Quaternary Science Reviews 21 (2002) 1023–1037 A Late Pleistocene vegetation history from the Australian semi-arid zone Judith H. Field a, *, John R. Dodson b , Ian P. Prosser c a Archaeology A14, University of Sydney, NSW, Sydney 2006, Australia b Department of Geography, University of Western Australia, Perth 6907, WA, Australia c Division of Water Research, CSIRO, GPO Box 1666, Canberra 2601, ACT, Australia Abstract Cuddie Springs is an ephemeral lake in central northern New South Wales, Australia. The upper 3 m of sediment consist of lacustrine clays containing a Late Pleistocene sequence of extinct and extant fauna, and in the upper 1.7 m, an associated archaeological record. Changes observed in the pollen sequence include: (i) a peak in charcoal values corresponding to a dramatic decline in Casuarina woodland to chenopod shrubland at 2.5 m, respresenting a climatic shift to more arid conditions; (ii) chenopod shrubland moved into decline with the spread of grasslands around 1.7 m, and the amelioration in climatic conditions persisted until approximately 28,000 BP. A regime emerged which resulted in extended lake dry periods and peak aridity by approximately 19,000 BP and (iii) at 1 m depth, around 19,000 BP a shift to peak arid conditions is observed with a return of Chenopodiaceae and a decline in grasses. The lake entered an ephemeral phase that has persisted until the present day. The broad palaeoenvironmental framework of lake history, climate and vegetation change spans the archaeological and faunal records from Cuddie Springs. The direct association enables a closer examination of causation in faunal extinctions and human subsistence activities in the Australian arid zone. r 2002 Elsevier Science Ltd. All rights reserved. 1. Introduction Towards the close of the Pleistocene a number of dramatic environmental changes are recorded in the fossil record for the Australian continent (as elsewhere). The culmination of the glacial period at the Last Glacial Maximum (LGM), the extinction of a suite of large animals (the megafauna) and the arrival of people have all provided a complex record for the Australian arid zone. In Australia, the range of sites where pollen records for the Late Pleistocene are found is restricted to the more temperate margins of the continent (Kershaw, 1985; Singh and Geissler, 1985; Colhoun and van de Geer, 1988; Pickett, 1997). The aridity of the Australian interior has precluded widespread formation of organic- rich deposits with pollen. Data comes mainly from salt lakes, springs and anomalous peat deposits that generally have only shallow time-depth records (e.g. Bell et al., 1989; Boyd, 1990). The exceptions to these examples are studies from north-west Australia, the Nullarbor Plain, Lake Frome, Ulungra Springs and Cuddie Springs (Martin, 1973; Wyrwoll et al., 1986; Dodson and Wright 1989; Singh and Luly, 1991; Dodson et al., 1993). Of these studies only Ulungra Springs and Cuddie Springs provide information for the pre-LGM. Tracking the course of people and fauna through time in the arid zone has been difficult also because there are few preserved stratified sites. Open sites are rarely found intact leading most investigations to be focused on rockshelters and caves, locations often devoid of palaeoenvironmental data. As a result, researchers have turned to charcoal, phytolith and faunal studies in an attempt to reconstruct vegetation histories (Vrba, 1981; Dortch, 1984; Smith et al., 1995; Bowdery, 1998). While providing some clues to the nature of the local environment, faunal deposits can be of limited value considering the ‘‘disharmonious’’ assemblages that are known from the Pleistocene with no modern analogues (e.g. Graham and Lundelius, 1984). The open archaeological site of Cuddie Springs is interesting because of the associated archaeological, palaeoenvironmental and faunal records (Field and Dodson, 1999). The latter two extend well beyond the *Corresponding author. Tel.: +61-2-9351-7412; fax: +61-2-9351- 5712. E-mail address: j.fi[email protected] (J.H. Field). 0277-3791/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII:S0277-3791(01)00057-9
15
Embed
A Late Pleistocene vegetation history from the Australian semi-arid zone
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
Quaternary Science Reviews 21 (2002) 1023–1037
A Late Pleistocene vegetation history from the Australiansemi-arid zone
Judith H. Fielda,*, John R. Dodsonb, Ian P. Prosserc
aArchaeology A14, University of Sydney, NSW, Sydney 2006, AustraliabDepartment of Geography, University of Western Australia, Perth 6907, WA, AustraliacDivision of Water Research, CSIRO, GPO Box 1666, Canberra 2601, ACT, Australia
Abstract
Cuddie Springs is an ephemeral lake in central northern New South Wales, Australia. The upper 3m of sediment consist oflacustrine clays containing a Late Pleistocene sequence of extinct and extant fauna, and in the upper 1.7m, an associatedarchaeological record. Changes observed in the pollen sequence include: (i) a peak in charcoal values corresponding to a dramatic
decline in Casuarina woodland to chenopod shrubland at 2.5m, respresenting a climatic shift to more arid conditions; (ii) chenopodshrubland moved into decline with the spread of grasslands around 1.7m, and the amelioration in climatic conditions persisted untilapproximately 28,000BP. A regime emerged which resulted in extended lake dry periods and peak aridity by approximately
19,000BP and (iii) at 1m depth, around 19,000BP a shift to peak arid conditions is observed with a return of Chenopodiaceae and adecline in grasses. The lake entered an ephemeral phase that has persisted until the present day. The broad palaeoenvironmentalframework of lake history, climate and vegetation change spans the archaeological and faunal records from Cuddie Springs. The
direct association enables a closer examination of causation in faunal extinctions and human subsistence activities in the Australianarid zone. r 2002 Elsevier Science Ltd. All rights reserved.
1. Introduction
Towards the close of the Pleistocene a number ofdramatic environmental changes are recorded in thefossil record for the Australian continent (as elsewhere).The culmination of the glacial period at the Last GlacialMaximum (LGM), the extinction of a suite of largeanimals (the megafauna) and the arrival of people haveall provided a complex record for the Australian aridzone. In Australia, the range of sites where pollenrecords for the Late Pleistocene are found is restricted tothe more temperate margins of the continent (Kershaw,1985; Singh and Geissler, 1985; Colhoun and van deGeer, 1988; Pickett, 1997). The aridity of the Australianinterior has precluded widespread formation of organic-rich deposits with pollen. Data comes mainly from saltlakes, springs and anomalous peat deposits thatgenerally have only shallow time-depth records (e.g.Bell et al., 1989; Boyd, 1990). The exceptions to theseexamples are studies from north-west Australia, the
Nullarbor Plain, Lake Frome, Ulungra Springs andCuddie Springs (Martin, 1973; Wyrwoll et al., 1986;Dodson and Wright 1989; Singh and Luly, 1991;Dodson et al., 1993). Of these studies only UlungraSprings and Cuddie Springs provide information for thepre-LGM.Tracking the course of people and fauna through time
in the arid zone has been difficult also because there arefew preserved stratified sites. Open sites are rarely foundintact leading most investigations to be focused onrockshelters and caves, locations often devoid ofpalaeoenvironmental data. As a result, researchers haveturned to charcoal, phytolith and faunal studies in anattempt to reconstruct vegetation histories (Vrba, 1981;Dortch, 1984; Smith et al., 1995; Bowdery, 1998). Whileproviding some clues to the nature of the localenvironment, faunal deposits can be of limited valueconsidering the ‘‘disharmonious’’ assemblages that areknown from the Pleistocene with no modern analogues(e.g. Graham and Lundelius, 1984).The open archaeological site of Cuddie Springs is
interesting because of the associated archaeological,palaeoenvironmental and faunal records (Field andDodson, 1999). The latter two extend well beyond the
0277-3791/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 2 7 7 - 3 7 9 1 ( 0 1 ) 0 0 0 5 7 - 9
limits of the radiocarbon technique and possibly beyondknown human occupation of the southern part of thecontinent (see O’Connell and Allen, 1998). The aim ofthis paper is to report a Late Pleistocene environmentalhistory of the semi-arid zone around Cuddie Springs.This study provides a record of vegetation and environ-ment covering a pre-LGM time period not previouslyreported for arid zone pollen deposits.
2. Cuddie Springs
Cuddie Springs (Fig. 1) was known as a fossil faunasite for over a century (Abbott, 1881; Wilkinson, 1885)but only recently has it also been established ascontaining important archaeological and palaeoenvir-onmental records. The results to date include: (i)establishing the presence of people in this part of thecontinent at B35,000 years ago; (ii) the co-existence ofhumans with a suite of extinct animals (megafauna); and(iii) the presence of seed-grinding implements in a pre-LGM context (Fullagar and Field, 1997; Field andFullagar, 1998; Field and Boles, 1998; Field andDodson, 1999). The preservation of palaeoenvironmen-tal information was established in earlier studies(Dodson et al., 1993) and revealed the potential forCuddie Springs to provide a substantial contribution toour knowledge concerning arid zone vegetation historiesfor the Late Pleistocene.
2.1. Environmental setting
The Cuddie Springs lakebed is located on the north-west plains of New South Wales between the Macquarie
River and Marra Creek channels and is not part of acurrently active river system. The site is on the southernboundary of the summer rainfall belt in the semi-aridzone, with a mean rainfall of approximately 400mm/year. The average temperatures for the winter monthsrange from 5.31C to 18.51C, with summer maxima oftenexceeding 401C. The lake bed is approximately 2 km indiameter and lies in a landscape of o20m relief. Thefossil and archaeological records are found in the centreof a claypan (approximately 200m in diameter) on thelake floor (Furby, 1995, Fig. 2).The area around Cuddie Springs supports vegetation
that is characteristic of semi-arid environments (Fig. 3).The lake is characterised by grey alluvial soils whichsupport Eucalyptus largiflorens and Eucalyptus micro-theca interspersed with Acacia stenophylla. Atriplexspecies and Muehlenbeckia occur with Rhagodia spines-cens, Goodenia sp. and some unidentified grasses. Themistletoe, Lysiana exocarpi sp. tenius is found on manyof the trees in the area. The surrounding red soil plainssupport vegetation that includes Callitris columellaris,Casuarina luehmanii, Eucalyptus populnea, Geigeraparviflora, Flindersia maculosa and a range of shrubsincluding Atriplex, Eremophila, Santalum, Senna, Lepi-dium and Piemelea species.Early last century Cuddie Springs was reported as
the only source of water between the Barwon andthe Macquarie River during dry seasons (Andersonand Fletcher, 1934). The claypan fills after local stormsfrom rainwater and local surface run-off, and thegreater lake floor may fill after exceptional rainfalltaking up to 12 months to dry (see Field and Dodson,1999) (Fig. 4). When the claypan floods (B1 in 3 years),large numbers of waterbirds are attracted to the
Fig. 1. Location of Cuddie Springs in south-eastern Australia and in relation to the current river systems of the north-western riverine plains of New
South Wales. (Illustration: Fiona Roberts).
J.H. Field et al. / Quaternary Science Reviews 21 (2002) 1023–10371024
area and many aquatic plant species emerge. Theclaypan coincidentally dried after the sinking of abore on the adjacent property of Moranding in 1902,
and it was proposed then that this was responsiblefor the drying of Cuddie Springs. The site has beendescribed in the literature as both a spring and a
Fig. 2. The Cuddie Springs claypan. Excavations at the centre of the claypan (arrow) have revealed a record of fossil fauna, environment and
archaeology. (Photo: J. Field).
Fig. 3. Vegetation on the lake floor at Cuddie Springs, characterised by Eucalyptus species that tolerate periodic inundation. This photo was taken
during a wet year, in dry years there is no understorey vegetation visible. (Photo: J. Field).
J.H. Field et al. / Quaternary Science Reviews 21 (2002) 1023–1037 1025
mound spring with interpretations of its formationranging from meteorite impacts to deflation hollows anderoded structural domes (Habermahl, 1980; Meakin,1991).On the north-eastern margin of the lake there is a
remnant source bordering dune approximately 30 cmhigh, comprised of coarse and fine sand (see Fig. 5).Numerous artefacts, including backed pieces (Hiscockand Attenbrow, 1996) and grinding stones have beenobserved eroding out of the dune.Cuddie Springs is located on Quarternary alluvium
and aeolian deposits, consisting of gravels, sand, siltand clay. These deposits are derived from fluvialsources to the west (Marra Creek), the north(Barwon River) and the east (Macquarie River).These fluvial systems no longer contribute sedimentsto the Cuddie Springs lake. The dark features on theaerial photos that appear to be palaeo-channels tothe north and south of the lake floor (Fig. 5) areof a wavelength and width that are many timeslarger than the present rivers in the region. Thepalaeo-channels and the feature called Cuddie Springsare a boundary (a low ‘‘back swamp’’) between theMarra Creek floodplain and the slightly higher red soilsto the east.
2.2. Stratigraphy of the claypan sediments
The stratigraphy of the Cuddie Springs claypan wasdocumented by the Geological Survey of New SouthWales in 1990 after they recovered a 54m core from theclaypan periphery (Meakin, 1991). Following archae-ological excavations undertaken by Field and associatesin 1991, the stratigraphy in the claypan centre was alsorecorded (Dodson et al., 1993; Furby et al., 1993; Fieldand Dodson, 1999; Fig. 6).
2.2.1. Geological survey of NSW DDH: claypanperiphery (Profile 1, Fig. 6)From the base of the DDH to approximately 38m
depth, the core consisted of gravels and medium tocoarse sands indicative of a fluvial sequence. Between 39and 38m depth, carbonaceous fragments and pieces ofpyrite were found and these have been interpreted byMeakin (1991) as indicating the presence of vegetation,possibly on a swampy sandbank. The Geological Surveyof NSW also obtained one TL determination of467,0007150,000 at a depth of 40–45m in theunconsolidated sand. This would appear to be aminimum date considering the presence of Cretaceoussediments at 54m depth (Meakin, 1991).
Fig. 4. View of the claypan (background) from the southern margins showing local runoff held back from the claypan by a low levee (foreground).
The claypan fills after local rainfall and may take several months to dry. (Photo: J. Field).
J.H. Field et al. / Quaternary Science Reviews 21 (2002) 1023–10371026
Above 38m depth to approximately 12.6m from thesurface, the sediments were described by Meakin aslacustrine silts deposited in a low energy depositionalenvironment. The upper section of this unit is inter-preted by Meakin to represent a shallow lake phase witha higher energy regime than in previous periods assuggested by the ‘‘angled cross beds and fine sandylaminae’’ (Meakin, 1991). From 12.6 to 1.5m depth,there are fine to medium grained, well sorted sands,which are overlain by lacustrine clays. These clays,comprising the upper 1.5m of sediments, are interpretedas intermittent flood deposits (Meakin, 1991).
2.2.2. Archaeological excavations: claypan centre(Profile 2, Fig. 6)In 1991, a pit at the claypan centre was excavated to a
depth of 3m (Dodson et al., 1993). The sediments in theclaypan centre are at the lowest point on the lake floorand are constantly damp below about 1m depth.Sampling of the deposits below 3m was undertakenusing a sand auger. Particle size analysis of thesediments was compiled to 5m depth for the 1991
excavations (Dodson et al., 1993; Fig. 7). Furthersampling was conducted in March 1994 and ceased at10m depth due to waterlogging and slumping of thedeposit. The results showed that the Cuddie Springsprofile consists of a number of lithologically distinctunits as described below.From approximately 10 to 7m depth at the claypan
centre, the sediments consist of gravels and coarse sandwith high concentrations of bone, both complete andfragmented. In situ deposition of the bone is suggestedby the absence of any significant abrasion. Some of thebone recovered had teeth marks on the surface and all ofthe bone exhibited various degrees of mineralisation.Nodules of waterlogged clays (gley) were also found butthey contained no pollen. These deposits are interpretedas a fluvial stage prior to the formation of the lake.From 7m to 3m depth there are clean, well sorted
medium to fine grained sands of possible aeolian origin.An aeolian source is inferred from the particle size andthe degree of sorting that is common for wind blownsands as compared with fluvially sorted sands (e.g.Watson, 1989). At 5m depth the sediments contain littlesilt and clay (o15%) most probably introduced by rootchannels from the overlying deposits. This horizonappears to correspond to the sand horizon identified byMeakin (1991) from 12.6 to 1.5m in the GeologicalSurvey drill hole (Profile 1, Geol. Survey DDH, Fig. 6).The variable depth of the sand suggests a dune. At thetop of the aeolian sand unit there is a sharp boundarywhere the lacustrine clays begin and this is also markedby an apparent increase in the concentration of fossilbone. The lacustrine facies consist of a sequence ofsandy clays and silts that extend to the surface (Fig. 6).These have been interpreted to represent pan andephemeral conditions and are mostly grey to greyishbrown in colour (Munsell colour 5Y 7/1), are massive instructure and highly cohesive. A series of auger sampleshas shown that the upper 3m of lacustrine clays arecontinuous to the edge of the lake. Two layers of boneidentified as occurring below the archaeological recordsediments have sharp upper boundaries, possiblyrepresenting erosion surfaces. In both layers (at B1.8andB2m depth), bones are found in a matrix of quartzgravel and coarse sands, interpreted as lake shore (beachlag) depositional environments.Overlying the bone horizons are lake deposits
comprised principally of clay with a minor componentof sand and numerous small root channels. The claydeposits are interpreted as a shallow lake or swamp,with decreasing clay content corresponding to ephem-eral lake conditions. Between B1.7 and 1.35m depththere is sharp increase in the clay and silt content of thesediments.From 1.35m to approximately 1.1m depth the
particle size analysis indicates a return to the ephemeralconditions seen in the lower levels. At approximately
Fig. 5. Aerial photograph of the Cuddie Springs lake and surrounds.
The lake floor is approximately 2 km in diameter and the claypan
(arrow) is where the fossil records are found. (Photo courtesy of the
NSW Government Department of Lands).
J.H. Field et al. / Quaternary Science Reviews 21 (2002) 1023–1037 1027
1.1–1m depth there is a concentration of stone, boneand charcoal interpreted as a deflation pavement thatdips to the south-west. At 0.47–0.45m there is a layer offinely laminated silts, divided in places into two layersindicating deposition under standing water. From0.45m to approximately 0.25m depth there is a bandof well sorted orange gravel and quartz sand withinwhich are found numerous exfoliated teeth, mineralisedbone fragments and artefacts. This unit is discontinuousover a wider area, is similar to the sands recovered atdepth (>5m), and appears to represent spoil from thewell digging in the 1950s. The upper 25 cm of deposit
comprises clay and silt similar to that observed between1 and 0.47m depth and is consistent with depositionfrom intermittent flooding such as the ephemeralconditions observed in the present day.Bone has been recovered from all stratigraphic levels
at the claypan centre and the concentration of boneappears to increase in the coarse gravel deposits ataround 7m depth. The colour of the bone was highlyvariable in most levels ranging from light brown friablebone to very dense black material. ICP-AES analysis ofthe bone has shown that manganese is the primarymineralising element throughout the profile and thecolour of the bone directly correlates to the manganesecontent (K. Privat and J. Field, unpublished results,2000). At the lowest levels of sampling with the auger,bone comprised about 80% of the auger bucket contentsand was found in varying degrees of fragmentation. Atthe base of the 1991 excavations, in the surface of theaeolian sands at 3m depth, a number of bone elements,including a Diprotodon sp. mandibular ramus werefound in a para-conformal orientation. Above 1.7mdepth the bone is unabraded indicating an in situdeposition in the claypan.
2.2.3. ChronologyThe upper lacustrine deposits have been dated using
conventional and AMS radiocarbon techniques (Fieldand Dodson, 1999). The fifteen radiocarbon datesindicate that the unit with the human/megafaunaoverlap (B1.7–1m depth) was deposited sometimebetween approximately 34,000 and 28,000BP (Fieldand Dodson, 1999). The disconformity at 1m representsa time-compressed unit of up to 10,000 years. While theupper metre of sediment is disturbed, the earliest datefrom this level is approximately 19,000 years. One OSLage has been obtained at 1.55m of 35,40075800(ANUOD118a.) (Field and Dodson, 1999). Additionalradiocarbon determinations for the AMS techniquewere prepared using the ABOX-SC procedure (Birdet al., 1999; Fifield et al., in press) and returned ages thatfall within the range of radiocarbon ages obtained forarchaeological levels 1–4 using conventional and stan-dard AMS methods (see Table 1).
3. Sampling and pollen preparation
Samples for pollen analysis (3 cm2) were collected at5 cm intervals from 3.05m depth to the surface. Theywere placed into glass vials in the field with a spatulathat was cleaned after each use and then transported tothe laboratory. These samples supplement a previouspollen analysis (Dodson et al., 1993) in which sampleswere collected at 5 cm intervals to a depth of 1.8m.
Fig. 6. Stratigraphy of the Geological Survey DDH core (Profile 1)
and the archaeological excavations at the centre of the Cuddie Springs
claypan (Profile 2). The DDH core was recovered approximately 60m
from the excavations. No bone was reported nor was pollen preserved
in the core sediments. DS: Deflation Surface; FS: Ferruginised Sands,
which are overlain by a concreted beach lag deposit. (Illustration:
Megan Mebberson).
J.H. Field et al. / Quaternary Science Reviews 21 (2002) 1023–10371028
3.1. Chemical preparation of pollen samples
The pollen was separated from the high clay contentof the sediments by heavy liquid separation using ZnBr2in bromoform (Faegri and Iversen, 1975). The organicfraction was collected onto an acetate based filter paper,which was dissolved in acetone and removed beforetreatment with HF and acetolysis (Moore et al., 1991).The samples were dehydrated through an alcohol seriesand mounted in silicone oil (4000 cs viscosity).Terrestrial and aquatic taxa were counted until a
minimum of 200 grains of dryland taxa had beenidentified for each level. These counts formed the pollensum used in the pollen diagram (Fig. 8). The charcoalwas analysed using the point count method of Clark(1982). Azolla spores were counted separately and arerepresented in the diagram as raw counts. Spores andmassulae of the small freshwater fern Azolla have beenused in this study as an indicator of relatively high-levellake conditions. Azolla today is generally found‘‘floating on still, fresh waters of stock tanks, lagoons,
swamps and back waters, often blanketing large areasby the massing of innumerable plants’’ (Aston, 1973; seealso Cunningham et al., 1992: 34). Foster and Harris(1981: 202) observe that the ideal growing conditions forAzolla are those ‘‘where the effects of turbulence andperiodic flooding will not fragment the colonies’’. TheTertiary age beds that were the subject of the study byFoster and Harris (1981) were interpreted as shallow,permanent lakes, based on the Azolla content and thefine grained enclosing sediment. The presence of Azollain the fossil pollen record is therefore interpreted as anindicator of freshwater conditions at Cuddie Springs.The pollen diagram was prepared using the TILIA
program (Grimm, 1991). The taxa in the diagram havealso been grouped as trees, shrubs, herbs, aquatics andfern taxa. The sequence has also been divided into anumber of zones based on observations of sedimentchanges and major pollen changes through the profile.The full age of the fossil pollen deposits is unknownsince the lower layers extend beyond the limits ofradiocarbon. However, the age determinations forthe beginning of the human/megafauna overlap, atabout 1.7m depth, are aroundB35,000 years (Field andDodson, 1999).
4. Pollen Analysis
4.1. Zone 1.3.10–2.50 m; age unknown
Zone 1 is defined by high levels of Casuarinaceae withrelatively high levels of Poaceae and Asteraceaethroughout this zone. Eucalyptus and Acacia are alsorepresented, although at lower frequencies comparedwith Casuarinaceae. Some herbaceous taxa (e.g. Mal-vaceae) are also present. Towards the upper boundaryof Zone 1, there is a sharp increase in charcoal valuescorresponding to a sharp decrease in the relativeabundance of Casuarinaceae, Malvaceae and herbac-eous taxa. Chenopodiaceae, Dodonaea and Asteraceaealso decrease at the top of Zone 1. Changing freshwaterlake levels are indicated by the fluctuating counts ofAzolla spores.
4.2. Zone 2.2.5–1.9 m; age unknown
The lower boundary of Zone 2 is defined by a sharpdecrease in Casuarinaceae, associated with a corre-sponding increase in Chenopodiaceae values. Poaceaeand Asteraceae values maintain low levels, but increasetowards the top of the zone. The aquatic taxaCyperaceae, Typha and Myriohphyllum are representedthroughout this zone. Azolla levels are high at the baseof Zone 2, a continuation of the high levels in Zone 1.Eucalyptus appears to be represented by consistentlylow numbers throughout the sequence. The semi-arid
Fig. 7. Particle size analysis of the Cuddie Springs claypan deposit
(Profile 2, Fig. 6) (after Furby, 1991; Dodson et al., 1993). The relative
percentages of clay, silt, fine and coarse sand and gravel with depth to
5m are shown.
J.H. Field et al. / Quaternary Science Reviews 21 (2002) 1023–1037 1029
Table 1
Summary of pollen, geomorphology, archaeology and faunal data for the Cuddie Springs lacustrine facies
aAfter Field and Dodson (1999)bDates listed in stratigraphic order (see Field and Dodson, 1999 for detail).cSamples prepared using the ABOX-SC technique (Fifield et al., in press).
J.H
.Field
etal./Quatern
ary
Scien
ceReview
s21(2002)1023–1037
1030
Fig. 8. Pollen and charcoal diagram showing relative percentages of pollen versus depth for the Cuddie Springs deposits. The radiocarbon dates are listed on the left-hand side of the diagram. The
asterisk on the right-hand side of the diagram indicates the level where stone artefacts first appear in the sequence. Where pollen values are too low to register on the diagram, a cross marks the levels
where these taxa are present. (Illustration: Fiona Roberts).
J.H
.Field
etal./Quatern
ary
Scien
ceReview
s21(2002)1023–1037
1031
conditions, as indicated by the increase in Chenopodia-ceae, persist throughout Zone 2. However, long termflooding of the lake was common. A slight decrease inChenopodiaceae and an increase in Poaceae, Asteraceaeand Cyperaceae at the top of Zone 2 suggests a changeto less arid local conditions.
4.3. Zone 3.1.9–1.7 m; (?) >60,000–B35,000 BP
Zone 3 represents the period just prior to theappearance of the first stone artefacts. Stratigraphically,it comprises a layer of coarse gravel overlain by a wellsorted and concreted layer of bone and stone interpretedas a lag deposit and may represent hiatus in sedimentaccumulation. Zone 3 is likely to be a period of lowdeposition or erosion due either to deflation or fluvialactivity. Some of the bone is slightly rounded andcemented into the silt and gravels indicating a hiatus islikely.Arboreal pollen types are rare, although values for
Chenopodiaceae and Asteraceae are high. The ephem-eral lake conditions persist with Cyperaceae andMyriophyllum still present, however Azolla is nowvirtually absent. The depositional environment isuncertain in Zone 3, however, the high percentage ofgravel suggests that the deposit may have been part ofan erosion channel infill.Throughout Zone 3, the vegetation around Cuddie
Springs appears to have been a chenopod shrublandwith scattered trees. Some herbaceous species arepresent and a number of aquatic taxa indicate apositive water balance. It is possible that there wasdeflation of some of the fine-grained sedimentsbefore Zone 3 was sealed and this zone is nowcapped by a band of ferruginised sands. There is weakiron staining with no evidence of an ironpan orlateritic crust.
4.4. Zone 4.B1.7–B1.2 m;B35,000 BP
A gradual decrease in levels of Chenopodiaceae isseen through Zone 4. Poaceae and Asteraceae numbersincrease with a broader range of other shrub and herbtaxa present (i.e. Zygophyllaceae, Solanaceae, Malva-ceae, Heterodendron, Brassicaceae and Apiaceae). Fewertrees are present and their representation in the pollenrecord may be masked by a greater local pollen influx.Azolla is found in greater numbers through Zone 4indicating lake full conditions. This interpretation is alsosupported by the particle size analysis showing increasesin silt and clay, with very little sand and gravel (Dodsonet al., 1993). The deposit has numerous fine roots andplant matter and the clay breaks up into peds.Around 35,000 years ago, Cuddie Springs was a
shallow and marshy freshwater lake in an arid environ-ment, the catchment supported both chenopod
shrublands and grasslands. The lower values of Casuar-inaceae may be a function of a distant source. Thepresence of Acacia polyads indicates a local source (asopposed to monads which are possibly from moredistance sources). Aquatic taxa are constantly repre-sented and towards the top of the zone an increase incharcoal is observed. A gradual decrease in Chenopo-diaceae and increase in grass, herbs and aquatic taxasuggest that conditions became slightly moister duringthis period.
4.5. Zone 5.B1.20–1.00 m;B31,000–19,000BP
The period leading up to the LGM is representedby Zone 5 which is sealed at the upper boundaryby a deflated surface comprising bone, stone andcharcoal. The pollen evidence indicates that around28,000BP, there was a wetter period and the highestlevels of aquatic taxa were found during this time.The ephemeral conditions returned following themore permanent nature of the previous zone. Casuar-inaceae levels increased and Poaceae and Asteraceaeand other herbs were present. Chenopodiaceae didnot dominate the vegetation as seen following28,000BP and in the older deposits. The deflationpavement, which forms the upper boundary ofZone 5, represents a period of approximately 10,000years. The concentration of stone and bone in thedeflation surface suggests a cessation of sedimenta-tion and/or sediment deflation (possibly by increasedwindiness).
4.6. Zone 6. 1–0.5 m; 19,000–(?) 10,000 BP
Zone 6 covers a time period that includes theLGM. Recent excavations have revealed extensivedisturbance through this horizon and inverted radio-carbon ages at the base suggest bioturbation orreworking of sediments may have taken place. Thepollen record for this zone is therefore regarded asunreliable. However, in broad terms, the absenceof Azolla massulae through this zone is an indicationof the extended dry periods and along with the generalChenopodiaceae representation, is consistent with theLGM as described in Dodson and Wright (1989) andDodson (1989).The Cuddie Springs site was well within the arid zone
during the Last Glacial period with a dramatic changein the vegetation as an arid climatic regime dominated.A Holocene record is not thought to be represented inthis deposit.
4.7. Zone 7. 0.5–0.05 m; the historic period
No pollen preserved.
J.H. Field et al. / Quaternary Science Reviews 21 (2002) 1023–10371032
4.8. Zone 8. Surface: present
The pollen spectra from the surface sample reflect thepresent day environment at Cuddie Springs, i.e., theimmediate environs on the lake floor and possibly acomponent of the vegetation from the surrounding redsoil plains. Some species, identified on the red soilplains, in particular Callitris sp. have not been identifiedin any of the sediment samples analysed. The Eucalyptuscomponent of the vegetation is higher than at anyother time in the pollen record with Casuarinalevels similar to those observed around 30,000BP(at approximately 1.1m depth). The Azolla levels arevery low when compared to pre-LGM levels, thoughthe recent flooding of the lake in January 1995 resultedin an extended period of lake full conditions whichmay see a significant rise in the Azolla input to thepollen record.The relative concentrations of pollen represented in
the surface sample from Cuddie Springs indicate thatthe source of pollen is predominantly from theimmediate environs of the lake floor. The regionalpollen taxa in the present day context are species such asCallitris sp. some species of Eucalyptus and Casuarina,which are found up to several kilometres from theclaypan (see Dodson, 1983).
5. Fire history
Three distinct peaks can be observed in the charcoalrecord at 2.5, 1.15 and 1m depth. The lowest peak(at 2.5m) occurs prior to the arrival of people andbeyond the limits of radiocarbon dating. Highcharcoal levels correlate with an abrupt change inthe local vegetation, i.e. a decrease in Casuarinaceaevalues and an increase in Chenopodiaceae. Thepeak in charcoal at this level is most likely to bethe result of local fires (e.g. as in Clark, 1982)although a human cause cannot be eliminated untilthese levels have been investigated by archaeologicalexcavation.
6. Discussion
6.1. Vegetation history
The vegetation history at Cuddie Springs showsdistinct changes and variation starting from thecommencement of sedimentation of the lacustrine faciesthrough to the LGM. There are four distinct phases inthe pollen record. The first two phases correspond to theperiod prior to the arrival of people at Cuddie Springsand the onset of the final two phases begin when anarchaeological record is present (see Table 1).
An open woodland of Casuarina dominates therecord in the opening stages. The understorey iscomposed mainly of Poaceae and Asteraceae species.Eucalyptus species are a minor component of therecord with contributions from Acacia. Azolla countsindicate varying lake levels with intermittent dryingof the lake floor. The high levels of Casuarinamay be fringing vegetation and thus an indicator thatCuddie Springs was a floodplain depression. Callitris,often viewed as an indicator of dryness in vegetationhistories, is noticeably absent from the Cuddie Springsrecord, possibly because of distance to source and itssusceptibility to destruction (Dodson, 1979). The declineof Casuarina at the top of Zone 1 is coincident with alarge peak in charcoal values. Casuarina has beendescribed as a fire sensitive species and as such mayhave suffered a decline through burning rather thanclimatic change.The second distinct vegetation phase at Cuddie
Springs (>35 ka) occurs prior to the arrival of peopleand is marked by the high levels of Chenopodiaceae, agradual decline in Casuarina, constant levels of Poaceaeand varying levels of Asteraceae. Some aquatic taxaare recorded in low frequencies through these
levels with an increase in Cyperaceae following thedecline of Azolla. The charcoal concentrations throughZone 2 are constant but very low, suggesting that nomajor fire events occurred during this time. Thedisconformity at 1.7m forming the upper limit of Zone2 is characterised as a high-energy depositional environ-ment, possibly a stream channel with rapid deposition ofsediments.The third broad phase of the Cuddie Springs
vegetation history (B35–B28 ka) begins at about thesame time as evidence of people are first detected at thesite. The local vegetation suggests a semi-arid climaticregime at this time, signalled by a shift to opengrasslands with scattered Eucalyptus, Acacia and Ca-suarina trees. Azolla values rise and are maintainedthrough this zone by permanent inundation of theclaypan, and possibly the larger lake floor on inter-mittent occasions as seen in the present day (Furby,1995). Particle size analysis for the third phase reveals amarked increase in fine silts and clays with the absenceof gravels and coarse sands. Sediments indicate marshyconditions with the formation of peds and the abun-dance of fine roots in the profile. The formation of pedsmay mean seasonal or occasional drying out of theclaypan. The presence of Nitella oogonia, Azollamassulae, and Chydorid Cladocera carapace, headshields and fragments combine to support an interpreta-tion of still, shallow freshwater conditions and incom-plete drying of the lake (Ralph Ogden, ANU, pers.comm., 1994). Rapid deposition of sediment mayindicate frequent storm events and/or possibly contin-uous filling of the lake from a sparsely vegetated
J.H. Field et al. / Quaternary Science Reviews 21 (2002) 1023–1037 1033
catchment. Fossil remains of Grus rubicundus (brolga)have been also identified from these levels (Furby, 1995).Brolga are now only seen when the claypan has beenflooded, as are many other species of water birds (e.g.pelicans, ducks and spoonbills). The compacted stonepavement interpreted as a deflation surface marks theupper limit of this phase.The eventual drying of the lake and development
of high wind activity, also seen in the aeolian duneactivity in central Australia (Wasson, 1989), mayhave led to the loss of sediments on the Cuddie lakefloor and the construction of the source borderingdune which is found on the north-east margin. Theperiod represented by the deflation surface does notnecessarily indicate a cessation of sedimentation butextended dry periods and high wind activity or strongvariability between high and low water conditions thatresulted in a time averaging of the record. Beach sandscontributed to the lunette at lake full conditions plus thedeflation of sediments off a contracted lake floor at lowlake levels.The fourth phase (B19–B6ka) can only be described
in the very broadest terms because of the extensivedisturbance which has been identified through archae-ological excavation and evidence in the apparentlyhomogeneous pollen spectra. Sometime around19,000BP, sedimentation recommenced with periodicinundation of the claypan. The homogeneous lake mudsdeposited since 19,000BP are in contrast to thestratigraphic profile developed after the ephemeralflooding events recorded in the earlier zones. The highlevels of Chenopodiaceae are consistent with extendeddrying of the claypan surface. The absence of Azollaspores supports this conclusion.There is no intact Holocene record at the centre of the
claypan that may be a function of the depositionalenvironment during the Holocene with warmer climatesleading to greater evaporation and wetting and drying ofthe upper sediments (precluding pollen preservation).The present day Cuddie Springs is marked by theirregular filling and drying of the claypan. There are alsoperiodic droughts and relatively high winds that mayhave prevented a constant sedimentation regime follow-ing the LGM.The Cuddie Springs record shows a shift from
Casuarina forest to Chenopodiaceae in the early phasesof the record, with a shift to grasslands and then markedaridity around the time of the LGM. Modern pollenspectra show a shift to semi-arid vegetation, probablydeveloping sometime during the Holocene. The absenceof a Holocene record makes the timing of this changeimpossible to define.In the present day setting, Cuddie Springs is an
ephemeral lake that irregularly fills with water. A recentflooding event of January/February, 1995 is the firsthistoric recorded filling of the ancient lake floor (see
Field and Dodson, 1999). Initially, Cuddie Springs waspart of a riverine floodplain, probably joined to theDarling River system and later followed by a number ofshort-lived lacustrine phases.
6.2. Stratigraphic associations for the human/megafaunaoverlap
The combined records for pollen, charcoal, lakehistory and archaeology, provide a picture of sequen-tially deposited horizons within a sealed stratigraphicunit (Zones 4 and 5), albeit formed in a relatively shorttime frame (ca.7000 years) (Field and Dodson, 1999).This interpretation is supported by the composition ofstone tool and fossil bone assemblages that exhibitdistinct changes through time corresponding to phasesin lake hydrology (see Fullagar and Field, 1997; Fieldand Dodson, 1999; Table 1). The observed changes invegetation through Zone 5 directly correlate to climaticand vegetation changes in time frames established forother sites from across the continent (Bowler et al.,1976; Dodson, 1989; Dodson and Wright, 1989;Kershaw et al., 1991).The age of the Cuddie Springs deposit below 1.7m
depth is beyond the limits of radiocarbon. The firstarchaeological evidence of human occupation at CuddieSprings is not accompanied by any significant increasesin microscopic charcoal and it is not until higher in thesequence that peaks of charcoal are observed. Thesepeaks coincide with an apparent intense period ofoccupation of the lake floor as supported by the densityof the archaeological evidence. The peak in microscopiccharcoal at 1.15m in Zone 5 is associated with anincrease in large pieces of charcoalFrecovered duringarchaeological excavation, >1 cm in maximum dimen-sion, angular and shows little abrasion or rounding. Thepeak in concentration of charcoal in Zone 5 (equivalentto archaeological levels 2, 3 and 4) is followed by amarked decrease at its upper boundary. The charcoalincreases in Zone 5 correlate with a decrease in Poaceae(grasses), Casuarinaceae and aquatic taxa. The devel-opment of chenopod shrublands at the expense ofgrasses and some tree taxa may have been accelerated bylocal burning during a period of developing aridity. Thepersistence of charcoal through Zone 6 during max-imum dry conditions, in conjunction with the presenceof a range of stone artefacts, suggests that CuddieSprings may not have been abandoned during the LGM,as has been observed in other arid zone sites (e.g.O’Connor et al., 1993).
6.3. The implications for the archaeological andmegafaunal records
The two main explanatory models for the Pleistocenefaunal extinctions revolve around the effects of climatic
J.H. Field et al. / Quaternary Science Reviews 21 (2002) 1023–10371034
change and the timing of the arrival, and subsequentactivities of people (Dodson et al., 1988; Flannery, 1990;Horton, 2000). For the arid zone, the combination offactors may have been very different from thoseacting in the more temperate coastal zones. Climatemay well have been the driving force behind thearid zone extinctions. As Horton (1984) argues, theloss of free water as a result of rainfall depressionduring the lead up to the LGM may have hadirreversible effects on the large animals such as theDiprotodon. Not only could it not migrate to better-watered areas, but the shift from shrublands to grass-lands at Cuddie Springs during this period would haveselected against it in favour of grazers such as the redkangaroo (Macropus rufus) (see Dodson, 1989). Thedisappearance of the large bird Genyornis newtoni fromthe arid zone around 50,000BP (Miller et al., 1999) maybe another example of an animal ill-equipped to dealwith climate change and reconfiguration of habitats.The Emu (Dromaius novaehollandiae) persisted in thesehabitats through the extinction period documented inthis study.Elements of Diprotodon and Genyornis make up a
significant proportion of the bone assemblage in theearly archaeological levels (Field and Dodson, 1999;Zone 4). The animals appear to have died at awaterhole: by predation from humans, perishing duringa local drought or a combination of both. Localconditions were certainly favourable for browsersduring this time. However, around this time thereis a shift from chenopod shrublands to grasslands(Zones 4–5), and in the archaeological record theintroduction of seed-grinding stones (Fullagar andField, 1997; Field and Fullagar, 1998). The appearanceof these specialised technologies implies an environ-ment of uncertainty developed. Seed-grinding stonesare only found in arid/semi-arid zone contextsand the exploitation of seeds is viewed as aresource choice when people wished to maintaina presence in a resource depleted environment(Tindale, 1977).By the time the deflation pavement formed at the
upper limit of the megafauna/human overlap, themegafauna had all but disappeared. The environ-ment was more arid and the lake had entered anextended dry period. Fire does not seem to haveplayed a major environmental role in the archaeo-logical levels at Cuddie Springs, and through thisperiod peaks in charcoal are correlated tocamp-fires with people camping around a possiblesoak, or digging for water. The apparent broadsubsistence base of the first inhabitants andthe increasing aridity associated with the leadup to the LGM would see the latter as the driving forcein the localised events at Cuddie Springs around30,000BP.
The Cuddie Springs vegetation history has confirmedthe trends in general aridity and temperature forlatitudes of 321S in eastern Australia observed forthe Ulungra Springs record further east (Dodsonand Wright, 1989). It also demonstrates that firehistories are not necessarily accurate indicatorsof a human presence on the landscape and finally, itshows that climate change is certainly a significantfactor in arid zone faunal extinctions during the LatePleistocene.
Acknowledgements
Funding for the project was provided by theUniversity of New South Wales, Australian Geographic,National Geographic and an ARC Project Grant. Fieldwas the recipient of an Australian Postgraduate Award.We wish to thank Chris Myers, the Johnstone family,Jan Currey, Doug Green, the Brewarrina Local Abori-ginal Land Council, the Walgett Shire Council andnumerous volunteers for help with the project. MeganMebberson and Fiona Roberts are thanked for produ-cing the line diagrams. Thanks are also due to EricColhoun, Jim Rose and an anonymous referee forconstructive comments.
References
Abbott, W.E., 1881. Notes of a journey on the Darling. Proceedings of
the Royal Society of NSW 15, 41–70.
Anderson, C., Fletcher, H.O., 1934. The Cuddie Springs bone bed. The
Australian Museum Magazine 5, 152–158.
Aston, H.I., 1973. Aquatic Plants of Australia. Melbourne University
Press, Melbourne.
Bell, C.J.E., Finlayson, B.L., Kershaw, A.P., 1989. Pollen analysis and
dynamics of a peat deposit in Carnarvon National Park, central
Queensland. Australian Journal of Ecology 14, 449–456.
Bird, M.I., Ayliffe, L.K., Fifield, K., Cresswell, R., Turney, C., 1999.
Radiocarbon dating of ‘old’ charcoal using a wet oxidationFstepped combustion procedure. Radiocarbon 41, 127–140.
Bowdery, D., 1998. Phytolith analysis applied to Pleistocene–
Holocence archaeological sites in the Australian arid zone BAR