Quaternary Science Reviews 26 (2007) 1621–1637 Pollen evidence for the transition of the Eastern Australian climate system from the post-glacial to the present-day ENSO mode Timme H. Donders a, , Simon G. Haberle b , Geoffrey Hope b , Friederike Wagner a , Henk Visscher a a Palaeoecology, Institute of Environmental Biology, Laboratory of Palaeobotany and Palynology, Utrecht University, Budapestlaan 4, 3584 CD, Utrecht, The Netherlands b Research School of Pacific and Asian Studies, Australian National University, Canberra, ACT 0200, Australia Received 10 April 2006; received in revised form 20 November 2006; accepted 23 November 2006 Abstract A review of Holocene climate patterns in eastern Australia is presented on the basis of a series of high-resolution pollen records across a north-to-south transect. Previously published radiocarbon data are calibrated into calendar years and fitted with an age-depth model. The resulting chronologies are used to compare past environmental changes and describe patterns of climate change on a calendar-age scale. Based on the present-day Australian climate patterns and impact of the El Nin˜o-Southern Oscillation (ENSO), the palynological data are interpreted and the prevalent climate mode throughout the Holocene reconstructed. Results show that early Holocene changes are strongly divergent and asynchronous between sites, while middle to late Holocene conditions are characterized by more arid and variable conditions and greater coupling between northern and southern sites, which is in agreement with increasing influence of ENSO. r 2006 Elsevier Ltd. All rights reserved. 1. Introduction Australian vegetation and wildlife is well adapted to climate fluctuations imposed by the El Nin˜o-Sourthern Oscillation (ENSO) system (Nichols, 1992). The high adaptation capacity implies that this region is well suited for paleoclimatic studies related to the history of ENSO dynamics. Investigating Holocene ENSO variability is particularly important because in this time interval the effects of changes in background climate, caused by orbital precession, can be studied independently of large, long- term variations in global ice cover and sea level (Markgraf and Diaz, 2000). Currently, ENSO causes significant interannual climate variability in Australia (Dodson, 2001). The early Holo- cene Australian environment was characterized by gener- ally lower variability compared to the present situation. It is therefore unlikely, that the ENSO system has continu- ously operated in the present-day mode throughout the entire Holocene (McGlone et al., 1992). More arid and variable conditions seen in Australian tropical monsoon- dominated areas after 4 14 Cka BP (3.7 cal ka BP) have been attributed to the establishment of modern-day ENSO dynamics (Shulmeister and Lees, 1995). Strongly ENSO- teleconnected regions in South-America (McGlone et al., 1992) and the Southeastern United States (Donders et al., 2005) have shown similar changes in past ENSO state. Holocene ENSO variability has likely caused synchro- nous changes across large parts of Australia, since it presently impacts a wide area. Numerous high-resolution palynological records of Holocene vegetation cover in Australia are available, which have the potential to recognize temporal and spatial patterns of past climate dynamics. In order to accurately resolve the development and role of ENSO, these records must be compared on a calendar age-scale. However, available reviews of con- tinental (Harrison, 1993; Hope et al., 2004) or regional (Kershaw, 1994; Dodson and Ono, 1997; Dodson, 1998, 2001) vegetation and climate patterns in Australian do not ARTICLE IN PRESS 0277-3791/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.quascirev.2006.11.018 Corresponding author. Tel.: +31 30 253 2631; fax: +31 30 253 5096. E-mail address: [email protected] (T.H. Donders).
17
Embed
Pollen evidence for the transition of the Eastern ...palaeoworks.anu.edu.au/pubs/Dondersetal2007.pdf · Pollen evidence for the transition of the Eastern Australian climate system
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
ARTICLE IN PRESS
0277-3791/$ - se
doi:10.1016/j.qu
�CorrespondE-mail addr
Quaternary Science Reviews 26 (2007) 1621–1637
Pollen evidence for the transition of the Eastern Australian climatesystem from the post-glacial to the present-day ENSO mode
Timme H. Dondersa,�, Simon G. Haberleb, Geoffrey Hopeb, Friederike Wagnera,Henk Visschera
aPalaeoecology, Institute of Environmental Biology, Laboratory of Palaeobotany and Palynology, Utrecht University,
Budapestlaan 4, 3584 CD, Utrecht, The NetherlandsbResearch School of Pacific and Asian Studies, Australian National University, Canberra, ACT 0200, Australia
Received 10 April 2006; received in revised form 20 November 2006; accepted 23 November 2006
Abstract
A review of Holocene climate patterns in eastern Australia is presented on the basis of a series of high-resolution pollen records across
a north-to-south transect. Previously published radiocarbon data are calibrated into calendar years and fitted with an age-depth model.
The resulting chronologies are used to compare past environmental changes and describe patterns of climate change on a calendar-age
scale. Based on the present-day Australian climate patterns and impact of the El Nino-Southern Oscillation (ENSO), the palynological
data are interpreted and the prevalent climate mode throughout the Holocene reconstructed. Results show that early Holocene changes
are strongly divergent and asynchronous between sites, while middle to late Holocene conditions are characterized by more arid and
variable conditions and greater coupling between northern and southern sites, which is in agreement with increasing influence of ENSO.
r 2006 Elsevier Ltd. All rights reserved.
1. Introduction
Australian vegetation and wildlife is well adapted toclimate fluctuations imposed by the El Nino-SourthernOscillation (ENSO) system (Nichols, 1992). The highadaptation capacity implies that this region is well suitedfor paleoclimatic studies related to the history of ENSOdynamics. Investigating Holocene ENSO variability isparticularly important because in this time interval theeffects of changes in background climate, caused by orbitalprecession, can be studied independently of large, long-term variations in global ice cover and sea level (Markgrafand Diaz, 2000).
Currently, ENSO causes significant interannual climatevariability in Australia (Dodson, 2001). The early Holo-cene Australian environment was characterized by gener-ally lower variability compared to the present situation. Itis therefore unlikely, that the ENSO system has continu-
e front matter r 2006 Elsevier Ltd. All rights reserved.
ously operated in the present-day mode throughout theentire Holocene (McGlone et al., 1992). More arid andvariable conditions seen in Australian tropical monsoon-dominated areas after 4 14C ka BP (�3.7 cal ka BP) havebeen attributed to the establishment of modern-day ENSOdynamics (Shulmeister and Lees, 1995). Strongly ENSO-teleconnected regions in South-America (McGlone et al.,1992) and the Southeastern United States (Donders et al.,2005) have shown similar changes in past ENSO state.Holocene ENSO variability has likely caused synchro-
nous changes across large parts of Australia, since itpresently impacts a wide area. Numerous high-resolutionpalynological records of Holocene vegetation cover inAustralia are available, which have the potential torecognize temporal and spatial patterns of past climatedynamics. In order to accurately resolve the developmentand role of ENSO, these records must be compared on acalendar age-scale. However, available reviews of con-tinental (Harrison, 1993; Hope et al., 2004) or regional(Kershaw, 1994; Dodson and Ono, 1997; Dodson, 1998,2001) vegetation and climate patterns in Australian do not
ARTICLE IN PRESST.H. Donders et al. / Quaternary Science Reviews 26 (2007) 1621–16371622
focus on the high-resolution detection of Holocenechanges. Most published pollen records have been radio-carbon-dated, but especially pre-1993 records were notcalibrated into calendar ages.
The present review provides a regional synthesis ofenvironmental change documented in new and earlierpublished, high-resolution palynological records along anorth-to-south transect in eastern Australia. Conversion tocalendar ages of previously published radiocarbon dataenables a more accurate comparison of the available records,allowing better assessment of both temporal and spatialpatterns of environmental change. Re-evaluation of palynolo-gical records may reveal information on the differentiation ofthe climate system from the early Holocene mode to present-day ENSO forcing. Further, accurate chronologies allow totest whether the known past changes in ENSO dynamics haveresulted in synchronous vegetation changes that are inaccordance with the modern impact pattern of ENSO.
2. Regional setting
2.1. Australian climate
The major patterns of the Australian climate aredetermined by a high-pressure belt positioned below thesubtropical jet-stream across southern Australia (Harrison,1993). Anticyclonic activity moves eastward across thecontinent, causing arid conditions. During winter the high-pressure belt shifts northward to �29–321S, allowingwesterly winds to bring winter rainfall to the southernpart of the continent. The pressure belt moves southwardto �37–381S during summer, allowing tropical lowpressure systems and the developing northern monsooninto the north and northeast. The southward movement ofthe pressure belt displaces moist westerly winds fromsouthern Australia. Only western Tasmania receives pre-cipitation all year round (Harrison, 1993; Dodson, 1998).
Winds developing on the equatorial side of anticyclonicspirals form the southeasterly trade winds (tropical easterlies),which are the dominant source of precipitation in thenortheast. Occasional northwesterly monsoonal flows andassociated tropical cyclones cause intensive but infrequentrainfall events during the austral summer when the inter-tropical convergence zone (ITCZ) is at its most southerlyextent (Godfred-Spenning and Reason, 2002). The tropicalcyclones are an important rainfall source for the arid interiorand for general summer rainfall across the continent. Due tothe monsoonal activity and migration of the high-pressurebelt, southern areas experience a winter precipitation peakwhile northern areas mainly receive summer precipitation.The subtropical areas are intermediate and generally muchless seasonal (Fig. 1A, after Magee et al., 2004).
2.2. ENSO impact
At present, significant precipitation variability is gener-ated by ENSO (Van Oldenborgh and Burgers, 2005),
especially across the eastern side of the Australiancontinent (Dodson, 1998, 2001). During El Nino episodes,an equator-ward movement of the ITCZ and a north-eastward migration of the South Pacific convergence zoneresult in a significant decrease of summer precipitation ineastern Australia. In the northeastern tropics the deficitamounts to 150–300mm below seasonal average (Dai andWigley, 2000). Strong Walker circulation or La Ninaconditions intensify moist monsoonal flow over the dryeastern interior, and to the north and east due to enhancedtrade winds. Variations in ENSO dynamics affect the tradewind system and therefore influence the eastern Australianclimate synchronously from north-to-south. Figs. 1B–Eshow the seasonal correlation between ENSO, expressed asthe NINO3.4 index, and precipitation.However, independent changes in the mean subtropical
high-pressure belt position also significantly influenceAustralian climate (Pittock, 1978; Harrison, 1993). Inrecords of past climate, variation caused by changingENSO dynamics can be distinguished from high-pressurebelt migration since the latter causes non-synchronouschanges from north-to-south (Dodson, 1998).
2.3. Last glacial to early Holocene atmospheric circulation
Lake-level reconstructions for the last glacial maximumshow that the high-pressure belt expanded, and displacedmoisture-laden westerlies southwards during glacial peri-ods, depriving coastal areas, including Tasmania, of moistwinter conditions (Harrison, 1993). Glacial dune patternsin the arid interior indicate stronger circulation andreduced monsoonal flow (Hesse et al., 2004). Climaticconditions were more homogeneous, with less contrast inwater balance between coastal and interior environments,and generally drier, while interior temperatures werestrongly reduced (Hope et al., 2004). Reduced evaporativewater loss due to cooler temperatures caused some aridregions to experience less severe drying (Harrison, 1993).Glacial/interglacial changes in the Walker Circulation,which would affect moisture transport by easterly tradewinds, were obscured due to the generally cooler and drierconditions.
2.4. Insolation during the Holocene
The dominant factor controlling Holocene climatevariation is the orbital or Milankovitch forcing ofinsolation. Estimates of solar irradiance and seasonalityfor the past 15 ka BP in summer and winter are given inFig. 2A for the Equator, 301N and 301S, based on theLaskar-90 solution (Laskar, 1990). Fig. 2B shows thewinter and summer gradient strength between 601S/301Sand the equator. During the late Holocene the SouthernHemisphere experienced increased seasonality and summerwarmth, but a reduced gradient between tropical andtemperate areas, which implies a reduction of westerlyairflow (Dodson, 1998). Northern winter insolation is an
ARTICLE IN PRESS
Correlation NINO3.4 index with CRU precipitation(3-month averaged)
A
B
D E
C
a) Monsoonal / N-tropical
1 Groote Eylandt, N.T.
(Shulmeister, 1992;
Shulmeister & Lees, 1995)
2 Lake Euramoo, Atherton
Tablelands, Qld. (Kershaw,
1970; Haberle, 2005)
3 Quincan Crater, Atherton
Tablelands, Qld. (Kershaw,
1971)
4 Whitehaven Swamp, Qld.
(Genever et al., 2003)
b) Subtropical
5 Lake Allom, Fraser Island,
Qld. (Donders et al., 2006)
6 Barrington Tops, N.S.W.
(Dodson et al., 1986)
c) Temperate - East coast
7 Bega Swamp, N.S.W.
(Green et al., 1988; Hope
et al., 2004)
8 Club Lake (Kosciuszko
National Park), N.S.W.
(Martin, 1986)
9 Sperm Whale Head, Vic.
(Hooley et al., 1980)
d) Arid interior
10 Lake Frome, S.A.
(Singh & Luly, 1991)
11 Middens, Flinders
Ranges, S.A. (McCarthy
& Head, 2001)
12 Lake Tyrrell, Vic. (Luly,
1993; 1995)
e) Mediterranean climate -
southern coast
13 Fleurieu Peninsula, S.A.
(Bickford & Gell, 2005)
14 Lake Leake, S.A.
(Dodson, 1974)
15 Tower Hill, Vic.
(D'Costa et al., 1989)
16 Lake Wangoom, Vic.
(Edney et al., 1990)
f) Southern temperate -
Tasmania
17 Lake Johnston, Tas.
(Anker et al., 2001)
18 Cynthia Bay, Tas.
(Hopf et al., 2000)
19 Mt. Field (Eagle Tarn),
Tas. (Macphali, 1979)
Fig. 1. (A) Location of the study sites (as discussed in Section 5), and seasonality of annual rainfall (after Magee et al., 2004) across Australia. (B–E)
ENSO impact on Australian precipitation, expressed as a seasonally averaged correlation between gridded precipitation and the NINO3.4 index. Non-
significant correlations are grey. Maps were made with KNMI Climate Explorer (Van Oldenborgh and Burgers, 2005).
important control for the ITCZ position, which determinesthe Australian monsoon intensity (Magee et al., 2004).Ocean–atmosphere modeling studies suggest that theseasonality changes caused by orbital precession are amajor control over long-term ENSO variability throughasymmetric heating of the Equatorial Pacific (Turney et al.,2004), whereby increased Northern Hemisphere seasonalcontrast terminates ENSO warm events through easterlywind forcing (Clement et al., 2000).
3. Materials and methods
3.1. Data selection
Localities were selected on the basis of sample resolution(o500 a/sample) and independent dating control (Table 1).In areas with abundant pollen records, such as Victoria andTasmania, well-dated continuous records were preferred.Records are categorized and discussed per climate region as(a) monsoonal/northern tropical, (b) subtropical, (c) tempe-rate/eastern coast, (d) inland arid, (e) Mediterranean-type
and (f) southern temperate types (Fig. 1). Records fromthe arid interior are important since they are particularlyresponsive to precipitation changes, and do notsuffer from orographic effects that cause local rainfallanomalies (Shulmeister and Lees, 1995). Coastal recordsare more abundant and located in areas with strongENSO control, but are often not independent of sea levelchanges.Three high-resolution records, Lake Euramoo (Haberle,
2005), Lake Allom (Donders et al., 2006) and Bega Swamp(Green et al., 1988; Hope et al., 2004), are used as a basisfor the data comparison, and are discussed in greater detailin Section 4. All three records are located in a highlyENSO-sensitive region (Fig. 1) and are well-dated, allow-ing detailed spatial and temporal comparison. The maintrends of these records are represented by principalcomponent analysis (PCA) output, based on major drylandtaxa. The three key-sites are subsequently compared withnewly calibrated records from other localities acrosseastern Australia to assess the patterning or consistencyof climatic changes throughout the Holocene.
ARTIC
LEIN
PRES
S
Table 1
Radiocarbon data of study sites
# Site & author Location Lab code Depth (cm) Range (cm) 14C age (a BP) s (a) Method Cal. age (a BP) s (yr) Calib. Modeled (a BP) s (a)
Pollen sites used in the present study and radiocarbon data with calendar ages of previously uncalibrated records. Radiocarbon ages were calibrated with OxCal 3.10 (Bronk Ramsey, 1995, 2001), using
the SHCAL04 Southern Hemisphere calibration dataset (McCormack et al., 2004). For ages older than 10 14Cka BP the INTCAL04 (Reimer et al., 2004) dataset was used.
Radiocarbon age ranges of the Bega Swamp record are mostly duplicate samples. In the upper 150 cm they show little scatter and are assumed to be in chronological order. A Bayesian statistical
function within OxCal combines duplicate samples and uses the prior assumption about sample order to reduce age ranges of overlapping radiocarbon data (Bronk Ramsey, 2000). Subsequently, the
reduced 1s intervals were used for construction of the age-depth model.
n.a.: not applicable, see main text.
n.s.: not specified in original publication.
N.T.: Northern Territory.
Qld.: Queensland.
N.S.W.: New South Wales.
S.A.: South Australia.
CRS: constant rate of supply model.
Vic.: Victoria.
intcal98: Stuiver et al. (1998).
intcal04: Reimer et al. (2004).
shcal04: McCormac et al. (2004).
conv..: conventional radiocarbon date.
AMS: Accelerator mass spectrometry radiocarbon date.
CIC: constant initial concentration model.
T.H
.D
on
ders
eta
l./
Qu
atern
ary
Scien
ceR
eviews
26
(2
00
7)
16
21
–1
63
71627
ARTICLE IN PRESST.H. Donders et al. / Quaternary Science Reviews 26 (2007) 1621–16371628
3.2. Chronology
Available radiocarbon data are calibrated into calendarages for all sites (Table 1). For each previously un-calibrated record an age–depth curve is plotted (Fig. 3).Depending on the fit to the data, a linear or higher-ordermodel is used to describe the age–depth relationship.Published age–depth models based on calibrated ages(Anker et al., 2001; Bickford and Gell, 2005; Haberle,2005; Donders et al., 2006) are applied unchanged (Fig. 3).Subsequently, the main conclusions and times of majorvegetation change as described in the original publicationswere adjusted according to the new age–depth models.Although the calibrated records allow a much bettercomparison of temporal patterns than records on a 14C -age base, significant differences in resolution and agemodel quality still exist.
4. High-resolution records of Holocene vegetation
The three principal detailed Holocene records fromregions with high ENSO-impact are summarized individu-ally. Figs. 4A–F shows the PCA output for the first andsecond axes of these records, indicating the main trends inthe dry land taxa.
4.1. Northern tropics—Lake Euramoo (Atherton
Tablelands)
The northern Queensland Atherton Tablelands havebeen studied extensively and yielded multi-proxy datasetsthat raise a picture of a highly dynamic landscape, sensitiveto both climate change and human activity on timescalesranging from millennia to decades (Haberle, 2005). Thearea has a climate dominated by easterly tradewinds and asecondary monsoonal influence (Hope et al., 2004). TheTablelands, which uplifted during the Tertiary, support themost significant tropical rainforest area of Australia.Precipitation is �1500mm/a with a distinctly dry australwinter season. Mean daily maximum and minimumtemperatures are �25.9 and 14.4 1C and rare frosts occurduring the austral winter months at times of weak tradewinds and low cloud cover.
The most comprehensively studied site on the AthertonTablelands is Lynch’s Crater (e.g. Kershaw, 1976, 1986;Turney et al., 2004.). However, the Holocene section inLynch’s Crater is highly condensed, and does not allowdetailed analysis (Kershaw, 1983). An alternative site, LakeEuramoo (2, Fig. 1A) is located in a double eruption crater,with a small catchment and no in- or outflows.
The first detailed pollen record from this lake reported arainforest maximum at �7.0 14C ka BP (�6.2 cal ka BP)(Kershaw, 1970), interpreted as the result of increasedeffective precipitation, with reduced seasonality relative tothe present day (Kershaw and Nix, 1988). A recent, moredetailed record, from Lake Euramoo (Fig. 1A) confirmedand extended the earlier findings (Haberle, 2005).
It documents vegetation change and fire history from23 cal ka BP to the present, with a high resolution of�100 a/sample during the last 9 ka BP. The Holocenechronology of the record is based on 210Pb and 11 AMS14C measurements (Fig. 3B).The main change at Euramoo occurs at 8.7 cal ka BP
when dominance of sclerophyllous vegetation changed torainforest, which reached maximum abundance anddiversity at 7.3 cal ka BP. A concurrent charcoal influxminimum occurred between 7.3 and 6.3 cal ka BP. After5.0 cal ka BP, rainforest changed into a more open anddrought/disturbance-adapted vegetation type, character-ized by increased Agathis, Elaeocarpus and Mallotus/Macaranga. Charcoal influx increased significantly after2.7 cal ka BP (Fig. 4G). Effects of human occupation areevident from 120 cal a BP, when forest clearing and burningcaused rapid rainforest destruction.The PCA results clearly show the attainment of modern
rainforest (Figs. 4A and B). Axis 1 contrasts Casuarina/Poaceae with rainforest taxa, which is precipitation related.Axis 2 contrasts sub-montane/secondary rainforest species,related to increased disturbance or burning, with lowermontane closed rainforest taxa such as Cunoniaceae andUrticaceae/Moraceae (Haberle, 2005).The trend toward increased disturbance or burning is
very gradual and values become stable after 4.8 cal ka BP,although the main increase in fire-frequency is after2.7 cal ka BP (Fig. 4G). The major Holocene shifts in LakeEuramoo are rather subtle changes between differentrainforest types, and are therefore not very pronouncedin the PCA diagram. An additional record from QuincanCrater (3, Fig. 1A, Kershaw, 1971) provides independentcorroboration of the changes recorded at Lake Euramoo,although no detailed comparison is made here.
4.2. Subtropics—Lake Allom (Fraser Island)
A recent detailed pollen record from Lake Allom (5, Fig.1A), Fraser Island reveals past vegetation changes in thetropical to temperate transition zone in Australia (Donderset al., 2006). Fraser Island is a large Pleistocene dunesystem off the south-eastern Queensland coast character-ized by numerous perched lakes, which act as very sensitiverainfall gauges (Longmore, 1997).South-eastern Queensland has as a warm subtropical,
slightly seasonal humid climate (Webb and Tracey, 1994),where temperatures range between �14 1C in winter and�29 1C in summer (Longmore, 1997). Regional annualrainfall varies between 1300 and 1700mm/year, and amoisture deficit occurs in the drier winter/spring season(Walker et al., 1981). Island vegetation is a mosaic of mixedsubtropical rainforest in moist patches, to dry sclerophyllforest, heath and coastal vegetation in drier areas.The Holocene section of the Lake Allom record is dated
by 11 AMS 14C ages and especially the upper half has ahigh temporal resolution of �65 a/sample. Low sea-levelsduring the early Holocene created dry continental
ARTICLE IN PRESS
A B
E
C
FD
G H I
LKJ
M N O
Fig. 3. Calibrated ages with 1s ranges for age and sample depth (A–O). Age-depth models are based on linear or second-order fits to the data
(top ¼ present, except for the Sperm Whale Head site). Originally published 2s ranges are given for the Fleurieu Peninsula (O), Lakes Euramoo (B) and
Johnston (N) sites. The Bega Swamp (D) age–depth model below 150 cm is preliminary due to the high scatter. Note use of different axis scales between
Fig. 4. Principal component analysis (PCA) results of well-dated, high-resolution sites in Australia, showing the first 2 axes against time with the originally
published pollen zonation. Bega Swamp and Lake Allom data was limited to dry land taxa exceeding 2% abundance at least once. Eucalyptus types were
combined to omit multiple-person counting errors for Bega Swamp. Lake Euramoo data was limited to taxa exceeding 5% abundance at least once
(Haberle, 2005). All data were square root transformed and PCA was performed with C2 software (Juggins, 2003). The Lake Euramoo PCA axis 1 (A)
contrasts Casuarina/Poaceae with rainforest taxa, which is precipitation related. Axis 2 (B) contrasts sub-montane/secondary rainforest species, indicative
of increased disturbance or burning, with lower montane closed rainforest taxa such as Cunoniaceae and Urticaceae/Moraceae (Haberle, 2005). Lake
Allom PCA axis 1 (C) contrasts Casuarinaceae with (araucarian) rainforest, while axis 2 (D) shows disturbance-related alternations between
Casuarinaceae/Araucaria and Eucalyptus/heath-taxa. Bega Swamp PCA axis 1 (E) contrasts forest with open vegetation, while axis 2 (F) mostly contrasts
dry Asteraceae/Casuarina/Chenopodiaceae with moist Pomaderris/heath and fern taxa. Charcoal accumulation rates for Lakes Allom and Euramoo (G)
conditions, with low accumulation rates, low lake levels,active dune formation and high fire frequencies (highcharcoal influx, Fig. 4G). A short hiatus between 6.5 and5.4 cal ka BP marks the onset of distinctly differentconditions with high lake levels, forest succession andreduced fire frequency between 5.5 and 3 cal ka BP.
After 2.7 cal ka BP, a large diversification occurredtowards the present-day heterogeneous subtropical rain-forest vegetation. Lake level was slightly lower and firefrequency increased after �2 cal ka BP (Donders et al.,2006). A small araucarian rainforest decline at 0.45 cal kaBP, well before European settlement of Australia, indicatessub-optimal growth conditions possibly caused by atemperature decrease (Donders et al., 2006).
The Lake Allom PCA results (Figs. 4C and D) reflect thediversification in the pollen record. The first axis docu-ments the shift to araucarian rainforest at �2.7 cal ka BP,while the second axis shows disturbance related alterationsbetween eucalypt forest and Casuarina from 5.5 cal a BPonwards. These changes are accompanied by increased firefrequencies (Fig. 4G). Relatively stable conditions areattained after 1.5 cal ka BP.
4.3. Temperate—Bega Swamp
Bega Swamp (7, Fig. 1A) is a mire located 50 km inland,on the south-eastern edge of the Southern Tablelands inNew South Wales and represents a more temperate record
ARTICLE IN PRESST.H. Donders et al. / Quaternary Science Reviews 26 (2007) 1621–1637 1631
influenced by westerly winds. The catchment has low reliefand gradient, favoring undisturbed accumulation. Annualrainfall is �1000mm/a and limits plant growth in thisregion. Winter is relatively dry although seasonal variationis small. The swamp is surrounded by wet tall, opensclerophyll forest, with a rich shrub and herb understorey,containing elements as Asteraceae, Acaena, Ranunculus,Hydrocotyle, Plantago, Wahlenbergia, Gonocarpus, Blech-
num and Pteridium (Green et al., 1988).A detailed study based on pollen traps and annual-
resolution analyses of peat from the site focused on annualpollen production variability and in-wash in the mire.Results showed that pollen deposition and influx into themire is controlled by spring and summer precipitation,respectively (Green et al., 1988). A 2.7m peat profile fromthe mire has been analysed at very high resolution, �30 a/sample during the last 15 ka BP, and has been dated by 39conventional 14C ages from 21 sample horizons (Hopeet al., 2004; Singh and Hope, unpublished results). Arevised age–depth model is given in Fig. 7.3D and is used inthe interpretation of the record.
The main early Holocene forest expansion occursaround 10.5 cal ka BP, with increased Eucalyptus, Casuar-
ina, Pomaderris and Cyathea. A mid-Holocene wet phase,characterized by expansion of wet heath and ferns, startedaround 7.5 cal ka BP, followed by eucalypt expansion anddrier conditions after 3.5 cal ka BP. Short centennialalterations between wet and dry taxa can be observed inthe record (Fig. 4F). After 1.7 cal ka BP assemblagesremained relatively stable until, around 100 cal a BP, forestcover slightly declined caused by European settlement anddeforestation. PCA results provide a good summary of thepollen record (Figs. 4E and F). The first axis correspondsto changes between forest and open vegetation, while thesecond axis mostly represents the dry–wet alternations.
5. Regional comparison of Holocene vegetation changes
5.1. Monsoonal and northern wet tropics
Lowland tropical sites potentially provide the bestsources of past climate change since upland areas, suchas the Atherton Tablelands, are climatically somewhatatypical. They receive orographic rain, which is susceptibleto wind direction (Shulmeister and Lees, 1995). The onlydetailed Holocene record from the seasonally humidlowland tropics of Northern Australia is the Four MileBillabong profile from Groote Eylandt (1, Fig. 1A),Northern Territory (Shulmeister, 1992). The site is aperennial lake within a dune field and is located withinthe strongly seasonal monsoonal tropics.
The pollen record reflects an early Holocene increase ineffective precipitation during the post-Glacial sea-level rise,reaching a maximum between 8.4 and �4.5 cal ka BP. Asubsequent change to drier conditions during the lateHolocene has previously been interpreted as indicative forthe onset of present-day ENSO dynamics (Shulmeister and
Lees, 1995). In the original study, the onset of dryconditions after 4 14C ka BP was based on a reconstructeddecrease in pollen influx. This interpretation stronglydepends on the accumulation rate that was used. Althoughthe record has been dated by seven 14C ages (Fig. 3A), theages show a relatively high scatter. The reconstructeddecrease in sedimentation rate was based on a single date,while organic content actually increases towards the coretop (Shulmeister, 1992). Although, further dating would beneeded to confirm the original conclusion, the record doesshow an expansion in swamp Restionaceae, highercharcoal and dry-land pollen during the late Holocene.Like at the Atherton Tableland sites (2, 3), these changesare indicative of more variable and/or drier conditions.A lowland tropical record from Whitehaven Swamp
(4, Fig. 1A), a semi-perched basin on Whitsunday Island,Queensland, confirms the general pattern of Holocenevegetation change in the Australian tropics (Genever et al.,2003). The record has a fairly high resolution of �140 a/sample, but only a single basal 14C date of 7735765 cal aBP. Similar to the Lake Allom record, early Holoceneexpansion of Casuarina is related to colonization of youngdunes. The record reveals moister than present conditionsbetween �7.5 and 4.5 cal ka BP (based on a linearage–depth relation), followed by an increase in Eucalyptus.Although the site contains evidence of human presence, thevegetation shifts occur at the time of no substantialarchaeological change. Most likely the indigenous popula-tion did not significantly disrupt the vegetation, butadapted to dynamic ecosystem changes (Genever et al.,2003).Evidence for the ecosystem change in the Australian
tropics is strong and, while some chronologies could beimproved, a consistent pattern emerges of increaseddisturbed and slightly drier conditions at 5 cal ka BP andfurther after 2.7 cal ka BP. Early Holocene changes aremuch more dissimilar and the attainment of the early tomiddle Holocene moisture maximum varies by up to 1000years between sites.
5.2. Subtropics
Apart from the recent Lake Allom (5) record (Donderset al., 2006), few detailed subtropical Australian recordsexist. The only datasets available for comparison are fromthe inland Barrington Tops region (6, Fig. 1A, Dodsonet al., 1986). Eight upland sites between 1160 and 1530maltitude provide a regional overview with a resolution of upto �150 a/sample. The site is located close to the presentwinter/summer rain boundary (Fig. 7.1A), and it istherefore sensitive to changes in moisture regime. Manyof the records have a discontinuous accumulation and,although all have been radiocarbon dated, they are notdiscussed individually. The main conclusions of Dodsonet al. (1986) were based on a radiocarbon age-scale, andhave been adjusted according to the calendar age-scale inthe present comparison.
ARTICLE IN PRESST.H. Donders et al. / Quaternary Science Reviews 26 (2007) 1621–16371632
Nothofagus moorei rainforest expanded on BarringtonTops from local refugia at 9 ka BP (10–10.2 cal ka BP).Between 6.5 and 3.5 14C ka BP (7.8–3.7 cal ka BP) cooltemperate rainforest and wet-sclerophyll forest covered alarger area than present, likely caused by increased summerrainfall relative to today. Forest retreat began at 5 andlasted until 1.6 14C ka BP (5.6–1.4 cal ka BP). The changeswere accompanied by increased fire intensity from3.1 cal ka BP, while open lakes disappeared in favor ofwetlands and bogs, which dramatically increased accumu-lation rates.
Similar to the tropical sites, there are some differences inthe early Holocene records of between Barrington Topsand Lake Allom. The moisture optimum started signifi-cantly earlier at Barrington Tops, while both show higherfire intensity and slightly drier conditions after 3 cal ka BP.Conditions apparently stabilized after 1.5 cal ka BP at bothsites, although some further small changes did occur.
5.3. Temperate—eastern coast
Bega Swamp (7) represents the most detailed pollenrecord available from temperate Australia. Other informa-tion is available from both coastal and alpine sites. Tworecords from a coastal lagoon system, Sperm Whale Headin eastern Victoria (9, Fig. 1A), document a clear mid-Holocene moisture maximum (Hooley et al., 1980).Although the barrier development is closely linked to sea-level changes, a late Holocene decrease in lake level andCasuarina abundance is in contrast with the effects of post-glacial sea-level rise and likely results from changes inclimate. Temporal resolution of the Hidden Swamp andLoch Sport Swamp records is up to �150 a/sample but thechronologies do not agree (Fig. 3F). The Hidden Swampradiocarbon data were obtained from a parallel core otherthan the one used for palynology, which might explain thetemporal offset between both records.
The upland Club Lake site in Kosciuszko National Park,N.S.W. (8, Fig. 1A), has a better-constrained chronology(Fig. 3E). The site documents vegetation changes from theLate-Glacial to the present at 1950 m altitude in the SnowyMountain range, which is an important climatic barrierbetween the westerlies and eastern tradewinds (Martin,1986). A fen section adjacent to Club Lake is the best-datedand most detailed record from the site, �200 a/sample, andis used for the data comparison. Main tree expansion wasbefore 7.5 cal ka BP, followed by a high moisture-relatedPomaderris maximum between 7.5 and 6.5 cal ka BP.Different alpine taxa increase after 5.8 cal ka BP andalternate dominance after 3 ka. After 3.4 cal ka BP,Pomaderris is absent, Pteridium declined and the sedimen-tary regime indicates more variable conditions. However,this vegetation change was most likely not caused by atemperature decline since the position of the timberline wasnot negatively affected and even went upslope in thetopmost samples (Martin, 1986).
Hence, temperature variability was likely small in theHolocene once modern conditions were reached. Inaccordance with late Holocene development at BegaSwamp, conditions became slightly drier after 6 cal ka BPand particularly drier and more variable after �3.5 and3 cal ka BP.
5.4. Arid interior
Two detailed Holocene records are available from thearid (eastern) interior of Australia, an area very sensitive tochanges in precipitation and circulation patterns. Furtherdata are available from stick rat (Leporillus) middens, butthese records are not continuous. Lake Frome is close tothe most arid centre of Australia (10, Fig. 1A), at thesummer-winter rainfall boundary (Singh and Luly, 1991).A Holocene pollen record from the lake indicates changesbetween arid Chenopodiaceae–Asteraceae shrub vegetationand grassland. Early Holocene grassland was moreadapted to summer monsoon rain, which implies apersistent positive SOI or La Nina-like conditions (Singhand Luly, 1991). After 5.5–5 cal ka BP, the grass coverdeclined, followed by increases in Acacia and Eucalyptus
and a decline in Casuarina after �3 cal ka BP. However, therecord has a low resolution, �350 a/sample, and is poorlydated (Fig. 2G), which precludes accurate comparisons.The high-resolution Lake Tyrrell site, with �80 a/
sample, is located southeast of Lake Frome (12, Fig. 1A)and has a reasonably accurate chronology (Fig. 3(H). EarlyHolocene changes in Lake Tyrrell are relatively small, but amoisture optimum is evident between 8.5 and 3.3 cal ka BP,characterized by increased Callitris. Around 3.3 cal ka BP,Casuarina increases significantly in Lake Tyrrell, indisagreement with the increase of Acacia and Eucalyptus
seen at Lake Frome (Luly, 1993). This contradiction can beexplained by the presence of extensive dunes around LakeTyrrell. Since Casuarina roots are capable of producingnitrogen through microbial associations, this genus cancolonize nutrient-poor soils (Ng, 1987). Therefore, the lateHolocene Casuarina increase likely reflects dune formationcaused by slightly drier conditions, which is confirmed bythe occurrence of evaporites in the sediment. In addition,several species of Casuarinaceae presently grow within theregion, each adapted to different moisture levels (Sluiterand Parsons, 1995). Although the differences in Casuar-inaceae abundance between Lake Frome and Tyrrell mightbe explained by expansion of a drought-adapted species(possibly C. pauper), this cannot be substantiated by pollenmorphological evidence, and therefore remains a hypoth-esis only (Luly, 1995). Lake Tyrrell is not located close toclimatic boundaries, which likely explains the low varia-bility in the record.Pollen and macrofossil data from stick rat middens
between Lakes Frome and Tyrrell (11, Fig. 1A) confirmincreased moisture relative to the present at 7–5 14Cka BP(�7.8–5.7 cal ka BP), and a more variable and drier climate
ARTICLE IN PRESST.H. Donders et al. / Quaternary Science Reviews 26 (2007) 1621–1637 1633
at 4–2 14C ka BP (�4.4–1.9 cal ka BP) (McCarthy andHead, 2001).
5.5. Mediterranean climate—southern coast
The southern coast of eastern Australia is one of the beststudied regions in Australia. Westerlies cause high pre-cipitation during winter, while hot and dry summers arecaused by southward expansion of the continental high-pressure system. Lake Leake, S.A. (14, Fig. 1A), is avolcanic crater lake perched by ash and clay with no long-distance water transport, which is highly sensitive tochanges in precipitation regime (Dodson, 1974). A high-resolution pollen record from the lake, �115 a/sample,reveals a water table rise from dry swamp before 10 cal kaBP to maximum levels between 7.2 and 5.4 cal ka BP.Levels were higher than present between 7.2 and3 cal ka BP (Dodson, 1974). The changes are mostlyrestricted to the aquatic vegetation, indicating that noenvironmental threshold was crossed for the regional forestvegetation.
The Tower Hill record in southwest Victoria (15,Fig. 1A) reveals changes that are concurrent with eventsat Lake Leake. Around 5 cal ka BP, Casuarina decreasedand sclerophyllous vegetation expanded, followed by aslight reduction of the fern cover at 3 cal ka BP (D’Costa etal., 1989). However, the onset of the early Holocenemoisture maximum occurred earlier than at Lake Leake,around 8.4 cal ka BP. The temporal offset between therecords is significant although the early Holocene chron-ology of Tower Hill is better constrained than at LakeLeake (Figs. 3I and J), despite lower sample resolution(�240 a/sample).
Between 5 and 2 cal ka BP, Casuarina declined as well atFleurieu Peninsula in eastern South Australia (13, Fig. 1A),favoring drier Myrtaceae, Chenopodiaceae and Asteraceaevegetation (Bickford and Gell, 2005). Although temporalresolution is low, �400 a/sample, the study further showsthat Aboriginal burning affected the vegetation far lessthan the impact of European settlement in the 19thcentury.
Similar to Lake Leake, Lake Wangoom is a closed craterlake sensitive to changes in precipitation (16, Fig. 1A).After the Late Glacial Maximum (LGM), sedimentationrecommenced at �10.4 cal ka BP at Lake Wangoom.Pollen data from the site reveal a lake-level maximumbetween 8 and 5 cal ka BP, followed by decreasing andmore variable lake levels after �3 cal ka BP (Edney et al.,1990). The lake-level change was accompanied by a middleto late Holocene increase in fire occurrence.
The early Holocene dominance of Casuarina and thedecline of the genus after 5 cal ka BP are consistent featuresin the southeastern pollen records, while most recordsindicate increased dry and variable conditions after 3 cal kaBP. However, the onset of early Holocene moist conditionsis much more variable between records.
5.6. Southern temperate—Tasmania
Westerly winds influence Tasmania throughout the yearas the island is located south of the high-pressure belt,creating moist conditions and low seasonality mainly inwestern parts of the island (Harrison, 1993). Nothofagus
rainforest characterizes the moist western side of the island,while Eucalyptus dominates the drier eastern side (Mac-Phail, 1979). Alpine vegetation in Tasmania is largelytemperature dependant due to high altitudinal gradients(Anker et al., 2001).A detailed pollen profile from Lake Johnston, at 875m
altitude (17, Fig. 1A), provides one of the westernmostrecords from Tasmania. Maximum Nothofagus rainforestexpansion was between 9.5 and 6.8 cal ka BP, withincreased eucalypt and heath vegetation after 3.8 cal kaBP, suggesting colder and slightly drier conditions (Ankeret al., 2001). Cynthia Bay lies halfway across theTasmanian west-to-east vegetation gradient at 737maltitude (18, Fig. 1A), and the site is more sensitive toprecipitation variability than Lake Johnston. Nothofagus
decreased at 4.9 cal ka BP at Cynthia Bay and Eucalyptus,Poaceae and Asteraceae expanded (Hopf et al., 2000).However, little aquatic vegetation change occurred, in-dicating that Holocene precipitation variability waslimited.A central Tasmanian montane site close to the timber-
line, Eagle Tarn on Mt. Field (19, Fig. 1A), experiencedrapid early Holocene warming and precipitation increaseprior to 11 cal ka BP (MacPhail, 1979). The Nothofagus
rainforest maximum was reached around 8.6 cal ka BP,raising the timberline outside its present range, whilePomaderris apetala expanded into eastern Tasmania. TheNothofagus/Eucalyptus ratio increased until 5.4 cal ka BP,and subsequently decreased. Nothofagus gunnii disap-peared after 3.4 cal ka BP with the development of moreopen forest (MacPhail, 1979).The heterogeneous conditions in Tasmania complicate
the correlation of climate patterns. In particular thevegetation development at western sites differed fromcentral or eastern Tasmanian sites and mainland Australia.However, more precipitation-limited sites do show changesconcurrent with temperate sites on the Australian main-land, particularly during the late Holocene.
6. Discussion
The data comparison reveals significant Holocenevariability at most, but not all, reviewed sites. Environ-mental changes, although considered to be relatively minorin the Holocene (Luly, 1993), are very pronounced at LakeAllom, due to its high rainfall dependence, heterogeneoussetting and small catchment buffer capacity. At the otherextreme, the arid interior Lake Tyrrell record shows verylittle variability (Luly, 1993). It is likely that environmentalchanges throughout the Holocene are only clearly detect-able at sites that are strongly dependant on precipitation,
ARTICLE IN PRESST.H. Donders et al. / Quaternary Science Reviews 26 (2007) 1621–16371634
such as crater lakes, or at the transition between climaticregimes. Arid inland areas are potentially very sensitive toprecipitation but a large part of the typical pollen types inarid records, such as Chenopodiaceae and Asteraceae, havebroad environmental ranges and are thus less suited todetect relatively small changes in moisture balance. Themost prominent changes are summarized in Table 2.
Common to all records is a clear mid-Holocene moistureoptimum, but especially the onset of the climate ameliora-tion differs widely between records. The temporal offsetsare most pronounced between upland sites like LakeEuramoo, where rainforest taxa occur from �8.5 cal kaBP, and lowland sites like Lake Allom, which showsincreasing moisture levels as much as 3 ka later. Themarginal seas apparently delayed the onset of moisterconditions at coastal sites until the shelf areas had beeninundated, so that the development of moist forests issignificantly delayed (Hope et al., 2004).
In temperate regions the early Holocene moisturemaximum starts as early as 8.5 cal ka BP, further northaround 7.5 cal ka BP, and in the subtropical regions theonset is evident only around 6 cal ka BP. However,tropical–monsoonal sites already show a moisture max-imum around 8.5 cal ka BP. Likely the early Holocenewarming caused shifts in the mean position of the high-pressure belt, increasing moist westerly wind flow whileeasterly tradewinds remained fairly constant. The earlyHolocene moist conditions at tropical sites are related toincreased monsoonal activity (Magee et al., 2004), acti-vated by high insolation across Asia (Fig. 2A), whichdisplaced the ITCZ southward.
Subsequent changes occurred more synchronously overdifferent climatic regions. At many sites the moistureoptimum ends or starts to decrease at 5.5–5.0 cal ka BP, asin the high-resolution Lake Euramoo site. Especially LakeAllom and Bega Swamp show evidence of more synchro-nous changes indicating increased climatic coupling be-tween northern and southern sites. The PCA axes for LakeAllom and Bega Swamp (Figs. 4D and F) reveal highly
Table 2
General timing of main moisture-related changes accross eastern Australia du
Region (see Fig. 1) (a) Monsoonal
wet tropics
(ka cal BP)
(b) Subtropics
(ka cal BP)
(c) Tempera
eastern coa
(ka cal BP)
Start early
Holocene
moisture optimum
8.7–7.0 9.0–6.0 �7.5
Initial drying
phase
5.0–4.5 �5.0 6.0
Second/intensified
drying phase
�2.7 3.0–(2.0) 3.5–3.0
Comments After �3.0 ka cal
BP increased
coupling with
south (Fig. 4)
similar centennial-scale changes from 5 cal ka BP onwards,while rainforest at Lake Euramoo becomes more adaptedto disturbance (Fig. 4B). Exceptions occur at the LakeTyrrell site and in Tasmania. At Lake Tyrrell no majorchange occurs in the mid-Holocene, and a late HoloceneCasuarina increase is reported (Luly, 1993). Given con-current sedimentological changes and a Callitris decreasethat indicate drier conditions at Lake Tyrrell, theCasuarina expansion after 3.3 ka Cal BP is not indicativeof moist conditions but possibly caused by increased duneformation, but this needs to be confirmed (Luly, 1995). Asobserved earlier (Anker et al. 2001), mid-Holocene changeswithin Tasmania are not synchronous. Western sites inparticular experience a very different climatic regimedominated by westerly airflow (Harrison, 1993), and aremost likely not influenced by changes in tradewind ormonsoon activity.All the climatic regions, except Tasmania, show further
changes around 3.0 cal a BP (Table 2). At the majority ofsites conditions become slightly drier and more variable(adapted to disturbance). Both Lake Allom and BegaSwamp records have a pollen zone boundary at �2.7 cal kaBP, concurrent with charcoal increases at Lake Euramooand, to a lesser degree, Lake Allom (Fig. 4G). Theincreased fire frequencies are confirmed across easternAustralia at northern (Groote Eylandt), southern (LakeWangoom) and upland (Barrington Tops) sites.The nature and amplitude of environmental changes
differ between sites but appear to be synchronous, withinthe uncertainties of the available age–depth models(Table 2). Increased burning may be caused by Aboriginalpractice but it cannot explain all variability. The environ-mental changes are highly synchronous, which is unlikelyto be a consequence of human activities. In addition,available archaeological data cannot be reconciled withperiods of vegetation change (Genever et al., 2003;Bickford and Gell, 2005).The observed changes after 3 cal ka BP could well reflect
changing ENSO dynamics. Increasing variability is
ring the Holocene (see Section 5)
te—
st
(d) Arid interior
(ka cal BP)
(e) Mediterranean
climate—southern
coast
(f) Southern
temperate—
Tasmania
�8.5 8.4–7.2 9.5–8.6 ka cal BP
5.5–5.0 5.4–5.0 (6.8) 5.4–4.9 ka cal
BP
3.3–3.0 �3.0 ka cal BP 3.8–3.4 ka cal BP
Decline at
Fleurieu penin.
(13) gradual;
between 5 and
2 ka cal BP
Heterogeneous;
region less
correlated to
ENSO (Fig.
1B–E)
ARTICLE IN PRESST.H. Donders et al. / Quaternary Science Reviews 26 (2007) 1621–1637 1635
consistent with the effect of intensified ENSO, whichcauses wet/dry anomalies in Australia (Van Oldenborghand Burgers, 2005). A gradual increase in ENSO frequencyand intensity has been proposed on the basis of modelingexercises (Clement et al., 2000) and paleoclimatic evidence(McGlone et al., 1992; Shulmeister and Lees, 1995; Rodbellet al., 1999; Gagan et al., 2004). Recent palynologicalevidence suggests a stepwise increase in ENSO activity(Donders et al., 2005) around 5 and 3 cal ka BP. The lateHolocene trend across Australia is in agreement with thisscenario, since the main periods of change are confirmed.An intensification of the ENSO cycle would affect tradewind strength (Dodson, 1998) and ITCZ position (Hauget al., 2001). Trade wind variability in turn affects theentire East-Australian coast and creates moisture anoma-lies that are consistent with the reconstructions(Figs. 1B–E).
The changes across Australia are synchronous but notequally prominent. This may be partly explained bydifferences in climate sensitivity between sites, however,differences in seasonality may also play a role. The majorENSO impact is during the southern winter/spring wetseason, which coincides with the Northern Australian dryseason (Magee et al., 2004). Consequently, increasedENSO-related dry events are likely to affect the temperateto subtropical regions where ENSO has a high impactduring the main period of moisture supply. These factorsexplain the pronounced changes observed in Lake Allomand Bega Swamp, which contrast with the relatively smallshifts at Groote Eylandt and Lake Euramoo.
The late Holocene orbital forcing shows rising Decemberinsolation at the equator, which increased after 6 ka BP, andreached about present-day levels at �3ka BP, (Fig. 2A).Since El Nino warm events develop from the Indo-PacificWarm Pool (IPWP) during the austral summer (Diaz andKiladis, 1992), the ENSO intensification is possibly causedby the increased equatorial IPWP warming, allowing warmevents to develop more frequently. Increased SouthernHemisphere seasonal contrast (Fig. 2A) likely added to theoccurrence of ENSO events after 5 cal ka BP (Markgrafet al., 1992).
7. Conclusions
Recalibrated pollen records provide a more consistentview on Holocene climate patterns in eastern Australia.The new chronologies allow better comparison betweenexistent and new terrestrial pollen records, and the resultsshow that changes in the dominant climatic mode duringthe Holocene can be detected by accurately comparingdifferent regions. The results and data summary (Table 2)will enable further integration of marine and terrestrialclimate records and comparison with other proxy data.
The trend towards aridity and more variable conditions,contrary to the early Holocene moisture optimum,documented here in a broad dataset confirms earlierconcepts about increased late Holocene ENSO activity
(McGlone et al., 1992; Shulmeister and Lees, 1995;Clement et al., 2000). Based on the better-constrainedchronologies, it is shown for the first time that the mainmid- to late Holocene climatic events in Australia are,within the dating errors, synchronous and occur around 5and 3 cal ka BP (Table 2). This stepwise increase towardspresent-day ENSO dynamics provides the best explanationfor the reconstructed changes in precipitation, and agreesin terms of temporal development with recent results fromFlorida, which is also strongly influenced by ENSO(Donders et al., 2005). The Australian reconstructions areimportant for model-data comparisons that rely on spatialpatterns of past climate changes. Furthermore, they areimportant for future research into the Holocene dynamicsof the coupled ocean/atmosphere ENSO system, whichneeds further integration of well-documented terrestrialand marine climate records.
Acknowledgements
The authors thank Sandy Harrison and James Shulme-ister for early support and providing data and Geert Janvan Oldenborgh for help with the KNMI ClimateExplorer. Edward Bryant and an anonymous reviewerare thanked for their detailed comments on the manuscript.The research was supported by the Council of Earth andLife Sciences, Netherlands organization for ScientificResearch (NWO). This is publication number 20061102of the Netherlands school of Sedimentary Geology (NSG).
References
Anker, S.A., Colhoun, E.A., Barton, C.E., Peterson, M., Barbetti, M.,
2001. Holocene vegetation and paleoclimatic and paleomagnetic
history from Lake Johnston, Tasmania. Quaternary Research 56 (2),