Warfare rather than agriculture as a critical influence on fires in the late Holocene, inferred from northern

Zhen Lia,1, Yoshiki Saitob, Phong X. Dangc, Eiji Matsumotod, and Quang Lan Vue

aState Key Laboratory of Estuarine and Coastal Research, East Normal University, Shanghai 200092, China; bGeological Survey of Japan, National Institute of Advanced Industrial Science and Technology, Central 7, Higashi 1–1-1, Tsukuba 305-8567, Japan; cInstitute of Geography, Vietnamese Academy of Science and Technology, 18 Hoang Quoc Viet, Cau Giay, , Vietnam; dGraduate School of Environmental Studies, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8601, Japan; and eNorthern Mapping Division, Department of Geology and Minerals of Vietnam, Ai Mo 1, Gia Lam, Hanoi, Vietnam

Edited by Graeme Barker, University of Cambridge, and accepted by the Editorial Board May 22, 2009 (received for review December 26, 2008) Fire has played an essential role in the development of human measures in areas of open forest in present-day monsoonal civilization. Most previous research suggests that frequent-fire environments (4, 21, 22). However, determining how these regimes in the late Holocene were associated with intensification activities have affected fire regimes is complicated, and there are of human activities, especially agriculture development. Here, we still insufficient reliable data to do so. analyze fire regimes recorded in the Song Hong delta area of The Song Hong delta is densely populated and most of the Vietnam over the past 5,000 years. In the prehistoric period, 2 land has been cultivated for rice fields; fire has certainly been long-term, low-charcoal abundance periods have been linked to used for deforestation and domestic purposes. This research was periods of low humidity and cool climate, and 5 short-term fire carried out to address the question of how often fire regimes regimes of 100–150 years in duration occurred at regular intervals occurred and what were the critical factors influencing the fire of Ϸ700 years. However, over the last 1,500 years, the number, regimes in during the late Holocene. frequency, and intensity of fire regimes clearly increased. Six intensified-fire regime periods in northern Vietnam during this Results and Discussion time coincided with changes of Vietnamese dynasties and associ- Charcoal records in sediments provide the only way to recon- ated warfare and unrest. In contrast, agricultural development struct fire records spanning more than a few hundred years. We supported by rulers of stable societies at this time does not show carried out charcoal analysis on samples with a resolution of a positive correlation with intensified-fire regime periods. Thus, several decades from 2 sediment cores, VN (20°24Ј37’’N, warfare rather than agriculture appears to have been a critical 106°22Ј39’’E) and GA (20°15Ј26’’N, 106° 30Ј57’’E), drilled in the factor contributing to fire regimes in northern Vietnam during the Song Hong () delta in northern Vietnam during a late Holocene. cooperative project between Japan and Vietnam (Fig. 1). In previous studies, we reconstructed the palaeoclimate and dep- fire regime ͉ Song Hong (Red River) ͉ charcoal ͉ human activities ositional environments from these 2 cores (24, 25). Depositional environments in core sections we analyzed were submarine delta ire regimes play an important role in nutrient cycling, the from delta front platform to prodelta on the basis of results from Fdevelopment of vegetation ecosystems (1–3), and global sediment facies analysis including molluscan fossils (25) (Fig. 2) concentrations of CO2 in the atmosphere (4). To understand the (see Materials and Methods). All of the charcoals in the subma- factors that influence them, fire regimes have been studied at rine delta were transported by river from the Song Hong (Red various time scales. Wildfire and climate are intimately linked River) drainage area and surroundings together with sediment (5–7). Many previous studies have shown that fire regimes at materials, not in situ. Our age model was based on 14 AMS decadal, centennial, and millennial time-scales are influenced by (accelerator mass spectrometry) 14C dates from core VN, and 13 climatic change at the similar pacing (8–12). Fire regimes have 14C dates from core GA, and was calibrated using CALIB5.01 been presumed to occur under hot and dry climatic conditions (24–26) (see Materials and Methods). (13–16). However, high fire frequencies have also been identified According to archaeological records, from Ϸ3000 BC most of during relatively cool and moist periods, which suggests fuel- the population of Vietnam migrated from mountainous areas to determined rather than directly climate-determined influences areas of low hills surrounding the midlands and fertile plains of on fire regimes (17, 18). Climate controls fire directly by large rivers valleys and coastal areas, shown by cultures of the providing opportunities for ignition and spread of fire, but also Phung Nguyen, Ha Long, and Mai Pha in the Song Hong delta indirectly through regulation of the accumulation and structure (27). Stone hoes, harvesting knives, stone spades, and stone of fuel at longer time scales (19). In addition to natural fire ploughs found at archaeological sites at Hoa Binh suggest that triggers, human activities can also play an important role in the rice was cultivated in these areas. Rice grains related to the occurrence of fire. In the advancement of human civilization, fire Phung Nguyen culture have also been found at the Dong Dao has played an essential role; for example, fire has been used for heat and light, for cooking, to fend off wild animals, and to clear land in preparation for planting (20–22). Author contributions: Z.L., Y.S., and E.M. designed research; Z.L. performed research; Z.L., Y.S., P.X.D., and Q.L.V. contributed new reagents/analytic tools; Z.L. analyzed data; and Z.L., Because human activities intensified in the late Holocene, fire Y.S., and P.X.D. wrote the paper. regimes became decoupled from climate, especially during the The authors declare no conflict of interest. last 2 millennia (4, 21). Campbell and Campbell (23) suggested This article is a PNAS Direct Submission. G.B. is a guest editor invited by the Editorial Board. that a sharp increase of fire regimes in the last 2 millennia can Freely available online through the PNAS open access option. be ascribed in large part to an increase in human population 1To whom correspondence should be addressed at: Zhen Li, State Key Laboratory of compared with prehistoric times. Human activity is believed to Estuarine and Coastal Research, East Normal University, North Zhongshan Road 3663, induce more fire regimes as a result of adaptation of the Shanghai, 200062, China. E-mail: [email protected]. landscape for human use, large-scale deforestation, burning, This article contains supporting information online at www.pnas.org/cgi/content/full/ agriculture, and warfare, even as a result of forest-conservation 0813258106/DCSupplemental.

11490–11495 ͉ PNAS ͉ July 14, 2009 ͉ vol. 106 ͉ no. 28 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0813258106 Downloaded by guest on September 28, 2021 Fig. 1. Geographical locations of cores VN and GA on the Song Hong delta (modified from ref. 25). DFP: an area shallower than Ϸ6 m below mean sea level with a gradient of 0.5/1,000; DFS: an area from DFP to 20–30 m below mean sea level, having a relatively steep slope with a gradient of 2.5/1,000; PD: located further offshore from DFS.

site. However, rice did not developed and clear wild grass in Vietnam, especially in areas upstream widely until the Dong Son culture, when bronze ploughs were from the delta. However, our data do not show a clear corre- used. Oscar (28) reported that, from 3000 BC to 2000 BC, wet lation between the amount of charcoal present and the abun- rice cultivation was carried out in the Song Hong delta on a small dance of Gramineae pollen larger than 40 ␮m, which is consid- scale, whereas became widespread after 2000 BC, when bronze ered to be an indicator of human cultivation (Fig. 2) (24). tools were introduced. Various kinds of bronze plough have been According to historic records, in stable societies rulers generally discovered at Co Loa, Son Tay, Lao Cai, and Bat Xat. Most of strongly encouraged people to farm and exploit the land for these harvesting tools are the ‘‘nhip’’ type, and were found agriculture by decreasing (or abolishing) taxes, and soldiers mainly in the Song delta at locations such as Tho Vuc returned home for agriculture in the seasons when warfare did and Vinh Quang in Ha Tay province, and Duong May, Xom Trai, not normally occur. The rulers also supported the building of and Ma Tre in Ha Noi province (29). The occurrence of high canals, dams, and roads to support irrigation and protect farm- content of Gramineae pollen of Ͼ40 ␮m provides additional lands. All of these initiatives helped to restore agriculture after evidence for the intensified rice agriculture in the Song Hong periods of warfare and unrest, expanded the area of farmland, delta area after Ϸ1050 BC (Fig. 2) (24). and improved agriculture, especially in the periods 110 BC-39 The combination of archaeological and palaeoecological re- AD, 544–906 AD, 1010–1127 AD, and 1231–1266 AD (28, 35). search has provided evidence of numerous confirmed instances Moreover, the abundance of Gramineae pollen larger than 40 of land clearance for agriculture by 2500 yr BP (Ϸ650 BC) ␮m clearly increased during these periods, reflecting agricultural (30–32), corresponding to major waves of human migration to development that corresponds to historic records. However, northern Vietnam from southern China and the introduction of charcoal concentrations were low during these periods (Fig. 2). Chinese and Indian influences on social organization to the In contrast, fires were frequent when societies were unstable or Asian mainland. Similarly, our research showed that during dynasties when agriculture was hindered, for example in charcoal concentrations increased considerably after 400 BC, the A Dozen Local Military Chiefs Period (944–967 AD), and when there were clear decreases of arboreal pollen and increases during the Ly Dynasty recessions in 12th century and early 13th of nonarboreal pollen (Fig. 2). This was perhaps associated with century, and in the 15th century. Thus, it is unlikely that the deforestation by fire regimes during the period of major human intensified-fire regimes after 450 BC were significantly influ- migration. enced by human agricultural activities. Many scientists link the increase in fire regimes in the latter The size of charcoal particles can help distinguish local fires half of the Holocene to intensive human impacts, especially rice from regional fires (36). Some studies of lake sediments have cultivation, deforestation, and other agricultural activity (33, 34). suggested that particles of Ͼ1,000 ␮m in diameter are deposited Even today, fire is used to clear lands in preparation for near a fire, particles of Ͻ100 ␮m travel well beyond 100 m from

agricultural planting in the Song Hong delta (20). According to the fire, and very small particles are carried even greater GEOLOGY our field investigation along the Song Hong drainage area in distances before settling (37, 38). The charcoal particles in our 2007, fires are currently set once or twice each year to kill pests cores are small and predominantly in 2 size ranges, of 5–20 ␮m

Li et al. PNAS ͉ July 14, 2009 ͉ vol. 106 ͉ no. 28 ͉ 11491 Downloaded by guest on September 28, 2021 C age 14 ) ) -3 -3 n cm n cm 5 4 Pollen concentration µm) (>40 Gramineae % of total pollen) Tropical/Temperate herb Aquatic/Total Core no. Conventional Arboreal pollen Arboreal pollen (% of total grains) Mud content (%) Sedimentary environment (x10 (yr BP) year Calendar concentration Charcoal (x10 1950 pollen Non-arboreal (% of total grains) 130±40 290±40 DFP 740±40 1450 1030±40 III DFS

1330±40 950 GA

1960±40 PD 450 1800±40

2310±40 0 II DFP

2590±40 BC AD 550

1050 VN DFS 3440±40 1550

3630±40 2050 3990±40 I PD 4420±40 2550

3050

4960±40 3550 20 40 60 20 40 20 40 60 80 20 40 20 1 2 1 50 100 ABCDEFGH

Fig. 2. Curves of charcoal and palynological records from 3550 BC to 1950 AD (mud content and the sedimentary environment modified from ref. 25).

and 20–50 ␮m, and that Ͻ5% of particles are larger than 50 ␮m The frequency of fire regimes intensified after 450 AD, (Fig. 3). All of the samples for our study were from submarine occurring periodically at intervals of 100–200 years, as indicated delta sediments, which are generally subjected to much stronger by charcoal peaks 6–10 (Fig. 3A). The record of charcoal hydrodynamic action than lake sediments. Thus, the relationship concentrations since 950 AD shows several intensified-fire re- between charcoal size and transport distance determined from gime periods (A-F of Fig. 3B), when dynasties changed fre- lake samples is not appropriate for use in this study. Charcoal quently and society were unstable in northern Vietnam. Periods analysis of surface samples from the whole Song Hong drainage A–C correlate with charcoal peaks 7, 8, 9, and periods D–F with basin including mountains, floodplain, and submarine delta charcoal peak 10 (Fig. 3). Charcoal peak 6 spans the period from shows that charcoals in submarine delta are dominated by 450 to 610 AD, when Ly Bon successfully revolted against subround-subangle grains, and both round and angular charcoal Chinese rule, occupied Long Bien (modern-day Hanoi), and grains are rare (See SI). The charcoals in our cores are mostly founded the Empire of Van Xuan (Anterior Ly Dynasty) that subround-subangle and rarely display round and angular shapes, remained in power until Chinese domination of the region suggesting these charcoals, are not local, but were transported returned in 602 AD (39). There were several episodes of warfare from the Song Hong drainage basin through rivers. along the Song Hong drainage area during the period 450–610 AD. These included the Ly Bon rebellion, which mobilized the Two long-term, low-charcoal periods (periods I and II of Fig. imperial troops and naval fleet of Giao Chau (in the area roughly 2) during 2550–2050 BC and 400 BC–200 AD correspond to corresponding to modern Hanoi) in 542, 544, and 545 AD; periods of low humidity and cool climate, as demonstrated by low retaliation by the Chinese Liang Dynasty against Emperor Van abundances of aquatic herb pollen and the low ratio of tropical Xuan in the areas of Hanoi, Vinh Phuc, and Viet Tri in 545–548 arboreal pollen to temperate arboreal pollen. In the cool humid AD; and Trieu Quang Phuc’s resistance and defeat of Liang climate period III (1350–1750 AD, Fig. 2), charcoal concentra- Dynasty troops in 548–550 AD (35, 39). tions show considerable variation, but generally increase toward Fire was used during warfare as a tactic in ancient times; for 1750 AD. A total of 10 short-term charcoal peaks were clearly example, the firing of Chibi, the firing of the Shangfang Valley, identified over the last 5,000 years (Fig. 3A). In prehistoric times, and the firing of the 800-mile Military Barracks in battles in each of the periods of high charcoal concentration lasted for Three Kingdoms in China (40). Sun Tzu specially summarized 5 Ϸ100–150 years and occurred at regular intervals of Ϸ700 years ways of attacking with fire on military strategy: to burn soldiers (peaks 1–5 of Fig. 3A). Most of the regional fires likely took place in their camp, to burn stores, to burn baggage trains, to burn in upstream forest areas, as suggested by the rarity of charcoal arsenals and magazines, and to hurl dropping fire among the particles larger than 100 ␮m (Fig. 3A). The regular cycles of these enemy, having a significant impact on ancient warfare in Asian charcoal records are perhaps indicative of natural fire regimes (41). Fire was also used in ancient warfare to destroy fortifica- induced by systematic effects associated with climate and fuel tions, to create confusion among opposing troops, to provide accumulation. However, in the late Holocene, intensified human communication signals between groups of allied troops, and as activity likely upset the natural balance of fire regimes. a signal of rebellion (35).

11492 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0813258106 Li et al. Downloaded by guest on September 28, 2021 Dynasty(from Tran, 1950 1958; Oscar, 1995) French Colony 1900 ear ear C age y

1850 Nguyen Dynasty 14 ) 1800 Tay Son Dynasty -3 1750 n cm 5 Calendar 1700 Restored Le Dynasty Charcoal (50-100µm) Charcoal (50-100µm) (%) Charcoal (>100µm) (yr BP) High-fire-regime period High-fire-regime µm) Charcoal (5-20 (%) Charcoal (20-50µm) (%) (%) Conventional 1650 1950 (AD) Calendar year Charcoal concentration (x10 1600 1900 F Mac Dynasty 1850 130±40 1550 (Southern & Northern Dynasties) 1800 1500 Early Le Dynasty 1750 E 1450 IV Chinese domination 1700 1400 Ho Dynasty 1650 D 1600 1350 290±40 1550 Tran Dynasty 740±40 1300 1500 C 1250 1450 B 1200 1400 1350 1150 1030±40 1300 Ly Dynasty 1100 1250 B 1050 1200 1000 Prior Le Dynasty 1150 Dinh Dynasty 1100 950 Ngo Dynasty 1050 Autonomy 1330±40 900 1000 A 850 950 20 40 60 20 40 60 80 20 40 60 5 5 800 750 III Chinese domination

700 C age 14

650 ) -3 600 n cm 550 Anterior Ly Dynasty 5 Charcoal (>100µm) Charcoal (>100µm) Calendar year Calendar year Charcoal concentration Peaks Charcoal (5-20µm) (%) Charcoal (20-50µm) (%) Charcoal (50-100µm) (%) (%) (yr BP) Conventional 500 (x10 1950 130±40 450 10 290±40 1450 9 740±40 400 1030±40 II Chinese domination 8 1330±40 350 950 7 300 1960±40 450 6 250 1800±40 200 0 2310±40

150 BC AD 2590±40 550 5 100 A 50 Trung sisters 1050 4 3440±40 0 1550

BC AD -50 3630±40 I Chinese domination 3 -100 2050 3990±40

-150 2550 2 4420±40 -200 Thuc Dynasty 3050 -250 1 4960±40 Prehistoric period -300 3550 20 40 60 20 40 60 80 20 40 60 5 5

Fig. 3. Schematic diagram showing the potential relationships between fire regimes and historic events reflected by charcoal records and dynastic history. The size of charcoal is measured in the longest dimension. (A) During the last 5,000 years: 10 peaks of charcoal concentration are identified. The high-charcoal periods lasted

for Ϸ100–150 years at the stable intervals of Ϸ700 years in prehistoric period. Charcoal concentrations increases rapidly, and shifts frequently with longer lasting time GEOLOGY at shorter return intervals. (B) During the last 1,000 years: 3 important high-fire-regime periods marked by A–C corresponding to charcoal peaks 7, 8, and 9 in A and another 3 periods (D–F) corresponding to charcoal peak 10 in A.

Li et al. PNAS ͉ July 14, 2009 ͉ vol. 106 ͉ no. 28 ͉ 11493 Downloaded by guest on September 28, 2021 The frequent dynastic changes from the end of the third period made fire torch with rice straw pouring oil, and lighted forming a of Chinese rule to Vietnamese autonomy, followed by the Ngo, fire net surrounding the enemy’’ (35). Dinh, and Prior Le Dynasties, correspond to the intensified-fire Besides the 2 clear intensified-fire regime periods E and F, the regime of period A (Fig. 3B), indicated by charcoal peak 7 (Fig. high charcoal concentrations of our data show intensified fire 3A). During this period, warfare recorded in Vietnam history regimes throughout the period since 1740 AD (Fig. 3B). Ad- includes rebellions led by Mai Thuc Loan in 722 AD, by Phung vanced weaponry and gunpowder introduced from western Hung in 767–791 AD, by Duong Thanh in 819–820 AD, and countries might have contributed to this. Period F should other rebellions in 828, 841, 858, 860, and 880 AD; battles with correspond to the period of the rebellions in the mountain areas the Nam Chieu in China in 816, 832, 846, 853, 858, 861, 862, 863, of Thai Nguyen (1917), Yen Bai (1930), and Nghe An/Ha Tinh and 865 AD; the Rebellion of A Dozen Local Military Chiefs (1930–1931). from 944 AD to 948; and battles driving out Chinese troops in 931, 938, and 981 AD. During this period, guns and gunpowder Conclusion were used in battle (35, 41, 42). In summary, our data show 2 long-term low-charcoal periods Intensified-fire-regime period B corresponds to charcoal peak (periods I and II of Fig. 2) that are probably linked to millennial- 8 and is perhaps associated with the rebellions of 1140, 1144, scale periods of low humidity and cool climate, as inferred from 1154, 1192, 1198, 1208, 1205, 1212, and 1218 AD in Thai Nguyen, comparison of charcoal concentrations and palynological Ninh Binh, Hai Duong, and Bac Giang during the later Ly records. Short-term fire regimes in prehistoric times burned at dynasty, and battles against the Dynasty troops in the areas regular intervals of Ϸ700 years and had durations of Ϸ100–150 of Hai Duong, Hanoi, Hung Yen, and Bac Ninh in 1257, years. These fire regimes can be reasonably explained as being 1284–1285, and 1287–1288 AD during the Tran Dynasty. controlled by balance systems in nature, with little or no input Intensified-fire-regime period C corresponds to a period when from humans. Within the last 2 millennia, high charcoal con- there were frequent changes of dynasty: from the Tran Dynasty centrations show 6 clear periods of intensified-fire regime. All of to the Ho Dynasty, followed by the fourth period of Chinese these periods of intensified-fire regime correspond to frequent domination, the Early Posterior Le Dynasty, the Mac Dynasty, dynastic changes in northern Vietnam and the human conflict and the Restored Le Dynasty. During the Southern and North- and civil unrest that accompanied them. In contrast, periods of ern Dynasties period, there was an ongoing civil war between the agricultural development supported by rulers in stable societies Northern Court (Mac Dynasty) and the Southern Court (Re- do not appear to show a positive correlation with periods of stored Le Dynasty) until 1592 AD, when the army of Trinh Tung intensified-fire regime periods. Thus, we conclude that human conquered Hanoi and executed king Mac Mau Hop (26, 38). Two conflict, rather than agriculture, was likely the critical influence charcoal peaks at Ϸ1550 and 1600 AD are perhaps associated on fire regimes in the Song Hong delta in the late Holocene. with 2 main periods of conflict in northern Vietnam, in 1546– Materials and Methods 1561 and 1583–1592 AD. In addition to annual battles at this time, frequent and other disasters created an unstable Age Model. 14 AMS (14)C dates from core VN and 13 from core GA were used for the age model after calibration using CALIB 5.01 (24–26). The relationship be- society for 200 years (Fig. 3B) (35). The high relative abundance tween age and depth was expressed by change-point model with linear inter- of aquatic herb pollen substantiates the occurrence of frequent polations between the adjacent points. Other nondirectly measured ages were during this period (Fig. 2). calculated by inserting points. Based on the age model, 58 samples from core GA Intensified-fire-regime period D spans from 1640 to 1670 AD, that were younger than 1720 cal. BP and 45 samples from core VN spanning from concomitant with the Trinh-Nguyen War (1627–1672 AD). It is 5270 to 1720 cal. BP were analyzed for pollen and charcoal contents at a sample not likely that this war contributed to the fires at this time interval of Ϸ30–80 years. Each sample is Ϸ45–50 mm thick and represents a time because the war occurred mainly in present-day Central Viet- span of approximately several years to a decade (24). nam. Other battles in the surrounding mountains, in which the Mac army attacked Trinh in the Thai Nguyen area in Ϸ1638 AD The Sedimentary Environment. Based on sedimentary facies analysis charac- and Trinh eliminated Mac’s army in the Cao Bang area in terized by sediment color, grain size variations, sedimentary structure, sedi- ment composition, and molluscan fossils, more than 10 sedimentary environ- 1667–1669 AD, should be the dominant events contributing to ments from river to coastal/shelfal systems were identified (25). According to this high-charcoal-value interval. the sedimentary facies and (14)C dating (Fig. 3), pollen and charcoal samples Historical records show that Vietnamese farmers in the 18th were analyzed only taken from submarine delta facies from prodelta to delta century suffered starvation, flood disasters, and oppression from front platform through delta front slope, as shown above (Fig. 2). their dynastic masters, such as a severe famine in 1741 AD, devastating floods in 1735 and 1746 AD, and the largest gap Palynological Analysis. HCl (15%), KOH (15%), and HF (30%) were used to between the rich and the poor because of oppressive government dissolve calcareous minerals, humic components and siliceous materials, re- (35). Many rebellions occurred, spreading widely in northern spectively. The pollen and spores were then concentrated by heavy liquid Vietnam, such as the rebellion led by Nguyen Tuyen, Nguyen Cu, flotation with ZnCl2 (specific gravity 2.2) (24). and Vu Trac Oanh in the Hai Duong area from 1739 to 1741 AD; Charcoal Analysis. We counted charcoal particles by using the slides prepared for the Ngan Gia rebellion in the Son Nam area in 1740 AD; the pollen analysis. Lycopodium spores were added to each sample for calculating rebellion led by Nguyen Danh Phuong from Tam Dao in 1740 AD, charcoal concentration. We counted only particles whose length is between 250 which spread to Viet Tri and Bach Hac in 1744 AD, and conquered and 5 ␮m. Four size groups were defined: 5–20 ␮m, 20–50 ␮m, 50–100 ␮m, and Thai Nguyen, Tuyen Quang, Viet Tri, and Vinh Yen from 1740 to Ͼ100 ␮m to allow discrimination between local fires and regional fires. Charcoal 1750 AD; and the rebellion led by Hoang Cong Chat in Hung Yen concentration diagrams were created with Tilia and Tilia-Graph software, as were and Hai Duong, which spread to Nam Dinh and Thai Binh (Khoai pollen concentration diagrams. Chau) in 1740–1768 AD. The Tay Son revolution was successful in 1771 AD, and moved northwards from present-day Central Viet- ACKNOWLEDGMENTS. This work was supported by the Chinese Natural Science Foundation (No. 40606018), 111 Project No. B08022, and the Creative nam to oppose the Trinh Lord in 1786 AD, and then to the Hanoi Research Groups of China (No. 40721004). We thank Ms. J. Li (State Key area, Ha Tay, Hai Duong, and Ninh Binh on the Song Hong delta Laboratory of Estuarine and Coastal Research, East China Normal University) plain, where several fierce battles broke out between Chinese Qing and Dr. T.N. Tuyen (Institute of Geography, Vietnamese Academy of Science troops and Vietnam Tay Son between 1788 and 1789 AD (35). and Technology) for their help in the field investigation and sampling along the Song Hong drainage area of northern Vietnam; the editor and 3 reviewers Fire-regime period E in our data reflects this period well. There is for their useful comments and suggestions; and Dr. Yongxiang Li (Tulane a published description of the use of fire in battle at this time: ‘‘They University, New Orleans) for English correction.

11494 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0813258106 Li et al. Downloaded by guest on September 28, 2021 1. Bond WJ, van Wilgen BW (1996) In Fire and Plants (Chapman & Hall, London), pp 21. Marquer L, et al. (2008) Late Holocene high resolution palaeoclimatic reconstruction 38–42. inferred from Sebkha Mhabeul, southeast Tunisia. Quat Res 70:240–250. 2. Dwyer E, Pinnock S, Gregoire, J-M, Pereira JMC (2000) Global spatial and temporal 22. Maxwell AL (2004) Fire regimes in north-eastern Cambodian monsoonal forests, with distribution of vegetation fire as determined from satellite observations. Int J Remote a 9300-year sediment charcoal record. J Biogeogr 31:225–239. Sens 21:1289–1302. 23. Campbell ID, Campbell C (1994) The impact of late woodland land use on the forest 3. Wright HE, Jr, Heinselman ML (1973) The ecological role of fire in natural conifer landscape of southern Ontario. Great Lakes Geo 1:21–29. forests of western and northern North America—Introduction. Quat Res 3:319–328. 24. Li Z, et al. (2006) Climate change and human impact on the Song Hong (Red River) 4. Carcaillet C, et al. (2002) Holocene biomass burning and global dynamics of the carbon Delta, Vietnam, during the Holocene. Quat Int 144:4–28. cycle. Chemosphere 49:845–863. 25. Tanabe S, et al. (2006) Holocene evolution of the Song Hong (Red River) delta system, 5. Clark JS, Stocks BJ, Richard PJH (1996) Climate implications of biomass burning since the northern Vietnam. Sediment Geol 187:29–61. 19th century in eastern North America. Global Change Biol 2:433–442. 26. Stuiver M, et al. (1998) Intcal 98 radiocarbon age calibration, 24,000–0 cal BP. Radio- 6. Maxwell AL (2001) Holocene monsoon changes inferred from lake sediment pollen and carbon 40:1041–1083. carbonate records, Northeastern Cambodia. Quat Res 56:390–400. 27. Vu QL (2004) In Conference Proceedings of Stratigraphy of Quaternary System in 7. Swetnam TW, Betancourt JL (1998) Mesoscale disturbance and ecological response to Deltas of Vietnam, eds Nguyen TV, Saito Y, Nguyen VQ, Ngo QT (Department of decadal climatic variability in the American Southwest. J Climate 11:3128–3147. Geology and Minerals of Vietnam, Hanoi), pp 77–78. 8. Carcaillet C, Richard PJH (2000) Holocene changes in seasonal precipitation highlighted 28. Oscar C (1995) A : From Hong Bang to Tu Duc (Greenwood Press, by fire incidence in eastern Canada. Clim Dynam 16:549–559. Connecticut), pp 11–12, 20–25. 9. Clark JS (1990) Fire and climate change during the last 750 yr in northwestern Minne- 29. Trinh S (2000) The early agriculture in Vietnam: 3 development stages. Agric Arch sota. Ecol Monogr 60:135–159. 1:40–46. 10. Long CJ, Whitlock C, Bartlein PJ, Millspaugh SH (1998) A charcoal study 9000-year fire 30. Maloney BK (1985) Man’s impact on the rainforests of West Malesia: The palynological history from the Oregon Coast range, based on a high-resolution charcoal study. Can record. J Biogeogr 12:537–558. J Forest Res 28:774–787. 31. Maloney BK (1992) Late Holocene climatic change in Southeast Asia: The palynological 11. Turner R, Roberts N, Jones MD (2008). Climatic pacing of Mediterranean fire histories evidence and its implications for archaeology. World Archaeol 24:25–34. from lake sedimentary microcharcoal. Global Planet Change, 10.1016/j.gloplacha. 32. Maloney BK (1996). New perspectives on possible early dry land and wet land rice 2008.07.002. cultivation in highland North Sumatra. (Centre for South-East Asian Studies, University 12. Swetnam TW (1993) Fire history and climate change in giant Sequoia groves. Science of Hull) Occasional Paper No. 29. 262:885–889. 33. Anshari G, Kershaw AP, van der Kaars S (2001) A Late Pleistocene and Holocene pollen 13. Bergeron Y, Archambault S (1993) Decreasing frequency of forest fires in the southern and charcoal record from peat swamp forest, Lake Sentarum Wildlife Reserve, West boreal zone of Quebec and its relation to global warming since the end of the ‘‘Little Ice Age’’. Holocene 3:255–259. Kalimantan, Indonesia. Palaeogeogr Palaeoecol 171:213–228. 14. Santos GM, et al. (2000) (14)C AMS dating of fires in the central Amazon rain forest. 34. Behling H, Bauermann SG, Neves PCP (2001) Holocene environmental changes in the Nucl Instrum Meth B 172:761–766. Sao Francisco de Paula region, Southern Brazil. J S Am Earth Sci 14:631–639. 15. Whitlock C, Bartlein PJ (2004) Holocene fire activity as a record of past environmental 35. Tran TK, Ming Z (1958) In History of Vietnam (Shanlian Press, Beijing), pp 10–294. change. Dev Quat Sci 1:479–490. 36. Clark JS, Patterson WA, III (1997) In Sediment Records of Biomass Burning and Global 16. Willis KJ, Bennett KD, Haberle SG (2008) In Charcoals from the Past: Cultural and Change, eds Clark JS, Cachier H, Goldammer JG, Stocks B (Springer, Berlin), pp 23–48. Palaeoenvironmental Implications, eds Fiorentino G, Magri D (Archaeo Press, Oxford), 37. Clark JS (1988) Particle motion and the theory of stratigraphic charcoal analysis: Source pp 308–318. area, transport, deposition, and sampling. Quat Res 30:67–80. 17. Green DC (1982) Fire and stability in the postglacial forests of southwest Nova Scotia. 38. Patterson WA, III, Edwards KJ, MacGuire DJ (1987) Microscopic charcoal as a fossil J Biogeog 9:29–40. indicator of fire. Quaternary Sci Rev 6:3–23. 18. Ritchie JC (1987) In Postglacial Vegetation of Canada (Cambridge Univ Press, Cam- 39. Woods S (2002) In Vietnam: An Illustrated History (Hippocrene Press, New York), pp bridge, UK), pp 169–179. 1–164. 19. Campbell ID, Flannigan MD (2002) In Fire, Climate Change, and Carbon Cycling in the 40. Sun T (1910) In The Art of War (The Puppet Press, Norwalk), pp 143. Boreal Forest, eds Kasischke ES, Stocks BJ (Springer, New York), pp 151–156. 41. Chen S (2002) In Records of Three Kingdoms (Ancient Book Press, Shanghai), pp 900, 20. De Vries B, Marchant R (2002) In Mappae Mundi: Humans and Their Habitats in a 1310, 1512. Long-Term Socio-Ecological Perspective: Myths, Maps and Models, eds De Vries B, 42. Guo Z, Zhang X (2001) The rise and fall of Ding and Li Dynasties of Viet Nam. J Huanghe Goudsblom J (Amsterdam Univ Press, Amsterdam), pp 111–146. S-T University 3:49–56. GEOLOGY

Li et al. PNAS ͉ July 14, 2009 ͉ vol. 106 ͉ no. 28 ͉ 11495 Downloaded by guest on September 28, 2021