Reviewing the evidence for palaeoenvironmental setting of northwestern South Asia for the mid to late Holocene, especially ca 4000—1000 cal BCE, Marco Madella and Dorian Q. Fuller conclude that Harappan urbanism emerged on the face of a prolonged trend towards declining rainfall. They also posit strategic local shifts in agriculture in response prolonged droughts at ca 2200 BCE which might have contributed to de-urbanisation process over the following 200 to 300 years.
The authors observe: “The early food-producing communities (7000 to 4300BCE, see Table 1) are found on the hills of Baluchistan, Northwestern Frontiers and Sindh (however , early sites in the plains of the Indus might be covered by thick alluvium). These are agricultural and pastoral villages. There is subsequently (4300 to 3200 BCE) a consolidation of the early communities with a continuous growth in settled life as well as geographical expansion further into the plains. These communities have a clear cultural continuity with the earlier ones accompanied by technological developments with, e.g., the introduction of the potter’s wheel (Possehl, 2002). The Early Harappan (3200 to 2600 BCE) is characterized by incipient urbanism, growth and further expansion of the farming communities into new territories. The transition between the Early and Mature Harappan (2600 to 1900 BCE) the Harappan Civilisation covers an enormous region with evidence for an elaborated and stratified society, the presence of complex architecture, a sophisticated material culture and an overarching ideology (Kennedy, 1998, Possehl, 2002). It is during the Mature Harappan that there is the setting of urbanization and a rapid growth of cities like Harappa and Mohenjodaro. The Post-uran Harappan sees a demise of the urban character of this civilization with many areas abandoned or more sparingly occupied and a proliferation of smaller, village-like, settlements…Another factor in the Holocene environmental history of the northwestern sub-continent, overlooked in discussions of Quaternary palaeoecology, is the changes in the river drainage system, especially the Ghaggar-Hakra system (see Fig. 1) flowing roughly parallel but separate to the Indus (Agrawal and Sood, 1982; Misra, 1984; Courty and Federoff, 1985; Courty et al., 1989; Agrawal, 1992, pp. 237-242; Possehl, 1997, 1998; Kenoyer, 1998, p. 173; Schuldenrein et al., 2004). Archaeological research in Cholistan has led to the discovery of a large number of sites along the dry channels of the Ghaggar-Hakra river (often identified with the lost Sarasvati and Drishadvati rivers of Sanskrit traditions) (e.g., Mughal 1982, 1997; Dikshit, 1984; Misra, 1984; Flam, 1986; Possehl 1997, 1999). Along the Ghaggar-Hakra there is a relatively high frequency of settlements during the Mature Harappan (2600 to 2000 cal BCE), which suggests a well-watered region that could support agriculture. This may be interpreted either as a river or an inland delta in the area around Derawar (see Fig. 1). By the time of the Painted Grey Ware period (ca 1200 to 500 cal BCE) the river must have been dry, because several sites of this period are found in river bed contexts. This change, thought to have been brought on by tectonic uplift and the capture of the Ghaggar-Hakra headwaters by the Yamuna atershed (Agrawal and Sood, 1982; Agrawal, 1992, pp. 237–242), led to gradual desiccation during the Holocene, which was well underway by the period of the Harappan Civilisation (Courty and Federoff, 1985; Courty et al., 1989). The final desiccation of some of these channels may have had major repercussions for the Harappan Civilisation and is considered a major factor in the de-centralisation and de-urbanisation of the Late Harappan period (Misra, 1984; Chakrabarti, 1995, p. 274; Allchin and Allchin, 1997; Mughal, 1997; Fuller and Madella, 2001). The capture of the Ghaggar- Hakra waters by the Yamuna would have also decreased the water outflow into the Gulf of Kutch, and thus offers an alternative, non-climatic explanation to changes in the sedimentation rate of varve-like sediments off the Karachi coast (cf. von Rad et al., 1999). Although, it is not clear how this would have affected d18O records, the palaeo- Ghaggar-Hakra with more easterly headwaters might be expected to have carried a higher proportion of monsoon derived water, heavier in d18O (cf. Staubwasser et al., 2003)….
“ It can be suggested that the increase in sites next to the Ghaggar-Hakra and Indus during the Mature Harappan period (Figs. 1 and 9) could have been in part due to a concentration of the settlements along river system as rain-fed cultivation became more difficult. Changing cultural practices involving somem kind of water management – even at very low scale – may also have encouraged this concentration of population along water courses, although such practices are conjectural and not yet documented with archaeological evidence. Increasingly concentrated agricultural populations in the river valleys and more intensive methods of cultivation, which included arid-tillage from at least the first half of the third millennium cal BCE (Lal, 1971), may have contributed to greater soil erosion rates, loss of natural vegetation, and movement of wind-blown sediments during the dry summer months (Schuldenrein, 2001)…
“The evidence for the adoption of millets may begin in some areas prior to the rice of the Harappan Civilisation, while rice may begin ca 2200 cal BCE at Harappa. It is, nevertheless, clear that from 2000 cal BCE smaller, more widespread Late Harappan sistes show consistent evidence for these crops (Fuller and Madella, 201; Madella 2003).”
Source: Marco Madella, Dorian Q. Fuller, 2006, Palaeoecology and the Harappan Civilisation of South Asia: a reconsideration, in: Quaternary Science Reviews, 25, pp. 1283-1301 http://www.homepages.ucl.ac.uk/~tcrndfu/articles/MadellaFullerQSR.pdf