The Holocene 13,4 (2003) pp. 465–476

A synthesis of abrupt changes in theAsian summer monsoon since the lastdeglaciationCarrie Morrill,* Jonathan T. Overpeck and Julia E. Cole

(Department of Geosciences, Gould-Simpson Building, University of Arizona,Tucson, AZ 85721, USA)

Received 20 November 2001; revised manuscript accepted 2 September 2002

Abstract: We have compiled 36 previously published palaeoclimate records to determine the timing and spatialpattern of century-scale abrupt changes in Asian monsoon precipitation since the last deglaciation. We identifyabrupt events from (1) the interpretations of the authors of these records and (2) the more objective movingt-test calculation. Our results indicate that abrupt climatic changes occurred at ~11.5 cal. ka, ~4.5–5.0 cal. kaand ad 1300. At the start of the Holocene (~11.5 cal. ka), Asian monsoon precipitation increased dramatically.This climatic change is synchronous with an abrupt warming in the North Atlantic. During the middle Holocene,there was a time of preferred and widespread weakening in monsoon strength (~4.5–5.0 cal. ka). This resultcontradicts previous notions of either a gradual trend towards drier conditions or a series of abrupt events thatoccurred in an unorganized fashion across space and time. The middle-Holocene abrupt event could have beensynchronous with an abrupt cooling event in the North Atlantic, as well as a warming and intensi� cation ofinternannual variability in the tropical Paci� c. In contrast to previous periods, precipitation changes at ad 1300have a heterogeneous spatial pattern. We � nd no conclusive evidence for a change in the Asian monsoon at~8.2 cal. ka, as suggested by several previous studies. More high-resolution data may be needed to observethis short-lived event. Overall, our results attest to the potential for rapid and major shifts in Asian monsoonprecipitation that may be triggered by variations in other components of the climatic system.

Key words: Environmental change, abrupt climatic events, palaeoclimate, climatic change, precipitation,variability, Holocene, Asia.

Introduction

Variability of the Asian summer monsoon impacts many aspectsof the earth system, both regionally and globally. On the regionalscale, extreme variations in the monsoon cause � ooding and cropfailures that impact nearly two-thirds of the world’s population(Webster et al., 1998). On a larger scale, the Asian monsoon isan important component of the climatic system and interacts withother components. Many previous studies suggest that the Asianmonsoon responds to changes in other parts of the climatic sys-tem, including the El Nino-Southern Oscillation (ENSO) andNorth Atlantic thermohaline circulation (e.g., Shukla and Paolino,1983; Yang, 1996; Overpeck et al., 1996). Researchers alsospeculate that � uctuations within the Asian monsoon can modifyclimate elsewhere in the world, by triggering changes in tropicalmethane production or in the transport of water and energy (Gasseand Van Campo, 1994; Liu et al., 2000; Kudrass et al., 2001).

Previous reviews of monsoon variability have focused on

*Author for correspondence. Current address: National Center for AtmosphericResearch, PO Box 3000, Boulder, CO 80307–3000, USA (e-mail: morrill@ucar.edu)

Ó Arnold 2003 10.1191/0959683603hl639ft

determining how and why the Asian monsoon varies on interan-nual and millennial timescales (e.g., Webster et al., 1998; Zhouet al., 1991; Winkler and Wang, 1993). Researchers have givenless attention to abrupt changes in the monsoon that occur oncentury timescales. Changes occurring on these timescales are ofspecial importance, however, because this timescale is relevant tohuman activities. Over the past several years, researchers havegenerated many new records of century-scale � uctuations in mon-soon strength, which allow us to develop a more comprehensivepicture of the spatial and temporal � uctuations of the monsoonon this timescale. In our research, we examine these records toanswer two questions: (1) when have abrupt variations in thestrength of the Asian monsoon occurred since the last deglaciationand (2) what is the spatial signature of precipitation anomaliesduring these time periods?

To answer these questions, we have extracted information fromthe palaeorecords in two different ways. First, we catalogued thetiming and direction of precipitation changes as reported by theauthors of these records. This approach was taken in several pre-vious reviews of monsoon variability (e.g., Shi et al., 1993; Over-peck et al., 1996). However, this approach has several limitations.

466 The Holocene 13 (2003)

Table 1 Records of Asian summer monsoon strength

Site name Agea Record type Proxiesb

1. Somali upwelling area PH Marine Diatom assemblage2. Arabian Sea 74KL PH Marine Carbonate, dolomite, barium3. Hoti Cave PH Speleothem d18O4. Arabian Sea SO90–56KA H, PH Marine Varve thickness5. Arabian Sea SO90–137KA PH Marine TOC, foram assemblage6. Arabian Sea 111KL PH Marine TOC7. Nal Sarovar PH Lake d13C, C/N8. Karwar Coast PH Marine; pollen d13C, savanna, evergreen9. Arabian Sea 3268G5 PH Marine d18O, d13C, carbonate

10. Lunkaransar PH Lake d13C11. Sambhar PH Pollen Gramineae, Artemisia12. Guliya H Ice core Accumulation13. Siddha Baba Cave H Speleothem Mineralogy14. Dasuopu H Ice core dD, dust15. Dunde PH Ice core NO3, total particles16. Qinghai Lake PH Lake Carbonate17. Yiema Lake PH Lake MS, TOC18. Baxie PH Loess MS, TOC19. Zoige Plateau PH Pollen Abies, Picea20. Shayema PH Pollen Deciduous, evergreen21. Panlong Cave PH Speleothem d18O22. Chinese lake database H Historical Lake-level observation23. Midiwan PH Loess TOC24. Yulin PH Loess MS25. Yangtaomao PH Loess MS26. China semi-arid region H Historical Weather observation27. Maili H Pollen Quercus28. China semi-wet region H Historical Weather observation29. Jianghu PH Pollen Evergreen30. China wet region H Historical Weather observation31. Taihu H Historical Weather observation32. Hanjiang Delta PH Pollen Evergreen33. Toushe Lake PH Pollen Tree34. Chia-Min Lake H, PH Lake TOC, C/N35. Great Ghost Lake H Lake TOC, C/N36. South China Sea 17940–2 H, PH Marine d18O, alkenone SST, clays

aPH = pre-historical (before ad 250); H = historical (after ad 250).bTOC = total organic carbon; MS = magnetic susceptibility; C/N = carbon/nitrogen.

For example, the focus and length of the published paper willdetermine which events the author mentions. Second, we used astatistical test to identify abrupt climatic change in each individualrecord. We use results from both of these methods, along withevidence from palaeoclimate records from other regions of theworld, to speculate about possible causes for these century-scaleabrupt events.

Methods

We � rst compiled palaeoclimate records from the published litera-ture. Records must meet several criteria in order to be includedin our compilation. First, study sites must be located in regionsthat were under the in� uence of the Asian summer monsoon dur-ing the Holocene (Figure 1). This includes sites in� uenced byeither of the two regional monsoons in Asia, the Southwest Asianmonsoon (or Indian monsoon) and the East Asian monsoon. Weused the millennial-scale palaeoclimate reconstruction of Winklerand Wang (1993) to determine which sites meet this criteria. Forsites that were affected by the monsoon for just part of the Holo-cene, we examined only the segment of the record that, accordingto the authors of the record, re� ected monsoon � uctuations.

Second, resolution of the proxy records must be suf� cient toobserve abrupt events that occur over ~100 to ~500 years. We

Figure 1 Map of Asian monsoon region showing locations of palaeo-climate records used in this study (numbered according to Table 1). Linesindicate modern extent of East Asian and Southwest Asian monsoons(dashed lines) and maximum areal extent of monsoons during Holocene(solid line), as mapped by Winkler and Wang (1993).

required at least 50-year resolution for records spanning the inter-val with many historical records (i.e., since ad 250) and at least250-year resolution for records from earlier periods (i.e., from~15.0 cal. ka to ad 250). Requirements differ for the historicaland pre-historical time periods because the temporal resolution ofmany records is much � ner for the historical time period. The low

Carrie Morrill et al.: Abrupt changes in the Asian summer monsoon 467

Table 1 Continued

Site no. Timespan used (yrs ad or BP) Average sample No. of datesc Referencespacing (yrs)

1 15 250–11 400 BP 140 5 Zonneveld et al. (1997)2 15 250–1750 BP 250 8 Sirocko et al. (1993; 1996)3 11 160–7140 BP 40 7 Burns et al. (1998)4 4850 BP–ad 1950 Annual 12 von Rad et al. (1999a)5 15 250–10 900 BP 170 7 von Rad et al. (1999b)6 15 250–1940 BP 180 3 Schulz et al. (1998)7 7520–1750 BP 170 7 Prasad et al. (1997)8 5120–1750 BP 80 4 Caratini et al. (1994)9 10 040–1750 BP 250 4 Sarkar et al. (2000)

10 11440–5700 BP 100 6 Enzel et al. (1999)11 11 440–3180 BP 110 6 Singh et al. (1974)12 ad 1000–1950 Annual L Thompson et al. (1995)13 ad 250–1950 Subdecadal 5 Denniston et al. (2000)14 ad 1000–1950 Decadal L Thompson et al. (2000)15 15 250–4000 BP 180 L Thompson et al. (1989)16 15250–1750 BP 30 6 Kelts et al. (1989); Lister et al. (1991)17 15 250–2580 BP 80 5 Chen et al. (1999)18 14 730–6100 BP 160 5 An et al. (1993); Zhou et al. (1994)19 11100–7500 BP 140 6 Yan et al. (1999)20 12 880–1750 BP 110 5 Jarvis (1993)21 9200–1750 BP 50 11 Li et al. (1998)22 ad 250–1950 Annual H Fang (1993)23 15 250–1940 BP 100 17 Zhou et al. (1996)24 10 780–2580 BP 80 6 Shi et al. (1993)25 13 170–7770 BP 110 6 Zhou et al. (1996)26 ad 250–1950 20 H Gong and Hameed (1991)27 ad 250–1950 30 6 Ren (1998)28 ad 250–1950 20 H Gong and Hameed (1991)29 11 610–1750 BP 100 9 An et al. (2000)30 ad 250–1950 20 H Gong and Hameed (1991)31 ad 1120–1935 30 H Wang and Zhang (1992)32 12 500–6400 BP 210 5 Zheng and Li (2000)33 15 250–1940 BP 100 13 Huang et al. (1997)34 3900 BP–ad 1950 50 6 Lou and Chen (1997)35 ad 250–1950 10 4 Lou et al. (1997)36 15 250 BP–ad 1950 20 20 Wang et al. (1999a,b)

cL = age model based on layer counts; H = age model based on historical records.

resolution for the pre-historical time period does not allow us tonote century-long wet or dry spells. Instead, this analysis focuseson step changes between climatic states that occur on centurytimescales.

Third, age control in each record must allow for precise datingof century-scale events. We did not include records of non-continuous deposits (e.g., moraines, river deposits, lake terraces),records from lakes with hardwater effects greater than 500 yearswhose age control is not obtained from terrestrial macrofossils,and records lacking precise radiocarbondates (i.e., errors less than250 years).

Fourth, the proxies measured must re� ect variations in summermonsoon rainfall. In some cases, records we used may also re� ect,to a lesser degree, summer temperature or winter rainfall� uctuations (e.g., records of lake-water balance). We used theserecords in order to have a larger sample size for our analyses. Wedid not include records that re� ect only temperature or winter pre-cipitation.

There were 36 records that satis� ed our selection criteria (Table1). The locations of these records are shown in Figure 1. Thirteenof these records provided information on the historical time periodand 26 records provided information about the pre-historical timeperiod. Both time periods contained a variety of different proxyrecords (e.g., marine, lake, ice core) and were not dominated byany one proxy type (Table 1).

We identi� ed century-scale events in two ways. First, we com-piled events noted by the authors of these palaeoclimate records.Second, we identi� ed events using a rigorous and more objectivestatistical test, the moving t-test. Many researchers have pre-viously used the moving t-test method in order to identify abruptclimatic change (e.g., Karl and Riebsame, 1984; Fraedrich et al.,1997; Cao, 1998). This method is based on a series of t-statisticscalculated for adjacent windows of time along the length of adata record. If the t-statistic indicates that the mean values of twoadjacent windows are statistically different, then this is evidencethat an abrupt climatic change occurred between the two timeperiods. The t-statistic is calculated as:

t = (x2 2 x1) * n1/2 * (s22 + s2

1)2 1/2

where x1 and x2 are the means, s1 and s2 are the standard devi-ations of the two adjacent windows and n is the number ofmeasurements in each of the two adjacent time intervals. Whenthe number of measurements (n) was different for the two adjacenttime intervals, we used the lower number for n. This provided amore conservative calculation for statistical signi� cance. Numericdata � les for each of the 36 records were obtained either fromthe World Data Center for Paleoclimatology or from digitizingpublished � gures.

We made t-test calculations using a series of overlapping win-dows along the proxy time series. The � rst t-test calculation was

468 The Holocene 13 (2003)

made for two windows that met halfway between the second andthird data points in the proxy time series. The second calculationwas made for two windows that met halfway between the thirdand fourth data points, and so on. This process was designed totest for the presence of an abrupt change between each pair ofadjacent data points in the proxy time series. We used windowlengths ranging from 100 years to 500 years for the historical timeperiod and 500 to 2500 years for the pre-historical time period.We identi� ed abrupt events that were consistent across more thanone window length and that were statistically signi� cant at the99% and 95% con� dence levels.

For records with more than one proxy re� ecting summer mon-soon precipitation, we performed the moving t-test on eachindividual proxy. For these multiproxy records, we included inour compilation abrupt events that occurred: (1) in two or moreproxies, with the proxies indicating precipitation changes in thesame direction; or (2) in only one proxy, with all other proxiesindicating no precipitation change. We feel it is valid to includeevents recorded in only one proxy of a multiproxy record becausesome proxies might be more sensitive to particular abrupt changesthan other proxies are. Events observed in just one proxy are lessconvincing than events observed in two more proxies of a multi-proxy record. However, if we ignored events observed in onlyone proxy, we might have ignored real climate events. We didnot include abrupt events that occurred in two or more proxiesbut whose inferred precipitation change was not consistent amongthe proxies.

Lastly, we re-examined the proxy records in order to ensurethat the abrupt events identi� ed by the moving t-test were realand were not artifacts of our statistical approach. We clearly sawwithin the proxy records each abrupt event identi� ed by themoving t-test at the 99% con� dence level. Abrupt events ident-i� ed at the 95% con� dence level were often less clear, however,and we therefore report only on results obtained at the 99% con-� dence level.

To determine the time intervals in which the most widespreadcentury-scaleevents occur, we constructed histograms of the num-ber of events observed through time for both the historical andpre-historical time periods. For these histograms, we divided thehistorical time period into 100-year intervals and, due to the lowerresolution of the earlier palaeoclimate records, divided the pre-historical time period into 500-year intervals. We made additionalcalculations using intervals with different lengths and midpointsto verify that our results did not depend on these factors (notshown).

We expressed our results from both methods of compilation asanomalies from the expected number of events per histogram bin,assuming a random distribution of events through time. We alsocorrected the expected number of events for several factors. First,we scaled the expected number of events in each bin accordingto the number of records spanning the bin so that intervals thatare spanned by fewer records have a lower number of expectedevents (Figure 2). Second, we corrected for errors resulting fromthe non-linear relationship between the radiocarbon and calendartimescales. Most of the early to middle Holocene records weredated according to the radiocarbon-year timescale and we con-verted these ages to the calendar-year timescale using CALIB v.4.1 (Stuiver and Reimer, 1993) and the calibration data set ofStuiver et al. (1998). Errors occurred because 500-year bins thatspan times when the calendar timescale is compressed (expanded)relative to the radiocarbon timescale will be associated with more(fewer) radiocarbon dates and will therefore have an increased(decreased) likelihood of events falling within them. We correctedfor this factor by scaling the expected number of events in eachbin by the number of radiocarbon years each bin spans (Figure 3).

We did not perform separate analyses for the East Asianmonsoon subregion and the Southwest Asian monsoon subregion

Figure 2 Number of records spanning each (A) 100-year bin within thehistorical time period and (B) 500-year bin within the pre-historicaltime period.

Figure 3 Number of radiocarbon years spanned by each 500-calendaryear bin in the pre-historical time period.

because there were too few records for separate analyses. There-fore, our study is most likely to identify abrupt events that arewidespread and occur in both subregions. As more palaeoclimaterecords become available, subregional analysis will become poss-ible.

When do abrupt events occur?

It appears that the most prominent abrupt shift in monsoonstrength during the historical period took place at ad 1300 6 50

Carrie Morrill et al.: Abrupt changes in the Asian summer monsoon 469

years (Figure 4, A and B). We calculated a signi� cance test todetermine if the anomaly number of events detected by the authorsand the moving t-test at ad 1300 can be explained by chance. Wefound that the anomaly number of abrupt events is signi� cantlydifferent from zero at the 95% con� dence level, indicating thatan abrupt event most likely occurred at this time. The anomalynumber of events during other periods (e.g., ad 1000, ad 700–600) may be statistically signi� cant at lower con� dence levels,but they are not signi� cant for both compilations.

The authors of records spanning the pre-historical period detectimportant abrupt events at ~11.5 cal. ka, ~12.0 cal. ka and ~13.0cal. ka (Figure 4C). The anomalous numbers of events duringthese periods are all signi� cantly different from zero at the 95%

con� dence level. In contrast, the moving t-test results indicate asigni� cant number of events at only ~11.5 cal. ka (Figure 4D).The moving t-test also detects an important abrupt climatic changethat occurred at ~4.5 to ~5.0 cal. ka, which the authors of theserecords do not note as frequently.

A century-scale climatic change occurring ~8.2 cal. ka in theNorth Atlantic region (e.g., Alley et al., 1997; von Grafenstein,1998) is not indicated in the Asian monsoon region either by theauthors of the palaeorecords we examine or by the moving t-test.Instead, most of the bins representing the early to middle Holo-cene have a negative anomaly number of events, suggesting thatthis period of the Holocene was marked by fairly stable monsoonconditions. The 8.2 cal. ka event was a short-lived (i.e., severalhundred year) excursion from the mean climatic state. Our analy-sis, which focused on step changes from one climatic state toanother, is not likely to detect this event.

What is the spatial pattern of climaticchange?

The spatial pattern of climatic changes occurring at ad 1300 isheterogeneous (Figure 5, A and B). According to records from

Figure 4 Histograms showing the number of abrupt events observed in the Asian monsoon region for historical and pre-historical time periods. Eventfrequencies were obtained by using interpretations of the authors of the palaeorecords (A, C) and by a moving t-test (B, D). Frequency of events for eachbin is expressed as anomalies from expected number of events assuming a random distribution of events through time and correcting for two factorsdescribed in text. Black (striped) shading indicates bins with a positive anomaly that is statistically different from zero at the 95% (90%) con� dence level.

Tibet and the Arabian Sea, conditions became drier in the South-west Asian monsoon region. In Taiwan, two lake records show atransition to colder and/or drier conditions at this time. Similarly,a pollen record in NE China indicates climate changed to drierconditions. In east-central China, on the other hand, four historicalrecords based on weather observations indicate a shift towardsmoister conditions.

In contrast to the change observed at ad 1300, the spatial pat-terns of climatic change during the early and middle Holoceneare largely homogeneous across the Southwest and East Asianmonsoon regions. At ~4.5 to ~5.0 cal. ka, climate became colderand/or drier across the region (Figure 5, C and D). Two exceptionsare shifts towards moister conditions indicated by a marine varverecord off the coast of Pakistan (detected by the authors of therecord) and a marine sediment record off the coast of southeastChina (detected by the moving t-test). At ~11.5 cal. ka, abruptchanges occur without exception towards moister and/or hotterconditions (Figure 5, E and F).

The spatial pattern of change is less clear at ~13.0 cal. ka,which is close to the time commonly assigned to the start of theYounger Dryas interval (Figure 5, G and H). There are severalreasons for this. First, there are fewer records spanning this inter-val (Figure 3), which provides us with less information fromwhich to deduce patterns. Second, age control of the proxy recordsgenerally worsens with age, which leads to more errors in ageassignment. Third, and perhaps most important, several proxy rec-ords from the monsoon region that record abrupt change at ~13.0cal. ka suggest that several short-lived alternations between drierand moister conditions occurred during this time period (Zhouet al., 1996; Wang et al., 1999b). An analysis with resolution � nerthan the 500-year bins that we use might be required in order tomake sense of the climatic changes observed during this part ofthe deglacial period. Therefore, in the following sections, we willfocus our discussion on the abrupt climatic changes occurringduring the Holocene.

470 The Holocene 13 (2003)

Figure 5 Maps of the spatial distribution of climatic changes detected by authors of the palaeorecords (left column) and by a moving t-test (right column).Letters indicate direction of climatic changes, with M indicating a shift towards moister conditions, D a shift towards drier conditions, CD a shift towardscolder and/or drier conditions, and HM a shift towards hotter and/or moister conditions. Dots indicate locations of palaeoclimate records that span theparticular time interval but show no abrupt climatic change.

The abrupt events observed by the authors and by the movingt-test are recorded in a variety of different proxies, which givesus con� dence that these events are real. In fact, the proxies show-ing abrupt change at a particular time interval tend to be a rep-resentative sample of the proxy types spanning that time period.We do note, however, that records of many different proxy typesspan the middle and late Holocene, while records spanning thedeglaciation and early Holocene are more likely to be either loessor marine records.

There is a good correspondence between the abrupt eventsdetected by the authors and by the moving t-test at ad 1300(Figure 5, A and B). There is less correspondence between theabrupt events noted by the authors and by the moving t-test atother time periods. Any lack of correspondence between the twomethods was due either to the methods assigning slightly differentages to the same event or to the t-test method detecting severalevents at only the 95% (rather than 99%) con� dence level.

Another, less important, reason is that the moving t-test is lesslikely to detect events near the end of a record, where there aretoo few data points to identify a century-scale change.

The authors tend to observe more events at ~11.5 cal. ka andfewer events at ~4.5 to ~5.0 cal. ka than the moving t-test does.This could be the case because previous review articles commonlysuggest that an abrupt change in the monsoon occurred at the startof the Holocene (e.g., Wang and Fan, 1987; Overpeck et al.,1996), and researchers have logically sought to con� rm or denythat this abrupt event occurs in their own records. In contrast, thisis the � rst study to suggest that a widespread abrupt eventoccurred at ~4.5 to ~5.0 cal. ka in the Asian monsoon region.Previous studies indicated either a gradual weakening of monsoonstrength due to slow changes in orbital forcing (Overpeck et al.,1996) or a series of abrupt events occurring in an unorganizedfashion across space and time during the middle to late Holocene(Wang and Fan, 1987; Shi et al., 1993).

Carrie Morrill et al.: Abrupt changes in the Asian summer monsoon 471

What causes abrupt change in theAsian monsoon?

The ultimate cause of the abrupt events we observe must be vari-ations in external forcing or natural, internal � uctuations eitherwithin the monsoon or in a remote system. External forcingsinclude � uctuations in solar radiation and the frequency of vol-canic events and internal forcings include changes in the NorthAtlantic thermohaline circulation and the El Nino-Southern Oscil-lation (Rind and Overpeck, 1993). Based on present knowledge,we are unable to eliminate any of these possibilities. Internaloscillations occur in several climate model simulations withoutexternal forcing (e.g., Stocker and Marchal, 2000; Walland et al.,2000; Hall and Stouffer, 2001). Model results only indicate thatthe climatic system may be capable of these variations, however,not that these � uctuations actually occurred.

Estimates of solar variations based on measurements of 14C and10Be (Stuiver et al., 1998; Finkel and Nishiizumi, 1997; Bardet al., 2000) and estimates of � uctuations in volcanic activitybased on sulphate preserved in the GISP2 ice core (Zielinski andMershon, 1997) do not show clear correspondence between vari-ations in external forcing and the abrupt changes we haveobserved (Figure 6). One possible exception is abrupt changes in14C and 10Be at the start of the Holocene. However, � uctuationsin 14C and 10Be are also in� uenced by climatic changes, and it isunclear to what extent these variables re� ect solar variations.Also, the inferred solar � uctuations are small, and ampli� cationof this forcing from processes acting within the climatic systemis crucial for explaining the abrupt change. Climatic changecaused by a single volcanic eruption is short-lived. Unless a shiftin the frequency of eruptions occurred, processes acting withinthe climatic system are also required to prolong the initial effectsof volcanism. It is also important to note that there are many

Figure 6 Time series of proxies for solar variations and volcanic activitysince the last deglaciation. (A) Radiocarbon data from Stuiver et al.(1998); (B) 10Be data from Finkel and Nishiizumi (1997) and (C) volcanicsulphate data from Zielinski and Mershon (1997).

� uctuations in both solar and volcanic activity during the Holo-cene that do not correspond to abrupt events within the Asianmonsoon. This indicates that, if external forcing was the initialforcing mechanism for the abrupt changes we observe, the cli-matic system must play an important role in determining theresponse to this external forcing. In the following sections, wereview changes observed in other parts of the climatic system thatwere synchronous with abrupt events in the Asian monsoon anddiscuss possible linkages between these changes and summermonsoon � uctuations.

Onset of the HoloceneThe abrupt event we observe at ~11.5 cal. ka corresponds withthe start of the Holocene and with abrupt climatic changes inmany regions of the world. Ice-core records from Greenland sug-gest that temperatures increased 5–10°C in a few decades or lessat the start of the Holocene (reviewed in Alley, 2000). Similarly,a wide variety of proxy records from Europe provide evidencefor a 4–7°C temperature increase within a few decades at the startof the Holocene (reviewed in Walker, 1995; Ammann et al.,2000). Decreases in ice-rafted debris and the foraminifera N.pachyderma in North Atlantic marine records also point towardsa rise in sea-surface temperatures at this time (Lehman andKeigwin, 1992; Bond et al., 1997). In many regions of NorthAmerica, pollen and lake isotopic evidence suggests a rapid tem-perature increase of about several degrees at the start of the Holo-cene (reviewed in Peteet, 1995; Yu and Wright, 2001). In thetropics, lake-level records from Africa provide strong evidencefor increases in moisture during the early Holocene (reviewed inGasse, 2000) and records from South America show spatially vari-able changes in temperature and moisture (Thompson et al., 1998;Betancourt et al., 2000; Baker et al., 2001). Evidence for climaticchanges in the temperate and high-latitude regions of the SouthernHemisphere is more equivocal. Pollen records from New Zealandand Southern Chile provide no evidence for climatic change dur-ing this time period (Markgraf, 1993; Singer et al., 1998; Bennettet al., 2000), while other records from these regions suggest anabrupt warming and glacier retreat (Hellstrom et al., 1998; Mor-eno et al., 2001). In general, little change is observed in recordsfrom Antarctica at this time (Sowers and Bender, 1995). Oneexception is the ice-core record from Taylor Dome in coastal EastAntarctica, which shows dD variations that are synchronous withchanges observed in Greenland (Steig et al., 1998).

In order to explain these climatic changes, researchers cite evi-dence for enhanced North Atlantic deep-water formation at thestart of the Holocene (e.g., Boyle and Keigwin, 1987; Marchittoet al., 1998; Hughen et al., 2000) and subsequent changes inatmospheric circulation. A modelling study by Overpeck et al.(1996) suggests that an increase in deep-water formation couldalso have affected the Asian monsoon. In their simulation, warmanomalies in the North Atlantic were advected by the westerliesover the Eurasian continent. Warmer temperatures over Eurasiareduced snow accumulation, which increased land temperatures inthe spring and summer and enhanced the land-to-sea temperaturegradient that drives the Asian monsoon.

Several pieces of evidence support the Overpeck et al. (1996)hypothesis.First, the timing of abrupt change in the North Atlanticand in the Asian monsoon is the same, given the limits of theprecision of radiocarbon dating. This similarity makes it likelythat the events are related and also suggests that any teleconnec-tion between the two regions occurred rapidly and throughchanges in the atmosphere. Second, the abrupt changes observedin these two regions are both of large magnitude. In the monsoonregion, the abrupt change at the start of the Holocene is nearlyalways the largest shift of proxy values observed in the records.Similarly, the temperature increase in the North Atlantic regionat this time is larger than any other observed during the Holocene.

472 The Holocene 13 (2003)

Other hypotheses propose that a strengthened Asian monsooncould cause a warming in the North Atlantic by either (1) increas-ing atmospheric methane concentration due to the expansion oftropical wetlands (Gasse and Van Campo, 1994) or (2) strengthen-ing the hydrologic cycle and causing a sea-level fall that wouldstabilize coastal glaciers in the North Atlantic region and allow fora strong thermohaline circulation (Kudrass et al., 2001). Severalquestions are raised about each of these mechanisms, however.Severinghauset al. (1998) suggest that the increase in atmosphericmethane concentrations at the start of the Holocene lags the tem-perature increase in the North Atlantic by several decades. Fair-banks (1989) documents a sea-level rise, rather than fall, at thestart of the Holocene. Therefore, we think the Overpeck et al.(1996) hypothesis describes the most likely causal link betweenchanges in the North Atlantic and in the Asian monsoon at thestart of the Holocene.

Middle HoloceneAbrupt climatic changes occurring ~4.5 to 5.0 cal. ka have alsobeen observed in many regions of the world. Lakes across Africaexperienced major regressionsat ~4.5 cal. ka and have never sinceexpanded to their previous levels (Gasse, 2000). Other evidencesuggests that a decrease in vegetation and an increase in aeoliandust transport in North Africa could have begun earlier at ~5.5cal. ka (deMenocal et al., 2000). A signi� cant increase in aridityin Northern Mesopotamia beginning at 4.2 cal. ka, similar to thetiming of the 4.5 cal. ka bin (4.25 to 4.75 cal. ka), led to thecollapse of the Akkadian empire (Weiss et al., 1993; Cullen et al.,2000). In the tropical Paci� c region, a variety of coral, pollen andlake records indicate that interannual variations associated withENSO increased and that more intense and more frequent ENSOwarm events began to occur during the middle Holocene(McGlone et al., 1992; Shulmeister and Lees, 1995; Gagan et al.,1998; Rodbell et al., 1999; Cole, 2001; Tudhope et al., 2001). Inthe North Atlantic, an abrupt change towards cooler conditionsduring this time is documented by several palaeoceanographicrec-ords (Keigwin, 1996; Bond et al., 1997; Jennings et al., 2002).

These observations of abrupt change at ~4.5 to 5.0 cal. ka sug-gest several possible causes for the change in monsoon strength.First, the abrupt shift towards drier conditions in the monsoonregion could have resulted from cold temperature anomalies inthe North Atlantic being advected over Eurasia, similar to themechanism described for the start of the Holocene, but with theanomalies in the opposite direction. This hypothesis has severalpotential weaknesses, however. The abrupt change observed inthe North Atlantic at this time was one of many changes thatoccurred through the middle and late Holocene and, in general,there does not seem to be anything about its magnitude or durationthat sets it apart from the others. One exception is the cold eventat ~4.7 cal. ka observed by Jennings et al. (2002), which appearsto mark the onset of a series of cold events in their record. How-ever, this event might merely re� ect changes in local glacierdynamics or the gradual southward movement of the polar frontrather than a shift in climate regime (Jennings et al., 2002). Thereare also discrepancies between the timing of the events in theNorth Atlantic, with the record of Bond et al. (1997) showing coldevents out-of-phase with those observed by Jennings et al. (2002)and Keigwin (1996). It is necessary to determine whether theseevents in the North Atlantic are synchronous with the abruptdecrease in monsoon strength before this hypothesis can beaccepted.

The abrupt shift we observe in the monsoon region could alter-natively be related to changes in the state of ENSO. An increasein the frequencyof strong El Nino events could cause the decreasein monsoon precipitation we observe during the middle Holoceneby altering the location of convection in the Paci� c Ocean. DuringEl Nino events, warmer SSTs in the central and eastern equatorial

Paci� c cause surface convergenceto occur further east, away fromthe region of the Asian monsoon (Shukla and Paolino, 1983;Yang, 1996). This is thought to draw moisture away from Asiaand cause a weaker summer monsoon.

This hypothesis has several uncertainties that need to beresolved before it can adequately explain the abrupt change inAsia. First, the exact timing of the change in ENSO is not wellde� ned, and it is not clear whether this change was gradual orabrupt. Second, the mechanism responsible for less intense andless frequent warm events in the middle Holocene is not clear.Authors have identi� ed several processes that are forced byinsolation changes and that could be responsible for the middle-Holocene ENSO state. These include coupled ocean-atmosphereprocesses occurring within the equatorial Paci� c (Clement et al.,2000), increased transport of zonal momentum from the extratrop-ics into the tropical Paci� c (Bush, 1999) and northward transportof SST anomalies into the equatorial Paci� c from the south Paci� c(Liu et al., 2000). However, it is also possible that variations inAsian monsoon circulation affect the state of ENSO. Liu et al.(2000) suggest that the intensi�cation of the Asian monsoon dur-ing the middle Holocene led to enhanced trade winds in the equa-torial Paci� c, which inhibited the development of warm El Ninoanomalies. It may be most accurate to consider the Asian mon-soon and ENSO as interacting systems, in which a change in oneinduces a change in the other, which in turn feeds back on the� rst. For example, a gradual decrease in insolation could havecaused a gradual reduction in monsoon strength, which in turntriggered an abrupt shift in ENSO and a subsequent abruptdecrease in monsoon strength.

A third possibility is that the abrupt change in the middle Holo-cene originated within the monsoon region, as the result of a non-linear land-atmosphere process responding to insolation change.Claussen et al. (1999) propose that the abrupt decrease in Africanmonsoon precipitation in the middle Holocene is due to the effectsof an abrupt decrease in vegetation cover in the Sahara, whichwas a response to slowly varying insolation changes. Our resultsindicate, however, that the abrupt climate shift ~4.5 to 5.0 cal. kaBP extended well beyond North Africa. One implication is thatregional land-atmosphere interactions may not have been the pri-mary or sole driver of the abrupt shift observed in Africa, butrather just one important player. This is consistent with the grow-ing number of papers (e.g., Kutzbach and Liu, 1997; Hewitt andMitchell, 1998) that indicate ocean feedbacks may also haveplayed an important role in the circulation of the African monsoonduring the middle Holocene. Additionally, although a similarregional land-atmosphere interaction could have occurred in Asia,it would be surprising (but not impossible) if this interactionoccurred synchronously in both Africa and Asia. Therefore,changes in either the North Atlantic or the tropical Paci� c appearto be the most likely cause of the abrupt change we observe. Bothof these explanations are consistent with physically based modelsand with palaeodata from these two regions.

Late HoloceneThe abrupt event we observe at ad 1300 falls within the timeperiod palaeoclimatologists sometimes describe as the transitionbetween the ‘Mediaeval Warm Period’ and the ‘Little Ice Age’.The use of these terms may be misleading, however, because cli-mate during these time periods was not consistent through eithertime or space. Records from these time periods show several sub-stantial decadal to multidecadal warm and cold temperature anom-alies and also disagree with one another regarding the timing ofthese climatic changes (Grove, 1988; Bradley and Jones, 1993;Hughes and Diaz, 1994; Mann et al., 1999; Crowley and Lowery,2000). Despite the climatic heterogeneity of this time period,several high-resolutionrecords show a convincing shift in climaticregimes at ad ~1300. A record of G. bulloides abundance in the

Carrie Morrill et al.: Abrupt changes in the Asian summer monsoon 473

Cariaco Basin suggests that a shift in the variability of NorthAtlantic SSTs occurred at about ad 1320, from a time of rapid(less than a decade) high-amplitude changes to a period of slower(10–20 years) small-amplitude changes (Black et al., 1999). Sev-eral records from the central United States indicate that droughtsin that area changed from extreme, multidecade events to lesssevere and less persistent events sometime between ad 1200 and1300 (Laird et al., 1996; Woodhouse and Overpeck, 1998). Lastly,Verschuren et al. (2000), using fossil diatom and midge assem-blages from Lake Naivasha in Kenya, conclude that East Africaexperienced a climatic shift from dry and more stable conditionsto wet and less stable conditions at ad ~1270.

Climatic changes at this time, both in the Asian monsoon regionand around the world, are more heterogeneous than earlier in theHolocene. Due to this heterogeneity, there are no clear connec-tions between the Asian monsoon region and other regions of theworld at this time. The pattern of change within the Asian mon-soon region may provide information about the source of this cli-matic change, however. We observe a shift towards drier con-ditions in the Southwest Asian monsoon area, Taiwan andnortheasternChina, while east-central China becomes wetter. Thisis similar to the pattern that is observed in modern meteorologicalrecords during years with increased snowcover over Eurasia(Hahn and Shukla, 1976; Yang and Xu, 1994). This anomaly pat-tern does not occur with changes in ENSO, another strong controlon monsoon variability on the interannual timescale. Changes inthe state of ENSO tend to correlate with rainfall anomalies inIndia and east-central China that have the same sign (Hu andNitta, 1996).

Conclusions

To summarize, we identi� ed abrupt changes in the strength of theAsian monsoon at four time periods since the last deglaciation.Variations in monsoon strength at ad 1300 are more spatially het-erogeneous than at the other time periods, perhaps because thisclimatic change was smaller than during other time periods. Thepattern of precipitation changes is similar to the pattern occurringtoday due to an increase in winter snowcover over Eurasia. Moredata are needed to determine if an increase in snowcover couldhave caused the change in monsoon strength at ad 1300.

This research is the � rst to identify a period of widespread andabrupt weakening in monsoon strength during the middle Holo-cene (~4.5 to 5.0 cal. ka). This � nding contradicts previousnotions of either a gradual weakening in monsoon strength dueto slow changes in orbital forcing (e.g., Overpeck et al., 1996) ora series of abrupt events occurring in an unorganized manneracross space and time during the middle to late Holocene (Wangand Fan, 1987; Shi et al., 1993). This disagreement might resultfrom the fact that we examine many records from the East Asianmonsoon region as well as the Southwest monsoon region andthat we also incorporate many new and well-dated records intoour compilation. The abrupt weakening of monsoon strength atthis time might be tied to either a cooling in the North Atlantic oran increase in the frequency and intensity of warm ENSO events.

We also observe a widespread strengthening in the Asian mon-soon at the start of the Holocene, as suggested in previousresearch (e.g., Wang and Fan, 1987; Overpeck et al., 1996). Thisabrupt change could be linked to the abrupt warming that occurredat the same time in the North Atlantic. Signi� cant changes in thestrength of the Asian monsoon also appear to occur at the startof the Younger Dryas interval, but data are insuf� cient to drawany conclusions about the direction of this climatic change. We� nd no strong evidence for an abrupt change in the Asian mon-soon at ~8.2 cal. ka. More higher-resolutionpalaeodata are needed

to determine if this short-lived event occurred in the Asian mon-soon region.

One motivation for this research was to identify possible causesand patterns of future abrupt climatic change in this region. Ourresults indicate: (1) teleconnections with other parts of theclimatic system, in particular the North Atlantic and the tropicalPaci� c, are likely to be important in determining the timing anddirection of climatic changes; (2) gradual changes in forcing (e.g.,insolation) may trigger abrupt shifts in monsoon strength, eitherdirectly or through these teleconnections;and (3) changes in mon-soon precipitationhave had both homogeneous and heterogeneousspatial patterns during the Holocene. The size of the climaticchange might determine which of these spatial patterns occurs.Most importantly, our results reveal the potential for future majorabrupt changes in the Asian monsoon, some of which can takeplace on time scales of 100 years or less.

Acknowledgements

This project was supported by a National Science FoundationGraduate Student Fellowship and a NASA Space Grant Fellow-ship (to Morrill) and a National Science Foundation Earth SystemHistory grant (to Overpeck and Cole). Discussions with J. Quade,D. Dettman and J. Shuttleworth improved this research. We thankF. Gasse and an anonymous reviewer for their helpful reviews.

References

Alley, R.B. 2000: The Younger Dryas cold interval as viewed from centralGreenland. Quaternary Science Reviews 19, 213–26.Alley, R.B., Mayewski, P.A., Sowers, T., Stuiver, M., Taylor, K.C. andClark, P.U. 1997: Holocene climatic instability: a prominent, widespreadevent 8200 yr ago. Geology 25, 483–86.Ammann, B., Birks, H.J.B., Brooks, S.J., Eicher, U., von Grafenstein,U., Hofmann, W., Lemdahl, G., Schwander, J., Tobolski, K. and Wick,L. 2000: Quanti� cation of biotic responses to rapid climatic changesaround the Younger Dryas – a synthesis. Palaeogeography, Palaeoclima-tology, Palaeoecology 159, 313–47.An, Z., Porter, S.C., Kutzbach, J.E., Wu, X., Wang, S., Liu, X., Li,X. and Zhou, W. 2000: Asynchronous Holocene optimum of the EastAsian monsoon. Quaternary Science Reviews 19, 743–62.An, Z., Porter, S.C., Zhou, W., Lu, Y., Donahue, D.J., Head, M.J.,Wu, X., Ren, J. and Zheng, H. 1993: Episode of strengthened summermonsoon climate of Younger Dryas age on the loess plateau of CentralChina. Quaternary Research 39, 45–54.Baker, P.A., Seltzer, G.O., Fritz, S.C., Dunbar, R.B., Grove, M.J.,Tapia, P.M., Cross, S.L., Rowe, H.D. and Broda, J.P. 2001: The historyof South American tropical precipitation for the past 25,000 years. Science291, 640–43.Bard, E., Raisbeck, G., Yiou, F. and Jouzel, J. 2000: Solar irradianceduring the last 1200 years based on cosmogenic nuclides. Tellus 52B,985–92.Bennett, K.D., Haberle, S.G. and Lumley, S.H. 2000: The last glacial-Holocene transition in southern Chile. Science 290, 325–28.Betancourt, J.L., Latorre, C., Rech, J.A., Quade, J. and Rylander,K.A. 2000: A 22,000-year record of monsoonal precipitation from North-ern Chile’s Atacama Desert. Science 289, 1542–46.Black, D.E., Peterson, L.C., Overpeck, J.T., Kaplan, A., Evans, M.N.and Kashgarian, M. 1999: Eight centuries of North Atlantic Ocean atmos-phere variability. Science 286, 1709–13.Bond, G., Showers, W., Cheseby, M., Lotti, R., Almasi, P., deMenocal,P., Priore, P., Cullen, H., Hajdas, I. and Bonani, G. 1997: A pervasivemillenial-scale cycle in North Atlantic Holocene and glacial climates.Science 278, 1257–66.Boyle, E.A. and Keigwin, L. 1987: North Atlantic thermohaline circu-lation during the past 20,000 years linked to high-latitude surface tempera-ture. Nature 330, 35–40.Bradley, R.S. and Jones, P.D. 1993: ‘Little Ice Age’ summer temperature

474 The Holocene 13 (2003)

variations: their nature and relevance to recent global warming trends. TheHolocene 8, 477–83.Burns, S.J., Matter, A., Frank, N. and Mangini, A. 1998: Speleothem-based paleoclimate record from northern Oman. Geology 26, 499–502.Bush, A.B.G. 1999: Assessing the impact of Mid-Holocene insolation onthe atmosphere-ocean system. Geophysical Research Letters 26, 99–102.Cao, M.S. 1998: Detection of abrupt changes in glacier mass balance inthe Tien Shan Mountains. Journal of Glaciology 44, 352–58.Caratini, C., Bentaleb, I., Fontugne, M., Morzadec-Kerfourn, M.T.,Pascal, J.P. and Tissot, C. 1994: A less humid climate since ca. 3500yr B.P. from marine cores off Karwar, western India. Palaeogeography,Palaeoclimatology, Palaeoecology 109, 371–84.Chen, F.-H., Shi, Q. and Wang, J-M. 1999: Environmental changesdocumented by sedimentation of Lake Yiema in arid China since the lateglaciation. Journal of Paleolimnology 22, 159–69.Claussen, M., Kubatzki, C., Brovkin, V. and Ganopolski, A. 1999:Simulation of an abrupt change in Saharan vegetation in the mid-Holo-cene. Geophysical Research Letters 26, 2037–40.Clement, A.C., Seager, R. and Cane, M.A. 2000: Suppression of El Ninoduring the mid-Holocene by changes in the Earth’s orbit. Paleoceanogra-phy 15, 731–37.Cole, J. 2001: A slow dance for El Nino. Science 291, 1496–97.Crowley, T.J. and Lowery, T.S. 2000: How warm was the Medievalwarm period? Ambio 29, 51–54.Cullen, H.M., deMenocal, P.B., Hemming, S., Hemming, G., Brown,F.H., Guilderson, T. and Sirocko, F. 2000: Climate change and the col-lapse of the Akkadian empire: evidence from the deep sea. Geology 28,379–82.deMenocal, P., Ortiz, J., Guilderson, T., Adkins, J., Sarnthein, M.,Baker, L. and Yarusinsky, M. 2000: Abrupt onset and termination ofthe African humid period: rapid climate responses to gradual insolationforcing. Quaternary Science Reviews 19, 347–61.Denniston, R.F., Gonzalez, L.A., Asmerom, Y., Sharma, R.H. andReagan, M.K. 2000: Speleothem evidence for changes in Indian summermonsoon precipitation over the last ~2300 years. Quaternary Research53, 196–202.Enzel, Y., Ely, L.L., Mishra, S., Ramesh, R., Amit, R., Lazar, B.,Rajaguru, S.N., Baker, V.R. and Sandler, A. 1999: High-resolution Hol-ocene environmental changes in the Thar Desert, Northwestern India.Science 284, 125–28.Fairbanks, R.G. 1989: A 17,000 year glacio-eustatic sea level record:in� uence of glacial melting rates on the Younger Dryas event and deep-ocean circulation. Nature 342, 637–42.Fang, J.-Q. 1993: Lake evolution during the last 3000 years in Chinaand its implications for environmental change. Quaternary Research 39,175–85.Finkel, R.C. and Nishiizumi, K. 1997: Beryllium 10 concentrations inthe Greenland Ice Sheet Project 2 ice core from 3–40 ka. Journal of Geo-physical Research 102, 26699–706.Fraedrich, K., Jiang, J., Gerstengarbe, F-W. and Werner, P.C. 1997:Multiscale detection of abrupt climate changes: application to River Nile� ood levels. International Journal of Climatology 17, 1301–15.Gagan, M.K., Ayliffe, L.K., Hopley, D., Cali, J.A., Mortimer, G.E.,Chappell, J., McCulloch, M.T. and Head, M. 1998: Temperature andsurface-ocean water balance of the mid-Holocene tropical westernPaci� c. Science 279, 1014–18.Gasse, F. 2000: Hydrological changes in the African tropics since theLast Glacial Maximum. Quaternary Science Reviews 19, 189–211.Gasse, F. and Van Campo, E. 1994: Abrupt post-glacial climate eventsin West Asia and North Africa monsoon domains. Earth and PlanetaryScience Letters 126, 435–56.Gong, G. and Hameed, S. 1991: The variation of moisture conditions inChina during the last 2000 years. International Journal of Climatology11, 271–83.Grove, J.M. 1988: The Little Ice Age. New York: Methuen.Hahn, D.G. and Shukla, J. 1976: An apparent relationship between Eura-sian snow cover and Indian monsoon rainfall. Journal of the AtmosphericSciences 33, 2461–62.Hall, A. and Stouffer, R.J. 2001: An abrupt climate event in a coupledocean-atmosphere simulation without external forcing. Nature 409,171–74.Hellstrom, J., McCulloch, M. and Stone, J. 1998: A detailed 31,000-year

record of climate and vegetation change, from the isotope geochemistry oftwo New Zealand speleothems. Quaternary Research 50, 167–78.Hewitt, C.D. and Mitchell, J.F.B. 1998: A fully coupled GCM simulationof the climate of the mid-Holocene. Geophysical Research Letters 25,361–64.Hu, Z.Z. and Nitta, T. 1996: Wavelet analysis of summer rainfall overNorth China and India and SOI using 1891–1992 data. Journal of theMeteorological Society of Japan 74, 833–44.Huang, C.-Y., Liew, P-M., Zhao, M., Chang, T-C., Kuo, C-M., Chen,M-T., Wang, C-H. and Zheng, L-F. 1997: Deep sea and lake records ofthe southeast Asian paleomonsoons for the last 25 thousand years. Earthand Planetary Science Letters 146, 59–72.Hughen, K.A., Southon, J.R., Lehman, S.J. and Overpeck, J.T. 2000:Synchronous radiocarbon and climate shifts during the last deglaciation.Science 290, 1951–54.Hughes, M.K. and Diaz, H. F. 1994: Was there a ‘Medieval Warm Per-iod’, and if so, where and when? Climatic Change 26, 109–42.Jarvis, D.I. 1993: Pollen evidence of changing Holocene monsoon climatein Sichuan Province, China. Quaternary Research 39, 325–37.Jennings, A.E., Knudsen, K.L., Hald, M., Hansen, C.V. and Andrews,J.T. 2002: A mid-Holocene shift in Arctic sea ice variability on the EastGreenland shelf. The Holocene 12, 49–58.Karl, T.R. and Riebsame, W.E. 1984: The identi� cation of 10- to 20-year temperature and precipitation � uctuations in the contiguous UnitedStates. Journal of Climate and Applied Meteorology 1984, 950–66.Keigwin, L.D. 1996: The little ice age and medieval warm period in theSargasso Sea. Science 274, 1504–08.Kelts, K., Chen, K.Z., Lister, G., Yu, J., Gao, Z., Niessen, N. andBonani, G. 1989: Geological � ngerprints of climate history: a cooperativestudy of Qinghai Lake, China. Eclogae Geologicae Helvetiae 82, 167–82.Kudrass, H.R., Hofmann, A., Doose, H., Emeis, K. and Erlenkeuser,H. 2001: Modulation and ampli� cation of climatic changes in the NorthernHemisphere by the Indian summer monsoon during the past 80 k.y.Geology 29, 63–66.Kutzbach, J.E. and Liu, Z. 1997: Response of the African Monsoon toorbital forcing and ocean feedbacks in the Middle Holocene. Science 278,440–43.Laird, K.R., Fritz, S.C., Maasch, K.A. and Cumming, B.F. 1996:Greater drought intensity and frequency before ad 1200 in the NorthernGreat Plains, USA. Nature 384, 552–54.Lehman, S.J. and Keigwin, L.D. 1992: Sudden changes in North Atlanticcirculation during the last deglaciation. Nature 356, 757–62.Li, B., Yuan, D., Lauritzen, S-E., Qin, J. and Lin, Y. 1998: The YoungerDryas event and Holocene climate � uctuations recorded in a stalagmitefrom the Panlong Cave of Guilin. Acta Geologica Sinica 72, 455–65.Lister, G.S., Kelts, K., Chen, K., Yu, J. and Niessen, F. 1991: LakeQinghai, China: Closed-basin lake levels and the oxygen isotope recordfor ostracoda since the latest Pleistocene. Palaeogeography, Palaeoclima-tology, Palaeoecology 84, 141–62.Liu, Z., Kutzbach, J. and Wu, L. 2000: Modeling climate shift of El Ninovariability in the Holocene. Geophysical Research Letters 27, 2265–68.Lou, J.-Y. and Chen, C-T.A. 1997: Paleoclimatological and paleoenvi-ronmental records since 4000 a BP in sediments of alpine lakes in Taiwan.Science in China 40, 424–31.Lou, J.-Y., Chen, C-T.A. and Wann, J-K. 1997: Paleoclimatologicalrecords of the Great Ghost Lake in Taiwan. Science in China Series D40, 284–92.Mann, M.E., Bradley, R.S. and Hughes, M.K. 1999: Northern hemi-sphere temperatures during the past millennium: Inferences, uncertainties,and limitations. Geophysical Research Letters 26, 759–62.Marchitto, T.M., Curry, W.B. and Oppo, D.W. 1998: Millennial-scalechanges in North Atlantic circulation since the last glaciation. Nature 393,557–81.Markgraf, V. 1993: Paleoenvironments and paleoclimates in Tierra delFuego and southernmost Patagonia, South America. Palaeogeography,Palaeoclimatology, Palaeoecology 102, 53–68.McGlone, M.S., Kershaw, A.P. and Markgraf, V. 1992: ElNino/Southern Oscillation climatic variability in Australasian and SouthAmerican paleoenvironmental records. In Diaz, H.F. and Markgraf, V.,editors, El Nino: Historical and paleoclimatic aspects of the SouthernOscillation, Cambridge: Cambridge University Press, 435–62.Moreno, P.I., Jacobson, G.L., Lowell, T.V. and Denton, G.H. 2001:

Carrie Morrill et al.: Abrupt changes in the Asian summer monsoon 475

Interhemispheric climate links revealed by a late-glacial cooling episodein southern Chile. Science 409, 804–808.Overpeck, J., Anderson, D., Trumbore, S. and Prell, W. 1996: Thesouthwest Indian monsoon over the last 18,000 years. Climate Dynamics12, 213–25.Peteet, D. 1995: Global Younger Dryas? Quaternary International 28,93–104.Prasad, S., Kusumgar, S. and Gupta, S.K. 1997: A mid to late Holocenerecord of palaeoclimatic changes from Nal Sarovar: a palaeodesert marginlake in western India. Journal of Quaternary Science 12, 153–59.Ren, G. 1998: Pollen evidence for increased summer rainfall in the Medie-val warm period at Maili, Northeast China. Geophysical Research Letters25, 1931–34.Rind, D. and Overpeck, J. 1993: Hypothesized causes of decade-to-cen-tury scale climate variability: climate model results. Quaternary ScienceReviews 12, 357–74.Rodbell, D.T., Seltzer, G.O., Anderson, D.M., Abbott, M.B., En� eld,D.B. and Newman, J.H. 1999: An ~15,000-year record of El Nino-drivenalluviation in southwestern Ecuador. Science 283, 516–20.Sarkar, A., Ramesh, R., Somayajulu, B.L.K., Agnihotri, R., Jull,A.J.T. and Burr, G.S. 2000: High resolution Holocene monsoon recordfrom the eastern Arabian Sea. Earth and Planetary Science Letters 177,209–18.Schulz, H., von Rad, U. and Erlenkeuser, H. 1998: Correlation betweenArabian Sea and Greenland climate oscillations of the past 110,000 years.Nature 393, 54–57.Severinghaus, J.P., Sowers, T., Brook, E.J., Alley, R.B. and Bender,M.L. 1998: Timing of abrupt climate change at the end of the YoungerDryas interval from thermally fractionated gases in polar ice. Nature 391,141–46.Shi, Y., Kong, Z., Wang, S., Tang, L., Wang, F., Yao, T., Zhao, X.,Zhang, P. and Shi, S. 1993: Mid-Holocene climates and environments inChina. Global and Planetary Change 7, 219–34.Shukla, J. and Paolino, D.A. 1983: The Southern Oscillation and long-range forecasting of the summer monsoon rainfall over India. MonthlyWeather Review 111, 1830–37.Shulmeister, J. and Lees, B.G. 1995: Pollen evidence from tropicalAustralia for the onset of an ENSO-dominated climate at c. 4000 BP. TheHolocene 5, 10–18.Singer, C., Shulmeister, J. and McLea, B. 1998: Evidence against asigni� cant Younger Dryas cooling event in New Zealand. Science 281,812–14.Singh, G., Joshi, R.D., Chopra, S.K. and Singh, A.B. 1974: Late Quat-ernary history of vegetation and climate of the Rajasthan Desert, India.Philosophical Transactions of the Royal Society of London 267, 467–501.Sirocko, F., Garbe-Shonberg, D., McIntyre, A. and Mol� no, B. 1996:Teleconnections between the subtropical monsoons and high-latitudeclimates during the last deglaciation. Science 272, 526–29.Sirocko, F., Sarnthein, M., Erlenkeuser, H., Lange, H., Arnold, M.and Duplessy, J.C. 1993: Century-scale events in monsoonal climate overthe past 24,000 years. Nature 364, 322–24.Sowers, T. and Bender, M. 1995: Climate records covering the lastdeglaciation. Science 269, 210–14.Steig, E.J., Brook, E.J., White, J.W.C., Sucher, C.M., Bender, M.L.,Lehman, S.J., Morse, D.L., Waddington, E.D. and Clow, G.D. 1998:Synchronousclimate changes in Antarctica and the North Atlantic. Science282, 92–95.Stocker, T.F. and Marchal, O. 2000: Abrupt climate change in thecomputers: is it real? Proceedings of the National Academy of Sciencesof the United States 97, 1362–65.Stuiver, M. and Reimer, P.J. 1993: Extended 14C database and revisedCALIB 3.0 14C age calibration program. Radiocarbon 35, 215–30.Stuiver, M., Reimer, P.J., Bard, E., Beck, J.W., Burr, G.S., Hughen,K.A., Kromer, B., McCormac, G., Van der Plicht, J., Spurk, M. 1998:INTCAL98 radiocarbon age calibration, 24,000–0 cal. BP. Radiocarbon40, 1041–83.Thompson, L.G., Davis, M.E., Mosley-Thompson, E., Sowers, T.A.,Henderson, K.A., Zagorodnov, V.S., Lin, P.-N., Mikhalenko, V.N.,Campen, R.K., Bolzan, J.F., Cole-Dai, J. and Francou, B. 1998: A25,000-year tropical climate history from Bolivian ice cores. Science 282,1858–64.Thompson, L.G., Mosley-Thompson,E., Davis, M.E., Bolzan J.F., Dai,J., Yao, T., Gundestrup, N., Wu, X., Klein, L. and Xie, Z. 1989:

Holocene-Late Pleistocene climatic ice core records from Qinghai-TibetanPlateau. Science 246, 474–77.Thompson, L.G., Mosley-Thompson, E., Davis, M.E., Lin, P.N., Dai,J., Bolzan, J.F. and Yao, T. 1995: A 1000 year climate ice-core recordfrom the Guliya ice cap, China: its relationship to global climate varia-bility. Annals of Glaciology 21, 175–81.Thompson, L.G., Yao, T., Mosley-Thompson, E., Davis, M.E.,Henderson, K.A. and Lin, P.-N. 2000: A high-resolution millenial recordof the South Asian monsoon from Himalayan ice cores. Science 289,1916–19.Tudhope, A.W., Chilcott, C.P., McCulloch, M.T., Cook, E.R.,Chappell, J., Ellam, R.M., Lea, D.W., Lough, J.M. and Shimmield,G.B. 2001: Variability in the El Nino-Southern Oscillation through a gla-cial-interglacial cycle. Science 291, 1511–17.Verschuren, D., Laird, K.R. and Cumming, B.F. 2000: Rainfall anddrought in equatorial east Africa during the past 1100 years. Nature, 403,410–14.von Grafenstein, U., Erlenkeuser, H., Muller, J., Jouzel, J. andJohnsen, S. 1998: The cold event 8200 years ago documented in oxygenisotope records of precipitation in Europe and Greenland. ClimateDynamics 14, 73–81.von Rad, U., Schaaf, M., Michels, K.H., Schulz, H., Berger, W.H. andSirocko, F. 1999a: A 5000-yr record of climate change in varved sedi-ments from the oxygen minimum zone off Pakistan, Northeastern ArabianSea. Quaternary Research 51, 39–53.von Rad, U., Schulz, H., Reich, V., den Dulk, M., Berner, U. and Sir-ocko, F. 1999b: Multiple monsoon-controlled breakdown of oxygen-mini-mum conditions during the past 30,000 years documented in laminatedsediments off Pakistan. Palaeogeography,Palaeoclimatology,Palaeoecol-ogy 152, 129–61.Walker, M.J.C. 1995: Climatic changes in Europe during the lastglacial/interglacial transition. Quaternary International 28, 63–76.Walland, D.J., Power, S.B. and Hirst, A.C. 2000: Decadal climate varia-bility simulated in a coupled general circulation model. Climate Dynamics16, 201–11.Wang, F.-B. and Fan, C.Y. 1987: Climatic changes in the Qinghai-Xizang(Tibetan) region of China during the Holocene. Quaternary Research 28,50–60.Wang, L., Sarnthein, M., Erlenkeuser, H., Grootes, P.M., Grimalt,J.O., Pelejero, C. and Linck, G. 1999a: Holocene variations in Asianmonsoon moisture: a bidecadal sediment record from the South China Sea.Geophysical Research Letters 26, 2889–92.Wang, L., Sarnthein, M., Grootes, P.M. and Erlenkeuser, H. 1999b:Millennial reoccurrence of century-scale abrupt events of East Asianmonsoon: a possible heat conveyor for the global deglaciation.Paleoceanography 14, 725–31.Wang, P.K. and Zhang, D. 1992: Recent studies of the reconstruction ofEast Asian monsoon climate in the past using historical literature of China.Journal of the Meteorological Society of Japan 70, 423–45.Webster, P.J., Magana, V.O., Palmer, T.N., Shukla, J., Tomas, R.A.,Yanai, M. and Yasunari, T. 1998: Monsoons: processes, predictability,and the prospects for prediction. Journal of Geophysical Research 103,14451–510.Weiss, H., Courty, M-A., Wetterstrom, W., Guichard, F., Senior, L.,Meadow, R. and Curnow, A. 1993: The genesis and collapse of thirdmillenium North Mesopotamian civilization. Science 261, 995–1004.Winkler, M.G. and Wang, P.K. 1993: The late Quaternary vegetationand climate of China. In Wright, H.E., editor, Global climates since thelast glacial maximum, Minneapolis: University of Minnesota Press,221–61.Woodhouse, C.A. and Overpeck, J.T. 1998: 2000 years of drought varia-bility in the central United States. Bulletin of the American MeteorologicalSociety 79, 2693–714.Yan, G., Wang, F.B., Shi, G.R. and Li, S.F. 1999: Palynological andstable isotopic study of palaeoenvironmental changes on the northeasternTibetan plateau in the last 30,000 years. Palaeogeography, Palaeo-climatology, Palaeoecology 153, 147–59.Yang, S. 1996: ENSO-snow-monsoon associations and seasonal-interannual predictions. International Journal of Climatology 16, 125–34.Yang, S. and Xu, L. 1994: Linkage between Eurasian winter snow coverand regional Chinese summer rainfall. International Journal ofClimatology 14, 739–50.Yu, Z.C. and Wright, H.E. 2001: Response of interior North America to

476 The Holocene 13 (2003)

abrupt climate oscillations in the North Atlantic region during the lastdeglaciation. Earth Science Reviews 52, 333–69.Zheng, Z. and Li, Q. 2000: Vegetation, climate, and sea level in the past50,000 years, Hanjiang Delta, Southeastern China. Quaternary Research53, 330–40.Zhou, S.Z., Chen, F.H., Pan, B.T., Cao, J.X., Li, J.J. and Derbyshire,E. 1991: Environmental change during the Holocene in western China ona millennial timescale. The Holocene 1, 151–56.Zhou, W., An, Z. and Head, M.J. 1994: Stratigraphic division of Holo-cene loess in China. Radiocarbon 36, 37–45.Zhou, W., Donahue, D.J., Porter, S.C., Jull, A.J.T., Li, X., Stuiver,

M., An, Z., Matsumoto, E. and Dong, G. 1996: Variability of monsoonclimate in East Asia at the end of the last glaciation. Quaternary Research46, 219–29.Zielinski, G.A. and Mershon, G.R. 1997: Paleoenvironmental impli-cations of the insoluble microparticle record in the GISP2 (Greenland)ice core during the rapidly changing climate of the Pleistocene-Holocenetransition. Geological Society of America Bulletin 109, 547–59.Zonneveld, K.A.F., Ganssen, G., Troelstra, S., Versteegh, G.J.M. andVisscher, H. 1997: Mechanisms forcing abrupt � uctuations of the IndianOcean summer monsoon during the last deglaciation. Quaternary ScienceReviews 16, 187–201.