Monday, 13 April 2009

Rice Watch: Rice, methane and an early start to global warming

Last July, Bill Ruddiman and Chinese colleagues from the Institute of Geology published a artcile looking at the correlation between a compilation of archaeological evidence for rice in China and the divergence of global methane levels, towards higher levels, from the decrease that is expected based on previous interglacial patterns. This article appeared in Quaternary Science Reviews, and attracted a news note in Science "Was China an Early Emitter", but this has important implications for archaeologist, or rather it highlights the importance of the archaeology of early agriculture to wider issues to do global climate change. Below are my first reactions at the time of publication:  

This is an important and compelling study, and represents the first attempt to ground truth that hypothesis of major prehistoric anthropogenic methane production may be linked to intensive rice cultivation. It is a clear indication of the importance of long-term history to better predictive modelling, and the role that archaeology can play in contributing to a more nuanced understanding of the human-ecosystem feedbacks in the long-term. I think it points towards a future of increasing cross-disciplinary research by palaeoclimatologists, climate modellers and archaeologists. I see my own research agenda as moving in this direction, and I put in a research grant on this topic (in June 2008) which has now been funded by NERC and starts in 2009. Prof. Ruddiman is acting as an international project partner on this grant, with a role of identifying the cross-disciplinary implication for palaeoclimatic studies. My recent participation at the Dec 2008 meeting of the American Geophysical Union, with a session on Holocene CO2 and Methane, is also representative of this growing need for increased communication and synergy between environmental archaeology, palaeoclimatology, and climate-modelling [abstract of poster here].

The overall trend that they find is that the spread of rice in China is in agreement with predictions of Ruddiman’s hypothesis, and shows how patterns can emerge from large databases even of rather uneven data. This indicates that the archaeological record of some regions, such as China, is just reaching the critical mass of past research that allows us to see robust patterns. I would note that the archaeological record for rice is still quite uneven: most of the sites on their list were not subjected to systematic archaeobotanical sampling (flotation), nor have they been published with a full analysis of crops and weeds. Some are based on the presence of rice husk as ceramic temper (e.g. Pengtoushan—and the date of this site make its dubious whether rice was domesticated and even pre-domestication cultivation could be open to argument: see recent re-assessment by Fuller et al. 2007 Antiquity; Fuller et al. 2009 Science). Ultimately what is needed is a quantitative study of well-sampled sites, to assess the relative importance of dry-land millets (and later wheat) versus rice, and whether rice was indeed grown in wetland conditions- which should be revealed by wetland weeds (for a first stab at this kind of analysis, see paper by Fuller and Qin (2009) in World Archaeology).

In support of their hypothesis and conclusions I would offer the following observations:

1. recent work [e.g. the Tianluoshan Science paper] shows clearly the emergence rice agriculture in the 5th millennium BC (by which time a large minority of rice was morphologically domesticated, and by the end of that millennium the vast majority was), indicates that during the 5th millennium BC this early agriculture was associated with wetland weed flora with a wide range of annual sedges and small-seeded annual grasses that can be associated with paddy cultivation. Excavations by the Suzhou museum at Chuodun and Caoxieshan from near the end of the 5th millennium preserve clear fieldsystems for wetland paddy cultivation. I was in Suzhou in April 2008 to help set up the first systematic sieving and floatation program on one of their excavations, at the site of Caoxieshan , and we will soon be able to document directly what weeds were growing in those fields , and which came back to settlement areas with the harvest (as well as the proportions of immature and wild rice that remained in the population).

2. Rice starts to become much more geographically widespread from about 3000 BC. There a few sites with rice in Middle Yangshao (>3000 BC) in central China, and few more in the late Yangshap (3000-2500 BC) and many more in the Longshan (2500-1800 BC). Where full sampling has been carried out and weed flora studies, there is some indication for the presence of wetland weeds—including some of the same sedges as the Lower Yangzte sites. This is clear in our first results from the Upper Ying Valley (preliminary results were published in by Fuller & Zhang (2007) bilingual, in a Chinese monograph [access a PDF from here]), and our ongoing research at Yangshao-Longshan Baligang site in southern Henan. The overall appearance of these samples is similar that reported from the Yiluo Valley (see the Lee et al 2007 PNAS paper), although details of identification and presence of weeds has not been fully reported yet for the Yiluo. Work other sites in ongoing, including by UCL PhD students Jixiang Song and Alison Weisskopf: for a list of current UCL Archaeobotany PhD projects, see here).

3. The spread of rice outside China should also be taken into account. By 3000-2500 BC rice reached Taiwan , then by ca. 2000 BC rice had reached central Thailand, and this implies a rapid spread of rice cultivation into SE Asia. As argued by Higham, and others, this may have followed the major river valleys. If so, then it may be that early cultivation only replaced natural wild rice wetlands in the initial stages. As such it would not add to net Methane. What future research needs to show is that the total extant of marshy environments and rice paddy over the area represented by wild rice and other marshes…. For SE Asia our archaeobotanical record is still highly inadequate!

4. Also India: Rice cultivation was established in the Middle Ganges by the Third Millennium BC, and the first finds of rice in Northwest India and Pakistan, behind the range of the wild progenitor, may date to this period. In the subsequent late 2nd millennium BC and the early first millennium BC rice spread southwards, such that by 500-200 BC it is found in the far south of India in Tamil Nadu and Andhra Pradesh. It is also clear from available archaeological weed flora that Iron Age sites in the Ganges has a high diversity of wetland weed species (see paper by Fuller and Qin (2009) in World Archaeology). Thus even if it is likely that the earliest Indian rice cultivation was based on annual riverine floods and rainfall, by the time rice spread to South India wet-field systems had been developed. This accords with archaeological evidence for the development of tank irrigation systems from ca. 500 BC as rice spread south—for example from recent archaeological fieldwork by my colleague Dr. Julia Shaw in Madhya Pradesh (Sanchi area). [a paper in Asian Perspectives 2007].

5. BUT. As indicated in 3, and 4. What is needed is to test this correlations more rigourously is clear evidence that when rice was present it was being grown in wet-field (paddy) systems rather than dry-cropped, e.g. in upland indica cultivation or shifting cultivation as dry-cropped systems would not increase methane output. Although it is likely the case that most early rice in the Yangzte and Yellow basins of china was paddy systems (the earliest well-documented Yangzte systems were), it would be good to have better data on weed flora to confirm this, and quantitative archaeobotanical samples to better estimate the relative extents of millet farming vs. rice farming. Archaeobotnically we needed better developed methodologies for sampling and for analyzing samples to determine the nature of cultivation systems they represent. The kind of archaeobotanical weed flora analyses which are established in the Near East and Europe need to be developed in China and India

6. Not just rice? We also need to consider the potential of human sources other than rice. The other big potential source is the spread of livestock, especially cattle. Early in their history domestic cattle were in the Near East, Pakistan and the Sahara, and spread to parts of Europe (all by 6000/5500 BC), but herd numbers, as too human populations, are likely to have been relatively low. After 3000 BC and by ca. 1000-500 BC, the scale of things changes. Only from ca. 2500 BC onwards to you get domestic animals on sub-Saharan West Africa [e.g. new direct AMS dates from NE Mali], from which you get dispersal southwards and eastwards through central Africa. In Ethiopia, Eritrea and the horn what little data we have put the earliest cattle at 3000-2000 BC but with clear southwards dispersal to southern Africa from 2000-500 BC. Similarly livestock break the Thar Desert barrier and spread through the savannah of India after 3500 BC and mainly from 3000-2500 BC, with pastoralism (and agriculture) becoming much more widespread and dense throughout India from 2000-1000 BC. Similarly domesticated cattle only enter China in the Longshan period (2500-2000 BC) and then become widespread there. In SE Asia it is clear that cattle (mainly buffalo probably) are widespread by 500 BC. In other wards between 3000 and 500 BC you have massive increase in Old World domestic ungulate populations over much larger areas. This, together with the expansion of rice, may help to account for the accelerating trend of increasing methane level from ca. 1000 BC.

No comments: