Paleovegetation dynamics in an alternative stable states landscape in the montane Western Ghats,India

Abstract

Peat deposits (>50 ka) in the montane Nilgiris (Western Ghats, India), have been central to the reconstruction of late Quaternary paleoclimate using paleovegetation changes in the forest-grassland vegetation mosaic that coexist here. However, it is well-known that short-term disturbances can also cause vegetation switches when multiple stable vegetation states exist. We studied paleovegetation changes within the alternative stable states framework using stable carbon isotopes (relative abundance of C3-C4 vegetation) on the cellulose fraction from two high-resolution radiocarbon-dated peat cores ~170 m apart in the Sandynallah valley: Core 1 closer to the hillslope (32,000 years old) and Core 2 from the centre of the valley (45,000 years old). Core 1 is located in an ecotone showing shola-sedgeland dynamics with vegetation switching at c.22 ka from shola (possibly due to fire) to a prolonged unstable state until 13 ka sustained by low waterlogging. Following a hiatus c.13 ka, sedgeland dominates, with a shift into shola at 3.75 ka driven by increasing aridity. Core 2 shows a stable sedgeland mixed C3-C4 composition responding to temperature, enriched in C3-vegetation in the last glacial with C4-dominance beginning c.18.5 ka, indicative of deglacial warming. The distinctive vegetation states at corresponding times in Cores 1 and 2 within the same valley, responding independently to disturbances and climate, respectively, is the first paleo-record from an alternative stable states landscape in the montane tropics. Thus, short-term disturbances and site attributes need to be accounted for before ascribing vegetation change to changing climate in such vegetation mosaics.

Keywords

carbon isotope C3-C4 mixed deglacial warming disturbance late Quaternary Nilgiris paleoclimate resistivity survey shola-grassland tropical peat

Introduction

The existence of forest and ‘non-forest’ mosaics in close proximity with abrupt transitions has been debated by ecologists for over a century (Pausas and Bond, 2020). It is now widely accepted that vegetation mosaics represent stable vegetation states whose dynamics can be explained under the Alternative Stable States framework (Hirota et al., 2011; Hoffmann et al., 2012; Pausas and Bond, 2020). Such vegetation mosaics in mountainous regions are seen across the globe – in Madagascar (Bond et al., 2008), Sri Lanka (Pemadasa and Amarasinghe, 1982), North America (Delcourt and Delcourt, 1997), South America (Overbeck et al., 2007), Australia (Moravek et al., 2013) and the shola forest-grassland mosaics of the Western Ghats in India (Meher-Homji, 1967; Ranganathan, 1938). Traditionally, the global distribution of vegetation has been explained by climate, but the existence of strikingly different vegetation types (forest and grassland) in the same environment shows that climate cannot fully explain global vegetation (Bond, 2005). In several of these biomes with montane forest-grassland mosaics, anthropogenic clearance of forests was considered to be a reason for grassland occurrence since grassland was believed to be an alien element in a climate suitable for the growth of forest vegetation; a tenet which dynamic global vegetation models support (Chaturvedi et al., 2011). It was believed that these anthropogenically degraded forests (montane grasslands), given enough time and protection from disturbances, would recover to a closed forest state (e.g. Pemadasa, 1990). But these expectations run contrary to various palaeoecological studies that point to the antiquity and abundance of montane grasslands in several of these mosaics much before the arrival and settlement of humans (Behling and Pillar, 2007; Meadows and Linder, 1993; Moravek et al., 2013; Rajagopalan et al., 1997; Sukumar et al., 1993; Sutra et al., 1997; Vasanthy, 1988). This has led to the acceptance of the alternative stable vegetation states framework to explain montane grassland-forest mosaics (Joshi et al., 2020).

According to the alternative stable states theory, one climatic regime can support two or more stable vegetation states such that each state shows resilience by returning to its original state after small disturbances due to stabilising feedback mechanisms (Pausas and Bond, 2020). But a strong stochastic perturbation to the system or gradual shifts in environmental drivers can cause one stable state to move to an alternative state. The alternative state is also stable, stabilised through feedback mechanisms, while intermediate states are unstable. But the system may not revert to its original stable state even if the driver causing the perturbation is removed, due to hysteresis. When adapted to describe vegetation dynamics, the stable states apply to vegetation states whereas the perturbations to the system are stochastic disturbances (e.g. fires), or gradual changes in environmental drivers (e.g. temperature). Not all processes facilitating a vegetation switch can be reversed, so a switchback does not happen even when the disturbance causing the switch is reversed – this characteristic is called hysteresis.

An important characteristic of the alternative stable states framework in vegetation states is to establish a potential disturbance/perturbation that pushes the system away from one stable state to either an intermediate unstable state or to the alternative stable state. The second characteristic is to establish maintenance pathways that provide positive feedback to make the state resilient despite the presence of small perturbations. The third characteristic is to establish stability, that these systems can survive longer than one generation. Maintenance mechanisms such as herbivory, periodic fires, precipitation changes have been proposed globally (Pausas and Bond, 2020).

The tropical grassland-shola (‘shola’ is a word of Tamil origin for stunted evergreen forest patches) mosaic is characteristic of the montane region (>1800 m asl), especially the Nilgiri and Anamalai hill ranges, of the Western Ghats (India) (Meher-Homji, 1967; Ranganathan, 1938). In the shola-grassland mosaics, frost occurring in the grasslands were observed to act as a barrier for trees (Meher-Homji, 1967; Ranganathan, 1938). Fires and mammalian herbivory (Bor, 1938) and potential differences in soil nutrients were also proposed as barriers to tree establishment (Ranganathan, 1938). Joshi et al. (2020) test soil and microclimate as drivers and propose that the shola-grassland represent alternative stable states with frost as the primary mechanism determining their spatial distribution. But establishing the stability of a vegetation state with long-lived species such as trees is difficult in practical terms (Bowman and Perry, 2017). Palaeoecology can provide historical information on disturbance regimes such as fires and help establish whether gradual changes in environmental drivers such as temperature would have direct impacts on the establishment and subsequent maintenance of alternate vegetation states.

One way to visualise alternative stable states would be to see strikingly different vegetation under the same climatic regime with sharp spatial boundaries. But how would such sharp spatial gradients vary temporally? Intuitively we can think of a mean vegetation state with small disturbances, shifting to a different mean state, is one way to demonstrate alternative stable states in the past. In Tasmania, where fire-mediated vegetation dynamics have been documented for over a century, four alternative stable states at the landscape level have been described: rainforest, eucalypt forest, sclerophyll shrubs and moorland (Murphy and Bowman, 2012). (Fletcher et al., 2014) propose a candidate profile which would fit the alternative stable states framework where palynological evidence for a shift in vegetation mean state from forested Cyperaceae-Sphagnum dominated sedgeland (and eucalypt forests) to non-forested Restionaceae (rush)-dominated wetland was observed after a catastrophic fire 7000 years ago. The new stable vegetation state gets maintained by a positive feedback mechanism: rhizomatous vegetation increases waterlogging, thereby preventing forest elements from establishing themselves. Another way of establishing alternative stable states would be to look at multiple profiles from adjoining sites (hence subject to the same climate, potential vegetation mosaics) and compare the paleovegetation, to find simultaneous existence of different vegetation states.

Although archives that preserve past vegetation to allow for palaeoecological investigations in tropical southern India are scarce due to high levels of decomposition, the high altitude (>1800 masl) and cool temperatures (annual average 13.5°C, mean maximum 18.5°C, mean minimum 8.5°C (von Lengerke, 1977)) typically occupied by the shola-grasslands have facilitated multiple studies of paleovegetation and paleoclimate (Caner et al., 2007; Raja et al., 2019; Rajagopalan et al., 1997; Sukumar et al., 1993; Sutra et al., 1997; Vasanthy, 1988). Prominent among the various archives used in these investigations is the Sandynallah valley (Nilgiris, Western Ghats) with one of the oldest peat accumulations in the world >50 kyr (Ramya Bala et al., 2016), that preserves Late Quaternary vegetation and climate (Rajagopalan et al., 1997; Sutra et al., 1997; Vasanthy, 1988). These studies have broadly demonstrated the response of the vegetation to well-known global climatic events such as the Last Glacial Maximum, the Holocene Optimum and an additional phase of relatively arid and fluctuating monsoon (about 6–2 ka). Recently this valley has also been reported to preserve signatures of paleofires and human presence in this montane region at c. 3.5 ka (Kavil et al., 2021). Accordingly, Caner et al. (2007) state that ‘The Nilgiri highlands have long been an oasis for paleoenvironmental research within an otherwise unrewarding South Indian craton in terms of Quaternary environmental archives’.

One of the main historical debates to which palaeoecology in the region has lent its use is on the origin of the grasslands in the region: are they anthropogenic or natural? Palynological studies from this site (Suryaprakash, 1999; Sutra et al., 1997; Vasanthy, 1988) show that the mosaic is indeed not anthropogenic in origin as evidenced by the abundance of grass pollen throughout the Holocene, extending into the last glacial (~35 ka). Sutra et al. (1997) found that Poaceae and Cyperaceae together constitute the most abundant taxa (>60% throughout the profile) with Poaceae constituting as much as 78% of total pollen counts in some parts of the deeper section, which were dated to circa 30 ka. They conclude that this vegetation mosaic indeed represents an apparent equilibrium of two natural climatic climaxes; with the grasslands representing the climatic climax of the frost zone, and the stunted evergreen forests: the forest zone (Vasanthy, 1988).

Palaeoecological investigations have utilised stable carbon isotopic differences in photosynthetic pathways – the C3 shola vegetation and predominantly C4 nature of grasslands (both C3 and C4 grasses exist here) – to trace relative changes in biomass contribution from these two vegetation states and by extension, changing climate. This is based on the principle that the stable carbon isotope composition (measured by δ¹³C) of global vegetation shows a bimodal distribution with significantly different values between the C4 and C3 photosynthetic pathways (O’Leary, 1988). Average values reported by Basu et al. (2015) for a tropical plant database from eastern India is −12.7 ± 1.4‰ and −29.6 ± 1.9‰ respectively. The differences in photosynthetic pathways confer a physiological advantage to C4 plants at times of higher temperatures, lower water availability, and low atmospheric CO₂ (Ehleringer, 1978). In the Nilgiris, a relatively less negative δ¹³C value has been interpreted to represent the expansion of grasslands in glacial times, as opposed to a relatively greater C3 biomass contribution in the interglacial period, representative of shola expansion (Caner et al., 2007; Rajagopalan et al., 1997; Sukumar et al., 1993). This is also supported by the experimental work of Joshi et al. (2020) that frost is the primary barrier to establishment of shola saplings in grasslands.

Thus far, no study has investigated the palaeoecology of the montane Western Ghats within the alternative stable states framework. Within such a landscape, changes in spatially-heterogeneous mean vegetation states arise even within a single climate regime due to disturbances such as fire or sustained drought. The small-scale and sharp boundaries across these vegetation states warrants a careful analysis of results from multiple peat cores at a temporal resolution fine enough to capture state changes. It is also important to point out that several authors have studied peatlands in this landscape as an archive of regional changes in the shola-grassland mosaic. The vegetation of the peatlands here is unique to the valley floor, comprising of sedges, grasses, herbs and mosses and is a third unique vegetation state in this landscape. In order to differentiate the vegetation states clearly, we refer to this peat-forming vegetation as sedgeland (since sedges are not found in the grassland or shola) in this study. We believe it is pertinent to include the sedgeland as a third stable vegetation state in this dual stability grassland-shola landscape to understand (paleo)vegetation dynamics more comprehensively, especially given the effects of climate on these states. Our objective therefore was to reconstruct vegetation dynamics at sufficiently high (~multicentennial-to-millennial-scale) temporal resolution, using stable carbon isotopes from two peat cores in the Sandynallah valley, and examine the evidence for alternative stable states.

Methods

Site details

Our study site, the Sandynallah valley, is located between 11°26′32″N 76°38′6″E and 11°26′37″N 76°38′8″E (Figure 1) at an elevation of ~2200 m asl., in the southern Western Ghats, India. This site receives annual average precipitation of about 1240 mm of which ~50% is from the South West and ~30% from the North East monsoon (von Lengerke, 1977). The valley is underlain by charnockite of age ~2527 ± 14 Ma (Samuel et al., 2014). The natural vegetation at Sandynallah is of grassland on the gentle slopes and rounded crests, shola vegetation (stunted evergreen forests) along some folds of the hills and in the toe slope, with distinctive peat-forming wetland vegetation in the valley floor. These distinct vegetation states are described below.

(i) The shola vegetation is typically composed of members of the Lauraceae, Rubiaceae, Symplocaceae, Myrtaceae and Euphorbiaceae (Sukumar et al., 1995). The trees that we observed in the fragmented shola patch near the sampling site are Lasianthus venulosus, Turpinia nepatensis, Rubus elipticus, Withania somnifera, Michelia nilagirica, Buxus sp., Olea sp., Syzigium densiflorum, Rapanea wightii, Mahonia leschenaultii, Rhododendron nilagiricum, Symplocos obtusa, Elaeagnus conferta, Ligustrum roxburghii and the invasive species Acacia mearnsii.

(ii) The grasslands in the montane Nilgiris are reported to have dominant grass species Andropogon lividus, Arundinella vaginata, Digitaria wallichiana, Arundinella mesophylla (Jose et al., 1994). The grassland at Sandynallah also supports ferns, Pteridium sp., herbs of Asteraceae, Auxalidaceae and Begoniaceae, the carnivorous Drosera sp. as well as the invasive Ulex europaeus at present.

(iii) The wetland in the valley floor (referred to as ‘sedgeland’ in this paper) is dominated by Cyperaceae (Fimbristylis spp., Cyperus spp.) and Poaceae (e.g. Eragrosus sp.), several herbaceous members of Scrophulariaceae, Asteraceae, Polygonaceae, Eriocaulaceae, Gentianaceae and a few mosses of Polytrichidae.

Figure 1.

(a) Coring locations in the main valley at Sandynallah showing grasslands on the ridges and gentle slopes, sedgeland in the valley floor and a large patch of woody shola vegetation (top right). Locations 1 and 2 where resistivity survey was conducted, locations of the pit for excavation samples reported in Kavil et al. (2020), Cores 1 and 2 (this study) and (b) a view from the head of the valley, facing North-East.

Caner et al. (2007) point out that the Sandynallah valley, due to its location in the central Nilgiris in an intermediate precipitation and wind strength zone, would be especially sensitive to small changes in temperature and precipitation as opposed to the western Nilgiris. This behaviour is typical of ecotones, whereby climatic changes are preserved more sharply near a vegetation boundary than in the ecological core zone where greater amplitudes of climatic change are necessary to result in vegetation changes. Two sites were chosen for coring from a relatively less disturbed broad valley at Sandynallah. The first site was closer to the edge of the hill surrounding the valley and the second site was ~170 m along the length of the valley from the first site and possibly less disturbed than the earlier site (Figure 1). The valley is part of the Sheep Breeding Research Station (SBRS), maintained by the Tamil Nadu Agriculture and Veterinary Sciences University (TANUVAS), from whom permissions were obtained to carry out this work.

Resistivity survey and Vertical Electrical Sounding (VES)

Vertical Electrical Sounding (VES) using resistivity measurements of the site was first undertaken to inform us about sub-surface structures that might create impediments for manual coring. The survey was conducted by Elite Engineering Consultancy, Bengaluru. The details from the report provided by the company are reproduced here. Four electrodes were driven into the earth along straight line at equal intervals (Wenner configuration). A current I was passed through the two outer electrodes and the earth and the voltage difference V, observed between the two inner electrodes. The current flowing into the earth produces an electric field proportional to its density and to the resistance of the soil. The voltage measured between the inner electrodes is, therefore, proportional to the field. Consequently, the resistivity will be proportional to the ratio of the voltage to current. Resistivity was then calculated based on the formula, ρ = 2π.a.R, where, a = Distance between two consecutive electrodes, R = Observed resistance. The survey was conducted in the main Sandynallah valley (>40 kyr old) locations 1 and 2, at distances of 0.5, 1, 1.5, 2, 5, 7.5 and 10 m distances in all four directions.

Coring and sub-sampling

For achieving a fine temporal resolution in an undisturbed core, peat sampling was carried out using a Belarussian Peat corer (Jowsey, 1966). This is a D-type corer, 5 cm in width and 50 cm in length. The corer is manually operated and has two extension rods which were graduated at the time of coring for sub-surface depths and peat sampled from the Sandynallah valley during the dry season in the months of January–February 2012 using a predefined sampling strategy (De Vleeschouwer et al., 2010). Each master core comprises of nine core units, with 10 cm overlap between successive core units at either ends, procured from bore holes ~30 cm apart to account for lateral variation and to ensure non-disturbance from the coring head (see Ramya Bala et al. (2016) for the detailed sampling design). Two master cores were obtained from two pairs of boreholes ~170 m apart, spanning a depth of 3.7 m (these cores will hereafter be referred to as Cores 1 and 2). The cores were preserved in a −20°C freezer near the sampling site after collection until further use to avoid disintegration during transportation. A band saw with a stainless steel (SS) blade of 1 mm thickness was used to slice the frozen cores at ~1 cm (Core 1) and ~2 cm (Core 2) each (Givelet et al., 2004). Precautions were taken to avoid contamination during these procedures. Gloves were washed between different core sections. The blade was washed well and rinsed with distilled water before proceeding to the next core section. If found soiled even after multiple washes, the band saw was opened, and all parts were cleaned meticulously before proceeding to next subsampling. The slices were placed into labelled sample containers and stored at −20°C.

Chronology

Radiocarbon dating

About 2–3 g of wet samples (Core 1 – 73, Core 2 – 40) were used directly for radiocarbon measurements. The standard Acid-Alkali-Acid pre-treatment method was used for chemical extraction of the desired carbon fraction followed by combustion, CO₂ purification and graphitisation (Nakamura et al., 2003). The ¹⁴C measurements were performed on a Tandetron accelerator mass spectrometer (Model-4130 AMS, HVEE), at the Division for Chronological Research, Nagoya University. We used the HOx-II standard (NIST new oxalic acid standard, SRM-4990C) as a reference of carbon isotope ratios and commercial oxalic acid containing no ¹⁴C (oxalic acid dihydrates, prod. No. 57952, produced from Wako Pure Chemical Industries Ltd., Japan) for ¹⁴C blank subtraction in the data analysis. We repeated three to five runs of ¹⁴C measurements on a group of samples loaded at one time in the ion source of AMS system. We calculated ¹⁴C content of the sample in each run, and an average value and an error were obtained for the results from several runs. This error value is reported as the internal error. We also experimented with pre-treatment protocols to arrive at a laboratory error for our radiocarbon ages. The details of the high-resolution chronology have been described elsewhere (Ramya Bala et al., 2016).

Age-depth modelling

For the radiocarbon ages obtained from the AMS laboratory, we calculated a modified error which accounts for both internal error and estimated laboratory error (described in detail in Ramya Bala et al. (2016)) following the method recommended in Calib (Stuiver and Reimer, 1993). We modelled the age-depth relationship using a published Bayesian age-depth modelling software Bacon v2.2 (Blaauw and Christen, 2011). In tropical peat, inversions are common in the radiocarbon profile, possibly due to deep-rooted vegetation (Ramya Bala et al., 2016). But the Bayesian model does not account for reversals in the profile and hence we restricted our analyses to the section of the profile where major inversions do not occur. In the case of Core 1, the 10 cm overlap between core units was also dated to get an estimate of lateral variation in the substratum. Since the lateral variation in dates was significant, modelling the units as one profile caused the model to reject the lateral variations as inversions. Hence core units in Core 1 were modelled individually to assign dates for the proxy results while in Core 2 we modelled the entire core as one profile since information on the overlap sections was not available.

Stable isotope analysis

Cellulose extraction

The core slices were freeze-dried and homogenised by hand using mortar and pestle. About 1 gm of the powdered bulk peat was used for cellulose extraction following the principles used in the paper pulp manufacture process. This method was first used by Brenninkmeijer et al. (1982), and subsequently adapted by Rajagopalan (1996). The bulk peat was first treated with distilled water to remove soluble impurities, then treated with acid to remove carbonates and acid-soluble organics. Following neutralisation, samples were treated with alkali for a long duration to remove the humic fraction. This was followed by acidified sodium chlorite treatment for delignification: several rounds of addition over 8 h under constant heat (~60°C) on a hotplate. The final solution was sieved (63 µm mesh) and the fibres accumulating on the mesh collected. The fibre-rich water was then boiled with alkali to remove hemi-cellulose from alpha-cellulose. The fibres were then washed with excess water and stored at −20°C. The frozen cellulose in water was subsequently freeze-dried to yield fibres for analysis.

Isotope Ratio Mass Spectrometry (IRMS)

Carbon isotope ratios were measured on cellulose fractions for Cores 1 and 2 on an isotope ratio mass spectrometer (Delta V advantage, Thermo Fisher Scientific, Inc.) connected to an elemental analyser (Flash EA1112, Thermo Fisher Scientific, Inc.) via an interface (Conflo IV, Thermo Fisher Scientific, Inc.) at the Research Institute for Humanity and Nature, Kyoto, Japan. Data were corrected by two internal standards, CERKU 03 (glycine, δ¹³C = −34.92‰) and CERKU 07 (corn starch, δ¹³C = −10.76‰) which were calibrated with multiple international standards (Tayasu et al., 2011). A routine precision of the internal standards was <±0.1‰. We analysed 174 unique cellulose extracts and five triplicates, six duplicates to estimate sample heterogeneity (cellulose is fibrous, hence there is the risk that a bunch of fibres pulled for analysis might be representative of one plant tissue). We found an average standard deviation of 1.06‰ among triplicates (maximum 2.91‰, minimum 0.15‰). All δ¹³C values reported are with respect to Vienna-Pee Dee Belemnite (VPDB).

Disturbance records

We use the information recorded in excavation samples from a pit in the Sandynallah valley close to Core 1 (Figure 1) reported by (Kavil et al., 2021). Fires are reported at ~22 ka (macrocharcoal) and ~3.5 ka (macro-, microcharcoal and charcoal/pollen ratio). They also report peat surface wetness using alkane biomarker derived proxy P_aq, defined as C23+C25/C23+C25+C27+C29. It is the proportion of hydrocarbons from submerged and/or floating aquatic macrophytes relative to the input from terrigenous and immersed plants. Higher P_aq value is an indicator of surface wetness in the peatland which can also be inferred as increased precipitation.

Results

Resistivity survey and Vertical Electrical Sounding (VES)

Up to a depth of 2 m, the peat substratum generally shows a higher resistance per unit depth (Figure 2). From 2 to 10 m, in all directions (except location 1 – East) the medium is homogenous and continuous in stratigraphy without any sudden jumps in values. The peat is homogeneous laterally except for Location 1 – East, which was taken close to the toe slope. Due to the slightly higher elevation in that direction, the profile was terminated at 5 m depth in both locations 1 and 2.

Figure 2.

Resistivity survey results from Sandynallah at two locations.

Age-depth models

A detailed discussion on the results from the high-resolution chronology can be found in Ramya Bala et al. (2016). Both Core 1 (Figure 3a) and Core 2 (Figure 3b) show slower accumulation rate in the glacial period followed by higher accumulation rate in the Holocene, evident from the change in slope in the age-depth models. We also observe a jump in radiocarbon ages in Core 1 from ~6100 cal yr BP at 129 cm depth to ~11500 cal yr BP at 133.5 cm depth. These depths bracket a hiatus in the accumulation.

Figure 3.

Composite Bayesian age-depth models for (a) Core 1 and (b) Core 2.

Stable carbon isotope measurements

Data from δ¹³C measurements on cellulose extracts from Cores 1 and 2 are provided in Supplemental Tables S1 and S2, available online, respectively. The δ¹³C values plotted against the median values from Bacon age-depth models are shown in Figure 4 (along with data on fires from Kavil et al. (2021). The δ¹³C profile for Core 2 is different from that of Core 1, indicating different paleovegetation composition, which is very clear in the glacial period before 10 ka.

Figure 4.

δ¹³C results from cellulose extracts, Cores 1 and 2. Fire events (charcoal) and high surface wetness events (alkane derived ratio P_aq) from Kavil et al. (2020), study conducted on excavation samples proximate to Core 1.

In Core 1, the base of the profile (before the inversion zone) at ~32 ka, shows a C3 vegetation-rich phase. From 25 to 22 ka, well into the Holocene at ~7.5 ka, the C3 vegetation gradually diminishes to give way to C4 dominance. During the Holocene, Core 1 indicates a generally C4 rich vegetation with a short phase of C3 vegetation increase in late-Holocene. In Core 2, we observe a significant glacial-to-interglacial change in mean δ¹³C values. The deepest part of the profile from 45 to 20 ka has vegetation which is relatively more C3-rich than the rest of the sequence. From ~20 to 10 ka there is a marked increase in C4 vegetation. In the Holocene (Figure 5), from 10 ka onwards, the profile shows a continual enrichment in C4 vegetation with minor variability until a practically 100% C4 value of ~−11.5‰ in the ~1000 years at the top of the sequence. At some periods – ~15, 4.9 and between 22 and 11.5 ka, the paleovegetation shows a remarkable 100% C4 composition at cellulose values less negative than −12.7‰.

Figure 5.

An excerpt from Figure 4 for paleovegetation changes in the Holocene.

Discussion

We show divergent δ¹³C values at corresponding times in the last glacial, from ~32 ka to the Holocene, at two adjoining sites within one valley in the montane Nilgiris of the Western Ghats, representative of distinctive vegetation states. We also show evidence for an underlying stratum of contrasting lithology beneath Core 1 (making it more prone to drying), which together with the observed presence of woody montane shola vegetation, makes this site more prone to colonisation. Why would Cores 1 and 2 that are only ~170 m apart in the same valley show such a major difference in past vegetation composition? What role do disturbances such as the fires reported by Kavil et al. (2021) play in maintaining these vegetation states? We explore the various reasons that might have resulted in these patterns within the alternative stable states framework.

Differences between Cores 1 and 2

Site characteristics, differences in topography

At locations 1 and 2, Vertical Electrical Sounding (VES) survey was conducted in four directions. The plot of apparent resistivity versus spacing is expected to be a smooth curve with resistance decreasing with depth, governed only by vertical variability in resistivity. Any reversals in resistance and irregularities in the apparent resistivity curve indicate lateral changes in the substratum (Environmental Protection Agency, USA). The general trend observed in both locations is of decreasing resistance with depth. In both the locations, the surface layers may be compacted/unsaturated due to seasonal drying and hence resistance is high. We know from borehole measurements that peat water levels can be as low as 0.5 m below surface in April when the survey was conducted (Ramya Bala (2015)). In the easterly direction (close to the hill slope) at location-1, the apparent resistivity increases sharply after 2 m, but the same is not observed at location-2 where it is gradually decreasing up to 5 m depth. Hence, in both locations, in the north, south and west directions, a homogenous peat substratum extends up to 7.5 m, while in the east direction at location 1 the peat substratum overlies a contrasting material of higher resistivity, possibly weathered bedrock charnockite extending in from the hill slope or unsaturated clay/gravel. While Core 1 is proximate to the location of this underlying unsaturated basement on the toe-slope, Core 2 is placed well into the middle of the valley away from the boundary.

There is also a difference in hydrology between the two sites. Whereas Core 1 is located to one side of the central channel in Sandynallah; upstream of Core 2 the channel splits into two creating an island-like microsite, insulating this slightly raised peaty substratum from changes to hydrology. While observations in hydrologic differences are qualitative, we contend that the effects of being proximal to a boundary on the toe slope could explain vegetation differences. Ecotones are known to be more sensitive to small changes in climate whereas greater amplitudes of climatic change are necessary to make an imprint on the paleo-record in the ecological core zone (Caner et al., 2007). And indeed, topography and hydrology might be related to the boundary/edge effects and could constructively interfere to explain the observed changes.

Alternative stable states as a framework to explain vegetation mosaics

What maintains ecotones/interface between vegetation mosaics?

The boundary between shola-grassland is maintained primarily by the occurrence of frost in the winter preventing shola saplings from establishing themselves (Joshi et al., 2020; Ranganathan, 1938; Vasanthy, 1988). Fletcher et al. (2014) show that after a catastrophic fire in Tasmania, the new vegetation state gets maintained due to the rhizomatous nature of the colonising species that promotes water logging, presenting a barrier for forest re-establishment. We believe a similar feedback mechanism is operating at Sandynallah, where waterlogging acts as a barrier to shola sapling establishment in the sedgeland. Ranganathan (1938) observed that sholas are very fastidious about soil moisture, but they are ‘not to be found at the bottoms of the valleys or on flat country where the movement of soil water is sluggish or where swampy conditions exist. Good drainage is as important as a sufficiency of soil moisture’. He also reasoned that the shola is able to produce for itself the edaphic conditions necessary for its extension. Using these two factors together, we can surmise that the likelihood for any part of the valley floor to be colonised by shola saplings is dependent on (a) hydrology (determined by climate and topography) and (b) proximity to a pre-existing patch of shola (self-promoting behaviour).

Extending the alternative stable states paradigm to the valley floor at Sandynallah – a conceptual model

In the montane Nilgiris, the shola-grassland vegetation mosaic is widely observed and discussed by ecologists and foresters for nearly a century (Bor, 1938; Meher-Homji, 1967; Ranganathan, 1938). While the grasslands occur on the gentler slopes and ridges, the sholas occupy ‘sheltered valleys, glens, hollows and depressions’ (Ranganathan, 1938). In Sandynallah and other sites that support peat-forming wetland vegetation, a unique niche is created where the wetland flora occupy the valley floor with woody shola members scattered in the boundaries in sheltered locations. The shola saplings are subject to frost-kill higher on the slopes on more open situations, while in the valley floor waterlogging prevents the saplings from establishing themselves. When ground water is lowered, the boundaries of the sedgeland can become dry, allowing for the shola saplings to establish themselves. If similar hydrology gets maintained in sustained drier states, the sholas can grow and expand further by a process involving deep-rooting and drying out the surrounding areas, providing positive feedback to shola establishment. The likelihood that the site can convert from sedgeland to shola is much higher in the case of boundaries as compared to the interior sedgeland due to the elevated topography and self-promoting behaviour of the shola vegetation.

Interpreting the divergent δ¹³C profiles at Sandynallah

Core 1

Our resistivity results show that Core 1 is more susceptible to surface dehydration (due to the underlying topography). Core 1 is also proximate to the boundary where a few woody shola members continue to persist on the slopes. We believe that Core 1 switches between the two vegetation states by virtue of satisfying both the necessary conditions (a) and (b) for shola colonisation of sedgeland. The Indian summer monsoon is documented to have been weaker during the glacial, and hence, the glacial period is believed to be a more arid state (Govil and Divakar Naidu, 2011). Starting at the base of the peat profile at ~32 ka, the C3 rich vegetation indicates that the site around Core 1 could have been occupied by shola vegetation due to drier conditions (Figure 4). From 25 to 22 ka, well into the Holocene at ~7.5 ka the shola vegetation gradually diminishes and gives way to sedgeland. This could be due to gradually ameliorating hydrology (gradual, climate-initiated) or a disturbance to the shola leading to an unstable transition state (disturbance-initiated). The latter interpretation is supported by results from Kavil et al. (2021) who studied sediments from a pit proximate to Core 1 and found charcoal at 22.1 ka. They also present Scanning Electron Microscopy images to show that the macrocharcoal has abundant woody and grass elements, an indicator that woody elements indeed were present at 22 ka (Figure 5 in Kavil et al. (2021)). These results strengthen our claim that shola vegetation existed at that location at 22 ka. The shola vegetation may have been maintained due to the arid climate during the last glacial, but the fires may have destroyed standing woody vegetation, pushing the system into a transition phase. The gradual opening up of the landscape was facilitated by climatic conditions in the last glacial which were not conducive for the immediate re-establishment of sedgeland; however, the forested shola state could not fully re-establish either.

We should also be aware of the hiatus that is observed in Core 1 from ~13 to ~7 ka within a 5 cm difference in depth (See section 3.2.). A similar hiatus is also observed by (Raja et al., 2019) who see a jump in dates from 8 to 16 ka in samples 10 cm apart) at the Parsons valley reservoir nearby. Hence, we do not wish to place emphasis on the samples in this hiatus phase, except that it marks the end of the transition phase leading into a complete sedgeland.

From ~7.5 up to ~4 ka, the sedgeland has re-established completely, dominated by C4 vegetation due to wetter conditions. This is supported by a layer of increased surface wetness at 4.7 ka reported by Kavil et al. (2021) through alkane biomarker proxies. From ~4 to 3.8 ka, C3 vegetation establishment has once again started indicating a gradual change in climate or a disturbance (such as a multi-year drought) that helped the shola colonise. Although the mean-state changes, the magnitude is not as large as the sustained shola establishment of the glacial period. Kavil et al. (2021) report a fire layer at ~3.5 ka with macrocharcoal (abundant grassy elements), which supports the idea that arid conditions prevailed which were favourable for a large fire. Since grassy elements burned, the shola vegetation continues to have established itself at the location for a short spell before returning to a sedge-dominated wetland. The return to sedgeland seems to be a result of increased surface wetness, going by the high P_aq at 1.32 ka.

Core 2

The location of Core 2 creates two distinct possibilities for what the organic material buried here represents: (a) an autochthonous organic signal from local vegetation, (b) an autochthonous signal overprinted by allochthonous material washed in by the central channel at Sandynallah. The location of Core 2 in a slightly raised substratum, distal from boundary effects, can result in a stable sedgeland state, responding primarily to climate and climatic change. Hence the autochthonous vegetation signal here would be of the sedgeland biomass growing and accumulating at the site. However, the island-like topography with the central channel forking just ahead of this location might result in deposition of organic material from the slopes and upstream parts of the valley. This allochthonous material would then imprint the signal from the sedgeland state, potentially altering accumulation rates too. Using stable carbon isotopes alone as evidence, we are unable to diagnose the presence of allochthonous deposit of organics at the site. Using supplementary measurements such as grain-size analysis and additional proxies to identify erosional or depositional regimes at the site may help resolve this complexity.

If we follow the possibility of an autochthonous vegetation signal, the δ¹³C variations reflect changing C3-C4 species composition in the sedgeland, responding primarily to climate. The disturbances reported by Kavil et al. (2021) seem to show clear impacts on Core 1 through vegetation switches, whereas the effects on vegetation are not apparent in Core 2. We would first like to discuss the change in mean carbon isotopic composition from −17‰ in the last glacial to −14‰ in the interglacial, what climatic factors could lead to a relatively higher C3 biomass in the glacial period, and a very strong C4 signal in the current interglacial?

The factors controlling the relative abundances of C3/C4 plants especially in the last glacial and now in the Holocene are widely debated (Huang et al., 2001; Liu et al., 2005; Schefuß et al., 2005; Sinninghe Damsté et al., 2011). Modelling studies suggest mean annual temperature, atmospheric pCO₂, seasonal water availability (distribution of rainfall in the C3 vs C4 growing seasons) or a combination of these determine relative C3/C4 abundances (Ehleringer, 1978; Ehleringer and Björkman, 1977; Ehleringer and Pearcy, 1983; Winslow et al., 2003). A review of relative C3/C4 abundance since the last glacial to Holocene by Rao et al. (2012) seems to indicate that relative C4 abundance increased (temperature-controlled) in the mid-latitudes, whereas relative C4 abundance decreased in the low-latitudes (precipitation-controlled).

In tropical regimes dominated by seasonal rainfall, changes in paleomonsoon variability are closely tied to vegetation changes. The enrichment in C3 vegetation in the montane grassland-forest belts in the well-studied Mt. Kenya in Africa is closely connected with strengthening monsoons in the Holocene, as is the C4 increase to the arid glacial period (Street-Perrott et al., 2004 and references therein). A similar conclusion using precipitation as a driver of changing C3-C4 composition is drawn by Sukumar et al. (1993) and Rajagopalan et al. (1997) who attribute higher C4 abundance in the last glacial maximum to reduced monsoons, which is also supported by Huang et al. (2001). In montane grassland-forest mosaics where frost acts as a barrier to forest establishment, the significantly lower temperatures in the glacial period could also have resulted in forests retreating and grassland advancing. Whether temperature or precipitation driven, these studies look at C3 tree-C4 grass dynamics at landscape levels. If we look at vegetation dynamics between plants of similar morphology, that is, C3-C4 sedges and C3-C4 grasses that dominate the sedgeland, what climatic factors could have driven vegetation changes?

C3-C4 altitudinal distribution profiles of grasses of the entire Neotropical Andes are governed by mean annual temperature (Bremond et al., 2012). Their simulations show that the maximum elevation for C4 grasses during the LGM was 2032 m, compared to 2650 m under modern-day conditions due primarily to temperature differences. The Sandynallah valley occupies a high elevation (~2200 m asl) and consequently cool temperatures (modern mean 14°C). This would support a temperature-driven forcing showing relatively C3 enriched vegetation in the last glacial and, subsequently, a warming trend as seen by the C4 enrichment in the Holocene. The positive feedback between waterlogging and rhizomatous vegetation in the middle of the valley floor may have insulated the vegetation from the effects of reduced precipitation. Following the possibility of temperature-forcing, 45–20 ka shows a relatively cooler glacial period. Beginning at ~18.5 ka there is nearly complete C4 dominance, followed by consistent C4 enrichment into the Holocene excepting for a short departure at ~10 ka. The last millennium preserved at the top of the sequence, the temperature is warm enough to support a nearly 100% C4 vegetation.

However, apart from temperature, distribution of precipitation in temperate grasslands with distinct growing seasons has been said to be an important secondary factor that determines relative C3/C4 abundance (Murphy and Bowman, 2007; Paruelo and Lauenroth, 1996). While this is yet to be established in tropical montane grasslands, we consider this a very significant factor to be investigated, since Sandynallah receives ~50% of its annual precipitation in the summer and ~30% in the winter. If the C3-C4 abundance in the sedgeland responds to temperatures and seasonality of precipitation in distinct growing seasons, this would need careful further analysis and interpretation, which is ongoing.

Conclusions

We propose a conceptual framework for the valley floor at Sandynallah, extending the well-known montane grassland-shola dynamics, to include the peat-forming vegetation ‘sedgeland’. While the shola-grassland is maintained by frost on the hill slopes, shola-sedgeland in the toe slopes and fringes of the valley floor is maintained by waterlogging. Using this framework, we conclude that the diverging stable carbon isotope signatures during corresponding periods in Cores 1 and 2 within the Sandynallah valley are a good candidate profile to study alternative stable states in the past. Core 1 is closer to the boundary of valley and hill slope with an underlying geomorphic feature that makes it more sensitive to changes in hydrology. Hence, Core 1 shows dynamic switches between sedgeland and shola, connected to disturbances (fire), and to changing climate (through hydrology). In contrast, Core 2, located in the centre of the valley floor, represents a stable sedgeland state, having been insulated from dynamic disturbance-based vegetation state switches due to its location in the core sedgeland zone. The C3-C4 vegetation within the core sedgeland are plants of comparable morphology (C3-C4 sedges and grasses) unlike vegetation dynamics observed in forest-grassland mosaics. Given that temperature controls relative vegetation abundance in mixed C3-C4 grasslands, the relative enrichment in C3-vegetation in the last glacial compared to the Holocene points to a cooler glacial period and a subsequent deglacial warming with complete C4 dominance at the top of the profile. Our framework with the sedgeland-shola-grassland stable states in the Nilgiris provides us with the necessary lens with which paleovegetation signals here can be reliably interpreted, since vegetation shifts are not always climate-driven in a vegetation mosaic representative of alternative stable states.

Supplemental Material

sj-docx-1-hol-10.1177_09596836211066592 – Supplemental material for Paleovegetation dynamics in an alternative stable states landscape in the montane Western Ghats, India

Supplemental material, sj-docx-1-hol-10.1177_09596836211066592 for Paleovegetation dynamics in an alternative stable states landscape in the montane Western Ghats, India by Prabhakaran Ramya Bala, Sarath Pullyottum Kavil, Ichiro Tayasu, Chikage Yoshimizu, Kaustubh Thirumalai, Krishnan Sajeev and Raman Sukumar in The Holocene

Footnotes

Acknowledgements

PRB would like to acknowledge the pioneering work of C.R. Ranganathan and his visionary publication in Indian Forester, 1938, for setting in ink his astute observations in the field and conjecturing and theorising the various processes responsible for the spatial distribution of shola forests. This publication left behind sufficient food for thought and a plethora of hypotheses that continue to be tested even after 80 years of its publication. PRB would like to thank Sandeep Pulla for the introduction to the alternative stable states framework, which has benefitted this work immensely. PRB would also like to thank Prof. Iyue and the team at Sheep Breeding Research Station, Sandynallah, for permissions and enthusisastic cooperation for sampling.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: IT would like to acknowledge that this study was partly conducted by the support of Joint Research Grant for the Environmental Isotope Study of Research Institute for Humanity and Nature, and Japan Society for the Promotion of Science KAKENHI grant number 16H02524. RS was a JC Bose National Fellow (supported by Department of Science and Technology, Government of India) and also Visiting Professor, Institute of Advanced Study, Kyoto University, Japan, during the tenure of this study.

ORCID iDs

Prabhakaran Ramya Bala

Chikage Yoshimizu

Supplemental material

Supplemental material for this article is available online.

References

Basu

Agrawal

Sanyal

, et al. (2015) Carbon isotopic ratios of modern C3–C4 plants from the Gangetic Plain, India and its implications to paleovegetational reconstruction. Palaeogeography Palaeoclimatology Palaeoecology 440: 22–32.

Behling

Pillar

(2007) Late quaternary vegetation, biodiversity and fire dynamics on the southern Brazilian highland and their implication for conservation and management of modern Araucaria forest and grassland ecosystems. Philosophical Transactions of the Royal Society B: Biological Sciences 362(1478): 243–251.

Blaauw

Christen

(2011) Flexible paleoclimate age-depth models using an autoregressive gamma process. Bayesian Analysis 6: 457–474.

Bond

(2005) Large parts of the world are brown or black: A different view on the ‘Green World’ hypothesis. Journal of Vegetation Science 16(3): 261–266.

Bond

Silander

Jr Ranaivonasy

, et al. (2008) The antiquity of Madagascar’s grasslands and the rise of c4grassy biomes. Journal of Biogeography 35(10): 1743–1758.

Bor

(1938) The vegetation of the Nilgiris. Indian Forester 64(10): 601–609.

Bowman

DMJS

Perry

GLW

(2017) Soil or fire: What causes treeless sedgelands in Tasmanian wet forests? Plant and Soil 420: 1–18.

Bremond

Boom

Favier

(2012) Neotropical C3/C4 grass distributions - present, past and future. Global Change Biology 18(7): 2324–2334.

Brenninkmeijer

van Geel

Mook

(1982) Variations in the D/H and ¹⁸O/¹⁶O ratios in cellulose extracted from a peat bog core. Earth and Planetary Science Letters 61(2): 283–290.

10.

Caner

Seen

Gunnell

, et al. (2007) Spatial heterogeneity of land cover response to climatic change in the Nilgiri highlands (southern India) since the last glacial maximum. The Holocene 17(2): 195–205.

11.

Chaturvedi

Gopalakrishnan

Jayaraman

, et al. (2011) Impact of climate change on Indian forests: A dynamic vegetation modeling approach. Mitigation and Adaptation Strategies for Global Change 16(2): 119–142.

12.

De Vleeschouwer

Chambers

Swindles

(2010) Coring and sub-sampling of peatlands for palaeoenvironmental research. Mires and Peat 7: 1–10.

13.

Delcourt

(1997) Pre-Columbian Native American use of fire on southern Appalachian landscapes. Conservation Biology 11(4): 1010–1014.

14.

Ehleringer

Björkman

(1977) Quantum yields for CO₂ uptake in C₃ and C₄ plants: Dependence on temperature, CO₂, and O₂ Concentration. Plant Physiology 59(1): 86–90.

15.

Ehleringer

Pearcy

(1983) Variation in quantum yield for CO₂ uptake among C₃ and C4 plants. Plant Physiology 73(3): 555–559.

16.

Ehleringer

(1978) Implications of quantum yield differences on the distributions of C₃ and C4 grasses. Oecologia 31(3): 255–267.

17.

Fletcher

M-S

Wood

Haberle

(2014) A fire-driven shift from forest to non-forest: Evidence for alternative stable states? Ecology 95(9): 2504–2513.

18.

Givelet

Le Roux

Cheburkin

, et al. (2004) Suggested protocol for collecting, handling and preparing peat cores and peat samples for physical, chemical, mineralogical and isotopic analyses. Journal of Environmental Monitoring 6(5): 481–492.

19.

Govil

Divakar Naidu

(2011) Variations of Indian monsoon precipitation during the last 32kyr reflected in the surface hydrography of the western Bay of Bengal. Quaternary Science Reviews 30(27–28): 3871–3879.

20.

Hirota

Holmgren

Van Nes

, et al. (2011) Global resilience of tropical forest and savanna to critical transitions. Science 334(6053): 232–235.

21.

Hoffmann

Geiger

Gotsch

, et al. (2012) Ecological thresholds at the savanna-forest boundary: How plant traits, resources and fire govern the distribution of tropical biomes. Ecology Letters 15(7): 759–768.

22.

Huang

Street-Perrott

Metcalfe

, et al. (2001) Climate change as the dominant control on glacial-interglacial variations in C3 and C4 plant abundance. Science 293(5535): 1647–1651.

23.

Jose

Sreepathy

Mohan Kumar

, et al. (1994) Structural, floristic and edaphic attributes of the grassland-shola forests of Eravikulam in peninsular India. Forest Ecology and Management 65(2–3): 279–291.

24.

Joshi

Ratnam

Sankaran

(2020) Frost maintains forests and grasslands as alternate states in a montane tropical forest–grassland mosaic; but alien tree invasion and warming can disrupt this balance. Journal of Ecology 108(1): 122–132.

25.

Jowsey

(1966) An improved peat sampler. New Phytologist 65(2): 245–248.

26.

Kavil

Bala

Ghosh

, et al. (2020) Climate change and the migration of a pastoralist people c. 3500 cal. Years BP inferred from palaeofire and lipid biomarker records in the montane Western Ghats, India. Environmental Archaeology. bioRxiv: 2020.08.15.252189. DOI: 10.1080/14614103.2021.1959188.

27.

Kavil

Bala

Ghosh

, et al. (2021) Climate change and the migration of a pastoralist people c. 3500 cal. Years BP inferred from palaeofire and lipid biomarker records in the Montane Western Ghats, India. Environmental Archaeology. Epub ahead of print 4 August 2021. DOI: 10.1080/14614103.2021.1959188.

28.

Liu

Huang

, et al. (2005) Summer monsoon intensity controls C4/C3 plant abundance during the last 35 ka in the Chinese loess plateau: carbon isotope evidence from bulk organic matter and individual leaf waxes. Palaeogeography Palaeoclimatology Palaeoecology 220(3–4): 243–254.

29.

Meadows

Linder

(1993) Special paper: A palaeoecological perspective on the origin of Afromontane grasslands. Journal of Biogeography 20(4): 345.

30.

Meher-Homji

(1967) Phytogeography of the South Indian hill stations. Bulletin of the Torrey Botanical Club 94(4): 230.

31.

Moravek

Luly

Grindrod

, et al. (2013) The origin of grassy balds in the Bunya Mountains, southeastern Queensland, Australia. The Holocene 23(2): 305–315.

32.

Murphy

Bowman

(2012) What controls the distribution of tropical forest and savanna?: Tropical forest and savanna distribution. Ecology Letters 15(7): 748–758.

33.

Murphy

Bowman

DMJS

(2007) Seasonal water availability predicts the relative abundance of c3and c4grasses in Australia. Global Ecology and Biogeography 16(2): 160–169.

34.

Nakamura

Oda

Tanaka

, et al. (2003) High precision 14C age estimation of bottom sediments of Lake Baikal and Lake Hovsgol by AMS. Gekkan Chikyu 42: 20–31.

35.

O’Leary

(1988) Carbon isotopes in photosynthesis. Bioscience 38: 328–336.

36.

Overbeck

Muller

Fidelis

, et al. (2007) Brazil’s neglected biome: The South Brazilian Campos. Perspectives in Plant Ecology Evolution and Systematics 9(2): 101–116.

37.

Paruelo

Lauenroth

(1996) Relative abundance of plant functional types in grasslands and shrublands of North America. Ecological Applications 6(4): 1212–1224.

38.

Pausas

Bond

(2020) Alternative biome states in terrestrial ecosystems. Trends in Plant Science 25(3): 250–263.

39.

Pemadasa

(1990) Tropical grasslands of Sri Lanka and India. Journal of Biogeography 17(4/5): 395.

40.

Pemadasa

Amarasinghe

(1982) The ecology of a montane grassland in Sri Lanka: I. Quantitative description of the vegetation. Journal of Ecology 70(1): 1.

41.

Raja

Achyuthan

Farooqui

, et al. (2019) Tropical rainforest dynamics and palaeoclimate implications since the late Pleistocene, Nilgiris, India. Quaternary Research 91(1): 367–382.

42.

Rajagopalan

(1996) Stable isotope paleoclimatology based on tropical peat deposits in Nilgiri hills, southern India. Unpublished Thesis, M.S. University of Baroda, Gujarat, India.

43.

Rajagopalan

Sukumar

Ramesh

, et al. (1997) Late quaternary vegetational and climatic changes from tropical peats in southern India - an extended record up to 40,000 years BP. Current Science 73(1): 60–63.

44.

Ramya Bala

(2015) Evaluating Geochemical Proxies for Paleoclimate Reconstruction in Tropical Montane Peat: A Case Study From the Nilgiris, Southern India. Bengaluru: Indian Institute of Science.

45.

Ramya Bala

Nakamura

Sajeev

, et al. (2016) High-resolution age-depth chronology from tropical montane minerotrophic peat in the Sandynallah valley, western Ghats, southern India: Analytical issues and implications. Quaternary Geochronology 34: 12–23.

46.

Ranganathan

(1938) Studies in the ecology of the shola grassland vegetation of the Nilgiri Plateau. Indian Forester 64(9): 523–541.

47.

Rao

Chen

Zhang

, et al. (2012) Spatial and temporal variations of C3/C4 relative abundance in global terrestrial ecosystem since the last glacial and its possible driving mechanisms. Chinese Science Bulletin 57(31): 4024–4035.

48.

Samuel

Santosh

Liu

, et al. (2014) Neoarchean continental growth through arc magmatism in the Nilgiri Block, southern India. Precambrian Research 245: 146–173.

49.

Schefuß

Schouten

Schneider

(2005) Climatic controls on central African hydrology during the past 20,000 years. Nature 437(7061): 1003–1006.

50.

Sinninghe Damsté

Verschuren

Ossebaar

, et al. (2011) A 25,000-year record of climate-induced changes in lowland vegetation of eastern equatorial Africa revealed by the stable carbon-isotopic composition of fossil plant leaf waxes. Earth and Planetary Science Letters 302(1–2): 236–246.

51.

Street-Perrott

Ficken

Huang

, et al. (2004) Late quaternary changes in carbon cycling on Mt. Kenya, East Africa: An overview of the δ13C record in lacustrine organic matter. Quaternary Science Reviews 23(7–8): 861–879.

52.

Stuiver

Reimer

(1993) Extended 14C data base and revised CALIB 3.0 14C age calibration program. Radiocarbon 35(1): 215–230.

53.

Sukumar

Ramesh

Pant

, et al. (1993) A δ13C record of late quaternary climate change from tropical peats in southern India. Nature 364(6439): 703–706.

54.

Sukumar

Suresh

Ramesh

(1995) Climate change and its impact on tropical montane ecosystems in southern India. Journal of Biogeography 22(2/3): 533.

55.

Suryaprakash

(1999) Plant communities in the montane region of Nilgiris, southern India: Analysis of present and past vegetation based on plant-pollen assemblages. PhD Thesis, Pondicherry University.

56.

Sutra

J-P

Bonnefille

Fontugne

(1997) Etude palynologique d’un nouveau sondage dans les marais de Sandynallah (massif des Nilgiri, Sud-Ouest de l’Inde). Géographie physique et Quaternaire 51(3): 415–426.

57.

Tayasu

Hirasawa

Ogawa

, et al. (2011) New organic reference materials for carbon- and nitrogen-stable isotope ratio measurements provided by Center for Ecological Research, Kyoto University, and Institute of Biogeosciences, Japan Agency for Marine-Earth Science and Technology. Limnology 12: 261–266.

58.

Vasanthy

(1988) Pollen analysis of late quaternary sediments: Evolution of upland savanna in Sandynallah (Nilgiris, south India). Review of Palaeobotany and Palynology 55: 175–192.

59.

von Lengerke

(1977) The Nilgiris: Weather-Climate of a Mountain Area in South India. Wiesbaden: Franz Steiner Verlag.

60.

Winslow

Hunt

Piper

(2003) The influence of seasonal water availability on global C3 versus C4 grassland biomass and its implications for climate change research. Ecological Modelling 163(1–2): 153–173.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB

Paleovegetation dynamics in an alternative stable states landscape in the montane Western Ghats,India

Abstract

Keywords

Introduction

Methods

Site details

Resistivity survey and Vertical Electrical Sounding (VES)

Coring and sub-sampling

Chronology

Radiocarbon dating

Age-depth modelling

Stable isotope analysis

Cellulose extraction

Isotope Ratio Mass Spectrometry (IRMS)

Disturbance records

Results

Resistivity survey and Vertical Electrical Sounding (VES)

Age-depth models

Stable carbon isotope measurements

Discussion

Differences between Cores 1 and 2

Site characteristics, differences in topography

Alternative stable states as a framework to explain vegetation mosaics

What maintains ecotones/interface between vegetation mosaics?

Extending the alternative stable states paradigm to the valley floor at Sandynallah – a conceptual model

Interpreting the divergent δ13C profiles at Sandynallah

Core 1

Core 2

Conclusions

Supplemental Material

sj-docx-1-hol-10.1177_09596836211066592 – Supplemental material for Paleovegetation dynamics in an alternative stable states landscape in the montane Western Ghats, India

Footnotes

Acknowledgements

Funding

ORCID iDs

Supplemental material

References

Supplementary Material

Interpreting the divergent δ¹³C profiles at Sandynallah