Water management and agricultural intensification of rice farming at the late-Neolithic site of Maoshan,Lower Yangtze River,China

Water management and hydrological changes through time

Water management for rice cultivation during the mid-Liangzhu Period at Maoshan was mainly reliant on the maintenance of the natural creeks across the field. A small amount of labour was invested in building ditch outlets to connect the field with the creeks for draining and watering (Figures 2a and 3a). This had caused some change in soil paedogenetic processes, among which the Fe oxide coatings and hypocoatings are most abundantly present (Figure 5c). They are mostly formed along very large plant root channels, suggesting that oxidation was taking place here when the soil was saturated, leaving the voids an aerobic environment for oxidation (Lindbo et al., 2010). Such features, however, also appear in natural/undisturbed soils as rhizomes of grassy plants often derived such features under natural wet–dry alternating conditions (Vepraskas, 2001). The more diagnostic features are those clay textural features, primarily including the so-called clay concentration features and/or dusty or silty clay features, which according to some recent research (Lee, 2011) are associated with textural changes of the groundmass because of irrigation or water management under cultivation. These features are indeed characterized by their great contents of coarse particles (silt-sized), diffuse boundaries with the soil groundmass and weak or no birefringence under XPL (Figure 7b). However, while the similarity in the textual features found in our study and those recognized by Lee (2011) from modern and ancient cultivated soils in South Korea does suggest that they would have resulted from water management or related activities, cautions should be applied when making such an interpretation as such textual features can be formed as long as there are suitable conditions, including broken down soil particles, physical mobility of particles because of quick water saturation and quick precipitation with poor sorting (Lindbo et al., 2010).

OPEN IN VIEWER

Figure 5.

Micromorphological images from Maoshan: (a) later period charcoal and organic aggregates (MSP10:1), 400 µm (scale bar length, same below), PPL (plane-polarized light, same below); (b) later period dusty clay concentration features as a result of irrigation (MSP10:2), 50 µm, XPL (cross-polarized light, same below); (c) early-period charcoal in a void and Fe oxide feature (MSP10:3); 2.5 mm, PPL; (d and e) later period Fe oxide features and concentric dusty clay features (MSP7:2) resulting from chemical reactions and physical movement of clay particles, respectively, because of multi-episodes of wet–dry alternations, note iron-depletion and enriching areas in (d), 500 and 200 µm respectively, XPL; (f) the transition from later period paddy field to weak peat growth (MSP6:1), note the presence of very-coarse-sand-sized quartz, 500 µm, XPL; (g) later period Fe oxide feature and multi-period dusty clay features (MSP2:3), 2 mm, XPL; (h) the transition from later period paddy field to peat growth (MSP2:1), note the brownish clayey layer sandwiched between the organic-rich peaty layer and the silty layer, 2.5 mm, XPL; and (i) the transition from the alluvial sediment to peat growth (off-site sample), note the vesicle voids with smooth walls, 1.5 mm, PPL.

In addition to these are the abundant crusts, which are less-commonly encountered in thin sections from the late-Liangzhu paddy field. These crusts are composed of dusty to silty clay, with clear tendency of sorting (Figure 7c). They are formed during short periods of surface drying (Pagliai and Stoops, 2010) in an otherwise prolonged waterlogged condition. The crusts are generally densely overlapped layer by layer, partly because the surface drying was taking place frequently, but might be also because of compaction after burial. It is also interesting to note that some of them have a serrated surface, probably resulting from bioturbation when the surface was dry.

The transition from the earlier rice field to the later one was accompanied with fundamental changes in water management. First, a few more artificial ditches were newly dug; between these are the paddy squares much larger in size (see above and Figure 2c). Second, and more importantly, redoximorphic features, especially the typical concentric features formed by repeated Fe depletion and re-precipitation are especially abundant (Figures 5d and e and 6g and h), indicative of much longer periods of water saturation (Lindbo et al., 2010; Richards and Vepraskas, 2001). These were nonetheless frequently punctuated by short-term drying periods because of deliberate draining for rice farming. Associated with these are the Fe-depleted areas in the groundmass. The more commonly present dusty or silty clay textural features, including both clay concentration features and coatings (Figure 5b), further suggest frequent wet–dry alternations resulting in chemical reactions and physical movements of particles within the soils, implying a much more controlled water regime under management.

OPEN IN VIEWER

Figure 6.

Scanned and micromorphological images from Maoshan: (a) scanned image of MSP2:1, image size 4.5 × 8.5 cm; (b) scanned image of MSP7:2, note the disturbed soil structure, 5 × 8 cm; (c) scanned image of MSP1:2 (off-site sample), note the homogeneous and less-disturbed soil structure, 4 × 7.5 cm; (d) later period anthropogenic inclusions (MSP10:1), dung fragment, 500 µm, PPL; (e) horizontally distributed thin lenses of organic materials (MSP2:1), 1.6 mm, PPL; (f) later period anthropogenic inclusions (MSP7:2), 500 µm, PPL; (g) early-period Fe oxide and dusty clay features (MSP2:4), 2 mm, XPL; (h) later period Fe oxide features, note the alternation of iron-rich and depletion areas (MSP5:2), 100 µm, XPL; (i) later period charcoal in void and Fe oxide feature (MSP7:3), 200 µm, XPL; (j) early-period charcoal with well-preserved plant tissue structure (MSP10:3), 250 µm, PPL and (k) early-period anthropogenic inclusions (MSP10:3), 2 mm, PPL.

Dramatic hydrological change was recorded towards the end of the late-Liangzhu paddy field. In the upper part of the late-Liangzhu cultivated soil immediately underlying the peat-like sediments (Figure 6a), horizontally distributed thin lenses of organic matter are interbedded with a silty to clayey soil groundmass (Figure 6e). These organic lenses probably formed during a controlled flooding season or irrigation. Their poor preservation suggests episodic wet–dry events, which encouraged organic decomposition. Further up in the profile (Figure 3a and c), in the transitional zone to the peat layer, organic-rich sediment is replaced by poorly sorted orange silty clay, which might be a sign of periodic flooding (Figure 5h). The presence of slightly laminated clay lenses within this horizon also indicates low water velocity. Then, there was the onset of peat accumulation alternating with silty clay fraction and very few micritic quartz and other minerals (Figure 5f and h). This suggests, on the one hand, that flooding events were continuing, but for most of the year, the groundwater was already high enough to facilitate the encroachment of wetland reed fen onto the already abandoned paddy field. This quickly gave way to a sedimentation regime dominated by peat growth. However, the accumulating organic matter was often disturbed and broken up as a result of seasonal surface degradation. This suggests seasonal drying and disturbance. Between the transitional zone of peat growth and the overlying clayey flood sediment, there was a return to interspersed horizons of peaty and silty mud fractions. This episodic peat development occurred under fluctuating water-table conditions.

Albeit providing insightful information concerning water management and rice farming, the above reflects a more localized picture of the relationship between rice farming and environmental change. This should be supplemented by the examination on samples collected from wider environmental contexts. The off-site peat sediment (Figure 3b) is more homogeneous than its counterpart in the paddy fields, implying less disturbance during peat growing (Figures 5i and 6c) and probably a more stable groundwater regime. Immediately on top of the alluvial materials which are the parent materials of the cultivated soils in the paddy fields, the sequence is dominated by alternating peat and micritic minerals. The organic matter is more homogeneous, mostly with horizontal distribution except for occasional bioturbation that generates some bow-like microstructures. Of particular interest are the distinct vesicle voids found here (Figures 5i and 7d), suggesting fast aeration after saturation.

OPEN IN VIEWER

Figure 7.

Scanned and micromorphological images from Maoshan: (a) scanned image of MSP10-1, showing later period ploughing and cutting marks, note the whitish aggregates been churned up from the lower layer; image length 5.5 × 8 cm; (b) earlier period clay concentration features (MS-S19), note the darkish colour and diffuse boundary with the groundmass, 500 µm, PPL; (c) earlier period crust features (MS-S10), note the voids within these crusts, 850 µm, PPL; (d) vesicle voids, MSP1:2, 1 mm, XPL; (e) off-field charcoal (MS-Park), note the dust clay coatings around the void, 500 µm, PPL; and (f) locations of Maoshan and other contemporary Liangzhu-Period sites, including cemeteries and settlements. Although the functions of particularly the settlement sites are not well understood, it can be seen from the map that most of the sites are indeed situated very close to small mountains in this area. (f courtesy of Dr L Qin).

Both the paddy field and its surroundings were covered by clayey flooding deposits up to 1 m in depth. It should also be noted that we are not able to know how the local environmental change observed here is related to the regional alluvial activities induced by sea-level and climatic (monsoonal) changes, which is a topic subject to systematic surveys in the future.

Development of farming practices

Systematic studies have revealed a clear developmental discourse of different type of stone agricultural tools throughout the Liangzhu Period. The Liangzhu agricultural tools include potuqi (tools for breaking up soils), yuntianqi (tools for cutting the whole rice plants as shown by experimental studies) and sickles (ongoing research by L Qin and S Nakamura). The former two are likely bound to wooden instruments. The excavation and geoarchaeology at Maoshan provide direct evidence for their use in rice farming. Between the two periods of paddy fields and the deposits underlying them, tool marks creating clear-cut boundaries are often found (Figure 7a). The boundaries between the late-Liangzhu field and the underlying alluvial deposits are particularly sharp, implying a more intensive ploughing strategy. Moreover, the groundmass of the late-Liangzhu cultivated soils is full of disrupted soil aggregates consisting of heterogeneous materials (Figure 7a). These are probably products of severe churning and mixing of the soils during ploughing.

While there is already clue of manuring in the early-Neolithic rice farming in this area (Zong et al., 2007), our micromorphological examination offers straightforward evidence of deliberate addition of dung fragments, night soil and so on to the cultivated soils as soil amendment (Figure 6d, f and k). These are, again, significantly more abundant in the late-Liangzhu cultivated soils than in the earlier ones (also see the Supplementary Table, available online). The addition of soil amendment is also suggested by the results of particle size distribution. As can be seen in Figure 4, bulk samples from the late-Liangzhu field have a noticeable contribution from sandy particles, while their counterparts in off-site contexts contain very few or no particles larger than medium or coarse silt size. With the construction of the artificial ditch in the north of the field which essentially stopped the deposition of eroded material from the nearby hill (Figures 2c and 3d), the presence of those coarse material can be best explained as deliberate addition to the soils.

These heterogeneous aggregates were probably burned before being added to the soils, as most of them have a dense, dark-brownish or reddish colour under both XPL and PPL (Figure 6f and k). Despite this, phytoliths, commonly present in dung fragments because of animals’ consumption and digestion of plants, are preserved in many aggregates.

Two harvesting methods may have been employed. The first one involves the use of aforementioned stone sickles. Rice plants were cut from the stem, or in some instances, only panicles were cut by sickles, leaving straw in the field (cf. Weisskopf, 2009: 372–374). The second one is revealed by micromorphological examination. The occasional presence of small hole-shaped microstructures in thin sections from the mid-Liangzhu paddy field was probably caused by uprooting whole rice plants while harvesting (cf. Fuller et al., 2007). Although the information currently available at Maoshan is not sufficient to support the argument that the application of sickle harvesting in rice farming may have significantly accelerated domestication process of rice (cf. Fuller et al., 2009), future research might reveal that the development of harvesting techniques throughout the Liangzhu Period may have indeed played a very important role in the evolution of rice farming in this area.

It appears that firing was widely applied in managing field ecology. Apart from the randomly distributed charcoal within the groundmass, charcoal derived from in situ burning is often found inside large root channels in both mid- and late-Liangzhu paddy fields (Figure 5c). This suggests in-field-ecological interference perhaps during the growing and after the harvesting seasons. Firing not only creates ideal material, ash, for improving soil nutrient, but it might also help in reducing the growth of weedy plants.

Population growth, water management and intensifying rice cultivation

The development of the Liangzhu Culture was coincident with population expansion. The construction of the large walled site at Liangzhu and the associated facilities (e.g. artificial platforms used for elite burials and as an observatory) is an indication of highly stratified society and may have been both the trigger and consequence of population expansion (Liu, 2007). Especially engrossing is the construction of a huge dam in front of a canyon to the north of the walled site in the mid-Liangzhu Period. This may have served as an important barrier to prevent the site from being inundated by serious floods during the monsoonal rainfall season. The walled site was surrounded by both artificial and natural water. The completion of these labour-consuming edifices and the political system that supported them could not have achieved and survived without a surplus of rice. The late-Liangzhu Period (c. 4500–4000 BP) witnessed a large population moving from highland to lowland areas (Wang, 2007), and the number of small-size sites reached a peak at this time, both as a direct result of population migration and re-organization for rice cultivation in the region. This intensification in the later period is corroborated by micromorphology at the Maoshan site very well. For instance, it is evidenced by the presence of those redoximorphic features resulting from frequent groundwater changes (cf. Lindbo et al., 2010) as described in the section ‘Water management and hydrological changes through time’. It is thus important to note that such an intensification in rice farming was probably achieved along with the success of water management. Archaeological discoveries in this region show that the later period of rice farming witnessed the increasing abundance and technological improvement of agricultural tools such as ploughing tools made of very large pieces of flat, grounded stone, as already mentioned above (ongoing research by L. Qin and S. Nakamura; Yu, 2007). But this technological transition is not very evidently seen at Maoshan. Whether this difference within the late-Liangzhu Culture suggests regional disparity of economic and political functions among these contemporary sites requires more thorough survey. But it appears to suggest that water management and associated administrative control of water resources and uses played a more central role in rice farming in later periods.

Some burials located in the toeslope of the Maoshan Hill are also revealed during the excavation, with very few burial goods. These are in marked contrast with those found at Yujiashan, a cemetery site located just c. 1 km away, where most burials are buried with beautiful jade objects (Zhao, 2012). Sedentary features are rare at Maoshan. This settlement structure suggests that the rice farming at Maoshan was probably practised by farmers who had connections with other groups of people and/or with the Liangzhu walled site. This might also indicate a close relationship between the cultural centre and small sites towards the management, control and redistribution of water by the time, a type of relationship often observed in other ancient civilizations and historic periods (Higham, 1996, 2001, 2004), although direct evidence of this relies on systematic investigation on regional palaeo-hydrology.

Managing the local landscape

As stated above, there is also growing evidence showing that hunting and gathering had been an important part of prehistoric economic subsistence in the Lower Yangtze River (cf. Fuller et al., 2007; Qin et al., 2010). For instance, the consumption of acorns may have been a large proportion in palaeo-diets at some site (Center of Chinese Archaeology, Peking University, 2011; Qin et al., 2010). The exploitation of these wild resources would have involved the management of local landscapes. Micromorphological examination of the off-sites samples enables a deeper scrutiny of the interaction between human activities and regional landscape changes in the Liangzhu Period.

It shows that by the Liangzhu Period not only was there intensified rice farming, but the management of rich natural niches for the procurement of wild rice and other resources may have been crucial for the Liangzhu communities. Evidence for this long-neglected argument can be sought on the following aspects. First, as mentioned above, wetlands and marshes rich in wild rice were widely present in the middle to early–late Holocene (Fuller et al., 2010; Wang et al., 2001; Zong et al., 2012). Owing to favourable temperature, humidity and environmental conditions, the Lower Yangtze River has been one of the central place for wild-rice distribution since at least the early Holocene, as revealed by some recent surveys (Atahan et al., 2008; Fuller et al., 2010). Micro-fossil plant remains derived from systematic survey show that O. sativa phytoliths are abundantly deposited not only in the paddy field but also in contemporary levels of its vicinities at Maoshan (ongoing research by Y Zheng), for instance, indicating large-scale wild-rice growth in the Liangzhu Period.

Second, the management of marginal environments may have been important to the Liangzhu farmers too. The Maoshan site, as mentioned above, is actually located in one of these places (Figure 1a). Situated between the toeslope of the Maoshan Hill and the adjacent floodplain, the paddy field was connected to both natural and artificial ditches and creeks. Compared with fields constructed on flat floodplains, the water system here, consisting of slopes with gentle gradients and developed ditches and creeks, was more optimal for water management, which is influenced by monsoonal climate and the resultant seasonal groundwater fluctuations. In addition, such intermediate zones would have also supported large biomass and biodiversity, essential for the continued hunting–gathering activities. One of the off-site sequences, where evidence of possible wild-rice gathering is found as supported by the micromorphological examination which confirms the presence of abundant charcoal probably derived from in-field burning (Figure 7e), is located in a similar environment but with a much larger mountain which would potentially have had richer biomass and diversity (Figure 7f). This indicates that the management of wild field was taking place and the view of managing marginal environments might have a wider implication.

Finally, landscape management would have resulted in ecological consequences. Palynological studies demonstrate a marked increase of secondary forest and herbaceous species associated with human intervention in pollen assemblages in the Lower Yangtze River and the Yangtze River Delta during the Liangzhu Period (5200–4200 cal. yr BP), most likely because of increased land clearance and deforestation (Atahan et al., 2008; Li et al., 2010; Xu et al., 2010; Yi et al., 2003). At some local sites, for instance, Guangfulin in the Yangtze Delta, there is noticeable increase in both micro- and macro-charcoal from 4700 cal. yr BP, which is attributable to increased human activities (Atahan et al., 2008). Evidence of soil erosion in this area (Atahan et al., 2008), however, is not so apparent compared with other areas in the same period (e.g. the middle Yangtze River; see Li et al., 2010). This, according to the authors, could be related to either relatively more favourable environmental conditions in the Lower Yangtze River or the lag-feedback effect in local ecosystems (Atahan et al., 2008; Li et al., 2010).

There are also speculations that local processes (e.g. tectonic movements and changing sedimentary processes) may have been partly responsible for the relative sea-level change in the Lower Yangtze River (Zong, 2004). Although direct connection is still missing, effects of long-term water management and rice farming, deforestation and consequent sedimentation change should be taken into account when assessing relative sea-level changes and their effects on local ecologies (cf. Fuller and Ling, 2009; Li et al., 2010).