Introduction

Evidences of first organized life-style and agrarian human subsistence are found since the beginning of the Indus Civilization that spread along the Indus and the Ghaggar-Hakra river systems (along the present day India-Pakistan international border) (Giosan et al. 2012; Levey and Burke 1959; Possehl 2002; Weber et al. 2010). While human subsistence during the Early phase was mainly dependent on agricultural and pastoral activities, the Mature phase witnessed a variety of artisan activities and trade of goods with other contemporary civilizations (Park and Shinde 2014). It has been argued that the Indus Civilization experienced significant shifts in the environmental conditions i.e. arrival of monsoonal dryness during the second half of the Mature phase which possibly led to migrations and decline of this civilization (Dixit et al. 2014; Enzel et al. 1999; Giosan et al. 2012; Kathayat et al. 2017; MacDonald 2011; Petrie et al. 2017; Pokharia et al. 2017; Possehl 2002; Sarkar et al. 2016; Sharma et al. 2020a; Singh et al. 1971; Staubwasser et al. 2003; Wright 2010). Did these environmental changes affect agricultural strategies and production, especially from the Early to Mature phase of the Indus era? While palaeo-climatologists are broadly in agreement about monsoonal dryness prevailing during the latter part of the Mature phase (Dixit et al. 2014; Giosan et al. 2012; 2018; Gupta et al. 2003; Kathayat et al. 2017; MacDonald 2011; Sarkar et al. 2016; Staubwasser et al. 2003 Wasson et al. 1984), archaeo-botanists continue to debate its impact(s) on agriculture (Petrie and Bates 2017; Petrie et al. 2017; Pokharia et al. 2011, 2014, 2017; Sharma et al. 2020a; Singh et al. 1971, 1974). These studies yield two schools of thoughts: (i) monsoonal climate played a major role in shaping up the Harappan life-style, especially its subsistence (Bates et al. 2017; Kaushal et al 2019; Petrie and Bates 2017; Petrie et al. 2016; Pokharia et al. 2011, 2014, 2017; Sharma et al. 2020a; Sarkar et al. 2016; Weber 2003) and (ii) Harappan farming ironically arrived during deteriorating monsoonal conditions and was adaptive in nature, since its inception. Geological records largely support the latter view (Berkelhammer et al. 2012; Dixit et al. 2014; Enzel et al. 1999; Giosan et al. 2012, 2018; Gupta et al. 2003; Kathayat et al. 2017; MacDonald 2011; Prasad and Enzel 2006; Staubwasser et al. 2003).

In this communication, we present stable C and N isotopic data of archaeological grains/seeds recovered from an archaeological site 4MSR of western Rajasthan (India) which tends to reinforce the latter hypothesis. In addition to these, we also measured C and N isotopes of soils derived from host habitational sediments. The study site is a rural settlement, it can be well expected that archaeological/ habitational soil-sediments recovered from different cultural layers would have come from nearby arable land. Combined usage of C and N isotopes of crops vis-à-vis host soil-sediments from the studied archaeological site was aimed to provide an integrated information about cropping pattern and agro-hydrological status in the arable Harappan agriculture fields (Agnihotri et al. 2021; Wang et al. 2008). Where macro-botanical grains are expected to provide information about past crop-hydrological status, C and N isotopic data (along with their contents viz. TOC% and TN%) of soils recovered from different cultural periods may provide information about contemporary ecological status i.e. vegetation type (C-3 versus C-4) and in turn dryness of soils (Agnihotri et al. 2021; Araus et al. 1997; Ma et al. 2012; Parker et al. 2011; Peukert et al. 2012; Pokharia et al. 2017; Rosen et al. 1999; Simpson et al. 1999; Styring et al. 2016; Wang et al. 2008). This rationale is predominantly based on the principle that C isotopic data is a good indicator of palaeo-hydrology and vegetation type (Ferrio et al. 2005; Ma et al. 2012; Parker et al. 2011; Wang et al. 2008) and N isotopic data of plants/crops and soils can be governed by both environmental factors as well as agricultural amendments employed in agricultural fields such as manuring, irrigation etc. (Araus et al. 1997; Aguilera et al. 2008, 2017; Bogaard et al. 2013; Ferrio et al. 2005; Lee et al. 2005; Riehl et al. 2014; Styring et al. 2016, 2017; Wang et al. 2008; Wallace et al. 2013, 2015).

The isotopic values (δ13C) can be influenced by different environmental and biological factors, such as light, altitude, temperature water-logging, salinity, atmospheric CO2 concentration, nutrient availability, genotypically and environmentally determined physiological characteristics of the species. (Jones et al. 2021 and reference therein). Semi-arid conditions of the studied site 4MSR (Binjor) situated on the bank of dried Ghaggar river channel (~160 km away from the major urban centre of Indus Civilization i.e. Harappa town) could be an ideal locale to investigate agricultural manifestations (advancements in agricultural practices) versus hydrological status of agricultural fields in the past. For the 4MSR site, twelve (12) cultural layers were identified by archaeologists based on the material culture (such as pottery type, tools, seals etc.) and these cultural layers cover a total time span from ~2900 BCE to ~1800 BCE. Scientific chronologies of cultural layers were established by a combination of AMS and conventional radiocarbon dating of macro-botanical remains and soil organic matter recovered from nine strata (Sharma et al. 2020a and b).

It has to be noted that the majority of the archaeological researches carried out on Indus valley sites provide a host of information about their material-culture, advent of metal technologies, architectural prowess, trade, and overall socio-economic status of ancient settlers (Agarwal 1971; Asthana 1993; Bhan et al. 2002; Giosan et al. 2012; Kenoyer and Miller 1999; Lal et al. 2003; Marshall 1931; Possehl 2002; Sana Ullah 1931, 1940; Sharma et al. 2020b; Shinde 2016; Vats 1940). To the best of our knowledge, studies on Indus agronomy, crop-diversity, strategies on intensification are relatively sparse (García-Granero et al. 2016; Miller et al. 2006, 2015; Petrie and Bates 2017; Petrie et al. 2016; Weber et al. 2010). This study provides the first combined set of C and N isotopic data of Indus crops from a well dated archaeological site spanning the beginning of Early phase to the end of Mature phase. We also made an attempt to contextualize the derived palaeo-agricultural information with available knowledge gleaned from contemporary European and Chinese archaeological sites.

Chronology, archaeological background and macro-botanical details

The site 4MSR (29°12'87.2"N; 73°9'421"E), is situated in the dry alluvial bed of the Ghaggar river in western Rajasthan (Fig. 1A). This area along the India-Pakistan border of western Rajasthan (District Anupgarh) is known to be a semi-arid region. Based on 14C dating of charcoal, archaeological grains (barley, wheat and rice) and habitational soil-sediments, chronology of all the twelve cultural layers were ascertained (please see supplementary Table 1S). The total time-span covered by the obtained chronology has been sub-divided into three time-windows following traditional Indus chronology (Kenoyer 1991a, b). According to this convention, three periods are: the Early phase ca. ~2900−2600 BCE; Transitional phase ~2600−2500 BCE and the Mature phase ~2500−1800 BCE. Fig. 1B, contour map of the site show an excavated trenches with stratigraphic details from natural bed sediment to modern humus i.e. decreasing depth in mean sea level unit. Section in the Fig. 1C and D show the location of soil-sediment sample collected from the Early, Transitional and Mature phase stratigraphic cultural layers. As stated earlier, there are twelve distinct identified cultural layers. The upper layers (one to four) stored remnants of the Mature phase, while the middle layers (five to seven) represented a Transitional phase followed by deeper cultural layers belonging to the Early phase of the Indus era.

Fig. 1
figure 1

A Map showing location of archaeological site 4MSR, Rajasthan along with other Indus sites (created by using mapping software ArcGIS 10.3). B Contour map of the site showing aerial view of the excavated trenches taken by drone. C Section shows the location of sample collection from the Mature Harappan and Transitional phase stratigraphic layers. D Trench representing the sample collection area from the Early Harappan phase deposit

Full size image

Several studies have been conducted along the palaeo-channel of the Ghaggar (erstwhile Saraswati) river, originating in the northwest Himalayas and is supposed to have flown southwest towards the Gujarat Kachchh region (Chatterjee et al. 2019; Ghose et al. 1979; Giosan et al. 2012; Gupta et al. 2004; Joshi et al. 1984; Kar et el., 2004; Lal 2002; Marshall 1931; Oldham 1886, 1893; Possehl 1999; Singh et al. 2017; Yashpal et al. 1980). This relict channel is thought to have catered to several phases of human settlements of the Indus culture in north-western India (Giosan et al. 2012; Lal et al. 2003; Marshall 1931; Mughal 1997; Possehl 2002, Sharma et al. 2020a; Vats 1940 and references therein). The site 4MSR (locally known as Binjor) evolved from a typical agricultural settlement to a major rural craft production centre that manufactured copper artefacts, beads from semi-precious stones and a wide variety of terracotta products and may have exported them to other urban Harappan sites nearby and far-flung areas. A large series of different shapes of hearths, furnaces and kilns with a cluster of multi-purpose workshops for industrial activities indicate a large rural metal-working settlement at the site 4MSR (Sharma et al. 2020b). A range of recovered artifacts of gold and copper, seashells and terracotta such as pendant frames, earrings, beads, spacers, chisels, bangles, needles, fish hooks, big storage pots, twin pots, broken perforated jars, terracotta beads and broken terracotta bangles, weights, seals, terracotta toys (humped bulls) confirmed the industrial nature of the site (Sharma et al. 2020b).

The macro-botanical assemblage recovered from the site is shown in supplementary Fig. 1S (data adopted from Sharma et al. 2020a). In total 199 macro-botanical samples were collected by floating 7122 L (990 L from Early phase, 1825 L from Transitional phase and 4307 L from Mature phase levels) of sediment volume, systematically from different cultural contexts (floors, hearths, pits) during the course of excavation. However, only 156 samples (Early phase=13, Transitional phase=32, and Mature phase=111) yielded the macro-remains (Sharma et al. 2020a). The assemblage was comprising of a variety of cereals and leguminous crops, viz., Hordeum vulgare, Triticum aestivum/durum, Oryza sativa, Setaria sp., Pisum arvense, Lens culinaris, Cicer arietinum, Lathyrus sp. and Vigna sp., etc. Besides these, oleiferous and fibrous crops, viz., Sesamum indicum, Linum usitatissimum and Gossypium sp. were also been recorded (Sharma et al. 2020a). Fig. 2 (data adopted from Sharma et al. 2020a) presents pie charts depicting relative proportions of Indus summer (Oryza sativa, Setaria sp., Vigna sp., Gossypium Sp., and Sesamum indicum) and winter (Hordeum vulgare, Triticum sp., Pisum arvense, Lens culinaris, Cicer arietinum, Lathyrus sp. and Linum usitatissimum) crop species during the Early, Transitional and the Mature phases. It is noteworthy here that summer crops appear to have introduced mainly during the Transitional phase (Fig. 2B). Intensification and diversification of various crops appear to dominate Indus croplands belonging to the Mature phase (Fig. 2C). Presence of Gossypium sp. show an exponential increase from Early to Mature phase is also noteworthy from this rural industrial site (Fig. 2). The abundance of Gossypium seeds during the Mature phase suggests their prolonged cultivation as a fibrous crop owing to favourable climatic conditions. Being an indigenous crop, Gossypium has great importance for agriculture, industry and trade, especially for tropical and subtropical regions. This huge proportion of Gossypium comparing to other cereals and legumes indicate its extensive usage plausibly for trade purposes. The exploitation of Gossypium fabric has also been recorded from several other Indus archaeological sites. Potsherd with the impression of a fabric corroborates the macro-botanical finds recovered from the site and also indicates intensive usage/trade, as this site was an industrial centre and might have involved in the trade activities. The 4MSR archaeological site has yielded various metal artifacts along with a variety of domestic hearths from the cultural layers belonging to the Mature phase of the Indus era (Sharma et al. 2020a and b).

Fig. 2
figure 2

Pie charts showing the relative proportion of crop species during (A) Early phase (absolute count n=109), (B) Transitional phase (absolute count n=438), and (C) Mature phase (absolute count n=1051) of Indus era recovered from the site 4MSR (modified from Sharma et al. 2020a)

Full size image

Materials and Methods

Macro-botanical remains are large enough to be recognized with the naked eye or low-powered microscope (Ford 1979; Fritz 2005; Pearsall 2000). The macro-botanical samples (archaeological grains/seeds; supplementary Fig. 1S) were collected from all the excavated trenches. Major collection areas were floor, hearths and pits (belonging to the Mature phase; from topmost cultural layer number 1 to number 4), soils of the Transitional phase (layer number 5 to 7) and the Early phase (deeper cultural layers number 8 to 12). Botanical grains/seeds were separated from soil using water floatation technique in the field itself (Sharma et al. 2020a). Segregated samples were identified as per taxonomical classification up to the genus and species level and photo-documented (Sharma et al. 2020a). For carbon and nitrogen stable isotopic measurements, segregated archaeological grains of barley (winter crop; n=10), seeds of cotton (summer crop; n=10) and tiny vetch (leguminous weed; n=10) were considered. Well-preserved grains and seeds (preservation grade P3) of aforementioned three taxa were selected for isotopic analysis in order to ensure accurate identification and to minimize any possible isotopic offset from badly preserved grains produced in higher temperatures (Hubbard and Al Azm 1990). Binning of grains/seeds was done to represent different phases of the Indus era i.e. Early, Transitional and Mature phase. It is noteworthy that even though the study site falls under the semi-arid region of western Rajasthan (District Anupgarh), the modern-day landscape appears to be significantly influenced by the Indira canal in the vicinity (since 1980s) which changed arable fields into lush green farm lands (Fig. 1B) (Anonymous 2021).

We also collected soil-sediments from all the twelve stratigraphic cultural layers along with the natural bed sediment (beneath the oldest (bottommost) cultural layer of Early phase) and sediment belonging to modern humus from the vicinity of the archaeological site for measuring their C and N isotopic compositions. The site 4MSR is situated in the dry alluvial bed of the Ghaggar river (Ali 1941; Oldham 1893) which is now extinct. During its active phase, the channel was receiving waters and alluvium from the Himalaya (Kar 2011; 2014). The alluvium of this dry bed is of variable texture mainly representing from sandy loam to clayey sediments (Shyampura and Sehgal 1995). The surface soil in the area is fertile for agriculture (Anonymous 2021).

Carbon and Nitrogen isotopic analyses of archaeological grains/seeds and soil-sediments

We measured the δ13C and δ15N values along with C and N contents of archaeological grains/seeds (and soil-sediments) using an Elemental Analyzer (EA; Pyrocube®, Elementar® Germany) coupled with Isotope Ratio Mass spectrometry (EA-IRMS; Precision®, Elementar® UK) in continuous flow mode. A total of seventy (70) archaeological grains/seeds (of barley, cotton, tiny vetch) were collected from different cultural layers representing all the three phases (Early, Transitional, and Mature). The well-preserved grains and seeds (n=10) of each taxon (barley, cotton and tiny vetch) have been extracted randomly from the different samples of each phase (Early, Transitional and Mature) to cover the broad cultural stretch. The archaeological grain/seed samples were mildly washed to remove any adhering dirt/soil because these samples were collected by the water floatation technique which has already removed the soil from the grain/seed samples. The washed grains/seeds were then dried at 60 °C for 48 hours and then stored in vials. The dried grain samples were mildly powdered using an agate mortar and pestle. ~0.2-1.0 mg weights of these were taken in clean tin cups, which were subsequently pressed into oval shaped pellets. Hence, due to sandy loam nature of soil and ease with which it was cleaned from grain/seed samples, no chemical pre-treatment was carried out to further clean them (such as etching). We followed the analysis with mild cleaning and without pre-treatment to avoid any alteration in the original isotopic compositions as it is reported that there is no significant effect of using pre-treatments to remove contamination from entire or powdered grains (Aguilera et al. 2017). Although Vaiglova et al. (2014) observed the mean differences of 0.2‰ for the archaeological barley between the untreated and treated samples. However, the acid treatment itself cause the alteration in the δ13C signals of cereal grains, e.g. the cereal grains treated with the lowest concentration of HCl (1 M) recorded a mean value of −23.17‰, while a strong acid treatment (6 M HCl) gave slightly more positive mean value −23.06‰ (Aguilera et al. 2017). Therefore, to avoid the alteration of the original signal we decided to not follow the chemical pre-treatment of the samples. It is also important to mention that the process of carbonization do not affect the δ13C values in cereal grains significantly within the range of temperatures between ~~200–400°C) (Araus et al. 1997; Aguilera et al. 2008; Ferrio et al. 2007). There are studies (Charles et al. 2015; Fraser et al. 2013; Kaushal et al. 2019) that have demonstrated that lower charring temperatures between 200-240°C (for about a duration of six hours) under reducing conditions produce morphologically intact carbonized grains. For δ15N signal also, Bogaard et al. (2007) reported insignificant changes for grains carbonized at 230°C for up to 24 hours. Therefore, we chose morphologically intact (well-preserved) archaeological carbonized grains/seeds only upto the preservation class three (Hubbard and Al Azm 1990) for isotopic analyses. It is important to underscore that earlier archaeo-isotope workers have not catagorically reported any significant differences between pre-treated and non-treated grains samples (Aguilera et al. 2017; Kaushal et al. 2019; Lightfoot and Stevens 2012; Masi et al. 2014).

Samples of host soil-sediments, however, were pre-treated i.e. subjected to decalcification to measure the δ13CSOM values (Agnihotri et al. 2021). For this, ~1-2 grams of dried/homogenized sediment aliquots (without sieving to avoid biasness) were treated with 5% Hydrochloric acid and kept for ~8-10 hours (overnight) at room temperature. Treated samples were then washed several times with deionised water to remove any excess of chloride ion. Washed sediment samples were then again dried in an oven at ~50-60°C followed by mildly re-powdering with a pestle in an agate mortar. Finally grounded powders were dried and dried powders were then transferred in clean plastic vials. A total of 51 samples were analysed on EA-IRMS. Data quality of measured C and N contents together with δ13C and δ15N values was checked throughout the analysis using a suite of in-house and international IAEA standards. Accuracy of measured δ13C and δ15N data was better than 0.2‰, and for elemental concentrations, it was better than 5% (based on duplicate analysis). Detailed methodology of the used EA-IRMS facility is published elsewhere (Agnihotri et al. 2020). The isotopic data are reported using the standard delta notations. Standard for determining C isotopic data is Vienna-PDB, while N isotopic data is normalized with 15N/14N ratio of atmospheric N2 (Tables 1 and 2). Chemical and isotopic data are arranged phase wise, constrained by cultural identification of habitational layers (based on materials such as pottery, seal, metal objects etc.) and the determined radiocarbon chronology (Sharma et al. 2020a; Tables 1 and 2).

Table 1 Measured δ13C and δ15N values along with the estimated Δ13C values of archaeological barley, cotton and common/tiny vetch (weed) grains/ seeds. Total number of grains/seeds n = 10 has been selected for each crop/weed from all the three phases. (Abbreviations: Total Nitrogen (TN%), Total Carbon (TC%), Carbon/Nitrogen (C/N weight ratios)
Full size table
Table 2 Measured δ13CSOM and TC, TC%, δ15NTN and TN% values along with calculated TOC/TN, TIC/TOC ratios of soil-sediments from different cultural phases at 4MSR, Rajasthan, India. (Abbreviations: SOM= Soil Organic Matter, TC= Total Carbon, TOC= Total Organic Carbon, TN=Total Nitrogen, TIC=Total Inorganic Carbon)
Full size table

Estimation of carbon isotope discrimination factor (Δ 13 C)

We computed the Δ13C values using measured δ13C values of crop grains/seeds to evaluate crop-water status (Aguilera et al. 2008; 2012; Farquhar et al. 1989; Ferrio et al. 2005; Styring et al. 2016; Wang et al. 2008). Δ13C values were computed following Farquhar et al. (1989) using the formula-

$${\Delta }^{13}{\text{C}} = ({\updelta }^{13}{{\text{C}}}_{{\text{air}}}-{\updelta }^{13}{{\text{C}}}_{{\text{plants}}})/ (1+{\updelta }^{13}{{\text{C}}}_{{\text{plants}}}/1000)$$
(1)

Past δ13Cair values were inferred by interpolating a range of data from Antarctic ice-core records together with modern data from two Antarctic stations (Halley Bay and Palmer Station) of the CU-INSTAAR/NOAA-CMDL network for atmospheric CO2 (ftp://ftp.cmdl.noaa.gov/ccg/co2c13/flask/readme.html). In fact, slightly different δ13Cair values for the Early and Mature Harappan phase (−6.3‰ and −6.4‰ respectively) were used. We used estimated Δ13Ccrop-remains values to ascertain crop-water status (Araus et al. 1997; Aguilera et al. 2008, 2017; Ferrio et al. 2005; Lee et al. 2005; Riehl et al. 2014; Styring et al. 2016, 2017; Wang et al. 2008; Wallace et al. 2013, 2015).

Indus landscapes were resource-rich (in terms of availability of fertile soil and seasonal rainfall in both summer and winters). Numerous archaeological evidences reveal Harappan settlers were using water management practices in dwelling areas as well as in farming (by channelling of river water to crop fields or using ground waters for irrigation purpose). For instance, Harappans apparently evolved a more developed irrigation technology than their predecessors that allowed to exploit the spacious and fertile Indus River basin (Kenoyer 1991b). The network of dams, canals and reservoirs at Indus site at Dholavira indicates elaborated water management system during the Indus time (Bisht 1998–1999). Hence, human efforts might have played a significant role to manage agriculture activities required to feed communities. Several recent studies attempted to glean clues about changes in past agronomy under the changing monsoonal climate (Pokharia et al. 2017; Sarkar et al. 2016; Sharma et al. 2020a). These studies remained devoid of any numerical data to compare hydrological status of Indus farmlands.

Statistical analyses

We conducted non-parametric One-way ANOVA test (Kruskal-Wallis test) which investigates population of data categorized among different phases by retaining or rejecting the null hypothesis. These statistical validation tests were performed for both C and N isotopic data generated for grains/seeds as well as sedimentary layers using SPSS software (version# 21).

Results

Measured δ13C and δ15N values along with the estimated Δ13C values of archaeological grains/seeds viz. barley, cotton, and tiny vetch from all the three phases (Early, Transitional and Mature) are presented in the Table 1. Tiny vetch is an associated leguminous weed of winter crops like barley, wheat, gram and lentil (Zohaib et al. 2014). The crop-water status of the winter and summer seasons was assessed using Δ13C values of barley and cotton crops, respectively. Measured δ13CSOM and TC, δ15NTN, Total Organic Content/Total Nitrogen (TOC/TN) or simply C/N weight ratios, Total Inorganic Content/Total Organic Content (TIC/TOC) values of soil-sediments from different cultural phases have been presented in the Table 2.

Fig. 3 shows the box-whisker plots of the measured δ13C values of archaeological grains of barley and seeds of cotton and weed (tiny vetch) during the Early, Transitional and Mature phases. Average δ13C values of barley for the Early, Transitional and Mature phases were found to be −21.34‰ ± 1.9, −22.55‰ ± 1.6 and −22.75‰ ± 1.7, respectively (Fig. 3A; Table 1). Thus, they were found to have varied in overlapping ranges which can be interpreted in terms of similar hydrological conditions throughout, albeit a marginally drier status for the grains of the Early phase. Summer crops appeared mainly during the Transitional phase (Sharma et al. 2020a) and average δ13Ccotton values during the Transitional phase (−23.44‰ ± 1.8) were found to be marginally depleted compared to those during the Mature phase −22.55‰ ± 2.5, indicating better hydrological status of the Transitional phase (with respect to those in Mature phase) (Fig. 3G). No significant differences were found for the δ13C values for both barley and cotton during all three identified phases. Statistical validations of the aforesaid statements were ascertained using non-parametric one-way ANOVA tests which retained the null hypothesis.

Fig. 3
figure 3

Box-Whisker plots showing the δ13C (A, D, G), δ15N values (B, E, H), and C/N ratios (C, F, I) of barley, Vicia (common/tiny vetch) and cotton respectively, recovered from different cultural phases from the archaeological site 4MSR. Data points (shown in grey dots) display number of seeds/grains measured (n=10; for each crop and weed). The line inside the box represents median, upper half of the box represents upper quartile, while the lower half represents lower quartile. (Abbreviations: EP= Early Harappan Phase; TP= Transitional Phase; MP= Mature Harappan Phase; MO= Modern)

Full size image

The δ13C values of the weed plant (not sown/ seeded crop) ‘tiny vetch’ were also used to gauge environmental conditions of Harappan agricultural fields. Average δ13Ctiny vetch values for the Transitional and Mature phase were −22.91‰ ± 1.4 and −23.97‰ ± 1.3 respectively (Fig. 3D; Table 1). These values also indicated more or less similar hydrological status during the Transitional and Mature phases as yielded by the winter crop barley (Fig. 3 (A).

For comparing the hydrological status of Harappan crops grown at the site 4MSR with the hydrological status of crops (isotopic data) from other archaeological sites around the world, we used estimated Δ13C values of barley, cotton and tiny vetch during different phases (Fig. 4; Table 1). Estimated average Δ13C values for barley from our study site were 15.37‰ ± 2.0, 16.63‰ ± 1.7, 16.74‰ ± 1.8 for the Early, Transitional and Mature phases respectively (Fig. 4A; Table 1). Intriguingly, values during the Early phase were found to be of drier hydrological status compared to those of the Transitional and Mature phases. For summer crop cotton, average Δ13C values for the Transitional phase (17.55‰ ± 1.9) indicated a slightly wetter status compared to that of the Mature phase (16.53‰ ± 2.6) (Fig. 4C; Table 1). In contrast, average Δ13C values of tiny vetch (weed) indicated a drier hydrological status of this winter crop for the Transitional phase compared to that of the Mature phase (17.00‰ ± 1.5 and 18.00‰ ± 1.3 respectively) (Fig. 4B; Table 1). It is important to mention here that these archaeobotanical samples were not pre-treated with any chemicals prior to analysis as there is no significant effect of using pre-treatments to remove contamination (Aguilera et al. 2017), however, there is study by Vaiglova et al. (2014) which reported the mean difference of 0.2‰ for the archaeological barley between treated and nontreated samples. Further, Masi et al. (2014) used estimated Δ13C values to assess the growing conditions of fossil cereal grains without chemically pre-treating them. They also compared this data with the stable carbon isotope data available for other Near Eastern sites. Therefore, our data could be used to compare with other sites accordingly.

Fig. 4
figure 4

Box-Whisker plots showing the Δ13C (A, B, C) of barley, Vicia (common/tiny vetch) and cotton respectively (n=10; for each crop and weed), recovered from different cultural phases from the archaeological site 4MSR. The line inside the box represents median, upper half of the box represents upper quartile, while the lower half represents lower quartile. (Abbreviations: EP= Early Harappan Phase; TP= Transitional Phase; MP= Mature Harappan Phase)

Full size image

Fig. 5A shows box-whisker plots of measured δ13C values of the soil organic matter along with sediment δ15N values and C/N weight ratios from different cultural phases of the same occupation during the Indus era (Fig. 5B, C). C/N ratios of soils primarily indicate nature of soil, but they could also mimic environmental dryness i.e. higher the C/N ratio drier the soil is (Jiao et al. 2016). Average δ13C values of soil organic matter during the Early, Transitional and Mature phases (−23.54‰ ± 1.4, −22.65‰ ± 0.3, −20.40‰ ± 1.9, respectively) depict a conspicuous enhancing trend from Early to Mature phase (Fig. 5A; Table 2). Statistical validation of aforesaid trend and distinctiveness of δ13C values of soil organic matter during different phases (the Early, Transitional and Mature phases) was ascertained by the non-parametric One-way ANOVA test which rejected the null hypothesis (p = 0.000). This enhancing trend could be interpreted as prevailing aridity in the region and it is well supported by δ15Nsoil-sediment values which also show an increasing trend (validated by One-way ANOVA test; p = 0.015) from Early (9.10‰ ± 2.5) to Mature phase (11.98‰ ± 2.8) via Transitional phase (9.11‰ ± 0.9) (Fig. 5B; Table 2).

Fig. 5
figure 5

Box-Whisker plots showing (A) δ13CSOM, (B) δ15NSediment values, and (C) C/NSediment ratios of soil-sediments from different cultural phases. The line inside the box represents median, upper half of the box represents upper quartile, while the lower half represents lower quartile. The values lies outside the min and max range are outliers. (Abbreviations: EP= Early Harappan Phase; TP= Transitional Phase; MP= Mature Harappan Phase; MO= Modern)

Full size image

This increasing trend in δ15Nsoil-sediment could, however, may be due to complex mixture of environmental and agricultural intensifications (crop rotation and manuring activities) (Aguilera et al. 2008; Boggard et al. 2007, 2013; Bol et al. 2005; Choi et al. 2006; Fraser et al. 2011; Senbayram et al. 2008; Styring et al. 2016, 2019; Szpak et al. 2014). To investigate various possibilities, δ15N values of archaeological grains were used in tandem with TN contents. Fig. 3B displays δ15Nbarley values during the Early, Transitional and Mature phases, (7.72‰ ± 1.8, 8.16‰ ± 2.5, 11.17‰ ± 7.2 respectively) fall within overlapping ranges (validated by one way ANOVA test; p = 0.196), but show an increase from Early to Mature phase. It could be, however, noticed that intra-phase variability in δ15Nbarley values (depicted by standard deviations) overall increased from Early to Mature phase. Total Nitrogen contents of barley grains showed overlapping ranges from the Early (1.81% ± 0.6) to Transitional (2.34% ± 0.8) and then Mature phase (2.42% ± 0.5) (Table 1). The enhancing intra-phase variability of δ15Nbarley values from the Early to the Mature phase may be a combined effect of changes in cropping pattern and agricultural intensification adopted by Harappan farmers.

For the summer crop, δ15Ncotton values during the Transitional phase show much larger variability (10.29‰ ± 7.8) compared to that of the Mature phase 11.00‰ ± 3.6, most likely due to crop-diversification adopted during the Transitional phase and its impact on arable soil (Fig. 3H; Table 1). However, δ15Ncotton values did not show any statistically significant difference (validated by one way ANOVA test; p = 0.799) in the Early, Transitional and Mature phases (p = 0.450). Average TN contents of cotton crops also do not show significant variation (validated by one way ANOVA test; p = 0.768) from the Transitional phase to the Mature phase (4.09% ± 1.2 in Transitional phase and 4.25% ± 1.2 in Mature phase; Table 1). The weed plant (tiny vetch) also showed a larger degree of variability in δ15N during the Transitional phase (4.18‰ ± 2.8) compared to that for the Mature phase (2.68‰ ± 2.0) (Fig. 3E; Table 1). Besides, tiny vetch showed the lower δ15N values as expected for the leguminous plants (Fig. 3E; Table 1). C/N values for archaeological barley grains shown in Fig. 3C indicate better degree of preservation for grains during the Transitional and Mature phases compared to those from Early phase. Two higher (outlier) values for instance, clearly due to poorer preservation. Similarly, the C/N ratios of tiny vetch and cotton seeds also showed the good degree of preservation (Fig. 3F, I).

Discussion

Average δ13C values of soil organic matter indicates growing aridity from the Early to Mature phase (Fig. 5A), which could also be contributed by the C4 crop proportion as indicated by the archaeobotanical assemblage i.e. relatively high proportion of Setaria sp. (Fig. 2C) (Sharma et al. 2020a). If the enriched δ13C values of sediments were due to higher cultivation of C4 millets then also it corroborates growing dryness during the later part of Mature phase which might have forced the farmers to grow a drought resistant crop rather than C3 barley as a crop-management practice at the time of prevailing dryness in the region. However, interestingly the carbon isotope data of archaeological seeds/grains grown during both winter and summer seasons (Figs. 3(A) and (G) do not show such aridity trend clearly indicating mitigation of water requirement by using water-management practices. Our earlier work at Khirsara (23º27'N, 69º03'E), a Harappan site in the semi-arid region of Kachchh, Gujarat also demonstrated a major shift in crop-assemblage (towards drought resistant millet-based crops) well supported by a significant enhancement in the δ13C values of soil organic matter at ~2250 BCE (Pokharia et al. 2017). Cultural continuity was maintained despite this significant crop-shift likely enforced by prevailing monsoonal aridity. Several regional geological records do present evidences for significant monsoonal dryness (Dixit et al. 2014, 2018; Prasad et al. 2014; Staubwasser et al. 2003).

The enriching trend in δ13CSOM values (Fig. 5A) indicate environmental conditions possibly varied from wetter to drier conditions. This inference could be supported by macro-botanical assemblage that shows the arrival of drought resistant crops (e.g. millets) during the late Mature phase as mentioned above (Sharma et al. 2020a). However, it is also noteworthy here that the proportion of water-demanding crop especially cotton is increased during the Mature phase despite of growing aridity in the region suggests cultivation of crop for the trade purposes. It seems that the Harappan farmers have used their efficient crop management skills for the shifting of crops as per their social need and basic requirements. During Mature phase, the food crop barley got reduced with the increasing proportion of millets, however, the proportion of fibre crop cotton got increased possibly for the trade indicates economic exploitation of crop. Plausibly the existing trade practices at the site would have also utilized cotton as a fibre trade. Further, the cultivation of cotton in growing aridity along with the isotope values of archaeobotanical remains together hints the crop-water management at the settlement. The farmers would have managed water requirement of crops through river channelling or irrigation practices. Earlier, geological repositories of north-western India also have shown gripping aridity during the increasing craft (metallurgical) activities of Harappans based on the recovered material-culture (Dixit et al. 2014; Kathayat et al. 2017; Sarkar et al. 2016; Sharma et al. 2020b). Taken together, it appears that there was a gradual transformation in the agriculture prowess of Harappan farmers from the Early to Mature phase which appears to have outplayed prevailing aridity in the region. All these evidences suggest that the agriculture continued to the end of the Mature phase (~1800 BCE) together with upcoming industrial activities. δ13C and δ15N data of studied (well preserved) archaeological grains, however, do not show any significant changes. This observation led us to infer that it is likely that Harappan farmers in the western Rajasthan area was able to manage their agriculture by the application of other means such as crop shifting, irrigation and similar interventions. Cultivation of cotton (water-demanding crop) in relatively high proportion during Mature phase also supports the faming strategies which has been followed to cope up the trade requirement despite of growing aridity in the region. More intrusive analyses of the Indus agronomical treasures are needed. Drawn inferences, however, appear to be well corroborated by the analysis of botanical assemblage from the site which noticeably shows (i) crop-diversification during the intermediate Transitional phase and (ii) the advent of millet-based crops during the later part of the Mature phase, attributable to the increased monsoonal aridity in the region (Sharma et al. 2020a).

Routson et al. (2019) have noted a decreasing trend in net precipitation during early to mid-Holocene globally (for mid-latitudes). This inference is well corroborated by regional hydro-climatic records of the Indian monsoon (Flietmann et al. 2003; Gupta et al. 2003). It has been demonstrated that the Δ13C values of archaeological agricultural grains (such as barley) recovered from the different archaeological sites (along with the sites falling under semi-arid to arid zones) typically ranging between 15.0 to 22.0‰ were grown under ~100-250 mm of mean annual precipitation (Araus et al. 1997; Aguilera et al.2008; Ferrio et al. 2005; Gron et al. 2017, 2021; Kanstrup et al. 2014; Wallace et al. 2013, 2015; Vaiglova et al. 2020). Araus et al. (1997) have shown lower Δ13Cbarley values from archaeological sites in the Fertile Crescent and western Mediterranean (Spain) as indicator of poorer water status during arid epochs. Numerous other studies also have demonstrated that the lower Δ13C values resulting from the poor to moderate water availability reflected by the grains grown in drier conditions (Ferrio et al. 2005; Riehl et al. 2014; Wallace et al. 2013, 2015). Studies have reported Δ13C values of barley ranging between 16.2 to 18.0‰ during mid to late Holocene (~3000−1800 BCE) (Aguilera et al. 2008, 2012; Ferrio et al. 2005; Flohr et al. 2011; Mora-Gonzalez et al. 2016; Reihl et al. 2014; Wallace et al. 2013, 2015). Araus et al. (1997) suggested Δ13C value ~18‰ for barley crop imply irrigation practices in the arable fields (Mora-González et al. 2016). However, Δ13C values below 16-17‰ are generally regarded as as a moderately poor water availability. Average Δ13C values for barley grains grown during the Early phase was found to be 15.37‰ ± 2.0, indicating marginally drier status of grains compared to those during Transitional phase (16.63‰ ± 1.7) and the Mature phase (16.74‰ ± 1.8) (Fig. 4A). The overlapping Δ13C values of archaeobotanical grains/seeds thus indicating the practice of water-management for agricultural produce irrespective of changing environmental conditions in the region (Dixit et al. 2014; Enzel et al. 1999; Giosan et al. 2012; Prasad et al. 2014; Sarkar et al. 2016; Singh et al. 1974). Δ13C values of cereals grown during the Mature phase were found to be higher (Fig. 4A), thus indicate that the farming strategies has been practiced in an arid Indus region probably anthropogenically managed (rain fed + irrigation) (Wallace et al. 2013) due to the increasing dryness in temporal domain (during end part of the Mature phase). The study of the oxygen isotope of two foraminifer species Neogloboquadrina dutertrei and Globigerinoides sacculifer from the Indus River delta in the Arabian Sea also provided the evidence for the seasonal changes in the Indus region (Giesche et al. 2019). This study recorded strong winter monsoon between 4.5 to 4.3 kyr and reduction in both winter and summer rainfall during 4.1 Kyr and also suggest the growth of Indus urban centres coincided with increased winter rainfall, however, the de-urbanism and change in subsistence strategies followed a reduction in both winter and summer season rainfall (Giesche et al. 2019). Thus, the Δ13C values of barley grown during the Mature phase in growing aridity along with the cultivation of water demanding crops such as cotton and linseed supports our hypothesis of farming strategies such as water management or river channelling practiced by the Indus farmers (for site chronology please see supplementary Table 1S). While Giosan et al., (2018) provided evidence for a stronger winter monsoon between ~4.5 and 3.0 Kyr. Kathayat et al. (2017) provided evidence for drier conditions between ca. 4.3 and 3.3 Kyr from a detailed reconstruction of the summer monsoon in the Harappan domain. Our macro-botanical data (Sharma et al. 2020a) clearly indicate that the Harappan farmers shifted their cropping strategy from winter crops to summer crops by growing millets, cotton and leguminous crops like Vigna sp.

Average Δ13Ccotton values varied in the range 17.55‰ ± 1.9 for the Transitional phase and 16.53‰ ± 2.6 for the Mature phase (Fig. 4C; Table 1), indicating well-watered conditions (rain fed via summer precipitation + irrigation) during the Transitional phase. It may be due to change in course of summer monsoon that reached the western Rajasthan Indus farmlands (~2600-2500 BCE) and supported crop-diversification. Summer monsoon however, appears to have subsequently declined toward late Mature phase as evidenced by arrival of millet-based cropping pattern (Sharma et al. 2020a). Similar observations have also been made in our earlier studies (Pokharia et al. 2017). We surmise that the Harappan farmers exploited both the seasonal rainfall and other means of water management in arable lands, such as steering river waters into their agricultural fields. More isotopic data of archaeological grains from other peripheral archaeological sites are needed to garner much deeper insights into the Indus farming especially for the later part of the Mature phase (~1900−1800 BCE) of the Indus era.

δ15N values of the crop-remains in conjunction with soil-sediment have been used to assess aridity in farmlands and agricultural intensifying efforts (e.g. manuring) (Agnihotri et al. 2021; Bogaard et al. 2007; Kanstrup et al. 2012; Styring et al. 2017). δ15N values of host soil-sediment largely reflect land-use history; larger the magnitude greater the reworking of arable soils (Peukert et al. 2012). Enhanced δ15N values of host soil-sediment could also be due to enhanced aridity (loss of water taking out lighter isotope bound nitrogen from soil). Manuring of farmlands with animal dung generally result in enhanced δ15N values of host soil-sediments (Bol et al. 2005; Szpak et al. 2014). For instance, the effect of cattle dung manure in farmlands recorded δ15N values ranging between +2‰ to +8‰, while the values for pig manure range between +15‰ to +20‰ (Bateman and Kelly 2007; Szpak et al. 2014).

To deduce the factual interpretation of δ15N data, Nitrogen concentration data (TN %) greatly help in deciphering the aforementioned processes. Aridity conditions would favour higher δ15N values with lower TN contents, while organic manuring would result in higher TN contents (due to richness of nitrates and ammonia in soils) along with higher δ15N values (Fuertes-Mendizábal et al. 2018; Gron et al. 2021; Szpak 2014). TN contents in farmlands could also be enhanced by cropping leguminous plants (e.g. pulses) that fix nitrogen from the atmosphere. In this kind of agricultural intensification, agricultural grains/seeds would show enhanced TN contents but lower δ15N values (as δ15N of atmospheric N2 is regarded as 0.0‰; van Klinken et al. 2000). Similar effect would be seen in case of application ammonia based fertilizers (urea) as these contain large pool of reactive Nitrogen that has been fixed from atmospheric source (using Haber process) (Bateman et al. 2005). This case, however, is applicable only in modern-day agricultural fields, as this practice was introduced in India after the 1970’s.

Archaeological barley grains from the studied site (4MSR) show an enhancing trend in δ15N values in terms of their intra-phase variability (as determined by standard deviation) from the Early to Mature phases. Macro-botanical data revealed a conspicuous crop-diversification was conducted during the short Transitional phase via adopting leguminous crops (pulses) (Sharma et al. 2020a). This short duration of Transitional phase also witnessed marginally enhanced TN contents (p = 0.084) (Table 1). This observation could be interpreted in terms of successful crop-shifting strategy adopted by Harappan farmers to enhance soil nutritional health and their agricultural produce during the Transitional phase which continued towards the end of Mature phase. As a matter of fact, a significant diversification of crops (towards pulses) was inferred with two-fold enhancement (in abundance) from the Early to Transitional phase, and further two-fold enhancement from Transitional to Mature phase (using macro-botanical data from this site earlier; Sharma et al. 2020a).

Earlier in controlled agronomic conditions, Bogaard et al. (2013) quantified ancient manuring practice from the measured δ15N values of modern bulk cereal samples (barley and wheat) cultivated in Europe and noted higher δ15N values (ranging from 6-9‰) correspond to high manuring rates (≥35+ tons/hectare) while lower values (≤3.0‰) corresponds to long-term unmanured cultivation. Fraser et al., (2011) also recoded δ15N values of modern bulk cereal samples (wheat and barley) grown in Germany under different manuring rates and quantified δ15N values (1) 0.0 to 3.0‰ as low manuring, (2) ~3.0 to 6.0‰ as medium manuring and (3) above 6.0‰ as high manuring. Nitsch et al. (2017) published δ15N values of barley ranging between 5.4 to 6.8‰ for Early Bronze Age and 4.0‰±1.2 for Late Bronze Age from Archontiko. In the context of the above mentioned data, our isotope dataset would possibly support a manifestation of crop-management through diversification, for instance, by cultivating more leguminous crops.

Conclusions

Following conclusions could be drawn from the study-

  • δ13C values of soil organic matter clearly indicate a progressive regional aridity i.e. a transition from wetter to drier conditions. The growing aridity was found to have well corroborated by the crop-assemblage data that recorded dominance of millet-based crops during the later part of Mature phase (Sharma et al. 2020a).

  • The δ13C values of both winter and summer crops (barley and cotton) do not indicate any significant difference in the hydrological conditions among different Indus phases (Early, Transitional and Mature phase) indicating adoptive water management practices of Indus farming.

  • Observed scenario indicate that the Harappan farmers were most likely managing the hydrological status by other means such as channelling river waters into their agricultural fields.

  • δ15N soil-sediments (along with Total Nitrogen contents) indicate progressive agricultural intensification from Early to Mature phase, which yielded better crop-quality and soil-fertility under prevailing aridity.

  • Lower δ15N values (with enhanced Total Nitrogen contents) of weed (tiny vetch) demonstrate plausible impact of crop-diversification and efficacy of crop-management by sowing pulses during the Mature phase.

  • It appears that Harappan farmers were capable of changing their farming strategies during the Mature phase.