NATURE Vol 4454 January 2007 LETTERS insights into monsoon dynamics,as the past includes larger ampli- with a 9 point triangular window.Magnetic susceptibility was mea tudes of climate change that may reveal more robust linkages. sured in 2.5 mm steps directly on the surface of the split core halves. Previous palaeoclimate reconstructions generally agree that the Additionally,sediment slices of 4 mm thickness(for the time interval Asian summer monsoon was weaker during cold phases in the of Termination I and the early Holocene)were analysed for their rock Northern Hemisphere1,when the intertropical convergence magnetic properties (see Methods) zone (ITCZ)tends to move southward'1,as it does during El Two rock magnetic parameters,magnetic susceptibility and the Nino years1 S-ratio (Fig.1),measure the concentration of magnetic minerals Here we present a new palaeoclimatic record with nearly annual and the mean oxidation state of iron in those minerals,respectively. time-resolution from a sediment core recovered in Lake Huguang The S-ratio is a (nonlinear)estimate of the abundance of magnetite Maar,southeast China(219'N,11017'E),which extends back to compared to that of antiferromagnetic minerals,mainly haematite22 16.2 kyr ago(Fig.1).The sedimentation rates range from 41cm kyr In Lake Huguang Maar,a high S-ratio indicates the availability of before the Bolling-Allerod to 112 cmkyrduring the past 4,000 yr. bottom water oxygen,and it is interpreted to reflect wind-driven lake The age model is based on 5 AMS(accelerator mass spectrometry) mixing.Magnetic susceptibility is sensitive to both lake redox con- 4C dates of leaves and 4 of bulk sediment,with dating errors of less ditions and the aeolian input,both of which are affected by wind than +160yr within the 1o interval of the AMS C method. strength (see Methods). Adjustments using the well-dated records from Cariaco basin are The Ti content of the sediment is used to reconstruct the aeolian within the error of the originalC-based age model (Fig.1;see also input into the lake(Fig.1).The main lithogenic source to Huguang Supplementary Information).Lake Huguang Maar today lies 23m Maar sediments is dust transported by the winter monsoon winds above sea level and has a water depth of 20 m.The surface area of the from the arid areas in the north-for example,the loess plateau-and lake is 2.25 km,and it drains an extremely small catchment of possibly local sources.Changes in Ti are interpreted as a measure of 3.2 km2.Because of its small catchment and a lack of stream inputs, winter monsoon winds,with stronger winds more effectively trans- the lake receives a minimal quantity of material by runoff and thus porting dense Ti-and Fe-rich grains(including the magnetic miner- acts as a natural sediment trap for dust delivered to the site by the als)over the lake.Ti(rather than Fe)is used here as the dust input northerly winds of the winter monsoon.The Huguang Maar sedi- indicator because of its lack of redox sensitivity;however,the two are ments record the strength ofthe winter monsoon in two independent highly correlated(data not shown). ways:(1)the accumulation of wind-blown material,and (2)the During cold climates,for instance,before 14.8 kyr ago and during redox-sensitive characteristics and total organic carbon (TOC)con- the Younger Dryas(between 12.8 and 11.6 kyr ago),Ticontent,mag- tent of the sediment as a result of changes in wind stress and water- netic susceptibility and S-ratio are high(Figs 1,2)while TOC content column mixing. is low's.In contrast,during the Bolling-Allerod and early Holocene Our palaeoclimate time series are based on continuous measure- (before 7.8 kyr ago),Ti content and rock magnetic amplitudes drop, ments of sediment elemental composition and magnetic susceptibil- and TOC increases(Fig.1).Over the course of the Holocene,mag- ity,augmented by discrete measurements of additional magnetic netic susceptibility clearly increases while the Ti content shows a properties and TOC content(Fig.1).Micro X-ray fluorescence ele- weaker trend towards higher values;over this same time interval, ment scanning was performed at a resolution of 0.5 mm,smoothed the TOC content decreases (Fig.1). 9 95 10 10.5 12 12.5 13 13.5 14.5 15 Figure 2 Comparison of the monsoon sensitive sedimentary reborea Younger Dryas Bolling-Allerod records from Lake Huguang Maar with other climate records.These are from the Cariaco basin",in the Dongge ca southern Caribbean off Venezuela, and Hulu and Dongge caves.The Bolling-Allerod,Younger Dryas and Preboreal are highlighted. 0.92 9-0L 200 500 North (-s squnoo) 25 Cariaco basin (ref.17) Icz (Cariaco basin) 15 10- South 10.5 11 11.5 12 12.5 13.5 145 Age (cal.kyr BP) 75 82007 Nature Publishing Groupinsights into monsoon dynamics, as the past includes larger ampli￾tudes of climate change that may reveal more robust linkages. Previous palaeoclimate reconstructions generally agree that the Asian summer monsoon was weaker during cold phases in the Northern Hemisphere1–3,9–15, when the intertropical convergence zone (ITCZ) tends to move southward16–18, as it does during El Nin˜o years19–21. Here we present a new palaeoclimatic record with nearly annual time-resolution from a sediment core recovered in Lake Huguang Maar, southeast China (21u 99 N, 110u 179 E), which extends back to 16.2 kyr ago (Fig. 1). The sedimentation rates range from 41 cm kyr21 before the Bølling–Allerød to 112 cm kyr21 during the past 4,000 yr. The age model is based on 5 AMS (accelerator mass spectrometry) 14C dates of leaves and 4 of bulk sediment, with dating errors of less than 6160 yr within the 1s interval of the AMS 14C method. Adjustments using the well-dated records from Cariaco basin are within the error of the original 14C-based age model (Fig. 1; see also Supplementary Information). Lake Huguang Maar today lies 23 m above sea level and has a water depth of 20 m. The surface area of the lake is 2.25 km2 , and it drains an extremely small catchment of 3.2 km2 . Because of its small catchment and a lack of stream inputs, the lake receives a minimal quantity of material by runoff and thus acts as a natural sediment trap for dust delivered to the site by the northerly winds of the winter monsoon. The Huguang Maar sedi￾ments record the strength of the winter monsoon in two independent ways: (1) the accumulation of wind-blown material, and (2) the redox-sensitive characteristics and total organic carbon (TOC) con￾tent of the sediment as a result of changes in wind stress and water￾column mixing. Our palaeoclimate time series are based on continuous measure￾ments of sediment elemental composition and magnetic susceptibil￾ity, augmented by discrete measurements of additional magnetic properties and TOC content (Fig. 1). Micro X-ray fluorescence ele￾ment scanning was performed at a resolution of 0.5 mm, smoothed with a 9 point triangular window. Magnetic susceptibility was mea￾sured in 2.5 mm steps directly on the surface of the split core halves. Additionally, sediment slices of 4 mm thickness (for the time interval of Termination I and the early Holocene) were analysed for their rock magnetic properties (see Methods). Two rock magnetic parameters, magnetic susceptibility and the S-ratio (Fig. 1), measure the concentration of magnetic minerals and the mean oxidation state of iron in those minerals, respectively. The S-ratio is a (nonlinear) estimate of the abundance of magnetite compared to that of antiferromagnetic minerals, mainly haematite22. In Lake Huguang Maar, a high S-ratio indicates the availability of bottom water oxygen, and it is interpreted to reflect wind-driven lake mixing. Magnetic susceptibility is sensitive to both lake redox con￾ditions and the aeolian input, both of which are affected by wind strength (see Methods). The Ti content of the sediment is used to reconstruct the aeolian input into the lake (Fig. 1). The main lithogenic source to Huguang Maar sediments is dust transported by the winter monsoon winds from the arid areas in the north—for example, the loess plateau—and possibly local sources. Changes in Ti are interpreted as a measure of winter monsoon winds, with stronger winds more effectively trans￾porting dense Ti- and Fe-rich grains (including the magnetic miner￾als) over the lake. Ti (rather than Fe) is used here as the dust input indicator because of its lack of redox sensitivity; however, the two are highly correlated (data not shown). During cold climates, for instance, before 14.8 kyr ago and during the Younger Dryas (between 12.8 and 11.6 kyr ago), Ti content, mag￾netic susceptibility and S-ratio are high (Figs 1, 2) while TOC content is low23. In contrast, during the Bølling–Allerød and early Holocene (before 7.8 kyr ago), Ti content and rock magnetic amplitudes drop, and TOC increases (Fig. 1). Over the course of the Holocene, mag￾netic susceptibility clearly increases while the Ti content shows a weaker trend towards higher values; over this same time interval, the TOC content decreases (Fig. 1). Age (cal. kyr BP) Magnetic susceptibility (10–6 SI) Ti (counts s–1) stronger stronger weaker weaker Bølling–Allerød Ti (counts s–1) –9 –8 –7 –6 –9 –8 –7 –6 (Cariaco basin) Preboreal Younger Dryas Dongge cave (ref.3) Hulu cave (ref.1) Cariaco basin (ref.17) Winter monsoon (Lake Huguang Maar) Summer monsoon (Hulu cave) Figure 2 | Comparison of the monsoon sensitive sedimentary records from Lake Huguang Maar with other climate records. These are from the Cariaco basin17, in the southern Caribbean off Venezuela, and Hulu and Dongge caves1–3. The Bølling-Allerød, Younger Dryas and Preboreal are highlighted. NATURE| Vol 445| 4 January 2007 LETTERS 75 ©2007 NaturePublishingGroup