Introduction
Paleoclimate reconstructions spanning timescales from past geological history provide fundamental insights into the Earth system’s climate dynamics. Reconstructing climate variability over the past three millennia is particularly valuable for understanding natural fluctuations on decadal to centennial timescales and distinguishing them from the increasing influence of human-induced climate change, thereby improving projections of future climate trajectories.
The Indian summer monsoon (ISM) is one of the most significant atmospheric systems on Earth. Changes in the monsoon have a major impact on both the environment and human societies across the Indian subcontinent and adjoining regions. The ISM exhibits variability on decadal, centennial, and millennial timescales as part of the broader Asian monsoon system (Rajeevan et al., Reference Rajeevan, Gadgil and Bhate2010). It provides around 75–90% of the total annual rainfall in the region, mostly between June and September (India Meteorological Department, 2023). To better understand these long-term changes, scientists use palaeoclimate records, which offer valuable information of monsoon behaviour in the past. These records help us to understand the natural drivers of monsoon variability and improve our ability to predict future changes, especially in the context of ongoing human influence on the climate (e.g., Gasse et al., Reference Gasse, Fontes, Van Campo and Wei1996; Thompson et al., Reference Thompson, Yao, Davis, Henderson, Mosley-Thompson, Lin, Beer, Synal, Cole-Dai and Bolzan1997; Fleitmann et al., Reference Fleitmann, Burns, Mudelsee, Neff, Kramers, Mangini and Matter2003).
The Holocene epoch (∼11,700 cal yr BP) represents the current interglacial period and is characterized by the progressive development of human civilizations under relatively stable yet variable climatic conditions. High-resolution multiproxy records from the Holocene indicate sudden shifts between warm and cold periods, reflecting the climate system’s sensitivity to multiple forcing mechanisms (Lamb, Reference Lamb1965; Hodell et al., Reference Hodell, Curtis and Brenner1995; Trivedi et al., Reference Trivedi, Saxena, Chauhan, Sharma, Farooqui, Nautiyal, Yao, Wang, Li and Tiwari2019; Nag et al., Reference Nag, Trivedi, Farooqui and Morthekai2025). Mid-Holocene (∼8200 cal yr BP) climate variability has been attributed to the redistribution of solar energy driven by orbital-scale changes, which in turn influenced the temporal and spatial migration of the Intertropical Convergence Zone (ITCZ) (Yancheva et al., Reference Yancheva, Nowaczyk, Mingram, Dulski, Schettler, Negendank, Liu, Sigman, Peterson and Haug2007).
The Late Holocene, spanning approximately the past 4200 years (Cohen et al., Reference Cohen, Finney, Gibbard and Fan2013), is a significant period for reconstructing past climate variability and its impacts on human and environmental systems (Trivedi et al., Reference Trivedi, Agrawal, Sharma, Ali, Manoj, Nag, Misra and Kawsar2024a). This interval encompasses several well-recognized global climatic phases, including the Roman Warm Period (RWP; ∼2500–1600 cal yr BP), the Dark Ages Cold Period (DACP; ∼1600–1050 cal yr BP), the Medieval Warm Period (MWP; ∼1050–600 cal yr BP), and the Little Ice Age (LIA; ∼600–150 cal yr BP). These alternating warm and cold episodes reflect complex interactions between natural climate forcings and regional environmental responses.
Palynology, the study of pollen and spores (Traverse, Reference Traverse2007), serves as a critical tool for taxonomic identification, typically at the family, genus, or species level and plays a significant role in reconstructing past vegetation dynamics and climate–culture interactions (Erdtman, Reference Erdtman1943; Nayar, Reference Nayar1990; Lu et al., Reference Lu, Li, Wang, Xu, Ferguson, Trivedi and Xuan2018; Trivedi et al., Reference Trivedi, Agrawal, Sharma, Ali, Manoj, Nag, Misra and Kawsar2024a).
As a major component of the Core Monsoon Zone (CMZ), Madhya Pradesh lies in a climatologically active Indian monsoonal belt, spanning latitudes 18°N to 28°N and longitudes 65°E to 88°E, where interannual rainfall variability during the peak monsoon months (July–August) closely reflects ISM intensity (Gadgil, Reference Gadgil2003; Rajeevan et al., Reference Rajeevan, Gadgil and Bhate2010; Sabeerali et al., Reference Sabeerali, Ajayamohan and Rao2012). This climatic context supports high ecological heterogeneity, rendering Madhya Pradesh one of central India’s key biodiversity reservoirs. Within this landscape, the Bandhavgarh Tiger Reserve (BTR), positioned in the northeastern region of the Madhya Pradesh state within the Vindhyan hill range, represents a site of both ecological and archaeological significance (Srinivas and Lahiri, Reference Srinivas and Lahiri2024). Geographically located between latitudes 23°30′12″N and 23°45′45″N and longitudes 80°47′15″E and 81°11′45″E (Nath, Reference Nath2000), Bandhavgarh harbours tropical deciduous forests, a diverse wildlife assemblage, and one of India’s highest tiger densities (Nath, Reference Nath2000). Additionally, the region is replete with cultural and archaeological relics, providing an ideal natural archive for examining human–climate–vegetation interactions across the Late Holocene.
Extensive studies of well-dated high-resolution palaeoclimate records are essential for understanding Late-Holocene climate variability and for refining future climate projections (Wilcox, et al., Reference Wilcox, Thompson, Booth and Nicholas2007). Such records provide valuable insights into past climatic behaviour, which strengthens the reliability of climate modelling and forecasting. Understanding the long-term interactions between environmental change and human societies is essential for interpreting contemporary ecological transformations and developing strategies for sustainable ecosystem management. A wide range of terrestrial archives, including lake sediments, peat deposits, speleothems, and fluvial records, have revealed pronounced decadal to centennial-scale oscillations in monsoon intensity (Yancheva et al., Reference Yancheva, Nowaczyk, Mingram, Dulski, Schettler, Negendank, Liu, Sigman, Peterson and Haug2007; Zhang et al., Reference Zhang, Kong and Yan2009; Cook et al., Reference Cook, Anchukaitis, Buckley, D’Arrigo, Jacoby and Wright2010; Sinha et al., Reference Sinha, Stott, Berkelhammer, Cheng, Edwards, Buckley, Aldenderfer and Mudelsee2011a; Quamar and Chauhan Reference Quamar and Chauhan2012; Saxena et al., Reference Saxena, Prasad and Singh2013; PAGES 2k Network, 2013; Quamar and Nautiyal, Reference Quamar and Nautiyal2017; Chauhan et al., Reference Chauhan, Bhandari and Sharma2020; Quamar, Reference Quamar2022; Sarkar et al., Reference Sarkar, Sengupta, Ambekar, Bhushan, Dimri, Deshpande-Mukherjee and Juyal2024; Trivedi et al., Reference Trivedi, Kotlia and Joshi2011, Reference Trivedi, Saxena, Chauhan, Sharma, Farooqui, Nautiyal, Yao, Wang, Li and Tiwari2019, Reference Trivedi, Chauhan and Sharma2020a, Reference Trivedi, Agrawal, Sharma, Ali, Manoj, Nag, Misra and Kawsar2024a). Complementary evidence from marine sediment cores further supports the presence of persistent variability in ocean–atmosphere interactions linked to monsoonal circulation (Anderson et al., Reference Anderson, Overpeck and Gupta2002; Newton et al., Reference Newton, Thunell and Stott2006; Oppo et al., Reference Oppo, Rosenthal and Linsley2009; Cobb et al., Reference Cobb, Charles, Cheng and Edwards2003).
Forested landscapes in ISM-dependent regions, such as central India, are particularly sensitive to both climatic variability and anthropogenic disturbances. This study reconstructs the environmental history of the Bandhavgarh region over the past ∼2700 years, focusing on vegetation dynamics inferred from palynological evidences, reflecting ecological conditions. The palaeoclimatic interpretations, primarily derived from pollen analysis, are corroborated using outputs from Earth system palaeoclimate simulation (ESPS) models, enhancing the robustness of the inferred climatic trends. By integrating palaeoecological data and climate model simulations, the study offers a comprehensive perspective on ecological transformations in the region.
Regional setting
Study site
The BTR is located in the Umaria district of Northeastern Madhya Pradesh, India (24°37′N–24°46′N; 80°54′E–81°7′E) (Fig. 1). A digital elevation model (DEM) derived from Cartosat-1 data shows that the reserve encompasses an area of approximately 1536 km2 and spans an altitudinal range from approximately 390–750 m above mean sea level (Fig. 2) (Rodgers et al., Reference Rodgers, Panwar and Mathur2002). The region is dominated by tropical moist deciduous forest formations, and represents one of the ecologically significant biogeographic zones of central India.
Digital elevation model (DEM) of the BTR derived from CGIAR-SRTM data, showing the location of the study site in the CMZ, Madhya Pradesh, India.
Figure 1 Long description
The map consists of two parts. The left part shows a map of India with a highlighted region in Madhya Pradesh, indicating the study site. The right part is a digital elevation model of the highlighted area in Madhya Pradesh. The elevation is categorized into ranges: 89 to 230 meters, 230 to 317 meters, 317 to 387 meters, 387 to 450 meters, 450 to 505 meters, 505 to 576 meters, 576 to 664 meters, 664 to 760 meters, 760 to 883 meters and 883 to 1208 meters above mean sea level. The study site is marked with a circle symbol. The map includes a scale bar indicating distances in kilometers and a compass rose for orientation.
Three-dimensional elevation model of the BTR, located in the CMZ, Madhya Pradesh, India, derived from Cartosat-1 data.
Figure 2 Long description
LEGENDS: 1) Elevation in AMSL: 390-430, 430-470, 470-510, 510-550, 550-590, 590-630, 630-670, 670-710, 710-750. 2) Study Site. The map is a three-dimensional topographic elevation map. A compass-style north arrow is shown near the top-right. The terrain is drawn as a raised surface with stepped elevation bands. A broad high area occupies the left half of the map, with multiple contour-like bands descending outward from its edges. A smaller high area appears toward the upper-right, also surrounded by stepped bands. . The Study Site is marked by a downward-pointing triangle placed near the right-of-center where the terrain has lowest elevation or local depression , on adjacent terrain. No place names are shown on the map.
The BTR lies along the northern margin of the Satpura ranges and is characterized by two prominent hill systems. The Bandhavgarh hill, named after the medieval fort located on its summit, is the highest point in the reserve, reaching about 750 m above mean sea level (Fig. 2). The Bandheni hill, comparable in size, is separated from Bandhavgarh by a distinct and narrow defile (Lahiri et al., Reference Lahiri, Rajani, Sanyal, Banerjee and Tiwari2023). The BTR terrain is marked by sandstone hills and mountains, with soils ranging from sandy to sandy loam, derived mainly from the Mahadeva and Supra Barakar sandstone formations (Hughes, Reference Hughes1885).
The reserve is drained by more than 20 seasonal and perennial streams, including the Johilla, Janadh, Charanganga, Damnar, Banbei, Ambanala, and Andhiyari Jhiria (Lahiri et al., Reference Lahiri, Rajani, Sanyal, Banerjee and Tiwari2023). These streams ultimately join the Son River, an important southern tributary of the Ganga River system. Since the declaration of Bandhavgarh as a national park in 1968, access for detailed field-based geological investigations has been restricted, resulting in limited geological documentation of the area (Phartiyal et al., Reference Phartiyal, Karthick, Lahiri, Tiwari, Prasanna, Trivedi and Manoj2025). The study area is surrounded by low to moderate relief hills (Fig. 2), from which streams drain into adjacent low-lying zones. These depressions are commonly occupied by meadows and meadow-associated wetland systems, which form a distinctive component of the local landscape (Srinivas and Lahiri, Reference Srinivas and Lahiri2024).
Climate
The BTR experiences a tropical monsoon climate (Köppen-Geiger climate classification; Peel et al., Reference Peel, Finlayson and McMahon2007), with three distinct seasons. Climate data were obtained from the CRU TS 4.08 dataset at spatial resolution of 0.5° × 0.5° and analyzed for the period 2011–2024 to represent recent climate patterns (Harris et al., Reference Harris, Jones, Osborn and Lister2014). The decadal annual temperature recorded in the study region ranges between 31.5°C and 32.5°C (Fig. 3). The summer months (March to June) are intensely hot, with daytime temperatures ranging from 30°C to 45°C and peaks above 42°C are commonly recorded in May. The monsoon season (July to September) accounts for nearly 85–90% of the annual precipitation. The decadal annual rainfall ranges from ∼1000 to 1100 mm (Fig. 4). July and August are the wettest months, frequently receiving more than 350 mm of rainfall (Mishra et al., Reference Mishra, Rai, Mishra and Cabral2015; India Meteorological Department, 2019), whereas rainfall during the summer period is minimal, often below 50 mm. (Rodgers et al., Reference Rodgers, Panwar and Mathur2002). Winter (October to February) is characterized by cooler conditions (5°C to 25°C) and low rainfall (<100 mm) (Jhala et al., Reference Jhala, Qureshi and Nayak2019).
Decadal annual temperature (2011–2024) map of BTR in the CMZ, Madhya Pradesh, India.
Figure 3 Long description
The map shows temperature ranges in Madhya Pradesh, India, with a focus on a specific study site. The temperature is categorized into ten ranges: 29.9–30.5°C, 30.5–30.9°C, 30.9–31.5°C, 31.5–31.9°C, 31.9–32.5°C, 32.5–32.9°C, 32.9–33.5°C, 33.5–33.9°C, 33.9–34.5°C, and 34.5–34.9°C. The study site is marked with a symbol. The map highlights regional temperature variations, with higher temperatures in the central and southern areas and lower temperatures in the northern regions. A scale bar indicates distances in kilometers and the map includes latitude and longitude markers for orientation.
Decadal annual precipitation (2011–2024) map of the BTR in the CMZ, Madhya Pradesh, India.
Figure 4 Long description
The map shows the annual precipitation distribution in Madhya Pradesh, India, with a focus on a specific study site. The precipitation is categorized into several ranges: 579 to 666 mm, 666 to 752 mm, 752 to 838 mm, 838 to 925 mm, 925 to 1011 mm, 1011 to 1090 mm, 1090 to 1184 mm, 1184 to 1270 mm, 1270 to 1356 mm and 1356 to 1442 mm. The study site is marked with a specific symbol. The map includes a scale bar indicating distances in kilometers and is oriented with a compass rose. The inset map shows the location of Madhya Pradesh within India, highlighting the study area. The precipitation distribution shows a gradient with higher values concentrated in the central and eastern parts of the region, while lower values are more prevalent in the western areas.
Vegetation
The present vegetation of the BTR provides an important modern analogue for interpreting fossil pollen assemblages and reconstructing past climatic conditions. Vegetation in the park is predominantly classified under Moist Peninsular Sal Forests (3B/C1) and West Gangetic Moist Mixed Deciduous Forests (5A/C1d) (Champion and Seth, Reference Champion and Seth1968). Several dominant tree taxa in the region are sensitive to variations in rainfall, temperature, and moisture availability, making them reliable indicator taxa in palaeoclimate studies.
Floristic inventories conducted in the BTR have recorded over 600 species of angiosperms, belonging to 87 families and 312 genera. The grass family (Poaceae) exhibits the highest diversity (47 species), followed by Fabaceae (42 species) and Euphorbiaceae (28 species) (Singh and Kushwaha, Reference Singh and Kushwaha2008). Notably, the park supports several rare and endemic taxa, such as the carnivorous Drosera peltata, found in ephemeral wetlands, and Acorus calamus, a medicinal species restricted to isolated patches in the Talarange.
The forest canopy is primarily dominated by Shorea robusta (sal), accounting for roughly 55–60% of the total forest cover (Rodgers et al., Reference Rodgers, Panwar and Mathur2002; Sharma et al., Reference Sharma, Behera, Das and Panda2014). This sal-dominated forest exhibits a vertically stratified structure with three distinct layers. Their regeneration strongly depends on the timely onset of the southwest monsoon, and approximately 1200 mm of rainfall is required for successful seed germination and establishment and regeneration of sal forests (Meher-Homji, Reference Meher-Homji2000; Nag et al., Reference Nag, Trivedi, Farooqui and Morthekai2025).
The upper canopy (20–30 m) is composed of Shorea robusta with co-dominant species like Terminalia tomentosa (saja), Terminalia arjuna (arjun), and Madhuca indica (mahua). The sub-canopy (10–20 m) is populated by Anogeissus latifolia (dhaora), Diospyros melanoxylon (tendu), and other deciduous associates (Troup, Reference Troup1921).
The dominant species in these forests also exhibit significant phytochemical and ethnobotanical relevance (Arulmozhi et al., Reference Arulmozhi, Mazumder, Sathiyanarayanan and Thakurdesai2007; Khare, Reference Khare2007; Jadhav et al., Reference Jadhav, Holkar and Kale2009). Lower elevation zones are dominated by Dendrocalamus strictus (bamboo), which significantly contributes to the forest’s vertical complexity. This species undergoes synchronous gregarious flowering roughly every 25–30 years, influencing successional trajectories and regeneration dynamics (Troup, Reference Troup1921). The herbaceous layer is dense and seasonally dynamic, supporting numerous grasses, herbs, and shrubs.
Approximately 12% of the BTR area consists of grassland habitats, mostly established on former agricultural lands. These grasslands are dominated by C4 grasses such as Themeda triandra and Chrysopogon zizanioides, which provide vital forage resources for herbivores, with crude protein levels ranging from 6.8% to 14.2% during the growing season (Ashokkumar et al., Reference Ashokkumar, Swaminathan and Nagarajan2021). Additionally, 37 perennial and seasonal water bodies including both lakes and streams within the park support a rich diversity of aquatic macrophytes (∼50 species). The adjacent low-lying zones and depressions form a distinctive component of the local landscape enhancing ecological heterogeneity and supporting aquatic flora. The surrounding area of the investigated site in the BTR is characterized by savanna-type vegetation, where grasses dominate the landscape with a sparse to moderate presence of trees.
Materials and methods
Sediment core and age–depth model
To establish the chronology of the 60-cm-long sediment core retrieved from Bandhavgarh by a polyvinyl chloride (PVC) pipe from the periphery of the wetland (Fig. 5). The wetland occupies an area of approximately 650 m2, with an estimated length of ∼45 m and width of ∼25 m. Sediments at the site are predominantly fine-grained and muddy in nature. The surrounding landscape comprises forested hills and low mountains, from which eroded material is transported by seasonal streams and deposited in adjacent low-lying areas. Owing to this sediment accumulation, the wetland was selected as a suitable site for core extraction. The wetland flank is characterized by a very shallow water column, typically limited to a few centimetres, which restricts bioturbation. The soils are seasonally flooded, resulting in reduced groundwater percolation, minimal earthworm activity, and negligible root penetration in the absence of trees in the immediate vicinity. In addition, the site is not influenced by bedrock carbonate, reducing the likelihood of hard-water effects on radiocarbon ages.
Sediment coring at Chakradhara wetland using a PVC pipe, from which a 60 cm sediment core was recovered (after Lahiri et al., Reference Lahiri, Phartiyal and Balasubramanian2024).
Figure 5 Long description
Four individuals are engaged in sediment coring in a wetland area. One person is holding a PVC pipe vertically, while two others assist in stabilizing it. The fourth person is observing the process. The setting is a grassy wetland surrounded by hills and vegetation, with tall grasses in the background.
Three bulk soil samples were selected from depths of 2.5 cm, 35.5 cm, and 59 cm for radiocarbon (1⁴C) dating. The dating was performed using accelerator mass spectrometry (AMS) at Beta Analytic Testing Laboratory, Florida, USA (after Lahiri et al., Reference Lahiri, Phartiyal and Balasubramanian2024). The radiocarbon ages obtained from bulk sediment were calibrated using the IntCal20 dataset (Reimer et al., Reference Reimer, Austin, Bard, Bayliss, Blackwell, Ramsey, Butzin, Cheng, Edwards, Friedrich and Grootes2020; McKay et al., Reference McKay, Emile-Geay and Khider2021). Age–depth modelling and Bayesian calibration were performed using the Bchron package (Itzstein-Davey et al., Reference Itzstein-Davey, Atahan, Dodson, Taylor and Zheng2007a; Haslett and Parnell, Reference Haslett and Parnell2008) in R (R Core Team, 2021). Age calibration was carried out using the CALIB 7.1 software (Stuiver et al., Reference Stuiver, Reimer and Reimer2020). The calibrated age ranges at the 2σ confidence interval (Table 1), along with their weighted means, are illustrated in Figure 6.
Lithological profile and age–depth model of the sediment core from the BTR (after Lahiri et al., Reference Lahiri, Phartiyal and Balasubramanian2024).
Figure 6 Long description
A vertical core column is shown with the y-axis labeled Depth (cm), marked from 0 to 60. Three circular dots beside the log showing absolute AMS dates such as 2001 CE, 1240 ± 30, 2600 ± 30 at the depth of 302.5, 35.5 and 59 respectively on the depth scale. The core column contains multiple stacked lithology patterns. A legend below the log shows pattern keys labeled Base not exposed, Clay, Silty-Sand, Sand with pebbles and Sand with rootlets. In the adjacent figure A graph is shown with the x-axis labeled Age Cal yr BP and the y-axis labeled Depth (cm). The x-axis shows labeled ticks at 3500, 3000, 2500, 2000, 1500, 1000, 500 and 0. The y-axis shows labeled ticks at 0, 20, 40 and 60. A shaded band forms an age–depth model. Three vertical shaded distributions align with the three dated depths and are labeled BG-49, BG-24 and BG-2. The BG-49 distribution is centered near depth about 59 and age near 3000. The BG-24 distribution is centered near depth about 35 and age near 1200. The BG-2 distribution is centered near depth about 2.5 and age near 0. Two text annotations on the shaded band read 0.01 cm/yr and 0.02 cm/yr.
Details of the radiocarbon (1⁴C) ages from AMS dating of the Bandhavgarh sediment (after Lahiri et al., Reference Lahiri, Phartiyal and Balasubramanian2024).
Table 1 Long description
The table reports AMS radiocarbon dating results for three bulk sediment samples, listing beta number, depth, radiocarbon age, and calibrated calendar age ranges with two sigma uncertainty. The shallowest sample at 2.5 cm has no radiocarbon age reported and calibrates to about 2001 to 2005 CE. At 35.5 cm, beta 673066 has a radiocarbon age of 1240 plus or minus 30 years before present and calibrates to about 1192 to 1070 years before present. At 59 cm, beta 673067 has a radiocarbon age of 2600 plus or minus 30 years before present and calibrates to about 2767 to 2709 years before present. Overall, the calibrated ages become older with increasing depth, consistent with normal sediment accumulation. The calibrated uncertainty is smallest for the deepest sample and largest for the mid depth sample.
* 107.35 ± 0.40 pMC.
The core profile comprises silty sand, coarse sand, and humified and blackish mud layers. The dated stratigraphic levels correspond to distinct lithological units: humified blackish mud at 2.5 cm depth yielded a modern age (calibrated between 2001 and 2005 CE), while blackish silty sand at 35.5 cm produced a date of 1200 ± 30 years BP, and blackish compacted mud at 59 cm was dated to 2600 ± 30 years BP (after Lahiri et al., Reference Lahiri, Phartiyal and Balasubramanian2024). These age determinations indicate variable sediment accumulation rates through time.
Based on the calibrated age data, the age–depth model reveals a relatively slow sedimentation rate of approximately 0.01 cm/yr in the lower portion of the core (59–35.5 cm), which increases to about 0.02 cm/yr in the upper section (35.5–2.5 cm). This shift suggests a temporal variation in depositional dynamics, possibly driven by improvement conditions of climate, vegetation cover, and land-use patterns by human activities.
Pollen extraction procedure of sediment
Twenty-two sediment samples from a 60-cm-deep core profile were chosen for pollen analysis. These sediment samples were subjected to standard palynological procedures at the Birbal Sahni Institute of Palaeosciences (BSIP), following the acetolysis technique outlined by Erdtman (Reference Erdtman1943) and Faegri and Iversen (Reference Faegri and Iversen1989). Approximately 5 g of each sample were treated with 10% aqueous potassium hydroxide (KOH) and boiled for 5 minutes to deflocculate pollen/spores and remove humic matter (Nag et al., Reference Nag, Trivedi, Farooqui and Morthekai2025). The residue was sieved through a 150-mesh (105 µm) sieve to remove coarse debris, followed by repeated washing and decantation three times with distilled water to eliminate excess KOH. Subsequently, the samples were treated with 40% hydrofluoric acid (HF) thereafter, to eliminate siliceous components and then rinsed and decanted two to three times with distilled water. The HF-treated samples were then rinsed and decanted two to three times using distilled water. Finally, the samples underwent acetolysis using a mixture (9:1 ratio) of concentrated sulfuric acid and acetic anhydride. This step removes impurities and cytoplasm and stain to improve the visibility of pollen and spore features for microscopic identification (Erdtman, Reference Erdtman1943; Faegri and Iversen, Reference Faegri and Iversen1989). The processed residue samples were again sieved with a 600-mesh sieve (20 µm) to remove finer material and pollen that get trapped at the top of the sieve. Finally, the samples are stored in 50% glycerine with a few drops of phenol to prevent microbial degradation.
Temporary slides were analyzed under an Olympus BX52 microscope fitted with an Olympus DP23 digital camera. Pollen and spore identifications were conducted at 400× magnification, with photographic documentation provided in Figure 7. Depending on the concentration and preservation of pollen, each sample yielded between 300 and 500 countable grains. Taxonomic identification was aided by modern reference collections and key literature sources, including Pollen Flora of Maharashtra State, India (Nayar, Reference Nayar1990), the Australasian Pollen and Spore Atlas (Australian National University, 2025), and previous regional palynological studies (Chauhan, Reference Chauhan2004, Reference Chauhan2005; Trivedi et al., Reference Trivedi, Saxena, Chauhan, Sharma, Farooqui, Nautiyal, Yao, Wang, Li and Tiwari2019, Reference Trivedi, Agrawal, Sharma, Ali, Manoj, Nag, Misra and Kawsar2024a).
Pollen and spore identification: (1) Acacia, (2) Bombax ceiba, (3) Holoptelea integrifolia, (4) Shorea robusta, (5) Embelica officinalis, (6) Lagerstroemia parviflora, (7) Aegle marmelos, (8) Fabaceae, (9) Syzygium cumini, (10) Tectona grandis, (11) Poaceae, (12) Cerealia, (13) Chenopodiaceae/Amaranthaceae, (14) Brassicaceae, (15) Malvaceae, (16) Alternanthera, (17and18) Rumex, (19) Aspidopterys (20) Xanthium, (21) Polygonum serrulatum, (22) Justicia simplex, (23) Cyperaceae, (24) Polygonum plebeium, (25) Fern (trilete), (26) Fern (monolete), (27) Nigrospora, (28) Glomus, (29) Alternaria.
Figure 7 Long description
Insufficient visual information to describe this element accurately.
The identified pollen taxa were classified into nine broad ecological categories: moist deciduous, dry deciduous, shrubby, cultural, heathland, aquatics, marshy, ferns, and non-pollen palynomorphs (NPPs). Arboreal pollen (AP) included moist deciduous, dry deciduous, and shrubby taxa, while non-arboreal pollen (NAP) comprised cultural, heathland, aquatic, and marshy elements. All pollen and spores were identified. The pollen sum, representing the total count of terrestrial pollen grains in a sample, was calculated after excluding aquatic, marshy, fern, algal, and fungal remains (Trivedi et al., Reference Trivedi, Agrawal, Sharma, Ali, Manoj, Nag, Misra and Kawsar2024a, Reference Trivedi, Ali, Manoj, Agrawal, Sharma, Phartiyal and Kumar2024b). The pollen sum is used as the baseline for determining the relative proportions of different pollen types in sediment analysis.
The Poaceae (grass) pollen was divided into two main types based on grain size: (a) Poaceae with an axis length of less than 50 µm (small-sized) and (b) Poaceae with an axis length greater than 50 µm (large-sized). Small-sized Poaceae pollen is commonly associated with wild grass species (non-cereal Poaceae), whereas large-sized Poaceae pollen is more frequently linked to cultivated grasses. When large-sized Poaceae pollen occurs together with other cultivated plant indicators, such as Chenopodiaceae/Amaranthaceae, Caryophyllaceae, Brassicaceae, Artemisia sp., it is interpreted as Poaceae (Cerealia) representing agricultural activity (Itzstein-Davey et al., Reference Itzstein-Davey, Atahan, Dodson, Taylor and Zheng2007b; Yang et al., Reference Yang, Zheng, Mao, Li and Chen2018).
The pollen AP, NAP and NPPs were divided into five distinct zones and four subzones (II-a, II-b, II-c, and II-d) (Fig. 8) using constrained incremental sum of squares (CONISS) analysis in Tilia software (Grimm, Reference Grimm1993).
Frequency distribution diagram from the BTR, in the CMZ, Madhya Pradesh, India.
Figure 8 Long description
The diagram presents a frequency distribution of pollen types across different zones and subzones in Madhya Pradesh, India. The x-axis is labeled 'Percentage' and the y-axis is labeled 'Age (cal yr BP)'. The diagram is divided into two main sections: arboreal taxa and non-arboreal taxa. Each section lists various pollen types such as 'Shorea robusta', 'Embelica officinalis', 'Aegle marmelos', 'Bombax ceiba' and 'Chenopodiaceae'. The right side of the diagram shows a dendrogram labeled 'CONISS' and 'Pollen Zones', which divides the data into five distinct zones labeled I, II, III, IV and V, with zone II further divided into subzones II-a, II-b, II-c and II-d. The dendrogram illustrates the hierarchical clustering of the pollen data. The diagram includes horizontal bars representing the percentage distribution of each pollen type across different age ranges, togetherwith a total sum of squares indicated at the extreme right of the diagram.
Earth system palaeoclimate simulation (ESPS) models
To reconstruct regional climate variability between ∼2500 and 50 cal yr BP, a timeframe encompassing the RWP through the LIA, we utilized outputs from the Community Climate System Model version 3.0 (CCSM3). This model incorporates the interactions between Earth’s major climate subsystems: the atmosphere, ocean, land surface, and sea ice, all linked through a unified coupling system (Collins et al., Reference Collins, Rasch, Boville, Hack, McCaa, Williamson, Briegleb, Bitz, Lin and Zhang2006; Meehl et al., Reference Meehl, Teng and Branstator2006; Otto-Bliesner et al., Reference Otto-Bliesner, Brady, Clauzet, Tomas, Levis and Kothavala2006).
We used transient simulation outputs from the TraCE-21ka experiment, which offers climate data at a horizontal resolution of 3.75° latitude × 3.75° longitude and spans from ∼2700 to ∼50 cal yr BP. The focus of this analysis was on precipitation trends associated with the ISM, particularly during the core monsoon months (June to September, JJAS). For this purpose, rainfall variability was assessed by calculating anomalies over 50-year intervals, expressed in precipitation rates (mm/day).
To interpret the potential environmental implications of long-term monsoonal fluctuations, simulation results from eight ESPS models were compared with proxy-based reconstructions derived from the region’s pollen record. This integrative approach allowed for a more robust evaluation of climate–vegetation interactions during significant climatic episodes.
Results
Temporal distribution of fossil pollen and spores
Pollen zone I (2700–2300 cal yr BP)
This zone is characterized by a relatively high representation of AP (35–41%), alongside a comparatively lower proportion of NAP. A total of 442 pollen grains were counted in this interval. Within the AP, moist deciduous forest elements are notably dominant (21–25%), while dry deciduous taxa exhibit lower frequencies (7–16%). Key constituents of the moist deciduous assemblage include Shorea robusta (5–8%), Lagerstroemia parviflora, and Madhuca indica (3–5%), in association with the members of Rubiaceae, Morus alba, Syzygium, and Moraceae (Figs. 7, 8). The dry deciduous taxa, though less prominent, are represented by moderate frequencies of Bombax ceiba (3–5%), Diospyros, Lannea coromandelica (0.6–2%), Terminalia (0.7–2%), and Sterculia (2–2.5%). The NAP is dominated by higher values of Cyperaceae (17–22%) and Poaceae (13–17%). Other non-arboreal elements such as Tubuliflorae, Cerealia, Cannabis, Caryophyllaceae, and Solanum appear at moderate levels (1–3%). Fern spores are frequently encountered, comprising 19–22% of the total assemblage, suggesting moist and shaded understorey conditions. Additionally, marshy taxa such as Justicia simplex, Polygonum plebeium, and P. serrulatum are consistently recorded (13–18%). Aquatic taxa contribute approximately 7–8%, while fungal spores (NPP) are also abundant, accounting for 13–18% of the assemblage.
Pollen zone II (2300–1600 cal yr BP)
The pollen assemblage indicates AP (29-41%), accompanied by a marked increase in NAP (41–56%), relative to the preceding pollen zone I. A total of 387 pollen grains were recorded from this interval. Among the tree taxa, dry deciduous elements show a relative increase (14–20%) compared to moist deciduous taxa (12–16%). Within the dry deciduous taxa, Bombax ceiba and Diospyros show prominent representation (5–6% each). Additionally, Lannea coromandelica, Acacia, and Terminalia species appear consistently (1–5%). Holoptelea and Aegle marmelos are recorded at moderate levels (1–4%) though less frequently, while Buchanania makes its first appearance in the sequence. Conversely, the moist deciduous taxa show a declining trend, particularly Shorea robusta (2–4%), along with Madhuca indica and Lagerstroemia parviflora (2–3%). Adina and Emblica officinalis occur sporadically (0.7–2%). Shrubby taxa including Fabaceae, Oleaceae, and Tiliaceae exhibit a slight reduction in frequency (1–2%) compared to the previous zone. Among the NAP components, Poaceae exhibits a significant rise (21–29%), while Cerealia, Cannabis sativa (1.5–3%), and Tubuliflorae (5%) show increased values. Other non-arboreal taxa, including Chenopodiaceae/Amaranthaceae, Caryophyllaceae (1–4%), and sporadically occurring elements such as Artemisia, Brassicaceae, Lamiaceae, and Euphorbiaceae (0.6–1.3%), also show an upward trend (Fig. 8). A decline is observed in marshy elements such as Cyperaceae (9–14%), Polygonum plebeium, and P. serrulatum (2–5%), as well as ferns, which decrease to 16–25%. However, NPPs remain well represented, ranging from 6% to 21%.
Pollen subzone II-a (2300–2200 cal yr BP)
This subzone is marked by a notable predominance of NAP, comprising ∼65% of the assemblage, while AP values decline. Within the AP group, Shorea robusta is most frequent (8.3%), followed by Madhuca indica and Syzygium (2.6% each), and Lagerstroemia parviflora (3.2%), all representing moist deciduous forest elements. The herbaceous landscape is dominated by Poaceae (12.8%), accompanied with Cerealia (1.9%), Chenopodiaceae/Amaranthaceae (2.6%), and Artemisia (1.3%). Marshy taxa are also prominent, particularly Cyperaceae (22.5%), along with Polygonum plebeium (4.5%) and P. serrulatum (2.6%).
Pollen subzone II-b (2200–2030 cal yr BP)
This subzone is marked by a further rise in NAP, reaching ∼67%. In contrast, AP values exhibit a corresponding decline. Notably, Shorea robusta, a representative moist deciduous taxon shows a significant reduction (3.3–3.7%). Meanwhile, dry deciduous elements such as Aegle marmelos (2.5–3.3%), Bombax ceiba (4.3–4.6%), and Diospyros melanoxylon (3.7–4%) display an increase compared to the preceding subzone. Within the herbaceous taxa, Poaceae values rise further (25–26%), and Cerealia (2.6%), Chenopodiaceae/Amaranthaceae (3.8%), and Caryophyllaceae (1.6%) show higher representation. A notable increase in fern spores (∼25%) is also recorded in this phase.
Pollen subzone II-c (2030–1680 cal yr BP)
This subzone records a slight decrease in NAP values (∼66%) compared to subzone II-b. Within the arboreal component, moist deciduous taxa regain prominence, with an increase in Shorea robusta (4.5–5%), and Madhuca indica (2.6–3.8%). Grass pollen (Poaceae) continues to dominate the herbaceous group, maintaining stable values between 21% and 29%. However, Cerealia (1.2–1.7%) and Chenopodiaceae/Amaranthaceae (2–3%) exhibit a slight decline, whereas Caryophyllaceae shows a noticeable increase (1.3–4.9%). Marshy taxa and fern spores reflect a minor reduction.
Pollen subzone II-d (1680–1600 cal yr BP)
This subzone marks a notable decline in NAP values, accompanied by a rise in AP to ∼61.5%, suggesting a shift towards dense forest cover. Moist deciduous elements become more prominent, comprising around 28% of the total pollen assemblage, while dry deciduous taxa decrease to ∼6%. Within the moist deciduous group, Shorea robusta increases to 7.5%, with other taxa such as Adina cordifolia, Madhuca indica, Syzygium, and members of Anacardiaceae also showing elevated values. In contrast, dry deciduous indicators like Bombax ceiba and Diospyros melanoxylon display a decline. Among the herbaceous taxa, Poaceae shows a significant reduction (11%), whereas Cerealia reaches a higher representation (up to 4.3%). Additionally, aquatic taxa, marshy herbs, and ferns exhibit increased frequencies in the pollen assemblage.
Pollen zone III (1600–820 cal yr BP)
This zone reflects a marked shift in vegetation composition. AP values remain relatively high (37–48%), while NAP shows a considerable decline (17–43%) compared to the preceding pollen zone II. A total of 439 pollen grains were counted in this zone. Within the arboreal assemblage, there is a notable increase in moist deciduous forest taxa (20–34%) alongside a decline in dry deciduous elements (4–17%). Key moist deciduous taxa such as Shorea robusta (7–11%), Madhuca indica (2–5%), and Syzygium (1–4%) display moderate to elevated representation. In contrast, dry deciduous elements including Bombax ceiba (2–2.5%), Holoptelea, Terminalia, and Lannea (1–3%) occur in lower but consistent frequencies. Other components such as Diospyros, Butea, Aegle, Buchanania, and Schleichera are observed sporadically in minor proportions. Among shrubs, Fabaceae (1–2%) is relatively well represented, whereas Ziziphus, Oleaceae, Tiliaceae, and Acanthaceae occur with limited frequencies (∼1% each). Non-arboreal taxa are dominated by Cyperaceae (10–27%) and Poaceae (9–23%). Cultivation type taxa (2–4%) with Cerealia (2–5%) reflecting moderate values are also prominent. Heathland elements (15–21%) show increased representation. Additional herbaceous taxa such as Tubuliflorae, Liguliflorae, Chenopodiaceae/Amaranthaceae, and Urticaceae (1%) are sporadic. Aquatics (2–8%), including Potamogeton, Eriocaulon, and members of Liliaceae, are well represented. Fern spores (10–24%) and NPPs also occur in relatively high abundances, directing to favourable moisture conditions during this interval.
Pollen zone IV (820–400 cal yr BP)
This pollen zone reveals a notable shift toward increased representation of dry deciduous elements, particularly Bombax ceiba, which registers a marked rise in frequency. A total of 461 pollen grains were recorded in this interval. Madhuca indica (1–3%) also shows a modest increase, accompanied by improved values of Diospyros and Lagerstroemia. In contrast, Shorea robusta (2–3%), a key moist deciduous taxon shows a distinct decline at the onset of this phase. Taxa such as Emblica officinalis, Terminalia, and Adina cordifolia maintain relatively stable frequencies without significant fluctuations (Fig. 8). The first sporadic appearances of Anogeissus, Syzygium, Acacia, and Lannea coromandelica are recorded in this zone. NAP assemblages are dominated by Poaceae (33–40%) and Cyperaceae (29–35%), though the latter shows a substantial decline relative to the preceding phase. Cultivated taxa such as Cerealia (3–8%) exhibit enhanced representation, while herbaceous taxa including Chenopodiaceae/Amaranthaceae, Caryophyllaceae, and Tubuliflorae (1–3% each), along with Artemisia (1%), are also better represented in this interval. Ferns and their allies occur with moderate frequencies (9–19%), while aquatic elements are present only in trace amounts.
Pollen zone V (400 cal yr BP to the present)
This zone is characterized by a significant increase in AP (46–54%), while terrestrial NAP shows a marked decline to 16–20%. A total of 377 pollen grains were counted in this interval. Within the arboreal assemblage, moist deciduous taxa dominate (31–45%), far exceeding dry deciduous elements (3–8%). Notably, Shorea robusta (12–16%) exhibits a pronounced resurgence, along with improved and consistent frequencies of Lagerstroemia (1–4%), Madhuca indica, Emblica officinalis, Adina cordifolia, Sterculia, and Bauhinia racemosa (2–5%). Conversely, dry deciduous elements such as Bombax ceiba (1–2%), Lannea, Holoptelea, and Diospyros demonstrate a steep decline. Shrubby taxa increase slightly (2–7%), with Fabaceae (1–2%), Tiliaceae, and Oleaceae (1–2% each) showing modest improvements over the previous zone. Among the herbaceous components, both Poaceae (5–10%) and Cerealia (1–3%) exhibit reduced frequencies, accompanied by a decline in heathland taxa (6–12%). In contrast, marshy taxa such as Cyperaceae (15–22%) show noticeable gains, along with higher values of Justicia simplex, Polygonum plebeium, and P. serrulatum (collectively 8–14%). Aquatic elements also reflect a rising trend, with Potamogeton, Typha, and Nymphoides appearing in elevated frequencies (5–8%). However, fern spore abundance declines markedly to 10–13%.
Discussion
Monsoon rainfall variability during the Late Holocene in regions dominated by the ISM across the Indian subcontinent has been extensively documented using a wide range of terrestrial and marine archives. Studies from north India (Himalayan region) (Trivedi and Chauhan Reference Trivedi and Chauhan2008, Reference Trivedi and Chauhan2009; Kotlia et al., Reference Kotlia, Sanwal, Phartiyal, Joshi, Trivedi and Sharma2010; Trivedi et al., Reference Trivedi, Kotlia and Joshi2011, Reference Trivedi, Tang, Qin, Farooqui, Wortley, Wang, Blackmore, Li and Yao2020b; Nag et al., Reference Nag, Trivedi, Farooqui and Morthekai2025), northeast India (Dixit and Bera Reference Dixit and Bera2012; Basumatary et al., Reference Basumatary, Swati, Thakur, Jalil and Rahman2018), northwest India (Pant and Maliekal Reference Pant and Maliekal1987; Kumari et al., Reference Kumari, Verma, Rishi, Mehta, Singh, Sharma, Defliese, Holmes and Dixit2025), central India (Chauhan et al., Reference Chauhan, Sharma, Phartiyal and Kumar2013; Behera et al., Reference Behera, Mishra, Sabale, Bhattacharya and Anoop2023), and south India (Rajagopalan et al., Reference Rajagopalan, Sukumar, Ramesh, Pant and Rajagopalan1997; Rajmanickam et al., Reference Rajmanickam, Achyuthan, Eastoe and Farooqui2017), including the Western Ghats (Limaye and Kumaran Reference Limaye and Kumaran2012; Nogué et al., Reference Nogué, Whicher, Baker, Bhagwat and Willis2017), provide clear evidence of spatial and temporal fluctuations in monsoon strength during the past few millennia. These records highlight the sensitivity of regional vegetation and hydrological systems to changes in monsoonal rainfall and offer an important framework for interpreting local palaeoenvironmental archives.
Late Holocene vegetation dynamics in the BTR
The pollen record from the BTR provides a site-specific perspective on vegetation changes in central India during the Late Holocene. Variations in pollen assemblages reflect shifts in forest composition, landscape openness, and hydrological conditions, which can be linked to changes in monsoon intensity and local moisture availability. The following discussion is organized into distinct local pollen assemblage phases to describe temporal changes in vegetation and their climatic implications.
Local pollen assemblage phase I (∼2700–2300 cal yr BP)
This zone reflects a phase of warm and humid conditions, indicating strengthened ISM within the Bandhavgarh forested wetland. The palynological assemblage is marked by a predominance of AP over NAP, with a notable abundance of moist deciduous taxa, aquatics, and ferns. Such a composition is indicative of a dense, moisture-rich forest cover and sustained high monsoonal precipitation, consistent with a period of strengthening of ISM activity. The presence of Cerealia pollen in the beginning of this phase indicates the human settlement and agrarian practices in the region (Chauhan and Quamar Reference Chauhan and Quamar2012; Maurya et al., Reference Maurya, Rai, Sharma, Rawat, Chandana, Dhabi, Bhushan and Sarangi2022; Trivedi et al., Reference Trivedi, Agrawal, Sharma, Ali, Manoj, Nag, Misra and Kawsar2024a) and a phase of relatively favourable monsoonal conditions that supported the expansion of agrarian livelihoods.
Comparable hydroclimatic conditions during this interval have been identified across multiple sites of the Indian subcontinent. Quamar et al. (Reference Quamar, Kar and Thakur2021) reported warm and humid conditions from the CMZ during 2600–2200 cal yr BP, while Saxena and Trivedi (Reference Saxena and Trivedi2017) identified a moderate monsoonal regime in the Central Ganga Plain (CGP). Similarly, records from Jagmotha Swamp indicate prevailing warm and humid climatic conditions in the region between ∼2900 and 1050 cal yr BP (Chauhan, Reference Chauhan2000). Further corroboration comes from Renuka Lake in the northwestern Himalaya, where a marked increase in moist deciduous pollen taxa between ∼2480 and 2270 cal yr BP has been interpreted as evidence of a strengthened ISM phase (Nag et al., Reference Nag, Trivedi, Farooqui and Morthekai2025). Khajjiar Lake from the northwestern Himalaya recorded wet and moist climate during ∼2845–2115 cal yr BP (Kotlia et al., Reference Kotlia, Kholia, Porinchu, Sharma, Kumar, Basavaiah, Bisht and Kukreti2023). Records from the Ladakh Himalaya further indicate strengthening monsoonal conditions during ∼2890–2630 and ∼2630–2130 cal yr BP (Maurya et al., Reference Maurya, Rai, Sharma, Rawat, Chandana, Dhabi, Bhushan and Sarangi2022).
Local pollen assemblage phase II (∼2300–1600 cal yr BP)
Marginal decline in the ISM is recorded by palynological imprints from this zone compared to the preceding phase, though favourable hydroclimatic conditions persisted. The assemblage is marked by an increased proportion of NAP relative to AP, alongside a shift in forest composition characterized by the dominance of dry deciduous taxa over moist deciduous elements. Among dry deciduous taxa, Bombax ceiba and Diospyros considerably increased while moist deciduous taxa such as Shorea robusta and Madhuca indica declined. Among NAP, the percentage of Poaceae increased. This pattern suggests a reduction in forest density and a progressive opening of the landscape, potentially reflecting a transition towards more xerophytic or savanna-like conditions. The increased representation of Cerealia during this phase indicates intensified agricultural activity (Maurya et al., Reference Maurya, Rai, Sharma, Rawat, Chandana, Dhabi, Bhushan and Sarangi2022; Trivedi et al., Reference Trivedi, Agrawal, Sharma, Ali, Manoj, Nag, Misra and Kawsar2024a), likely supported by sufficient water availability under prevailing monsoonal conditions.
During this phase (∼2300–1600 cal yr BP), the region underwent four distinct climatic sub-trends, reflecting fluctuations in ISM strength. A warm and humid interval with intensified precipitation is evident between ∼2300 and 2200 cal yr BP, indicating strong ISM activity. Similar patterns have been reported in the Ladakh Himalaya, where warm and humid conditions prevailed between 2630 and 2130 cal yr BP (Maurya et al., Reference Maurya, Rai, Sharma, Rawat, Chandana, Dhabi, Bhushan and Sarangi2022), and at Renuka Lake in the northwestern Himalaya, where an intense wet phase was recorded between 2480 and 2270 cal yr BP (Nag et al., Reference Nag, Trivedi, Farooqui and Morthekai2025). This was followed by a phase of reduced precipitation and increased aridity from ∼2200 to 2030 cal yr BP, suggesting a weakening of the monsoon. Palynological investigations from Tso Kar Lake also suggest weakening of monsoon from ∼2800 to 1300 cal yr BP (Demske et al., Reference Demske, Tarasov, Wünnemann and Riedel2009). Similar patterns have been recorded from Dangxiong peatland, southeast of the Nyainqentanglha Mountain Range, China, where δ18Ocell values suggest weakened ISM conditions around 2000 cal yr BP (Wang et al., Reference Wang, Peng, Uchida, Zhao, Ding, Yao, Li, Hong and Hong2024). Subsequently, an ameliorating trend in monsoonal intensity is recorded between ∼2030 and 1680 cal yr BP, marked by gradual climatic recovery. The final sub-interval, from ∼1680 to 1600 cal yr BP, is characterized by a return to intensified monsoonal conditions consistent with records from Renuka Lake (1965–940 cal yr BP; Nag et al., Reference Nag, Trivedi, Farooqui and Morthekai2025) and Nonia Tal (2000–980 cal yr BP; Kumar et al., Reference Kumar, Agrawal, Sharma and Pandey2019). These oscillations reflect short-term climatic variability within an overall moderate monsoonal regime. The interval between ∼2300 and 1600 cal yr BP aligns with the RWP.
Comparable palaeoenvironmental records from the Chhattisgarh region points to a warm and humid climatic regime with enhanced rainfall between 2300 and 1600 cal yr BP (Quamar et al., Reference Quamar, Kar and Thakur2021). These regional counterparts support the inference of moderately strong monsoonal conditions during this phase, albeit with increased ecological variability and signs of climatic instability towards the later part of the period. Similarly, evidence from the Lake Lunkaransar in the Thar Desert, western Rajasthan, document high lake levels between ∼2500 and 1600 cal yr BP, indicating sustained humidity and effective monsoonal precipitation (Enzel et al., Reference Enzel, Ely, Mishra, Ramesh, Amit, Lazar, Rajaguru, Baker and Sandler1999). In contrast, records from the northern Andaman Sea indicate dry conditions from 2342 cal yr BP to the present (Achyuthan et al., Reference Achyuthan, Nagasundaram, Gourlan, Eastoe, Ahmad and Padmakumari2014; Banerji et al., Reference Banerji, Arulbalaji and Padmalal2020). Furthermore, at Batal, near Konjom La, dry alpine steppe elements were present around 2000–1800 cal yr BP, indicating a cold and dry climate (Bhattacharyya, Reference Bhattacharyya1988). The expansion of grasslands in Takche Lake, Lahaul-Spiti region, supported cold and dry conditions around 2000 cal yr BP (Mazari et al., Reference Mazari, Bagati, Chauhan and Rajagopalan1995). Weak southwestern monsoons have been reported from the Chandra valley during ∼2032 and 1158 cal yr BP (Rawat et al., Reference Rawat, Gupta, Sangode, Srivastava and Nainwal2015). Ameliorating trend with warm and moist climate is recorded from the Dewar Tal in Gharwal Himalaya during 2300–2000 cal yr BP (Chauhan and Sharma, Reference Chauhan and Sharma2000).
Local pollen assemblage phase III (1600–820 cal yr BP)
The pollen assemblage from this zone indicates a warm and humid climate with the persistence of tropical deciduous Shorea robusta-dominated forests sustained by a strengthened ISM. The increase in aquatic taxa and sedges during this interval points to a substantial expansion of swampy environments, while the abundance of fern spores suggests the prevalence of moist and shaded microhabitats within the forest understory. Collectively, these indicators reflect intensified monsoonal precipitation and a widespread wetland ecosystem, coinciding with the mature phase of the MWP. The absence of a distinct imprint of the DACP implies that local climatic and geographic controls modulated the regional response (Chauhan et al., Reference Chauhan, Sharma, Phartiyal and Kumar2013)
Comparable records from central India support this interpretation. For instance, pollen sequences from Madhya Pradesh also reveal sal-dominated forest communities and moderately strong monsoonal conditions between ∼1500 and 800 cal yr BP (Chauhan, Reference Chauhan2004, Reference Chauhan2005; Chauhan and Quamar, Reference Chauhan and Quamar2010). Evidence from Kerha, Kiktiha, Amgaon swamps in Madhya Pradesh, where the region was dominated by tropical deciduous sal forests from ∼1600 to 700 cal yr BP, indicates a warm and moist climate (Chauhan, Reference Chauhan1995, Reference Chauhan2005; Chauhan and Quamar, Reference Chauhan and Quamar2010). Similarly, records from Dongar-sarbar swamp in Madhya Pradesh document the presence of sal since ∼1800 cal yr BP (Chauhan, Reference Chauhan2002), suggesting that the region was not influenced by DACP due to local climatic and geographic conditions (Chauhan et al., Reference Chauhan1995, Reference Chauhan2000, Reference Chauhan, Sharma, Phartiyal and Kumar2013). Furthermore, Nonia Tal (Lake) of Madhya Pradesh recorded an abrupt increase in TOC, TN, χlf, and sand percentage during ∼2000–980 cal yr BP indicates a period of high productivity and enhanced ISM conditions. These changes suggest increased precipitation in the lake catchment area, likely influenced by intensified monsoon conditions (Kumar et al., Reference Kumar, Agrawal, Sharma and Pandey2019).
Records from Ropan Chappra lake in the CGP which also suggested a warm and humid condition with intensified ISM from 1070 to 700 cal BP (Singh et al., Reference Singh, Gupta, Sangode, Clemens, Prakasam, Srivastava and Prajapati2015). In contrast, the Sarsapukhra lake within CGP experienced fluctuating monsoon intensity, with relatively improved ISM conditions between ∼1200 and 1030 cal yr BP, followed by a weaker phase during ∼1030–900 cal yr BP (Trivedi et al., Reference Trivedi, Agrawal, Sharma, Ali, Manoj, Nag, Misra and Kawsar2024a). Furthermore, records from the Mahanadi delta region, Odisha suggests the warm and humid climate with high precipitation between 1300 and 800 cal yr BP (Samal et al., Reference Samal, Subramanian, Srivastava, Kawsar, Manoj, Gurumurthy and Chauhan2023). Additionally, humid climatic conditions recorded between 1200-1000 cal yr BP in Chilika Lagoon (Shah et al., Reference Shah, Khan, Rahman, Kumar, Achyuthan, Shukla, Kumar and Dash2022; Amir et al., Reference Amir, Paul and Malik2021) and the presence of tropical deciduous forest dominated by Shorea robusta suggests strong ISM conditions in the Deosila Swamp, Assam, during 1510 to 540 cal yr BP (Dixit and Bera, Reference Dixit and Bera2011). In contrast, Muval Pond, Gujarat, records low moisture conditions during 1400–1150 cal yr BP, followed by relatively higher moisture between 1150 and 850 cal yr BP (Kumar et al., Reference Kumar, Sharma and Agrawal2025).
The proliferation of moist deciduous taxa at Renuka Lake in the northwestern Himalaya between ∼1965 and 940 cal yr BP has been interpreted as evidence of a robust ISM phase (Nag et al., Reference Nag, Trivedi, Farooqui and Morthekai2025). Strengthened ISM conditions were reported at Benital Lake in lesser Himalaya between ∼1800 and 1000 cal yr BP (Bhushan et al., Reference Bhushan, Sati, Rana, Shukla, Mazumdar and Juyal2018). Additionally, the warm climate was recorded from North Pulu Lake, Ladakh, during 1182–958 cal yr BP (Phartiyal et al., Reference Phartiyal, Singh, Joshi and Nag2020). Improved moisture content was recorded from TsoKar Lake in Trans Himalaya during 1300 cal yr BP to the present (Demseke et al., Reference Demske, Tarasov, Wünnemann and Riedel2009). Subsequently, warm and humid conditions were recorded in Nachiketa Tal, Garhwal Himalayas, between 1872 and 767 cal yr BP (Trivedi et al., Reference Trivedi, Kotlia and Joshi2011). Onset of Warm and moist climate recorded from Chandra valley, Northwest Himalaya and Gharwal Himalaya region during 1158‒647 cal yr BP (Rawat et al., Reference Rawat, Gupta, Sangode, Srivastava and Nainwal2015) and 1400‒400 cal yr BP (Chauhan and Sharma, Reference Chauhan and Sharma2000), respectively. Additional evidence from the Ladakh Himalaya also indicates climatic deterioration between ∼1660 and 1280 cal yr BP (Maurya et al., Reference Maurya, Rai, Sharma, Rawat, Chandana, Dhabi, Bhushan and Sarangi2022). Speleothem δ1⁸O records from central India suggest a weakened ISM around ∼1050 cal yr BP (Sinha et al., Reference Sinha, Berkelhammer, Stott, Mudelsee, Cheng and Biswas2011b), a trend that is also reflected in proxy records from the Lesser Himalaya (Kotlia et al., Reference Kotlia, Singh, Joshi and Dhaila2015).
Local pollen assemblage phase IV (820–400 cal yr BP)
The pollen assemblage from this zone reveals weakening of the ISM, resulting in reduced precipitation and an overall warm and drier climatic regime. The reduced frequency of Shorea robusta and its associated taxa such as Holoptelea integrifolia, Emblica officinalis, and Meliaceae suggests a degradation of the moist deciduous forest structure. The poor regeneration of Shorea robusta during this interval may be linked to the monsoonal delay, as the species is known to require immediate post-dispersal rainfall to facilitate seed germination due to its low seed viability (Meher-Homji, Reference Meher-Homji2000). Concurrently, the expansion of Bombax ceiba, along with increased representation of Diospyros and the emergence of xerophytic elements such as Aegle marmelos, Anogeissus, and Lannea coromandelica, points toward the development of more open and drought-tolerant woodland communities. The concurrent rise in Poaceae and other herbaceous taxa further underscores the transition toward open habitats and grassland expansion. Increased frequencies of culture pollen (Cerealia) and ruderal taxa including Chenopodiaceae/Amaranthaceae, Caryophyllaceae, and Artemisia indicate intensified anthropogenic disturbances, possibly related to land clearance, agricultural activity, and settlement expansion. Comparable evidence from Sarsapukhra Lake in the CGP shows increased Cerealia pollen between ∼780 and 620 cal yr BP, indicating enhanced agrarian practices and human settlement during this period (Trivedi et al., Reference Trivedi, Agrawal, Sharma, Ali, Manoj, Nag, Misra and Kawsar2024a).
Overall, this region experiences warm and dry phase that broadly coincides with the dry phase of LIA, and is consistent with palaeoclimatic records from other regions of the CMZ, which similarly documented the forest degradation and monsoon weakening between ∼800 and 400 cal yr BP (Chauhan, Reference Chauhan2004, Reference Chauhan2005; Chauhan and Quamar, Reference Chauhan and Quamar2010; Quamar and Chauhan, Reference Quamar and Chauhan2015). Pollen evidence from Kerha swamp in Madhya Pradesh further reveals the expansion of Bombax ceiba between ∼850 and 400 cal yr BP, indicating a period of warm and dry climate in the region (Chauhan, Reference Chauhan2005). Additionally, records from the Mahanadi delta region also indicate weakened ISM conditions between 800 and 300 cal yr BP (Samal et al., Reference Samal, Subramanian, Srivastava, Kawsar, Manoj, Gurumurthy and Chauhan2023). Similarly, records from Muval Pond, mainland Gujarat, show a warm and relatively less humid phase between 850 and 650 cal yr BP, followed by a transition to comparatively higher moisture conditions during 650–400 cal yr BP (Kumar et al., Reference Kumar, Sharma and Agrawal2025). Comparable trends are also documented at Renuka Lake in the northwestern Himalaya, where a dominance of dry deciduous species between ∼940 and 540 cal yr BP has been interpreted as evidence of a weakened ISM phase (Nag et al., Reference Nag, Trivedi, Farooqui and Morthekai2025). Records from Mansar Lake in lower Siwalik also recorded cold and dry conditions with reduced monsoon precipitation during 750 cal yr BP to the present (Trivedi and Chauhan, Reference Trivedi and Chauhan2008). Furthermore, records from the Himalayan lakes such as Benital, Badanital and Din Gad peat profile suggest weakened summer monsoon during 800 cal yr BP (Phadtare, Reference Phadtare2000; Demske et al., Reference Demske, Tarasov, Leipe, Kotlia, Joshi and Long2016; Bhushan et al., Reference Bhushan, Sati, Rana, Shukla, Mazumdar and Juyal2018). Decline in the precipitation condition recorded from Gharwal Himalaya region during 400 cal yr BP to the present (Chauhan and Sharma, Reference Chauhan and Sharma2000). The moisture patterns derived from PCA-2 data at Basomtso Lake on the southeastern Tibetan Plateau point to a dry climate from about 867 to 330 cal yr BP, reflecting weakened ISM activity (Li et al., Reference Li, Liu, Wang, Herzschuh, Ni, Liao and Xiao2017). The multiproxy records from southwest China indicate a dry climate, as evidenced by the reduction in the size and depth of lakes (Guo et al., Reference Guo, Huang, Huang and Cheng2022) during this phase. Lake Shuanghaizi in southwest China also recorded cool and dry conditions with the weakest monsoon during 970–470 cal yr BP (Trivedi et al., Reference Trivedi, Tang, Qin, Farooqui, Wortley, Wang, Blackmore, Li and Yao2020b).
Local pollen assemblage zone V (400 cal yr BP – present)
This zone marks a phase of ecological recovery and forest regeneration, coinciding with ameliorated climatic conditions and a strengthening of the ISM. The renewed dominance of Shorea robusta and a noticeable expansion of moist deciduous elements such as Terminalia, Madhuca indica, Lagerstroemia, Emblica officinalis, and Adina cordifolia suggest the re-establishment of dense tropical deciduous forests under warmer and more humid conditions. This recovery phase is interpreted as a consequence of strengthened summer monsoon precipitation, which created favourable ecological conditions for the resurgence of Shorea robusta forests. The marked decline in Poaceae supports this interpretation, pointing toward reduced disturbance and a transition away from open grass-dominated landscapes. The improved monsoonal regime not only promoted forest regeneration but also likely enhanced seedling recruitment and canopy closure, resulting in increased forest biomass and structural complexity. Similar monsoon-driven vegetation dynamics have been reported from other parts of central India, particularly in Madhya Pradesh, where increased representation of Shorea and associated taxa during the late Holocene has been linked to improved hydroclimatic conditions (Quamar and Chauhan, Reference Quamar and Chauhan2014). Furthermore, records from Kerha swamp in Madhya Pradesh, when a good monsoon favoured proliferation of sal forests, suggest a warm and moist climate during 400 cal yr BP (Chauhan, Reference Chauhan2005; Chauhan and Quamar, Reference Chauhan and Quamar2010). Additionally, maximum total organic carbon (TOC) and total nitrogen (TN) values along with higher magnetic susceptibility (χlf) values suggest a period of enhanced hydrological conditions and increasing precipitation in Nonia Tal of Madhya Pradesh (Kumar et al., Reference Kumar, Agrawal, Sharma and Pandey2019). Likewise, evidence from Muval Pond in Gujarat indicates a warm and humid phase from ∼400 cal yr BP to the present (Kumar et al., Reference Kumar, Sharma and Agrawal2025). Comparable trends from Renuka Lake in northwestern Himalaya demonstrate a balanced presence of dry and moist elements from ∼540 cal yr BP to the present, reflecting a period of relatively stable and moderate monsoon activity (Nag et al., Reference Nag, Trivedi, Farooqui and Morthekai2025). A slight advance in the oak trees reflects an ameliorating trend of the climate in the Surinsar Lake of Jammu region between 800 cal yr BP to the present (Trivedi and Chauhan, Reference Trivedi and Chauhan2009). Expansion of Juniperus in the greater Himalayas also suggest a warm and moist climate during 800-496 cal yr BP (Bhattacharyya, Reference Bhattacharyya1988). Higher C/N ratios and increased grain size values suggest more humid conditions in Lugu Lake, southwest China characterized by increased rainfall (Sheng et al., Reference Sheng, Yu, Xu, Lan, Liu and Che2015). Additionally, records from Lake Erhai in southwest China indicate an increasing trend in precipitation during the period of ∼400 to 60 cal yr BP (Chen et al., Reference Chen, Wan, Zhang, Chen, Xu, Xiao and Huang2005). Lake Shuanghaizi in southest China recorded an ameliorating trend in the monsoon between 470 cal yr BP to the present (Trivedi et al., Reference Trivedi, Tang, Qin, Farooqui, Wortley, Wang, Blackmore, Li and Yao2020b). Meanwhile, around ∼400 years ago (1614 CE), the Baghel rulers abandoned Bandhavgarh Fort and shifted their capital to Rewa (Lahiri et al., Reference Lahiri, Rajani, Sanyal, Banerjee and Tiwari2023), likely reducing permanent settlement and landscape disturbance. This decline in political occupation may have decreased deforestation and land-use pressure in the region. Consequently, reduced anthropogenic disturbance likely facilitated natural forest regeneration and the expansion of monsoon-dependent vegetation.
Climate variations derived from ESPS models
The simulation results of the ESPS models shown in Figures 9 and 10 reveal five distinct phases I–V and four sub-phases (II-a–IId) of climatic variability in the region during the Late Holocene. The models indicate that during ∼2550 cal yr BP the study site experienced a strengthened ISM (Fig. 9A), consistent with enhanced precipitation along with the warm and humid conditions observed in phase I (∼2700‒2300 cal yr BP). Figure 10 shows the simulations from a sequence of four sub-phase models, beginning with intensified monsoon activity ∼2280 cal yr BP (Fig. 10A), followed by a notable reduction in rainfall ∼2110 cal yr BP (Fig. 10B). This is succeeded by a transitional amelioration phase ∼1810 cal yr BP (Fig. 10C) and a renewed phase of monsoonal strengthening ∼1680 cal yr BP (Fig. 10D). These sub-phases correspond broadly with phase II of the pollen record (2300‒1600 cal yr BP). The next phase around ∼1560 cal yr BP (Fig. 9B) marks a shift towards warmer and wetter conditions, suggesting a revival of strong monsoonal dynamics in agreement with the pollen-based data of phase III (1600‒820 cal yr BP). The subsequent pollen zone (820‒400 cal yr BP) is marked by weakening of monsoon as validated by the ESPS model with a pronounced decline in precipitation around 690 cal yr BP (Fig. 9C), reflecting weakened ISM conditions and increasing aridity. Finally, the model for phase V ∼90 cal yr BP (Fig. 9D) depicts a gradual recovery of climatic conditions, with signs of improved moisture availability and monsoon rejuvenation during the most recent period, consistent with the ameliorating trend observed in the pollen record from ∼400 cal yr BP to the present.
Simulations of ESPS models of the ISM region showing climatic variability in the study area over the past ∼2700 cal yr BP to the present. (A) Periods of intensified monsoon activity; (B) a return to strong monsoonal conditions; (C) a phase of weakened ISM; and (D) an ameliorating trend, suggesting gradual recovery in precipitation patterns.
Figure 9 Long description
Four thematic maps (ESPS models) illustrate changes in mean precipitation during the ISM season relative to 1950 AD, with a study site marked by a star. Panel A (2550 cal yr BP) shows increased precipitation in the central region, indicated by warm colors and decreased levels in the surrounding areas, shown by cool colors. Panel B (1560 cal yr BP) displays moderate increases across the region, with warm colors dominating. Panel C (690 cal yr BP) highlights a central decrease in precipitation, with cool colors, surrounded by areas of increase. Panel D (90 cal yr BP) shows a strong increase in the north and a decrease in the south. The legend indicates that warm colors represent increased precipitation, while cool colors indicate decreases. The maps are oriented with north at the top and boundaries and coastlines are visible. The study site is centrally located, marked by a star symbol.
Simulations of ESPS models of the ISM region showing climatic variability in the study area over the past ∼2300 to 1600 cal yr BP. (A) Periods of intensified monsoon activity; (B) a phase of weakened ISM; (C) an ameliorating trend, suggesting gradual recovery in precipitation patterns; and (D) strengthened monsoon activity.
Figure 10 Long description
Thematic precipitation anomaly maps depict changes in mean precipitation during the ISM season (JJAS) relative to 1950 AD for Indian Subcontinent. North is oriented at the top. The study site is marked with a star. A) The map for 2280 cal yr BP shows increased precipitation in central areas, . Wetter conditions are indicated by higher values. B) The map for 2110 cal yr BP displays a mix of wetter and drier conditions, . Central regions show significant variability. C) The map for 1810 cal yr BP indicates predominantly drier conditions. Drier areas are more extensive. D) The map for 1680 cal yr BP shows a return to wetter conditions. Wetter areas are more prominent. The legend explains that the colors represent changes in precipitation, with specific numeric ranges indicating wetter or drier conditions. The study site is consistently marked across all maps.
Conclusions
This study presents a reconstruction of vegetation and climate dynamics in the Bandhavgarh region over the past ∼2700 years, derived from palynological analyses supported by radiocarbon dating. The findings emphasize the dominant influence of the ISM in shaping the ecological conditions of the region. Phases of robust monsoonal activity, corresponding to pollen zones I (2700–2300 cal yr BP), III (1600–820 cal yr BP), and V (400 cal yr BP to the present), favoured the proliferation of moist deciduous forests dominated by Shorea robusta. In contrast, a weakening of monsoonal intensity during phase IV (820–400 cal yr BP) led to a reduction in forest cover and an increment in dry deciduous taxa. Further, pollen zone II experienced four distinct climate episodes with intensified monsoon between 2300 and 2200 cal yr BP, reduced precipitation during 2200 and 2030 cal yr BP, and an ameliorating trend between 2030 and 1680 cal yr BP, and again intensified monsoon between 1680 and 1600 cal yr BP. Additionally, a progressive rise in anthropogenic pollen indicators, including Cerealia and other cultivated taxa from 2700 cal yr BP, reflects growing human influence and the emergence of agricultural activity in the landscape. Thus, palaeoecological datasets underscore the sensitivity of central India’s forest ecosystems to long-term monsoonal variability and demonstrate the role of climate in influencing historical land use, settlement dynamics, and cultural evolution within the CMZ.
Acknowledgments
The authors (AT, AN) are grateful to Professor M.G. Thakkar, Director, Birbal Sahni Institute of Palaeosciences, Lucknow, for his sustained support and encouragement throughout this study. The authors also extend their gratitude to Professor Nayanjot Lahiri and Dr Binita Phartiyal for providing the core sediment samples.
