr BP, has been attributed to a tectonic uplift (Rajendran et al., 2008).The Interview Island yielded older coral terraces of 7500 cal yr BP, with an elevation of 7 m (Rajendran et al., 2008).Both tectonics and eustatic sea-level rise have contributed to the formation of these terraces.Further studies in Andaman and Nicobar Islands are required to accurately constrain the sea-level histories and to further add to the global dataset.

6.Conclusions

Proper understanding of the past sea-level variability and the dominant driving processes should be useful for developing mitigation strategies in the scenario of rising sea-level.This study presents the first holistic compilation of the sea-level fluctuations (RSL plots)during the Holocene across the Indian Peninsula.Even though the Indian Peninsula is currently a passive margin, a high degree of spatial and temporal variability in Holocene sea-level positions has been observed.The spatial and temporal variability in the relative sea-level curves is summarized below.

1) The RSL plots across the Indian Peninsula is highly variable and extends from-45 m to+5 m MSL.The first transgression is identified between 8.5-8 ka BP in Gujarat(Zone 1),～5.5 ka BP in Maharashtra(Zone 2),8-7 ka BP in Tamil Nadu(Zone 4)and 8-7.5 ka BP along the Bengal coast (Zone 6).

2) Early-mid Holocene transgression and mid-Holocene highstand over present MSL is identified along the Gujarat, Maharashtra, Tamil Nadu and Bengal coasts (Zones 1, 2, 4, and 6), whereas no transgression or mid-Holocene highstand in Kerala and Andhra Pradesh coasts has been identified(Zones 3 and 5).

3) The RSL plots for Zones 1, 2, 4 and 6 demonstrate eustatic and uplift control (tectonics + GIA),whereas Zone 5 demonstrates eustatic and subsidence (due to delta) control during the Holocene(Zone 3-insufficient data).

Author statement

SM and PK wrote the manuscript.SB wrote parts of the manuscript.SM developed the sea level indicator database.PK edited and reviewed the sea level indicator database.PK and SB reviewed the manuscript.SM and PK are joint first authors.

Conflict of interest

The authors express no conflict of interest.

Acknowledgements

The authors would like to acknowledge financial support from IIT Gandhinagar — grant number: IP/IITGN/ES/PK/2122/31 and SCIENCE & ENGINEERING RESEARCH BOARD (SERB) project number SRG/2022/000514 for undertaking this research.Without their support this research wouldn't have been possible.