Variations in Holocene fire activity and its controls in the Ningshao Plain, eastern China

Extensive fires pose catastrophic threats to both human and natural ecosystems. Understanding the history of fire, particularly Holocene palaeofire activity in densely populated areas, is essential for predicting future fire risks and developing effective fire management policies. The complexity of fire activity is influenced by various factors, including climate and anthropogenic activities. In this study, we analysed microcharcoal from the top 35.36 m of a well‐dated sediment core HMD1401 in Ningshao Plain, eastern China. We combined our findings with phytolith and diatom evidence to obtain a comprehensive understanding of variations in Holocene fire activity and its controls. The results showed that there was higher fire activity during the early and late Holocene and less fire activity during the mid‐Holocene. More frequent fire occurred from c. 10 000–7000 cal. a BP and was primarily caused by abundant biomass and high seasonal flammability due to increased annual temperature and precipitation and warm but dry winter climate. Fire occurrences between c. 7000–2000 cal. a BP remained at a low level, except for the periods c. 5900–5600 cal. a BP and c. 5300 cal. a BP, which may have been caused by extreme climate events. The impact of fire caused by human activity was significantly enhanced during the last two millennia.

Fire is an essential part of both natural ecosystems and human communities, as it affects vegetation composition, carbon cycles, aerosols, and human social systems (Andreae & Merlet 2001;Carcaillet et al. 2002;Bond et al. 2005;Kelly et al. 2020;van der Velde et al. 2021).Extreme fire events pose a great threat to human safety and can cause enormous socioeconomic and environmental losses (Johnston et al. 2012;Bowman et al. 2017).The prediction of fire occurrence trends and formulation of fire management policies are crucial for the environment and human societies, especially in areas with high population densities.This necessitates the understanding of the fire pattern through the analysis of palaeofire events (Bowman et al. 2009;Marlon 2020).Long-term data such as multidecadal-to-millennial scale fire records from sedimentary archives may help to understand the controls of variable fire activity (Power et al. 2008;Marlon et al. 2013).
Distinguishing human-induced fires from climatedriven fires is one of the main difficulties in palaeofire studies, especially in high-density population areas, as both climate change and human activity are thought to have great influences on fire activity (Marlon et al. 2008), and there is no explicit method to distinguish humaninduced fire from climate-driven fire in palaeorecords.Studies often use human activity evidences, such as regional archaeological evidence, crop pollen evidence, and/or historic records, to explain fire occurrence, when fire activity and climate changes asynchronously (Zong et al. 2007;Li et al. 2009;Pei et al. 2020).Collecting data on human activity is certainly important for regional fire studies, but there remains the question of how the use of fire by humans impacts on the natural environment (Dietze et al. 2018).Another question is to what extent human-induced fire can influence natural environments, and when the impact region becomes large.Answering these questions is definitely challenging but there are some ways to increase our understanding of fire processes, such as comparing fire records from archaeological culture layers with those from natural sediments to test the impact of human-induced fire, and comparing fire records from different natural sediments to find regional natural characteristics of fire variations.
The effects of climate change on fire are another controversy.The most important effects of climate on fire are changes in temperature and precipitation, which influence fire activity through impacting on the accumulation and flammability of biomass (Marlon et al. 2009;Jiang & Rao 2018).However, the exact relationship remains unclear.Some regions show a positive correlation between fire activity and climate (mainly increased precipitation and increased temperature) (Li & Wang 2020;Pei et al. 2020), whereas others show a negative correlation (Zhang et al. 2015;Xue et al. 2018).Spatial analysis reveals a non-linear relationship between fire and moisture (Bradstock 2010).
Clarifying this debate requires several considerations.One consideration involves regional and seasonal fire activity differentiation.Decadal-scale studies based on monitoring data have shown that fire activity exhibits different trends in different regions under global climate change, and is typically limited to specific months in a year (fire season; Fang et al. 2021;Senande-Rivera et al. 2022).In addition, sedimentary processes also need to be considered in fire records, as palaeofire indicators such as microcharcoal can be transported by wind and/or water after burning (Sun et al. 2000;Butler 2008).A comprehensive study of fire variations with considerations of various controls, like fire season climate change and sedimentary processes, may advance the understanding of fire-climate relationships.
The Ningshao Plain, containing one of the highest population density in China (Xu 2017), is located in the south of Hangzhou Bay.During the Holocene, the Ningshao Plain was an area of intense human activity and was the centre of the Neolithic Hemudu culture (7000-5000 cal. a BP).It was also the birthplace and centre of development for rice agriculture (Institute of Archaeology, China Academy of Social Sciences 2010; Silva et al. 2015;Ma et al. 2016;Zhao 2020).In recent years, there has been a significant amount of research on fire history in the Ningshao Plain (e.g.Zong et al. 2007;Shu et al. 2010;Li et al. 2021).However, these studies largely focused on archaeological sites without comparison to natural deposits, which has hindered the understanding of wildfires and the extent and intensity of human fire activity in this area.Microcharcoal is an important indicator of fire occurrence, as it is widely distributed and well preserved in various Holocene sediments (Li et al. 2010;Marlon 2020).In this study, we analysed the Holocene microcharcoal of a well-dated natural sediment core from Yaojiang Valley, Ningshao Plain (Shao et al. 2021b).Furthermore, we compared our results with regional archaeological fire records and natural fire records.By comprehensive consideration of sedimentary processes, vegetation changes and hydrological environment changes through sedimentary facies, and phytolith and diatom records, we aim to reconstruct variations in Holocene fire activity in eastern China and explore its controls.

Geographical and archaeological background
The Ningshao Plain is located in the subtropical coastal area of eastern China, with an average elevation of <5 m a.s.l. in the plain and a low hilly region (~100-500 m a.s.l.).In the north is Cinan Mountain and in the south is Siming Mountain.Between the two mountain areas is Yaojiang Valley, filled with Holocene sediments (Liu et al. 2018;Fig. 1).The Ningshao Plain is mainly influenced by the East Asian Monsoon, with a mean temperature of ~4.7 °C in January and ~28 °C in July, and the mean annual precipitation, concentrated from June to September, is ~1480 mm (Ningbo Weather 2022; Fig. 1).Fire records from 2005 to 2018 showed that wildfire events were concentrated in the relatively cold months (Fang et al. 2021;Fig. 1C).Extreme climate events, such as storms and floods, are common in this area, as recorded by monitoring data and numerical simulations (Xu 1997;Wang et al. 2020;Wu et al. 2022).The zonal vegetation of the Ningshao Plain is a subtropical mixed forest consisting of evergreen and deciduous trees (Wu 1980), with plain areas currently exploited as farmland.
The flourishing Neolithic archaeological culture of the Ningshao Plain can be traced back to the early Holocene.The Kuahuqiao culture (8300-7200 cal. a BP; Jiang 2014) was located at the west end of the Ningshao Plain and the Jingtoushan (8300-7800 cal. a BP; Sun & Wang 2020) archaeological site in the Yaojiang Valley.The Ningshao Plain was also the core area of the Hemudu culture (7000-5000 cal. a BP; Zhejiang Provincial Institute of Cultural Relics and Archaeology 2003).There are also archaeological sites indicating the presence of the Liangzhu culture, which succeeded the Hemudu culture in the study area.

Material and methods
In 2014, sediment core HMD1401 (latitude 29°59 0 04 00 N, longitude 121°21 0 23 00 E) was retrieved from a rice paddy in the centre of the Yaojiang Valley, Ningshao Plain (Fig. 1), at an elevation of 1.09 m a.s.l., using real-time kinematic positioning.The total core length was 43.88 m, and samples were taken at 1-cm intervals.The detailed results for the sediment lithology, chronology, and 87 diatom and 348 phytolith samples from the top 35.36 m of core HMD1401 were published by Shao et al. (2021b).After Shao et al. (2021b), another five phytolith and three diatom samples were selected to analyse from the relatively low-resolution sections.The variation trends were not altered by these newly added samples, so we do not give another description here.The microcharcoal result of these 35.36 m is reported in this study.
The sedimentary facies include homogenous dark greyish clayey silt (35.36-4.61m), interbedded layers between dark greyish silty and peaty layers (4.61-0.28m), and a modern tillage layer (0.28-0 m) (Fig. S1).A modern surface rice paddy soil sample from this site was also collected and analysed to represent the modern environment, and it was plotted at a depth of 0 m relative to the top of the core for comparison.
The chronology of core HMD1401 was based on AMS 14 C dating of five plant remains samples, six peat samples and eight bulk samples, using bulk sample radiocarbon date correction and the Bacon model (Figs S1, S2, Table S1).
Microscopic charcoal particles were obtained and identified synchronously with phytolith and diatom preparation and identification, firstly with H 2 O 2 (30%) to remove organic matter, secondly with HCl (10%) to remove carbonates and lastly with heavy liquid (ZnBr 2 , 2.35 g cm À3 ) to extract microfossils (Piperno 1988; Runge 1999).After microfossils were permanently mounted on glass microscope slides with Canada balsam, identification, measurement and counting were carried out using a Leica DM 750 microscope at 4009 or 6309 magnification.Samples were selected for microfossil identification according to chronology with an average resolution of ~30 years for phytoliths and microcharcoal rather than at equidistant lengths.At least 400 phytoliths and 200 diatoms were counted for each sample.Microcharcoal was identified according to Li et al. (2010).Microcharcoal identification was considered complete when phytolith identification of more than 400 particles was completed or when the identified microcharcoal exceeded 2000 particles.Lycopodium spores (Stockmarr 1971) were also used to calculate microcharcoal and phytolith concentrations (particles g À1 ).
The microcharcoal was divided into three categories according to size: 5-50, 50-100 and >100 lm, as small

Results
A total of 353 samples from core HMD1401 and a surface rice paddy soil sample of microcharcoal were analysed (Fig. 2).In general, the concentration (particles g À1 ) of fine particles (5-50 lm) was higher than that of relatively coarse particles.The general tendency of microcharcoal of different sizes was the same.Variations in 5-50, 50-100, and >100 µm microcharcoal saw a similar trend.High level of microcharcoal and phytolith concentration showed synchronously (Fig. 2).
• Zone I (c. 10 000-7000 cal. a BP, 35.36-4.3m).The microcharcoal concentration in this zone was generally high, with averages of 3.5 9 10 6 particles g À1 of 5-50 lm, 2.5 9 10 4 particles g À1 of 50-100 lm and 5.5 9 10 3 particles g À1 of >100 lm.The maximum values of 5-50, 50-100 and >100 lm in this zone all occurred at c. 9500 cal. a BP.The C/F ratio showed a decreasing tendency.Phytolith concentration was also relatively high in this zone.• Zone II (c.7000-2000 cal. a BP, 4.3-0.4m).The microcharcoal concentration in this zone was relatively low, except for the period c. 5900-5600 cal. a BP and c. 5300 cal. a BP, with averages of 1.4 9 10 6 particles g À1 of 5-50 lm, 1.1 9 10 4 particles g À1 of 50-100 lm and 2.4 9 10 3 particles g À1 of >100 lm.The C/F ratio also remained at a low level, except at c. 6400 cal. a BP and c. 5900 cal. a BP, when the diatom concentration also showed an abrupt increase.The concentration of phytolithswas relatively low, except at c. 5900 cal. a BP.• Zone III (after approximately 2000 cal. a BP, 0.4-0 m).The microcharcoal concentration of this zone increased, with averages of 2.4 9 10 6 particles g À1 of 5-50 lm, 1.8 9 10 4 particles g À1 of 50-100 lm and 7.1 9 10 3 particles g À1 of >100 lm.The C/F ratio was relatively high, and the concentration of phytoliths synchronously increased.
Since microcharcoal particles of 5-50, 50-100 and >100 lm showed similar trends in this study, they are discussed together below.

Impact of human activity on fire events
A similar trend of fire history to that in core HMD1401 with more fire activity during the early Holocene and lower fire activity during the mid-Holocene was also recorded in surrounding areas, such as sediment cores in a relatively closed basin located 1.5 km from the Kuahuqiao archaeological site (Hu et al. 2020;Dai et al. 2022b), Luojiang site in Yaojiang Valley located 3 km from Hemudu archaeological site (Atahan et al. 2008), GDP Core 1 from Guxu Lake near the north shore of Taihu Lake (Qiu et al. 2020), Zk01 from an ancient river (Shu et al. 2007) and CH-1 from Chaohu Lake (Wu et al. 2019;Fig. 3).These kinds of sediments are all from lakes or flat areas, some distance from archaeological sites (at least >1.5 km).
At the same time, other studies in the region showed different trends.Sediment profiles located in or very close to archaeological sites (<100 m) show frequent fire activity synchronizing with high intensity of human activity (Zong et al. 2007;Innes et al. 2009;Liu et al. 2020;Li et al. 2021;Dai et al. 2022a;He et al. 2022), with active fire events mainly in the mid-Holocene, differing from the fire records from natural sediments mentioned above and from this study (Fig. 3).Clearly, the high level of biomass combustion of archaeological sites was related to human daily life or rice agriculture.The mismatch between the fire records in natural sediments, including HMD1401, in this study and nearby archaeological sites indicates the influence of human-related biomass burning is very limited.This is most likely because human-induced fire only has an impact on their living sites and has little impact on sites more than 1.5 km away.The low concentration of ricerelated phytoliths of HMD1401 in the mid-Holocene also backs up the opinion that the impact range of human activity is small (Fig. 3).In addition, pollen evidence and archaeological assessment also show that vegetation was less affected by human impacts before c. 3000-2000 cal. a BP (Giesecke et al. 2019;Stephens et al. 2019;Zheng et al. 2021), consistent with the microcharcoal evidence from this study of a limited influence of human activity from the early to middle Holocene.Concentration of <125 lm microcharcoal from core CH-1 in Chaohu Lake (Wu et al. 2019).C. Concentration of total microcharcoal from core Zk01 in Boyiqiao Town (Shu et al. 2007).D. Concentration of total microcharcoal from Guxu Lake GDP Core 1 (Qiu et al. 2020).E. Concentration of total microcharcoal from core HMD1401, this study.F. Influx of >100 lm microcharcoal from core TJA near the archaeological site Tongjia'ao (Dai et al. 2022a).G. Influx of >100 lm microcharcoal from core YJ1503 near the archaeological site Jingtoushan (Liu et al. 2020).H. Percentage of rice-related phytoliths from core HMD1401 (Shao et al. 2021b).I. Summed probability distributions of archaeological radiocarbon dates in Ningshao Plain (Shao et al. 2021a).
However, there was an exception during c. 5900-5600 and c. 5300 cal. a BP in core HMD1401.Diatoms from this site indicate that coastal flooding may have occurred at approximately 5900 cal. a BP (Shao et al. 2021b; Fig. 2).Other studies have also indicated the possibility of extreme coastal storm events at c 6000 cal. a BP and 5300 cal. a BP (Huang et al. 2020;Wu et al. 2022).As the core site HMD1401 is surrounded by archaeological sites, it is possible that the high concentration of microcharcoal in these sites was transported to the sediment site by extreme flood events.The high C/F ratio may also be the result of strong hydrodynamic forces (Fig. 2).Although the possibility of other driving forces, such as wind or intensive human activity, cannot be excluded, there is no evidence of enhanced wind power at that time and archaeological evidence shows lower levels of human activity during 6000-5000 than 7000-6000 cal. a BP (Zhejiang Provincial Institute of Cultural Relics and Archaeology 2003; Shao et al. 2021b).
According to historical records, the population of the Ningshao Plain underwent a sudden increase during the Jin Dynasty (AD 265-420), with the second largest population growth occurring during the Song Dynasty (AD 960-1279) due to war in northern China.Population pressure has led to large amounts of land being cleared for rice farming (Zhang 1990).This corresponds to the significant increase in rice phytoliths during the last two millennia (Fig. 3), which indicates that the study site may have also been transformed into paddy fields.Intensive human activity also influenced natural vegetation during this period (Cao et al. 2022;He et al. 2022).As a rice cultivation site, the microcharcoal record during this time relates mostly to human behaviours.Human-  (Dykoski et al. 2005;Wang et al. 2005).B. Moisture index of subtropical evergreen forest (Zhao et al. 2009).C. Pollen-based quantitative reconstruction of annual precipitation and contribution of extended summer precipitation from May to September to annual precipitation at Xinjie site (Lu et al. 2018).D. Pollen-based quantitative reconstruction of annual temperature trends in Northern Hemisphere (purple) and southern Asia (blue) (Zhang et al. 2022).E. Pollenbased quantitative reconstruction of winter temperature trends in Northern Hemisphere (purple) and southern Asia (blue) (Zhang et al. 2022).In D and E, dashed lines indicate zero values.F. Microcharcoal concentration from core HMD1401, this study.induced fire used for rice cultivation between c. 1500 and 1200 cal. a BP was also recorded in the Taihu region, which is not far from Ningshao Plain (Innes et al. 2019).

Impact of climate on fire events
Generally, in previous studies, the Holocene fire activity in the monsoon-dominated region of China has been regarded as drought-driven because the aboveground fuel load was sufficiently high and thus the amount of fuel moisture (and hence flammability) was the decisive factor (Xue et al. 2018;Li & Wang 2020).Conversely, in this study, a similar tendency of fire variationsmore active fire activity during the early and late Holocene and lower fire activity during the mid-Holoceneis found in core HMD1401 and other natural fire records in eastern China, which is generally positively correlated with the moisture records in this region (Dykoski et al. 2005;Wang et al. 2005;Zhao et al. 2009;Fig. 4).The Holocene fire activity in this area of eastern China seems to be different from that in the northern or southern monsoonal regions of China (Wang et al. 2013;Zhao et al. 2017;Xue et al. 2018).However, it is not surprising that there is a difference in this tendency, as Holocene syntheses and monitoring data all show variations in different regions (Marlon et al. 2013;Senande-Rivera et al. 2022).
Yet, there remains the question of why fire activity is higher in the humid early Holocene rather than in the relatively dry middle Holocene in HMD1401 and other fire records from eastern China.To better interpret the Holocene fire dynamics, it is crucial to detect modern fire-climate relationships.Based on modern observation, in eastern China, particularly the Ningshao Plain, fire occurrence is typically concentrated in the period from winter to early spring, when the precipitation is relatively low (Fang et al. 2021;Fig. 1C).That is to say, seasonal drought is the important factor for fire events.According to Holocene seasonal temperature reconstructions, winter temperatures rose rapidly during the early Holocene, and the increase in magnitude was larger in winter than annually (Dong et al. 2022;Zhang et al. 2022; Fig. 4D, E).In addition, precipitation in eastern China is concentrated in summer, and the increase in precipitation was more concentrated in summer during the early Holocene than middle Holocene (Lu et al. 2018;Fig. 4C).Evaporation in spring showed a relatively higher level during the early Holocene than during other periods (Zhang et al. 2017).Other isotopic evidence also indicates that increased temperatures might have resulted in intense evaporation and decreased the effective humidity during the early Holocene, which led to the early Holocene's high frequency of fire events (Pang et al. 2021;Dai et al. 2022b).
According to modern relationships between climate and fire events and Holocene seasonal climate changes, it can be assumed that the high level of fire activity during the early Holocene compared with the middle Holocene in HMD1401 and other records in eastern China may be the result of both annual and seasonal climate changes.In the early Holocene, higher annual temperature and precipitation facilitated biomass growth and therefore added to the fuel load (Pei et al. 2020) and, at the same time, warmer but drier winter climates were conducive to biomass burning (Li & Wang 2020), which resulted in more frequent fire activity than in the mid-Holocene.Certainly, studies about Holocene regional seasonal climate changes are insufficient and more Holocene fire-seasonal climate change relationship studies are needed to support this inference.
Regarding the late Holocene, as discussed above, the study site may have converted into farmland, so human activity was the main factor for the high level of fire activity that time.Furthermore, not only in China but also on other continents all reported this phenomenon that anthropogenic land-use played a dominant tole in the increase in fire activity during the late Holocene (Cui et al. 2015;Sch€ upbach et al. 2015;Xu et al. 2021).

Conclusions
Based on the Holocene microcharcoal analysis of core HMD1401, combined with the phytolith and diatom results and the natural and archaeological fire records in the areas around Ningshao Plain, the major findings are as follows: • The sediment core in the Ningshao Plain, with abundant Neolithic archaeological sites, recorded mainly natural wildfire changes during the Holocene until c. 2000 cal. a BP.• The high frequency of fire events during the early Holocene and the relatively low frequency of fire events during the middle Holocene might be related to biomass abundance and seasonal flammability changes influenced by Holocene annual and winter (fire season) climate changes, respectively.• More frequent fires at the study site during the last two millennia were caused by human-induced burning.

Fig. 4 .
Fig. 4. Comparison between microcharcoal concentration and climate.A. d 18 O from Dongge Cave stalagmite(Dykoski et al. 2005;Wang et al. 2005).B. Moisture index of subtropical evergreen forest(Zhao et al. 2009).C. Pollen-based quantitative reconstruction of annual precipitation and contribution of extended summer precipitation from May to September to annual precipitation at Xinjie site(Lu et al. 2018).D. Pollen-based quantitative reconstruction of annual temperature trends in Northern Hemisphere (purple) and southern Asia (blue)(Zhang et al. 2022).E. Pollenbased quantitative reconstruction of winter temperature trends in Northern Hemisphere (purple) and southern Asia (blue)(Zhang et al. 2022).In D and E, dashed lines indicate zero values.F. Microcharcoal concentration from core HMD1401, this study.