500‐Year Periodic Vegetation and Monsoonal Climate Oscillations During the Last Deglaciation in East Asia

Fluctuations in monsoonal climate on orbital‐ to millennial‐timescales have had a major impact on vegetation change in East Asia since the last deglaciation. However, the influence of centennial‐scale oscillations on the spatio‐temporal characteristics of vegetation and climate during the last deglaciation remains controversial. We present a well‐dated, decadal‐resolution pollen record from the sediments of annually laminated Lake Xiaolongwan in Northeast Asia, which reveals major orbital‐/millennial‐scale and minor quasi‐periodic, that is, ∼500‐year (yr) cyclic, vegetation changes from ∼19.96 to ∼10.79 Kyr ago (covering the last deglaciation). The ∼500‐yr cycles are characterized by alternations in broadleaved forest/boreal coniferous forest with tundra steppe representing warm and humid/cold and dry phases of the monsoonal climate. This cyclicity may be related to El Niño‐like/La Niña‐like state and low‐latitude atmosphere‐ocean processes, against the background of the long‐term trend of the deglacial climatic evolution. Our findings demonstrate a close relationship between ecosystem succession and climate change on centennial scale.


of 11
Oscillation (ENSO) variations affect the intensity of the East Asian summer monsoon (EASM), and thus the vegetation, by controlling the position of the subtropical high (Park, 2017;Park et al., 2014;Xu et al., 2020). The studies referenced above have proposed the existence of teleconnections between high-and low-latitude climatic forcing and various environmental responses or feedbacks within the Northern Hemisphere (Chapman & Shackleton, 2000;Cheng et al., 2015;Xu et al., 2020). However, whether these ∼500-yr quasi-periodic oscillations occurred during the last deglaciation, and the nature of the driving mechanism, are unresolved issues in understanding the relationship between climate change and ecosystem evolution .
So far, some studies have revealed the existence and driving mechanism of centennial-scale vegetation and climate cycles, but those studies were limited to the Holocene (Cheng et al., 2015;Xu et al., 2020;Zhu et al., 2017). Furthermore, many climate studies of the last deglaciation have focused on the structure, timing, phase relation of orbital-scale warming and millennial-scale climatic events (Cheng et al., 2015;Sun et al., 2016;Wang et al., 2001). The latter involves the Dansgaard-Oeschger, Heinrich Stadial 1 (HS1), Bølling-Allerød (BA), and Younger Dryas (YD) events (Andersen et al., 2004(Andersen et al., , 2006Sun et al., 2016). Although there is evidence of centennial-scale climatic cycles in high-resolution ice core records from Greenland as well as in climate model simulations (Adolphi et al., 2014;Braun et al., 2005), there has been a general lack of in-depth investigations into the presence, process and mechanism on centennial scale climate cycles in the Asian monsoonal regions and the related ecological responses during the last deglaciation. This is largely owing to the scarcity of long-term vegetation and climate records with a decadal resolution in East Asia.
We present the results of a pollen-analytical study of the last deglaciation from the sediments of annually laminated Lake Xiaolongwan (42°18.0′N, 126°21.5′E) (Figure 1) in Jilin Province, Northeast (NE) China. This lake has recently yielded a decadal-resolution pollen and long-chain n-alkane carbon isotope record spanning the past 9.26 Kyr Xu et al., 2019Xu et al., , 2020. Lake Xiaolongwan now provides the opportunity to establish decadal-resolution vegetation records for comparison with other high-resolution climate records for the last deglaciation (Sun et al., 2016). Here, we focus on the timing, cyclicity, and phase relationships between vegetation and monsoon climate changes on centennial scale during the last deglaciation. Time-series analyses are used to reveal the phase relationship between ∼500-yr cycles of vegetation and monsoonal climate change. Finally, we consider the possible processes and mechanisms responsible for these cycles.

Study Site
Lake Xiaolongwan, a Maar Lake, is located within the Longgang region, on the northwest margin of the Changbai Mountains in East Asia ( Figure 1). The lake has a surface area of 0.079 km 2 , a maximum depth of 15 m, and a catchment area of 0.16 km 2 ; it is a closed basin with no inflows or outflows (Chu et al., 2008). The lake has seasonal dinoflagellate blooms and is annually laminated (Chu et al., 2009) (Figure S1 in Supporting Information S1). An 845-cm-long piston core was raised from a water depth of 14.5 m near the center of the lake in early spring 2006.

Regional Vegetation and Climate
Surveys of the modern vegetation show that temperature controls the altitudinal and latitudinal distribution of vegetation zones in Northeast Asia (Binney et al., 2017;Sun et al., 2003;Zheng et al., 2014), as follows.
1. Temperature decreases with increasing latitude in Northeast Asia, resulting in the following distribution of vegetation zones, from south to north: temperate broadleaved deciduous forest, cool-temperate mixed deciduous and coniferous forest, Taiga forest, and tundra with forbs (Binney et al., 2017 (Sun et al., 2003;Xu et al., 2014Xu et al., , 2019. The climate of the study region is mainly controlled by the East Asian monsoon system Zhao et al., 2009). High rainfall is usually coupled with high temperatures driven by the transport of heat and water vapor from the mid-to low-latitude Indo-Pacific Ocean during the summer season, whereas the winter monsoon is controlled by the Siberian High and results in a cold and dry climate during the winter season Zhao et al., 2009).

Chronology
In this study, we built a cross-validated age-depth model for 413-752 cm using three steps. First, we compared accelerator mass spectrometry (AMS) 14 C ages of leaf and bulk samples at 430 cm in the slump bottom to determine the anchor point. The bulk samples at 430 and 674 cm were ∼210 and ∼720 yr older than leaf samples respectively, indicating a reservoir effect. Thus, we chose the calibrated median age (11.40 Kyr b2k) of the leaf sample based on Bayesian age-depth modeling as the anchor point age at 430 cm (Figures S2-S4 in Supporting Information S1). Second, laminations between 413 and 752 cm were recounted four times based on this anchor point.  Figure S5 in Supporting Information S1 and Figure 1e). Comparison of the Bayesian model with the lamination-based chronology showed that their mean ages overlapped or were very close; however, the Bayesian model had larger 95% confidence intervals, demonstrating the high reliability and precision of the lamination-based chronology (Texts S1 and S2 in Supporting Information S1).

Pollen Analysis
Samples of weight 0.3 ± 0.05 g at 1-cm intervals were prepared for pollen analysis. This interval was selected for thin section preparation for lamination counting to ensure the optimal temporal correlation between the pollen and lamination data. All samples were treated with potassium hydroxide (KOH), hydrochloric acid (HCl), hydrofluoric acid (HF) and a hot acetolysis mixture (Faegri et al., 1989). Lycopodium spores (27,637 per tablet) were added to each sample to calculate pollen concentrations. Sample residues were suspended in glycerin and analyzed using a Leica DM750 microscope at 400X magnification. Pollen preservation was excellent, and the pollen concentration was high [ranging from 56,380 to 3,979,980 N/g (grains per gram of dry sediments), 505,870 N/g on average]. An average of 743 terrestrial pollen grains were counted for each sample ( Figure S6 in Supporting Information S1). A total of 311 samples with the total of 231,127 pollen grains were analyzed, with the average temporal resolution of ∼29-yr. TILIA 3.0.1 (Grimm, 1987) was used to calculate pollen percentages and to plot the pollen diagram. Pollen percentages were calculated based on the sum of total terrestrial pollen, excluding the pollen of aquatic taxa and fern spores. In this study, we extend the pollen record from 413 to 752 cm (from ∼19.96 to 10.79 Kyr b2k), using 311 fossil pollen samples with a ∼29-yr resolution ( Figure S3 and Table S1 in Supporting Information S1), with the objective of determining the nature, timing, and periodicities of vegetation succession.

Time-Series and Their Quantitative Analysis
The pollen time series was interpolated to a uniform 3-yr interval before the application of time-series analysis. Then, we performed Hilbert-Huang Transform-Ensemble empirical mode decomposition (HHT-EEMD) (Huang et al., 1998;Wu & Huang, 2009) (Text S3 in Supporting Information S1) on the pollen percentages of broadleaved tree, boreal conifer with herb, Betula (henceforth, "Betula" indicates Betula, excluding B. nana) and Artemisia to remove the long-term trend and decadal-scale noise. Redfit spectral (Schulz & Mudelsee, 2002), wavelet (Torrence & Compo, 1998) Figure S7 in Supporting Information S1) were used to determine the periodicities and their temporal stability.

Pollen Record of Lake Xiaolongwan From ∼19.96 to 10.79 Kyr b2k
The pollen diagram of the dominant pollen types (Figure 2), including the elements of boreal coniferous forest (e.g., Abies, Larix, and Picea), temperate broadleaved forest (e.g., Corylus, Fraxinus, Juglans, Viburnum, Quercus, Tilia, and Ulmus)-boreal broadleaved forest [e.g., Alnus, Betula, B. nana], and herbs (e.g., Artemisia, Amaranthaceae, Cyperaceae, and Poaceae), can be divided into five pollen zones (CONISS) (Grimm, 1987) from ∼19.96 to 10.79 Kyr b2k, which are described below. The warming trend from ∼19.96 to ∼10.79 Kyr b2k in Northeast Asia led to the overall northward and upward migration of the vegetation belts (Binney et al., 2017;Park & Park, 2015). Superimposed on these trends there was a series of millennial-scale vegetation changes in response to HS1 and the BA,YD and early Holocene, likely reflecting the impact of the Northern Hemisphere ice sheets and the Siberian High via the winter monsoon on both the altitudinal and latitudinal vegetation succession in Northeast Asia (Park & Park, 2015;Stebich et al., 2009;Wang et al., 2012;Wu et al., 2016).

Orbital and Millennial-Scale Vegetation and Climate Change
Broadleaved tree and boreal conifer with herb show almost antiphase orbital-and millennial-scale changes, with large amplitudes (∼20%-∼85%) during the interval of ∼19.96 to 10.79 Kyr b2k (Figure 2). In addition, the altitudinal gradient of the Changbai Mountain regions within 100 km of Lake Xiaolongwan is occupied by broadleaved forest, boreal coniferous forest with alpine tundra steppe, and their ecotone (Sun et al., 2003(Sun et al., , 2020. Furthermore, along latitudinal or hydrothermal gradients, the regions in the temperate to subarctic Far East are occupied by broadleaved forest and boreal coniferous forest with alpine arctic tundra (Binney et al., 2017;Müller et al., 2010).
Thus, the anti-phase fluctuations between broadleaved tree and boreal conifer with herb represent changes in the ratio of broadleaved forest to boreal coniferous forest with (alpine) tundra steppe, reflecting the response of the  vegetation community to changes in the hydrothermal gradient (Texts S4-S6, Figure S8, and Table S2 in Supporting Information S1) (Binney et al., 2017;Müller et al., 2010;Sun et al., 2003Sun et al., , 2020.

Origin of the ∼500-yr Cycles in Vegetation and Monsoonal Activity
In addition to the orbital-and millennial-scale pattern of vegetation succession revealed by our pollen records (Figure 2), the most striking feature is that the pollen percentages of broadleaved tree and boreal conifer with herb exhibit a series of centennial oscillations superimposed on this general trend. Broadleaved tree and boreal conifer with herb reflect the operation of a "see-saw" pattern of fluctuations on centennial scale. As the pollen percentages of broadleaved tree decrease to a minimum, the pollen percentages of boreal conifer with herb increase to a maximum. The interval between two broadleaved tree maxima (and two valleys in Artemisia) is ∼500 yr (Figures 2 and 3).
The HHT-EEMD results (Figure 3, Text S7, Figure S7, and Table S3 in Supporting Information S1) reveal both general climatic trend of deglacial warming and millennial-scale events (contributing 52%-74% of the total variance of East Asian monsoon changes) and well-defined centennial oscillations in the c3-c7 summed components of broadleaved tree (Figure 3a: c3-c7) and boreal conifer with herb ( Figure 3d: c3-c7). Furthermore, the c6 component directly reflects ∼500-yr oscillations (contribute 6%-13% of the variance). The spectra of the c3-c7 summed components of the pollen percentages of broadleaved tree and boreal conifer with herb reveal peaks with a periodicity of ∼500-yr that are significant at the 99% confidence level (Figures 3b and 3e). The results of the wavelet analysis show that these ∼500-yr quasi-periodic oscillations in broadleaved tree and boreal conifer with herb are near-stable and dominant from ∼19.96 to ∼10.79 Kyr b2k (Figures 3c and 3f). The c6 component of broadleaved tree and boreal conifer with herb reveals a total of 18 ∼ 500-yr oscillations from ∼19.96 to ∼10.79 Kyr b2k (Figure 3a: c6, Figure 3d: c6, and Figure 4a).
This pattern of vegetation changes suggests that the northward advance of the EASM rainbelt led to warm and humid conditions that favored broadleaved forest, whereas a southward retreat of the EASM rainbelt resulted in a cold and arid environment in East Asia that favored coniferous forest with tundra steppe (Mingram et al., 2018;Stebich et al., 2009;Xu et al., 2020;Yang et al., 2015). Moreover, these changes in climate and environment were driven by ∼500-yr periodic oscillations of the EASM.
The ∼500-yr cycle of high proportion phases of broadleaved tree (Figure 3 and Figure S9 in Supporting Information S1) is nearly in-phase with, or lags by several decades behind, the ∼500-yr cycle of La Niña phases (Sulu SST-c3: low SSTs = El Niño phases, high SSTs = La Niña phases) (Rosenthal et al., 2003) (Figure 4 and Figure  S10 in Supporting Information S1). However, the ∼500-yr cycle of high proportion phases of broadleaved tree and the ∼500-yr cycle of La Niña phases exhibit nearly anti-phase oscillations (e.g., cycles 1 and 5), which may have been associated with dating uncertainties with different materials from different regions. Thus, the ∼500-yr cycle of vegetation changes in Northeast Asia during the last deglaciation was almost synchronous and highly coherent with the EASM fluctuations triggered by ENSO.
Holocene climate records from East Asia indicate that the ∼500-yr monsoon cyclicity is triggered by ENSO variability on the centennial scale (Xu et al., , 2020Zhu et al., 2017). This phenomenon can be explained by ENSO changes on the interannual scale, as revealed by modern observations and models (Chen et al., 2019;Shi et al., 2019;Wang et al., 2000Wang et al., , 2017. The mechanism responsible for this forcing is that high-frequency El Niño/La Niña events on the interannual scale can: (a) weaken/strengthen the trade winds and lower/raise the sea surface temperature (SST) of the Indo-Pacific warm pool and northwest Pacific; (b) trigger south-east/north-west movements of the western Pacific subtropical high (WPSH), with changes in the western Pacific anticyclone; and (c) weaken/strengthen the EASM intensity (Huang, 2004;Wang et al., 2000Wang et al., , 2017. Thus, the combined effects of ENSO changes, SST anomalies in the Western Pacific, and changes in the WPSH location together drive movements of the EASM rainbelt (Chen et al., 2019;Huang, 2004;Wang et al., 2000Wang et al., , 2017Xu et al., 2020;Zhu et al., 2017).
This mechanism is supported by modern observations on interannual scales ( Figure S11 in Supporting Information S1): 2015-2016 CE were super El Niño years, and the SST in the Indo-Pacific warm pool and northwest Pacific was low (Chen et al., 2017). Similarly, 2009-2010 CE were also strong El Niño years (Central Pacific type-II), and the SST in the central Pacific was high (Chen et al., 2019). The weakened summer monsoon triggered by southeastward movements of the WPSH results in more rainfall in Central China, and drought in Northeast China (Ma et al., 2018). 2010-2011 CE were La Niña years, and the SST in the Indo-Pacific warm pool and northwest Pacific was high (Zhang et al., 2013). The strengthened summer monsoon triggered by northward movements of the WPSH caused drought in Central China, but increased precipitation in Northeast China (Orsolini et al., 2015).
The cumulative effect of high frequency ENSO changes on the interannual scale can ultimately cause a phase change of El Niño and La Niña on the centennial scale (Moy et al., 2002;Xu et al., 2020;Zhu et al., 2017). Consequently, high-frequency ENSO events on the centennial scale have a major impact on the distribution of monsoonal precipitation between the northern and southern parts of East Asia (Xu et al., , 2020 ( Figure S12 in Supporting Information S1).
Several pollen and speleothem records from northern (Park, 2017;Wu et al., 2019;Xu et al., 2020), central Zhu et al., 2017) and southern (Sheng et al., 2017;Wang et al., 2007;Zhang et al., 2020) East Asia during the last deglaciation and Holocene demonstrate the impact of ENSO events on rainfall and vegetation changes, indicating that ENSO events on centennial scale drove the ∼500-yr cyclic EASM oscillations and the resulting vegetation succession. Furthermore, we hypothesize that the signal of ∼500-year ENSO variability was superimposed on the long-term deglacial warming trend and millennial-scale climatic events, although few examples of ∼500-yr environmental cyclicity have been reported during the last deglaciation. Moreover, the characteristics of the ∼500-yr monsoon cycles were not obscured by the longer-term (orbital-and millennial-scale) climatic background, and they exhibit a different pattern of vegetation change to the millennial-scale warm/ cold events of the HS1, BA, and YD events. During the Holocene  and BA interstadial, the ∼500-yr cycles exhibited a "see-saw" pattern of fluctuation between temperate deciduous broadleaved forest and temperate mixed forest in Northeast China ( Figure 4). However, during the last deglaciation (excluding the BA interstadial), the ∼500-yr cycles in this region exhibited a "see-saw" pattern between boreal broadleaved forest and boreal coniferous forest with (alpine) tundra steppe (Figure 4).

Conclusions
Our pollen record from Lake Xiaolongwan reveals millennial-scale events and a ∼500-yr cyclicity of vegetation and East Asian monsoonal changes during the last deglaciation. Intervals of broadleaved forest corresponded closely to warm and humid phases of ∼500-year monsoonal cycles, whereas intervals of boreal coniferous forest with (alpine) tundra steppe corresponded to the cold and dry phases of these 500-yr monsoonal cycles. Significantly, this ∼500-yr cyclicity in vegetation and climate change across East Asia persisted against the background of deglacial warming and millennial-scale cold and warm events, which was not previously reported. We suggest that for both millennial-scale events and the ∼500-yr cyclic oscillations, the vegetation response to the changes in the East Asian monsoon was related to an atmospheric transmission mechanism. This mechanism involved fluctuations in ENSO frequency on the centennial scale which impacted trade wind intensity, Western Pacific warm pool SST anomalies, the location of the WPSH, and the strength of the EASM, which drove movements of the summer monsoon rainbelt between the northern and southern parts of East Asia, and thus the resulting vegetation oscillations. This mechanism implies that cyclical changes in the regional rainfall redistribution in East Asia remained under the control of low-latitude ocean-atmosphere dynamics during the last deglaciation.
Overall, our findings reveal a close correspondence between monsoon climate and ecosystem changes in East Asia, and they provide insights into the long-term interrelationship between climate change and vegetation succession linked to high-and low-latitude climatic interactions on both millennial and centennial timescales. Importantly, the physical mechanisms of the ∼500-yr cyclicity in the ecological and climate systems of East Asia are yet to be incorporated in coupled land-ocean-atmospheric models. The incorporation of these ∼500-yr cycles of linked climate and vegetation could substantially improve the predictive skill of these models. Hence, it is important to provide an improved understanding of East Asian monsoon dynamics on different timescales associated with the linkage of high-and low-latitude land-ocean-atmosphere processes.

Data Availability Statement
All data generated or analyzed during this study are included in the open access database: https://doi.org/10.5281/ zenodo.8053801.