Do Stable Isotope Data from Calcrete Record Late Pleistocene Monsoonal Climate Variation in the Thar Desert of India?
Julian E. Andrews
School of Environmental Sciences, University of East Anglia, Norwich, NR4 7TJ, United Kingdom
Ashok K. Singhvi and Ansu J. Kailath
Physical Research Laboratory, Ahmedabad 380009, India
Ralph Kuhn
Forschungstelle Archaeometrice, Max Planck Institut fur Kernphysik, P.O. 103980, 69029 Heidelberg, Germany
Paul F. Dennis
School of Environmental Sciences, University of East Anglia, Norwich, NR4 7TJ, United Kingdom
Sampat K. Tandon
Department of Geology, University of Delhi, Delhi 110007, India
and Ram P. Dhir
Central Arid Zone Research Institute, Jodhpur 3420003, India Received September 29, 1997
Late Pleistocene terrestrial climate records in India may be preserved in oxygen and carbon stable isotopes in pedogenic cal- crete. Petrography shows that calcrete nodules in Quaternary sediments of the Thar Desert in Rajasthan are pedogenic, with little evidence for postpedogenic alteration. The calcrete occurs in four laterally persistent and one nonpersistent eolian units, sepa- rated by colluvial gravel. Thermoluminescence and infrared- and green-light-stimulated luminescence of host quartz and feldspar grains gave age brackets for persistent eolian units I–IV of ca.
70,000 – 60,000, ca. 60,000 –55,000, ca. 55,000 – 43,000, and ca.
43,000 –;25,000 yr, respectively. The youngest eolian unit (V) is
<10,000 yr old and contains no calcrete. Stable oxygen isotope compositions of calcretes in most of eolian unit I, in the upper part of eolian unit IV, and in the nonpersistent eolian unit, range between24.6 and22.1‰ PDB. These values, up to 4.4‰ greater than values from eolian units II and III, are interpreted as repre- senting nonmonsoonal18O-enriched “normal continental” waters during climatic phases when the monsoon weakened or failed.
Conversely, 25,000 – 60,000-yr-old calcretes (eolian units II and III) probably formed under monsoonal conditions. The two peri- ods of weakened monsoon are consistent with other paleoclimatic data from India and may represent widespread aridity on the Indian subcontinent during isotope stages 2 and 4. The total
variation ind13C is 1.7‰ (0.0 –1.7‰), andd13C covaries positively and linearly withd18O.d13C values are highest whend18O values indicate the most arid climatic conditions. This is best explained by expansion of C4 grasses at the expense of C3 plants at low latitudes during glacial periods when atmosphericpCO2was low- ered. C4dominance was overridingly influenced by global change in atmospheric pCO2 despite the lowered summer rain- fall. © 1998 University of Washington.
Key Words: calcrete; paleoclimate; stable isotopes; monsoon;
Pleistocene; India.
INTRODUCTION
It is well established that the Indian Ocean summer monsoon dominates seasonal wind and precipitation patterns and the character of land vegetation in southern Asia (Fein and Ste- phens, 1987). Global climate models (GCMs) have shown that on million-year timescales the summer monsoon is sensitive to land elevation and radiation (orbital) changes (Prell and Kutzbach, 1992). On Quaternary glacial–interglacial time- scales, the strength of the monsoon is apparently affected by differences in sea-surface temperature, sea level, albedo of the
0033-5894/98 $25.00
Copyright © 1998 by the University of Washington.
All rights of reproduction in any form reserved.
240
land surface, and extent and elevation of large ice masses (Prell and Kutzbach, 1992).
The record of paleo-monsoonal variation over the last 40,000 yr in the Indian subcontinent and surrounding Arabian Sea and northwestern Indian Ocean is reasonably well estab- lished from marine records (e.g., Duplessy 1982; Van Campo 1986). Limited data are also available from a southern Indian terrestrial peat (Sukumar et al., 1993; Rajagopalan et al. 1997).
In the Thar Desert (Rajasthan), palynological and geochemical data suggest that monsoonal rainfall was weak during the last glaciation, but was re-established in the Holocene (Singh et al., 1974; Bryson and Swain 1981; Wasson et al., 1984). However, little pre-20,000 yr B.P. terrestrial paleoclimatic information is available from the Indian subcontinent. In Rajasthan, for ex- ample, the distribution of eolian dune sands, associated fluvial deposits, and gypsum lakes suggests a semiarid climate during the last 200,000 yr (Dhir et al., 1992); however, the details of late Pleistocene climate change are unknown.
In this paper we attempt to construct a detailed terrestrial late Pleistocene climatic record in India by studying oxygen and carbon stable isotopes in pedogenic calcrete from the Thar Desert. The potential for using stable isotopes as climatic proxies in pedogenic calcrete has been recognized elsewhere, for example, in Nevada, USA (Amundson et al., 1988), and the Negev Desert, Israel (Goodfriend and Magaritz, 1988). Previ- ous paleoclimatic studies of Quaternary calcrete from the In- dian subcontinent have been limited to alluvial sequences in Pakistan (Pendall and Amundson, 1990) and Bangladesh (Alam et al., 1997). However, these sequences have short stratigraphic ranges or poorly constrained chronology, respec- tively. The exciting aspect of our Thar site is the nearly continuous calcrete record in eolian intervals and the chronol- ogy from thermoluminescence, infrared- and green-light- stimulated luminescence dates. This combination gives us a good opportunity to examine the isotopic data for paleoclimatic variability and relate this to other marine and terrestrial records.
GEOLOGICAL AND CLIMATIC SETTING
The studied section at Shergarh Trijunction in Rajasthan (Fig. 1) is in an abandoned gravel quarry, approximately 70 km west of Jodhpur (Dhir et al., 1992). The section dissects the lower slope of a rocky pediment associated with a prominent rhyolite hill. During the late Quaternary, the edge of this pediment was covered from time to time, and to varying degrees, by veneers of eolian sand (Fig. 1), in which pedogenic calcrete carbonates formed. The colluvial units, therefore, rep- resent sheetwash events, which punctuate the eolian sediment record.
The field area has a semiarid monsoonal climate with annual rainfall between 300 and 400 mm. Potential evapotranspiration is 1800 –2000 mm/yr and mean annual temperature is 26.7°C.
FIELD RELATIONS
The studied section is 4 m thick and contains five dm-scale colluvial units, separated by eolian units with abundant calcrete nodules (Fig. 1). The calcrete nodules have formed in four eolian sand units (units I–IV of Dhir et al., 1992), similar to the youngest, nodule-free uppermost eolian sand unit (Unit V of Dhir et al., 1992). In addition, colluvial unit V of Dhir et al.
(1992) contains a laterally nonpersistent eolian unit not previ- ously recognized. Each eolian unit is divided, with sharp upper and lower contacts, by colluvial units. The colluvium consists of angular gravel, typically rhyolite clasts, 2–5 cm in size, in places coated and weakly cemented by calcium carbonate. The colluvial units contain some reworked calcrete nodules, indi- cating that nodule formation was complete before each stage of colluvial deposition.
The nodules are creamy white, 1–3 cm in size, and irregular in shape. They have grown within millimeters of one another but do not coalesce except at the base of eolian unit I, where coalesced nodules formed a weakly cemented sheet. In this sense the calcretes are mainly at development stage III of Machette (1985), although at the base they are transitional between stage III and IV. The nodular calcrete probably formed at some depth (tens of centimeters) in the soil vadose zone.
CALCRETE PETROGRAPHY
Description
The groundmass of the calcareous nodules is a brown, nonferroan, nonluminescent calcite microspar with subhedral to anhedral fabric. In patches, and often around detrital grain margins, darker, nonferroan, nonluminescent micrite is present, with indistinct cavities or possible tubules. Micritic peloids occur but are not common.
Coarse silt to very fine sand-sized (typically 50 –100 mm) quartz (non- or dully luminescent), feldspar (blue luminescent), and lithic (variably luminescent) detrital grains, typically sub- angular to subrounded, are common. Calcite (orange lumines- cent) silt particles are much less abundant and typically are smaller (,10mm). Detrital grains are dispersed in the micro- sparitic groundmass (Fig. 2a). All detrital grains show micrometer-scale-microspar re-entrants or crenulations on the grain margins (Fig. 2b), and most detrital grains have either a micritic coating (Fig. 2c) or a microsparitic corona (typically ,10mm wide; Fig. 2d). In some cases the micritic grain coat appears to have been partially replaced by microspar. Partial replacement of detrital grains by microspar can leave a ghost of the former grain (Fig. 2a), and in patches, microspar appears to have totally replaced detrital grains. Some detrital grains have been totally or partially dissolved to leave a mold, while the largest detrital grains (1–2 mm) are typically cracked or ex- ploded (Fig. 2e).
Interpretation
Dispersed detrital grains in a microsparitic groundmass (floating grain fabric) is a common calcrete fabric (Tandon and Friend, 1989; Wright and Tucker, 1991). The groundmass precipitated in situ, and the rounded margins of the siliciclastic detrital grains, micrometer-scale microspar re-entrants, micro- sparitic coronas, grain “ghosts”, and molds are all fabrics diagnostic of grain dissolution. The calcite groundmass is thus replacive, and the presence of cracked and exploded grains also implies displacive growth (Buczynski and Chafetz, 1987; Tan- don and Friend, 1989). The darker micrite patches and micritic grain coatings are calcans or calcitans (Fitzpatrick, 1984), thought to be constructed by soil microorganisms, probably fungi (Calvert and Julia, 1983; Wright, 1986), while the indis- tinct cavities and possible tubules may be former root-hair channels or microbial tubes.
These calcretes have mainly alpha fabrics (Wright and Tucker,
1991), i.e., they show few biogenic features, although the pres- ence of cutans and a few root channels suggests a pedogenic origin. The dominantly microsparitic groundmass, partial replace- ment of micrite by microspar, and nonluminescence of all the groundmass components suggest that diagenetic alteration (post- calcrete nodule formation) has been minor.
LUMINESCENCE DATING
Methods
Luminescence dating of the eolian units was done on 125- to 150-mm quartz and feldspar mineral separates. These were extracted following a sequential pretreatment of the samples with 1 N HCl and 30% H2O2, followed by sieving, Franz magnetic separation, and sodium polytungstate heavy liquid separation of the two mineral fractions (,2.58 and 2.58 to 2.7 gm/cm3for K-feldspars and quartz, respectively). An 80-min
FIG. 1. Maps showing the location of the study site (A) in India and (B) in Rajasthan. Position of map B is indicated on map A. (C) A schematic sketch of the study profile, which is on the feather edge of a colluvial pediment. Unit EV contains no calcrete nodules.
FIG. 2. Photomicrographs of typical calcrete fabrics, all in plane polarized light. (a) Floating grain fabric. Note re-entrants on grains (light) indicative of corrosion. In the lower right part of the photograph, grains have been completely replaced by microspar, leaving indistinct grain ghosts. Sample from middle of eolian unit II. (b) Re-entrants on quartz grain, typical of corrosion and replacement by microspar. Sample from upper part of eolian unit II. (c) Micritic coatings (cutans) on quartz grains. Sample from upper part of eolian unit I. (d) Microspar grain coronas on quartz grains. Sample from upper part of eolian unit II. (e) Exploded grain in microspar matrix. Sample from upper part of eolian unit II.
40% HF etch was done on the quartz fraction to remove the alpha skin, but for K-feldspars the limited sample availability precluded this and a calculated correction for the alpha dose was used in the estimation of the equivalent dose.
For K-feldspar grains, the infrared-stimulated luminescence (IRSL) and thermoluminescence (TL) measurements were done sequentially on the same discs, with the premise that a limited duration shine down (12 sec) would not measurably deplete the TL signal (Duller and Botter-Jensen, 1993). IRSL stimulation was done using TEMT 484 infrared diodes emit- ting at 880 nm. The detection system comprised a photon- counting system interfaced with a PC having a multiscaling card. The optical detection window comprised EMI 9635 QA coupled to Corning 7-5915-58 filters. The same set of filters were used for TL recordings.
For quartz, the green-light-stimulated luminescence (GLSL) and TL measurements were done with a xenon lamp filtered with Schott GG 475 1 BG 39 1 KMZ 50/2 filters that provided an excitation maximum at 520 nm (Lang et al., 1993).
The detection window used Schott U 340 1 Hoya MUG 2 filters.
Short shine normalization was used for the K-feldspar IRSL and TL data, whereas weight normalization was used for quartz samples. The K-feldspar IRSL and TL growth curves were fitted to saturating exponentials using the prescription dis- cussed in Felix and Singhvi (1997). The TL growth curves were fitted using standard daybreak equipment software. The early onset of saturation-like behavior in the quartz data and large disc to disc variability precluded application of additive dose methods; hence, the Australian slide method (Prescott et al., 1993) was used to estimate the paleodoses. In view of this, we used the K-feldspar data for final age computation. How- ever, concordance between the ED estimates on quartz and feldspars was good.
The radioactivity assay was done using the thick source ZnS(Ag) alpha counting, and a radioactive equilibrium was used for computation of the elemental concentrations of U and Th and their dose rate. As the principal component of the dose is the beta dose, the samples were finely powdered before counting. Estimation of K1was done by wet chemical methods and ICP AES. A water content of 5% by weight was assumed for all samples. Errors were computed using the methods of Aitken (1985).
Results
Experimental data on equivalent dose estimates for quartz and feldspar mineral separates are shown in Figures 3 and 4, and data for age estimations are shown in Tables 1 and 2. The concordance between the two mineral separates, despite their different photo bleaching behavior, is encouraging and sug- gests that these minerals were well bleached before deposition and were not contaminated by the overlying colluvium. These results also suggest that the K-feldspars do not suffer any age underestimation effect or anomalous fading. This is supported
by the field observations and the inference that calcrete nodule formation was complete before the deposition of the overlying colluvial debris.
Because the calcrete nodules formed postdepositionally in a pre-existing sand, and because the colluvial debris contains no datable material, it is reasonable to suggest that the age limits for calcrete carbonate formation are set by the underlying host sands and the overlying sands. Thus, the calcretes in units I, II, III, IV, and V are assigned age brackets of ;70,000 – 60,000 (taking age 1 sigma and age 2 sigma of the basal and the overlying sands),;60,000 –55,000,;55,000 – 43,000, ;43,000 –;25,000 yr (assuming that precipitation was at its lowest during the last glacial maximum), and,10,000 yr (Fig. 5).
STABLE ISOTOPES
Methods
Samples for isotopic analysis were drilled from the center of individual nodules that had been checked for potential contam-
FIG. 3. Typical IRSL shine down curves, IRSL vs beta dose growth curve and plot of equivalent dose vs IRSL illumination time for ARD-2 (Tables 1 and 2).
inants (e.g., weathered rinds, spar-filled cracks, macroorganic matter). The isotopic chemistry procedures followed Andrews et al. (1993). Isotopic data are reported in delta notation rela- tive to the Vienna Pee Dee Belemnite (VPDB) international scale (Coplen, 1994). The mass spectrometer was calibrated using the NBS 18 and NBS 19 standards. Outliers in the isotopic data were replicated to ensure that quoted values are
precise. Precision for the laboratory standard was better than6 0.06‰ ford13C and60.10‰ ford18O.
Results
Pedogenic nodules from Shergarh Trijunction have a total variation ind18O of 4.4‰, (22.1 to 26.5‰) (Fig. 5). How- ever, within an individual eolian unit, variation ind18O is not more than 2‰ (Fig. 5). The total variation ind13C is 1.7‰ (0.0 to21.7‰). Within an individual eolian unit, variation ind13C is not more than 1.0‰ (Fig. 5). Moreover, d18O and d13C values covary positively and linearly (Fig. 6).
Interpretation
Our interpretations assume that (1) soil carbonate precipi- tated in equilibrium with soil CO2below 25–30 cm in the soil profile (Cerling, 1984); (2) minimal postdepositional diage- netic alteration of the calcrete carbonate occurred, in accor- dance with our petrographic data; and (3) the presence of some cutans and a few root channels implies biologically active soils.
Interpretation of oxygen-isotope data. Thed18O values for calcrete carbonate in eolian units II, III, and the base of IV have a range (24.5 to 26.4‰) similar to that of late Pleistocene (,19,000 yr B.P.) to Holocene (2000 yr) calcretes elsewhere in the Thar desert (Singhvi et al., 1996). These authors suggested that the values are consistent with calcrete carbonate formation from monsoonal rainfall with d18O values in the range of modern monsoonal rainfall in this area. Ramesh et al. (1996) report oxygen and hydrogen isotope compositions of24.1 to 24.8‰ VSMOW and228 to237‰ VSMOW, respectively, for shallow, fresh groundwaters in the Thar Desert. These waters plot on an evaporation trend between the Global Mete- oric Water Line (GMWL) and terminal salt lakes in the region.
Extrapolation of the compositions back to the GMWL suggests original precipitation compositions in the range 25 to 26‰
VSMOW. This compositional range is also consistent with monsoon-derived groundwaters and precipitation compositions in New Delhi to the north northeast (Krishnamurthy and Bhat-
FIG. 4. Typical TL glow curve, TL vs beta dose growth curve and equivalent dose vs temperature growth curve for quartz extracted for sample ARD-2 (Tables 1 and 2). The region between;300° and 350°C was used for age computation.
TABLE 1
Radioactivity, Luminescence, and Age Data
Sample
Count rate
(c/ksec/cm2) K% Dose ratea(Gy/1000 yr) ED Gy Age (103yr)
ARD 1 (eolian V) 0.58 1.56 3.6560.21 296 2 96 1
ARD 2 (eolian V) 0.58 1.49 3.0960.21 286 2 96 1
ARD 3 (eolian IV) 0.46 1.79 3.0860.22 133613 436 5
ARD 4 (eolian III) 0.49 1.35 2.7360.18 150614 556 6
ARD 5 (eolian II) 0.48 1.28 2.6360.18 168611 646 6
ARD 6 (eolian I) 0.55 1.98 3.5060.25 211633 60610
aThis assumes average water content of 5% by weight and a cosmic ray dose of 150mGy/yr. ED, equivalent dose; Gy, gray (1mjoule/mg); K%, wt% potassium ion. Typical errors inacounting and AAS are 3% and 5%, respectively. Eolian units I–V after Dhir et al., 1992, and Figure 5.
tacharya, 1991; Rozanski et al., 1993). Based on the temperature-dependent isotopic fractionation between water and calcite (Kim and O’Neil, 1997; O’Neil et al., 1969), converted to the paleotemperature expression of Hays and Grossman (1991) and using rainfall compositions of 25 to 26‰, the calcretes are calculated to have formed at 15° to 20°C. In comparison, modern vegetated desert soil tempera- tures at 30 –50 cm depth in western Rajasthan vary from 22° to 30°C (November to March) and 30° to 40°C (April to Octo- ber), with diurnal variations at this depth of up to 4°C (Gupta, 1983, 1986).
Pedogenic calcrete carbonate precipitation in monsoon set- tings typically occurs after the rainy season (i.e., after July to August in the modern Indian summer monsoon), following downward leaching of soluble CaCO3and its accumulation at the evapotranspiration front (Courty et al., 1987). Thus, Octo- ber to March is the probable time of formation of these Indian calcretes. During this time period, Rajasthan soil temperatures at 30 –50 cm depth vary between 32°C in October and 22°C in January (Gupta, 1983). Based on this temperature range, the calcrete oxygen isotope composition, and the temperature- dependent isotope fractionation between water and calcite, the isotopic composition of the calcrete precipitating water is con- strained to between21.1 and25.1‰ VSMOW. As this range of values is the same as the freshwater well data reported by Ramesh et al. (1993; Table 1 and Fig. 2) that typically have somewhat evaporated compositions, it implies that the calcrete carbonate probably precipitated from similarly, somewhat evaporated soil water. Moreover, the 4°C diurnal temperature variation could change the isotopic composition of precipitat- ing calcrete by up to 0.85‰ VPDB.
These findings suggest that shifts in calcrete d18O of ,0.85‰ could be generated by diurnal temperature change and should not, alone, be used to infer paleoclimatic change.
The broad paleoclimatic implication of thed18O data for eolian units II, III, and the base of IV, and their similarity with modern calcretes reported by Singhvi et al. (1996), is that the calcrete probably formed under monsoon conditions in the period between 60,000 and 25,000 yr B.P.
In most of eolian unit I, in the upper part of unit IV, and in the nonpersistent colluvial unit V (Fig. 5),d18O compositions range between24.6 and22.1‰ VPDB. These values, up to 4.4‰ greater than values from units II and III, are unlikely to have been caused by a large decrease in precipitation temper- ature (.9°C). One possible explanation for these larger values is that these calcretes formed from nonmonsoonal rainfall (Cerling, 1984) during climatic phases when the monsoon weakened or failed. A characteristic feature of the isotope composition of monsoon rain is a negative correlation between the amount of rainfall and the isotope composition, the so- called “amount effect,” i.e., low d18O compositions in rainy months and highd18O compositions in months where there is little rainfall (Dansgaard, 1964). For example, the magnitude of the shift between dry and wet months for tropical island stations in the IAEA network is, on average, 5‰ (Rozanski et al., 1993). This is similar to the shift in isotope composition of precipitation required to explain the difference between car- bonate compositions of eolian units II, III, the upper part of IV, and units I, lower part of IV, and V.
We cannot rule out the possibility that the calcretes in units I, lower part of IV, and V formed from waters that had undergone increased surface evaporation. However, increased evaporation is also likely to be an indicator of a more arid environment. The general paleoclimatic implications, there- fore, are that arid conditions existed about 70,000 – 60,000 yr ago and sometime after 25,000 yr ago, but before 10,000 yr B.P. (date for eolian unit V; unfortunately, the nonpersistent eolian unit in colluvium V was not dated). The data of Singhvi TABLE 2
Comparison of Equivalent Doses (GLSL/IRSL and TL) for Quartz and Feldspar Mineral Separates
Sample
Equivalent dose
Quartz Feldspar
EDOSL(Gy) EDTL(Gy) EDOSL(Gy) EDTL(Gy)
ARD 1 (eolian V) 256 5 326 8 296 2 2863
ARD 2 (eolian V) 25610 31614 286 2 2763
ARD 3 (eolian IV) 135625 134616 133613 NM
ARD 4 (eolian III) .160a 287634a 150614 NM
ARD 5 (eolian II) 247625b 292644b 168611 NM
ARD 6 (eolian I) ;240b ;250b 211633 NM
Note. NM, not measured; ED, equivalent dose; Gy, gray (1mjoule/mg).
aTreat as estimates only, as the data scatter was large and the luminescence vs beta dose growth curve was nearly parallel to the dose axis precluding more precise estimation of equivalent dose.bEstimates are based on the Australian slide technique (Prescott et al., 1993).
Eolian units I–V after Dhir et al., 1992, and Figure 5. For quartz, the equivalent dose was measured using green-light-stimulated luminescence and thermoluminescence. The equivalent dose for the K-feldspar extracts was measured using infrared-stimulated luminescence and thermoluminescence.
et al. (1996) suggest that wetter conditions developed after 19,000 yr B.P., probably sometime after 14,300 yr B.P. when the monsoon was reestablished following the last glacial max- imum (Van Campo, 1986; Sirocko et al., 1993).
Interpretation of carbon-isotope data. The principal con- trols ond13C compositions in most soil carbonates probably are (1) the relative proportion of C3to C4biomass in the local ecosystem, which have very different mean d13C values (Deines, 1980), and (2) changes in thed13C of soil CO2related to soil respiration rate (Cerling, 1984; Cerling and Quade, 1993).
Cerling (1984) has correlated thed13C of soil carbonate with the fraction of C4flora in the local ecosystem. On this basis the Thar calcrete data imply a flora with 75– 88% C4composition.
The positive covariation ind13C andd18O is important in this context, because thed13C values are highest (ca. 0‰) when the d18O values are also highest, interpreted as indicating the most arid climatic conditions. At first sight, this positive covariation appears inconsistent with empirical data showing that C4plants are favored over C3 plants when night-time temperatures are high (Cerling, 1984) and when rainfall increases (Hattersley, 1983). C4plants would thus be favored by hot, wet summers, i.e., stronger monsoonal conditions, and not by arid conditions.
FIG. 5. Sketch section of the stratigraphic profile (after Dhir et al., 1992, pp. 120 –124), the calcrete isotopic data, and the TL and IRSL ages for the eolian units. EI–EV are eolian units and CI–CV are colluvial units (after Dhir et al., 1992). Black-and-white scale bars represent 1 m. Relative aridity (based on the interpretation of thed18O values) is expressed in the right-hand column.
FIG. 6. Cross plot to show covariation in calcrete stable carbon and oxygen isotopes. Increasingly negatived18O values indicate a stronger mon- soonal climate and vice versa. Times of weak monsoon are interpreted as representing glacial or cold stage conditions when lowered atmospheric pCO2 allowed expansion of C4grasses expressed as less negatived13C values.
The implications are that (1) our interpretation of thed18O data is wrong; (2) another factor is influencing the relative abun- dance of C4versus C3plants; or (3) changes in soil respiration principally control thed13C of soil CO2.
Assuming that our interpretation of the d18O data is correct, an important factor that may be affecting C4dom- inance is change in atmospheric pCO2 during glacial and interglacial times (Cerling et al., 1997; Ehleringer et al., 1997). The photosynthetic efficiency of C3grasses relative to C4grasses varies with both atmospheric pCO2and tem- perature, with C4plants favored by low atmospheric pCO2 accompanied by high temperature (Cerling et al., 1997).
Modeling of this pCO2–temperature relationship (Cerling et al., 1997) suggests that expansion of C4grasses would be expected at low latitudes during glacial periods when atmo- spheric pCO2was lowered. If the Thar C4dominance were overidingly influenced by global change in atmospheric pCO2, allowing expansion of C4 grasses during glacial (cold) stages, despite lowered summer rainfall, it also im- plies that the aridity, caused by weakening of the monsoon, should coincide with glacial cold stages.
Interpreted on this basis, the Thar isotope data suggest that during times of weakened monsoon (e.g., the last glacial maximum) the local ecosystem was dominated by C4 grasses (up to 88%), whereas during times of strengthened monsoon conditions, i.e., between 25,000 and 60,000 yr B.P., C3 plants became more important, reducing the C4 component to about 75%. Dominance of C4 plants during the last glacial maximum and C4 decline during deglacia- tion, as recorded by d13C in soil carbonate, has been sug- gested for semiarid areas in the USA (Cole and Monger 1994). Carbon isotope values from younger calcretes (19,400 –2000 yr B.P.) elsewhere in the Thar (Singhvi et al., 1996) are more negative (27 to 27.8‰) than any of our data suggesting that the C4 component of vegetation was much reduced to 28 –35%, responding to re-establishment of higher atmospheric pCO2after the last glacial maximum.
We noted earlier that changes in thed13C of soil CO2related to soil respiration rate could affectd13C compositions of soil carbonate. In climates fluctuating between increasing and de- creasing aridity, the most arid conditions may have caused a decrease in soil respiration rates, thereby allowing ingress of atmospheric CO2deeper into the soil profile. This mechanism could cause the positive covariation ofd13C andd18O (Fig. 6), although it is difficult to evaluate the relative importance of this effect. If respiration rate is related to effective moisture con- ditions, as discussed by Quade et al. (1989), then the high average annual rainfall (300 – 400 mm) suggests that respira- tion rates were quite high, making the likely impact of this effect rather small. It is also clear that this effect would reinforce that caused by glacial/interglacial changes in C4 versus C3plant dominance.
Other Causes of Positive Covariation ind13C andd18O?
While we think our interpretations of thed18O andd13C data above are most convincing in the light of both the probable controls on isotope chemistry and the general paleoclimatic context discussed below, we note that the data can be inter- preted in other ways that would not implicate glacial/
interglacially driven climatic changes. For example, if evapo- ration of soil water were increased, perhaps by increasing and intensifying the spring dry season, this might explain the shift to more positive d18O values. The increased heating of the landmass might then intensify the monsoonal season, which would be accompanied by an increase in C4 plants as a re- sponse to the warmer and wetter conditions (Hattersley, 1983).
Alternatively, if the late Pleistocene winter dry seasons were at times wetter than present, this might have increased humid- ity and decreased evaporation, leading to a decrease in the mean d18O value of annual rainfall. These cooler but wetter conditions may have favored C3plants, because only combined warmer and wetter conditions appear to favor C4plants (Hat- tersley, 1983).
PALEOCLIMATIC CONTEXT
Our favored interpretation, that Thar paleoclimates were predominantly arid at some time before 10,000 (presumably ca.
18,000 yr B.P. based on other data discussed below), and after 25,000 yr B.P. is consistent with other independent Indian paleoclimatic indicators based on (1) pollen in Thar lake sed- iments (e.g., Singh et al., 1974, 1990), (2) dune activity in the Thar desert (Sarnthein, 1978; Chawla et al., 1992), (3) d13C values for peat from peninsular India (Sukumar et al., 1993;
Rajagopalan et al., 1997), (4) pollen composition in nearshore Indian Ocean sediments (Van Campo, 1986), and (5) d18O values of northern Indian Ocean foraminifera (Duplessy, 1982). The aridity indicated by these climate proxies is thought to have been caused by weakening of the summer monsoon associated with the last glacial maximum (Duplessy, 1982;
Prell and Kutzbach, 1992, and references therein) when surface albedo increased over the snow- and ice-covered Tibetan Pla- teau (Manabe and Hahn, 1977).
There is very little paleoclimatic information from mainland India to compare with our older (pre-25,000 yr B.P.) calcrete isotope record. A period of fluvial aggradation, sediment red- dening, and pedogenesis on the Gujarat alluvial plain between 39,000 and 59,000 yr B.P. is thought to have occurred under moist semiarid climatic conditions (Tandon et al., 1997).d13C data from 20,000 – 40,000-yr-old peat from Nilgiris also show a decrease in C4 components relative to 18,000-yr-old peat, consistent with our inferred monsoonal climatic period be- tween 25,000 and 60,000 yr B.P. These results also agree with rainfall reconstructions based on magnetic susceptibility vari- ations in Chinese loess deposits. Maher and Thompson (1995) inferred that rainfall was relatively high in the early Holocene
and between 55,000 and 25,000 yr B.P., suggesting that the summer monsoon was relatively strong at these times. During glacial times (25,000 –11,000 yr ago) and (75,000 –55,000 yr B.P.), however, rainfall was much decreased across the loess plateau, presumably due to a weakening of the summer mon- soon. If this tentative regional correlation between Indian and Chinese paleoclimate can be corroborated, it could be very important to understanding the driving forces behind Asian climate variation.
There is also clear biogeochemical and lithogenic evidence in marine sediment cores from the Oman continental margin (northwestern Arabian Sea) that the strength of (1) marine upwelling and hence productivity, and (2) wind strength is linked to variations in monsoon activity (Clemens et al., 1991).
These authors noted that the monsoon was strong;50,000 yr B.P., which is within the range of our inferred monsoonal climatic period between 60,000 and 25,000 yr based on the Thar calcrete data. Recently, biogeochemical data from a 500,000-yr record from ODP site 723 in the northwestern Arabian Sea (Emeis et al., 1995) has confirmed that during interglacial periods the strong summer monsoon caused en- hanced upwelling and productivity, and was accompanied by relatively high sea-surface temperatures. The converse was true during glacial periods. The monsoon was clearly weak- ened during isotope stages 2 and 4, the latter linked with the 60,000 –70,000 yr B.P. arid period in the Thar calcrete record.
Finally, we note that calcrete isotopic profiles from north- western Bangladesh, reported by Alam et al., (1997), show two clear shifts ind18O values toward larger values in their weakly calcretized geosols 2 and 4. These shifts are of similar magni- tude to those reported here and are accompanied by positive covariation ind13C, like ours; they are interpreted to represent relatively arid climatic conditions. These undated horizons may correlate with ours and therefore represent a record of widespread aridity on the Indian subcontinent during isotope stages 2 and 4.
CONCLUSIONS
Carbon and oxygen stable isotopes in calcrete carbonate are potentially important recorders of Pleistocene terrestrial paleo- climatic conditions. For a well-dated Thar desert sequence, our preferred interpretation suggests that indicators of nonmon- soonal rainfall, and hence relative aridity (relatively highd18O values), correspond with increasing presence of C4plants (rel- atively highd13C values). The proposed arid periods are con- sistent with paleoclimatic and paleoceanographic data from elsewhere in India and the Indian Ocean, and appear to record widespread aridity on the Indian subcontinent during isotope stages 2 and 4.
C4plant dominance during glacial times is not unexpected because C4plants are favored by low atmospheric pCO2ac- companied by high temperature (Cerling et al., 1997). Expan- sion of C4grasses is thus likely at low latitudes during times of
lowered atmospheric pCO2(e.g., glacial periods) despite the decreased summer rainfall that normally favors C3plants.
Our isotope data, when compared with empirical soil tem- perature data, suggest that shifts in calcreted18O of ,0.85‰
could be generated by diurnal temperature change. This sug- gests that changes in calcrete d18O of ,0.85‰ should not, alone, be used to infer paleoclimatic change.
ACKNOWLEDGMENTS
JEA thanks the INSA–Royal Society Exchange Program for funds to visit the University of Delhi and the field site in India. We thank Sarah Dennis for helping with stable isotope analyses at the UEA laboratory and Ian Marshall, Steve Bennett, Phil Judge, Sheila Davies and Rosie Cullington (UEA) for technical reprographic and secretarial support. This paper is a contribution to the project Quaternary Stratigraphy and Paleoenvironmental History of Thar Desert funded by the Indian Department of Science & Technology (Project ESS/CA/A3-08/92). This work is also a contribution to IGCP-349 on Pale- omonsoons and Desert Margins. We thank Dr. M. M. Sarin for his help with the potassium analysis. A.K.S. thanks the German Research Foundation for a fellowship, and the Max-Planck Institute for Nuclear Physics (Heidleberg) and Professor G. A. Wagner for providing facilities for GLSL analysis. Finally, we thank Thure Cerling and an anonymous referee for their very helpful sugges- tions for interpreting the carbon isotope data and broadening our outlook.
REFERENCES
Aitken, M. J. (1985). “Thermoluminescence Dating.” Academic Press, Lon- don.
Alam, M. S., Keppens, E., and Paipe, R. (1997). The use of oxygen and carbon isotope composition of pedogenic carbonates from Pleistocene palaeosols in NW Bangladesh, as palaeoclimatic indicators. Quaternary Science Reviews 16, 161–168.
Amundson, R. G., Chadwick, O. A., Sowers, J. M., and Doner, H. E. (1988).
Relationship between climate and vegetation and the stable carbon isotope chemistry of soils in the eastern Mojave Desert, Nevada. Quaternary Re- search 29, 245–254.
Andrews, J. E., Riding, R., and Dennis, P. F. (1993). Stable isotopic compo- sitions of Recent freshwater cyanobacterial carbonates from the British Isles: Local and regional environmental controls. Sedimentology 40, 303–
314.
Bryson, R. A., and Swain, A. M. (1981). Holocene variations of monsoon rainfall in Rajasthan. Quaternary Research 16, 135–145.
Buczynski, C., and Chafetz, H. S. (1987). Siliciclastic grain breakage and displacement due to carbonate crystal growth: An example from the Lenders Formation, Permian of North Central Texas, USA. Sedimentology 34, 837–
843.
Calvert, F., and Julia, R. (1983). Pisoids in the caliche profiles of Tarragona, north east Spain. In “Coated Grains” (T. M. Peryt, Ed.), pp. 456 – 473.
Springer, Berlin.
Cerling, T. E. (1984). The stable isotopic composition of modern soil carbon- ate and its relationship to climate. Earth and Planetary Science Letters 71, 229 –240.
Cerling, T. E., and Quade, J. (1993). Stable carbon and oxygen isotopes in soil carbonates. In “Climate Change in Continental Isotopic Records” (P. K.
Swart, K. C. Lohmann, J. McKenzie, and S. Savin, Eds.), pp. 1–36. Geo- physical Monograph 78. American Geophysical Union, Washington DC.
Cerling, T. E., Harris, J. M., MacFadden, B. J., Leakey, M. G., Quade, J., Eisenmann, V., and Ehleringer, J. R. (1997). Global vegetation change through the Miocene/Pliocene boundary. Nature, 153–158.
Chawla, S., Dhir, R. P., and Singhvi, A. K. (1992). Thermoluminescence chronology of sand profiles in the Thar Desert and their implications.
Quaternary Science Reviews 11, 25–32.
Clemens, S., Prell, W., Murray, D., Shimmield, G., and Weedon, G. (1991).
Forcing mechanisms of the Indian Ocean monsoon. Nature 353, 720 –725.
Cole, D. R., and Monger, H. C. (1994). Influence of atmospheric CO2on the decline of C4plants during the last glaciation. Nature 368, 533–536.
Coplen, T. B. (1994). Reporting of stable hydrogen, carbon and oxygen isotopic abundances. Pure and Applied Chemistry 66, 273–276.
Courty, M. A., Dhir, R. P., and Raghavan, H. (1987). Microfabrics of calcium carbonate accumulations in arid soils of Western India. In “Micromorpholo- gie des Sols” (M. Fedoroff and M. A. Courty, Eds.), pp. 227–234. Assoc.
Francaise pour l’Etude du Sol.
Dansgaard, W. (1964). Stable isotopes in precipitation. Tellus 16, 436 – 468.
Deines, P. (1980). The isotopic composition of reduced organic carbon. In
“Handbook of Environmental Isotope Geochemistry” (P. Fritz and J. C.
Fontes, Eds.), Vol. 1, pp. 329 – 406. Elsevier, Amsterdam.
Dhir, R. P., Kar, A., Wadhawan, S. K., Rajaguru, S. N., Misra, V. N., Singhvi, A. K., and Sharma, S. B. (1992). “Thar Desert in Rajasthan.” Geological Society of India, Bangalore.
Duller, G. A. T., and Botter-Jensen, L. (1993). Luminescence from potassium feldspars stimulated by infrared and green light. Radiation Protection Do- simetry 47, 683– 688.
Duplessy, J. C. (1982). Glacial to interglacial contrasts in the northern Indian Ocean. Nature, 494 – 498.
Ehleringer, J. R., Cerling, T. E., and Helliker, B. R. (1997). C4photosynthesis, atmospheric CO2, and climate. Oecologia 112, 285–299.
Emeis, K.-C., Anderson, D. M., Doose, H., Kroon, D., and Shulz-Bull, D.
(1995). Sea-surface temperatures and the history of monsoon upwelling in the northwest Arabian Sea during the last 500,000 years. Quaternary Re- search 43, 355–361.
Fein, J. S., and Stephens, P. L. (1987). “Monsoons.” Wiley, New York.
Felix, C., and Singhvi, A. K. (1997). Study of non-linear luminescence dose growth curves for the estimation of paleodose in luminescence dating.
Radiation Measurements 27, 599 – 609.
Fitzpatrick, E. A. (1984). “Micromorphology of Soils.” Chapman and Hall, London.
Goodfriend, G. A., and Magaritz, M. (1988). Palaeosols and Late Pleistocene rainfall fluctuations in the Negev Desert. Nature 332, 144 –146.
Gupta, J. P. (1983). Some studies on hydrothermal regime and daytime heat fluxes in a desert sandy soil with and without vegetation. Archives for Meteorology, Geophysics and Bioclimatology Series B 32, 99 –107.
Gupta, J. P. (1986). Moisture and thermal regimes of the desert soils of Rajasthan, India, and their management for higher plant production. Journal des Sciences Hydrologiques 31, 347–359.
Hattersley, P. W. (1983). The distribution of C3and C4grasses in Australia in relation to climate. Oecologia 57, 113–128.
Hays, P. D., and Grossman, E. L. (1991). Oxygen isotopes in meteoric calcite cements as indicators of continental palaeoclimate. Geology 19, 441– 444.
Kim, S. T., and O’Neil, J. R. (1997). Equilibrium and nonequilibrium oxygen isotope effects in synthetic carbonates. Geochimica et Cosmochimica Acta 61, 3461–3475.
Krishnamurthy, R. V., and Bhattacharya, K. (1991). Stable oxygen and hy- drogen isotope ratios in shallow ground waters from India and a study of the role of evapotranspiration in the Indian monsoon. In “Stable Isotope Geo- chemistry: A Tribute to Samuel Epstein” (J. H. P. Taylor, J. R. O’Neil, and I. R. Kaplan, Eds.), pp. 187–193. The Geochemical Society, Special Publi- cation No. 3.
Lang, A., Lindauer, S., Khun, R., and Wager, G. A. (1993). Procedures used
for optically and infrared stimulated luminescence dating of sediments in Heidelberg. Ancient TL 14, 7–11.
Machette, M. N. (1985). Calcic soils of the south western United States. In
“Soils and Quaternary Geology of the Southwest United States” (D. L.
Weide, Ed.), pp. 1–21. Geological Society of America Special Paper.
Maher, B. A., and Thompson, R. (1995). Paleorainfall reconstructions from pedogenic magnetic susceptibility variations in the Chinese loess and pa- leosols. Quaternary Research 44, 383–391.
Manabe, S., and Hahn, D. G. (1977). Simulation of the tropical climate of an ice age. Journal of Geophysical Research 82, 3888 –3911.
O’Neil, J. R., Clayton, R. N., and Mayeda, T. K. (1969). Oxygen isotope fractionation in divalent metal carbonates. Journal of Chemical Physics 51, 5547–5558.
Pendall, E., and Amundson, R. (1990). The stable isotope chemistry of pedo- genic carbonate in an alluvial soil from the Punjab, Pakistan. Soil Science 149, 199 –211.
Prell, W., and Kutzbach, J. E. (1992). Sensitivity of the Indian monsoon to forcing parameters and implications for its evolution. Nature 360, 647–
652.
Prescott, J. R., Huntley, D. J., and Hutton, J. T. (1993). Estimation of equiv- alent dose in thermoluminescence dating—The Australian slide method.
Ancient TL 11, 1–5.
Quade, J., Cerling, T. E., and Bowman, J. R. (1989). Systematic variations in the carbon and oxygen isotopic composition of pedogenic carbonate along elevation transects in the southern Great Basin, United States. Geological Society of America Bulletin 101, 464 – 475.
Rajagopalan, G., Sukumar, R., Ramesh, R., Plant, R. K., and Rajagopalan, G.
(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, 60 – 63.
Ramesh, R., Jani, R. A., and Bhushan, R. (1993). Stable isotopic evidence for the origin of salt lakes in the Thar desert. Journal of Arid Environments 25, 117–123.
Rozanski, K., Araguas-Araguas, L., and Gonfiantini, R. (1993). Isotopic pat- terns in modern global precipitation. In “Climate change in continental isotopic records” (P. K. Swart, K. C. Lohmann, J. McKenzie, and S. Savin, Eds.), pp. 1–36. Geophysical Monograph 78, American Geophysical Union, Washington DC.
Sarnthein, M. (1978). Sand deserts during glacial maximum and climatic optimum. Nature 272, 43– 46.
Singh, G., Joshi, R. D., Chopra, S. K., and Singh, A. B. (1974). Late Quater- nary history of vegetation and climate of the Rajasthan Desert, India.
Philosophical Transactions of the Royal Society of London Series B 267, 467–501.
Singh, G., Wasson, R. J., and Agrawal, D. P. (1990). Vegetational and seasonal climatic changes since the last full glacial in the Thar Desert, northwestern India. Review of Palaeobotany and Palynology 64, 351–358.
Singhvi, A. K., Banerjee, D., Ramesh, R., Rajaguru, S. N., and Gogte, V.
(1996). A luminescence method for dating ‘dirty’ pedogenic carbonates for paleoenvironmental reconstruction. Earth and Planetary Science Letters 139, 321–332.
Sirocko, F., Sarnthein, M., Erlenkeuser, E., Lange, H., Arnold, M., and Duplessy, J. C. (1993). Century-scale events in monsoonal climate over the past 24,000 years. Nature 364, 322–324.
Sukumar, R., Ramesh, R., Pant, R. K., and Rajagopalan, G. (1993). Ad13C record of late Quaternary climate change from tropical peats in southern India. Nature 364, 703–706.
Tandon, S. K., and Friend, P. F. (1989). Near surface shrinkage and carbonate replacement processes, Arran Cornstone Formation, Scotland. Sedimentol- ogy 36, 1113–1126.
Tandon, S. K., Sareen, B. K., Rao, M. S., and Singhvi, A. K. (1997).
Aggradation history and luminescence chronology of Late Quaternary semi- arid sequences of the Sabarmati basin, Gujarat, Western India. Palaeogeog- raphy, Palaeoclimatology, Palaeoecology 128, 339 –357.
Van Campo, E. V. (1986). Monsoon fluctuations in two 20,000-yr B.P.
oxygen-isotope/pollen records off south west India. Quaternary Research 26, 376 –388.
Wasson, R. J., Smith, G. I., and Agrawal, D. P. (1984). Late Quaternary
sediments, minerals and inferred geochemical history of Didwana Lake, Thar desert, India. Palaeogeography, Palaeoclimatology, Palaeoecology 46, 345–372.
Wright, V. P. (1986). The role of fungal biomineralization in the formation of Early Carboniferous soil fabrics. Sedimentology 33, 831– 838.
Wright, V. P., and Tucker, M. E. (1991). Introduction. In “Calcretes. Interna- tional Association of Sedimentologists Reprint” (V. P. Wright and M. E.
Tucker, Eds.), Vol. 2, pp. 1–22.