Non-marine evaporites with both inherited marine and continental signatures: The Gulf of Carpentaria, Australia, at ∼ 70 ka

This article was published in an Elsevier journal. The attached copyis furnished to the author for non-commercial research and

education use, including for instruction at the author’s institution,sharing with colleagues and providing to institution administration.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

http://www.elsevier.com/copyright

Author's personal copy

Non-marine evaporites with both inherited marine and continentalsignatures: The Gulf of Carpentaria, Australia, at ∼70 ka

Elisabet Playà a,⁎, Dioni I. Cendón b,c,⁎, Anna Travé a, Allan R. Chivas b, Adriana García b

a Departament de Geoquímica, Petrologia i Prospecció Geològica, Facultat de Geologia, Universitat de Barcelona,Martí Franqués, s/n, 08028 Barcelona, Spain

b GeoQuEST Research Centre, School of Earth and Environmental Sciences, University of Wollongong, NSW 2522, Australiac ANSTO (Australian Nuclear Science and Technology Organisation), Institute for Environmental Research, Menai, NSW 2234, Australia

Received 27 July 2006; received in revised form 17 April 2007; accepted 17 May 2007

Abstract

Changes in sea-level and associated climatic fluctuations resulted in extreme and cyclic changes in depositional environments inthe Gulf of Carpentaria region (N. Australia). Disconnection from the sea led to the establishment of a “Lake Carpentaria”, perchedabove sea-level. In this environment, evaporitic conditions at about 70 ka produced a repetitive alternation of μm to mm-thickevaporitic and micritic laminae with a varve-like appearance. These precipitates are interpreted as primary features, deposited in ashallow lake that retained limited water in its centre (core MD-32). Elemental and isotope geochemistry of gypsum and micritelaminae show a complex evaporitic environment where initially marine waters evaporated with the input of continental waterscompensating for evaporative losses. Reduced continental input could not support a lake of the initial dimensions and the lakecontracted to the deepest part of the basin along the north-eastern side of the basin. In a lake with smaller water volume, continentalsolutes became apparent.While Sr contents and sulfur isotopes indicate marine contributions, strontium isotopes and oxygen isotopesin sulfates reveal continental inputs and other processes such as recycling of previously precipitated evaporites, sulfate reduction andpotential reservoir effects. Carbonate-δ13C and δ18O values in micritic levels also reveal a continental influence and perhapsvariations in organic matter signatures associated with climatic variations and vegetation changes. The REE-normalized patterns ingypsum samples are like those found in northern Cape York rivers, restricting the potential continental inputs into the evaporatic basinto a limited geographical area. The small depletion in LREE-normalized patterns between gypsum and river samples is interpreted as amarine influence while depletions in HREE are considered to be the result of fractionation of HREE during gypsum crystallization.

The thickness of the calcite-gypsum couplets is consistent with those precipitated annually in modern evaporitic environments.This and the marked fluctuation between dry (gypsum laminae) and wet (micritic layer) environments, suggests a reducedmonsoon-like rainfall pattern operating in northern Australia during evaporite precipitation.© 2007 Elsevier B.V. All rights reserved.

Keywords: Gulf of Carpentaria; Evaporites; Gypsum-micrite cycles; Trace element contents; Rare-earth contents; Isotopes

1. Introduction

The Gulf of Carpentaria is a shallow (~50 m deep)epicontinental sea (~600,000 km2) between Australiaand New Guinea (Fig. 1). Its proximity to the equator,

Sedimentary Geology 201 (2007) 267–285www.elsevier.com/locate/sedgeo

⁎ Corresponding authors.E-mail addresses: [email protected] (E. Playà), [email protected]

(D.I. Cendón), [email protected] (A. Travé), [email protected](A.R. Chivas), [email protected] (A. García).

0037-0738/$ - see front matter © 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.sedgeo.2007.05.010


and its stable geological setting on the Australian Plate,make it a key area that records sea-level oscillations andclimatic (monsoonal) variations (Jones and Torgersen,

1988; Torgersen et al., 1988; McCulloch et al., 1989;Chivas et al., 2001). Communication with the Coral Sea(Pacific Ocean) and Arafura Sea (Indian Ocean) is

Fig. 1. Map showing the location of studied core (MD-32) and other cores in the area, general age distribution of lithologies in the catchment,bathymetry and outline of the Gulf of Carpentaria catchment in Australia. Cape York peninsula expands north of Karumba to the Torres Strait andnorthern Cape York rivers represent rivers north of Weipa. The − 53 m contour line represents the maximum extent of Lake Carpentaria (Chivas et al.,2001). Geological and bathymetrical information from Geoscience Australia (This figure incorporates data which are Copyright of theCommonwealth of Australia, 2006).

268 E. Playà et al. / Sedimentary Geology 201 (2007) 267–285


constrained by seafloor sills. To the east, Torres Strait hasa water depth of 12 m and to the west the Arafura Sill hasa maximum depth of 53 m. Sea-level fluctuations duringthe Quaternary have repeatedly transformed the marineenvironment into an estuary, open to the west, or a lake(Lake Carpentaria) perched above sea-level. Climaticvariations associated with fluctuating sea-levels haveaffected rainfall patterns as well as the sediment record.The intensity of the present Australian SummerMonsoon (wet season; December–March) duringlower sea-levels and its re-establishment after sea-levelrise still requires research (Bowler et al., 2001; Wyrwoll

and Miller, 2001). The potentially continuous sedimen-tary record of the area extends probably to the middleMiocene (Edgar et al., 2003). Presently available corematerials extend to approximately 130 ka (Chivas et al.,2001).

Recent palaeoclimatic studies and interpretationsfrom six shallow sediment cores obtained in the Gulfof Carpentaria (Chivas et al., 2001; Reeves et al., 2007,in press) have highlighted the necessity to understandthe nature of waters reaching the Gulf/Lake duringlower sea-level and their evolution when the area was ashallow continental lake. No evaporites had been

Fig. 2. Log of core MD-32 from the Gulf of Carpentaria (modified after Chivas et al., 2001). The studied evaporitic–micritic level is enlarged. Thevarious available dates are shown on the left, by 14C (AMS; expressed as calibrated ages), thermoluminescence (TL), optically stimulatedluminescence (OSL), and amino-acid racemization (AAR). The numbered facies, units 1 to 7, indicate the sedimentary facies and environments (seeReeves et al., 2007). T refers to the transitions between marine and non-marine facies, as deduce by faunal evidence.

269E. Playà et al. / Sedimentary Geology 201 (2007) 267–285


previously identified in shallow cores (Torgersen et al.,1988) spanning the past 40 ka.

The syn-sedimentary evaporitic levels reported inthis paper derive from core MD-32 (Figs. 1, 2; coredepth 853 to 873 cm, age approximately 70 ka) in thedeepest part of the basin (Chivas et al., 2001). Evaporitelayers have not been found in any of the other five coreswhere drier intervals are represented by disseminatedgypsum crystals or paleosols indicative of subaerialexposure. This suggests a restricted areal extent of Lake

Carpentaria at this time and perhaps times of nearly totaldryness (Reeves et al., 2007).

This paper considers the geochemical signature ofthe Carpentaria evaporitic levels using minor, trace andrare-earth elements and Sr, O, C and S isotopes. Theresults are interpreted with respect to the regional settingand linked to the climatic evolution of the area and inparticular to the suggestion of a strongly attenuated ornon-monsoonal climate. The chemistry of these evapor-ites also reveals important lessons in the search for

Fig. 3. Resin-embedded fragment of core and thin-section photomicrographs of the gypsum-carbonate alternations in the evaporite-micrite level fromcore MD-32. A, B) General view of the evaporitic (G, gypsum) and carbonate (M, micrite) cycles; sample 863–864; crossed-polarized light (cpl) in3B. C, D) Fining-upwards sequences (the arrows indicate the decreasing gypsum crystal size within the laminae); samples 865–866 (C) and 872–873(D); cpl. E) Detail of a gypsum lamina and gypsum crystals; sample 863–864; cpl. F) Micrite laminae under cathodoluminescence; sample 863–864.



criteria for distinguishing between ancient marine andnon-marine evaporites (Cendón et al., 2004a; Chivas,2007).

2. Study area

2.1. Climate and hydrology

The Carpentaria region covers three climatic zones.The north-eastern portion is an equatorial savannah,while most of the south and west are considered tropicalsavannah, with a small section of hot climate (drywinter) grassland in the southern catchment area. Theconnection between the three zones is that they reflectthe influence of a classical monsoon climate (Australiansummer monsoon). The majority of the annual rainfallfor the region occurs between December and April (wetseason). However the monsoon season can last fromtwo weeks up to four months, with lower rainfall in thesouth and generally higher in the northernmost areas.The main hydrological event of the year occurs duringthe monsoon rains with stream peak dischargesconcentrated during that time. Rivers reach their peakdischarge at any time during the January–April.

Bowler et al. (2001) interpreted variations of theAustralian summer monsoon during the last 300 ka basedon the paleohydrological record of Gregory Lake (NWAustralia). They recognised three phases ofmaximum lakeexpansion, and therefore maximum monsoonal rainfall,broadly occurring during interglacial stages. It is interest-ing to remark that each successivewet phasewas relativelyless intense than the previous one. The second wet phasereached its maximum at ∼100 ka with a termination atabout 90 ka. The onset of drier conditions in Gregory Lakebroadly coincides with the age of the evaporite layers incore MD-32 of ∼70 ka as bracketed by Amino AcidRacemisation (AAR), Optically Stimulated Luminescence(OSL) and Thermoluminiscence (TL) ages (Fig. 2) andgeneral trends of the Global sea-level curve of the past140 ka (Chappell et al., 1996; Waelbroeck et al., 2002).

2.2. Regional geology

The deeper sedimentary units of the Gulf of Carpen-taria form a series of stacked intracratonic basins anddepressions (Smart et al., 1980; Passmore et al., 1993a,b)namely the Bamaga Basin (of possible Palaeozoic age),the Carpentaria Basin (Jurassic and Cretaceous), and theKarumba Basin (Neogene and Quaternary). The areasurrounding the Gulf waters is important in understandingthe chemical signature of potential run-off inputs. Most ofthe catchment is characterized by a lowland perimeter of

Cenozoic alluvial fan deposits, with its maximum extentin central to south Cape York Peninsula (Fig. 1). This rimof sediments progressively narrows in a NW direction.The headwater lithologies vary around the Gulf. Thenorthern Cape York rivers drain through mono-litholog-ical catchments within the Late Jurassic Helby Beds andEarly Cretaceous Rolling Down Groups. Both aredominantly quartzose sandstones. The MD-32 cored siteis closest to the mouth of these rivers (Jardine, Dulhuntyand Wenlock) although 200 km offshore. Lithologicaldiversity increases to the south along Cape YorkPeninsula with abundant metamorphic and igneousrocks outcroping.

3. Methodology

The thickness of the evaporative laminae in MD-32(Fig. 2) governed their sampling. Sufficient gypsum wasrecovered from the top and central part of the evaporiteinterval, with the lower laminae being too thin tophysically separate. Micritic carbonate laminae weresampled in the whole evaporitic interval. Fragments ofcore (1 cm thick) were embedded in resin in order toperform petrological studies. Polished slabs and standardthin sections of samples were examined using optical andcathodoluminescence microscopy. A Technosyn ColdCathodoluminescence Model 8200 MkII was used,operating at 12–17 kV and 350 μA gun current. X-raydiffraction (XRD) was also used for identification ofmajor minerals in the collected samples.

Gypsum samples for geochemical analyses weretaken at 1-cm intervals in the core. Minor and traceelements (Mn, Ni, Sr, Ba, Li, V, Co, Cu, Zn, Rb, Y, Pb,Th, U) and rare-earth element (REE) contents of thepurest gypsum samples were obtained by InductivelyCoupled Plasma Mass Spectrometry (ICP-MS). Thesamples (N63 μm fraction) were first reacted with HCl5 M to remove carbonate and subsequently dissolved in18 mΩ cm−1 Milli-Q water. To avoid gypsum saturationin the final solution and to ensure a similar range ofconcentrations, a constant dilution ratio of 1:500 wasmaintained for all samples. The samples were thenfiltered through a 0.45 μm cellulose acetate membranefilter to remove any quartz or other insoluble mineralparticles. The concentrations of REE were found to behigh enough to avoid a pre-concentration step. ICP-MSanalysis was performed using a combined isotopedilution and internal standard method (Eggins et al.,1997), with external calibration to a multi-elementstandard. An enriched isotope standard of 145Nd–173Yb(87% and 92% respectively) previously calibrated usingMultiple Collector-ICP-MS (MC-ICP-MS) was added



Table 1Trace element and REE contents and isotope compositions of gypsum samples (oxygen, sulfur and strontium isotopes) and micrite samples (oxygen and carbon isotopes) from core MD-32, Gulf ofCarpentaria. Ce/Ce⁎ values are calculated based on Ce/Ce⁎=2[CeN / (LaN+PrN)] where N refers to MUQ normalized values

Depth(cm)

Samples Subunits Gypsum

Mn Ni Sr Ba Li V Co Cu Zn Rb Y La Ce Pr

Gypsum Micrite (ppm) (ppm) (ppm) (ppm) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb)

854–855 854–855 854.3 1.2 3.7 1325 2.7 119 132 449 648 284 133 24.83 25.58 56.17 8.42855–856 855–856 855.7 2.7 3.0 1353 16.2 173 153 390 828 388 141 29.83 33.73 76.61 11.66856–857 856–857 856.4 8.4 3.4 1175 2.4 104 129 365 1198 393 105 27.95 27.96 67.10 9.09857–858 857–858 857.2 3 1.1 3.3 1155 2.1 187 169 387 563 180 193 27.73 29.45 69.23 9.69858–859 858–859 858.3 1.0 2.3 691 1.3 143 119 275 421 275 122 23.59 27.02 64.71 8.20859–860 859–860 – 3.2 3.4 1090 4.4 194 168 433 638 259 178 44.27 56.18 130 16.31860–861 – 860.8 – – – – – – – – – – – – – –861–862 – 861.5 – – – – – – – – – – – – – –

862–863 862–863 862.6 2.0 3.4 1328 5.1 137 154 420 603 228 182 29.49 36.07 81.51 10.75863–864 – 863.3 – – – – – – – – – – – – – –

863.7864–865 864–865 864.2 2 1.3 3.6 1101 1.8 118 116 426 592 292 125 26.68 28.55 67.43 8.84

864.8865–866 865–866 865.3 3.1 3.4 1023 2.0 102 150 426 806 547 131 25.60 30.75 73.38 9.66

865.6866–867 866–867 866.7 2.9 3.9 937 3.7 357 310 524 788 419 317 42.19 54.75 126.38 16.64

867–868 – – – – – – – – – – – – – – – –868–869 – 868.4 – – – – – – – – – – – – – –869–870 – – – – – – – – – – – – – – – –870–871 – – 1 – – – – – – – – – – – – – –871–872 – 871.3 – – – – – – – – – – – – – –

871.6 – – – – – – – – – – – – – –872–873 – 872.3 – – – – – – – – – – – – – –

872.8 – – – – – – – – – – – – – –

Note: Differentiation of subunits (1–3) is mainly based on geochemical information; these subunits are described in the Palaeogeographic implications.

272E.Playà

etal.

/Sedim

entaryGeology

201(2007)

267–285


Micrite

Nd Sm Gd Dy Er Yb Lu Ce/Ce⁎ Pb Th U δ18OVSMOW δ34SVCDT87Sr/86Sr δ18OVPDB δ13CVPDB

(ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (ppb) (‰) (‰) (‰) (‰)

39.18 8.92 7.83 4.72 1.81 1.52 0.07 0.89 92 9.90 5.06 +15.2 +22.0 0.70956 (±9) −1.21 −12.9752.16 11.64 10.22 5.88 2.45 1.55 0.12 0.89 88 7.36 10.34 +15.3 +22.0 0.70949 (±10) −0.60 −12.8243.02 9.32 8.39 5.85 2.22 1.31 0.08 0.98 242 – 5.45 +15.8 +22.1 0.70954 (±14) −2.34 −15.3942.37 10.17 8.40 5.77 2.14 1.73 0.09 0.95 114 1.11 8.47 +16.9 +22.3 0.70941 (±12) −2.07 −13.7337.07 8.35 7.10 5.14 2.30 1.73 0.13 1.01 49 4.94 5.88 +16.2 +22.5 0.70956 (±6) −3.25 −14.6668.48 15.17 13.45 9.64 3.81 2.69 0.34 1 136 14.80 10.39 +15.6 +21.8 0.70957 (±20) – –– – – – – – – – – – – – – −2.93 −13.21– – – – – – – – – – – – – +0.09 −12.71

50.26 11.19 9.62 5.65 2.38 1.82 0.17 0.96 193 15.11 20.29 +14.1 +21.9 0.70943 (±10) −1.26 −16.14– – – – – – – – – – – – – +0.21 −13.64

−2.67 −12.2138.27 9.09 7.80 5.31 2.40 1.66 0.15 0.99 107 11.86 7.24 +14.9 +22.1 0.70932 (±19) −3.04 −12.81

−4.07 −14.3440.62 9.29 7.86 5.50 2.70 1.68 0.16 0.99 157 16.33 9.74 +15.5 +22.1 0.70943 (±31) −1.97 −16.59

−0.36 −11.7770.79 15.50 13.48 9.51 4.30 3.31 0.34 0.97 289 26.07 16.38 +16.5 +22.3 0.70977 (± 11) −2.24 −17.09

– – – – – – – – – – – – – – –– – – – – – – – – – – – – −2.01 −12.80– – – – – – – – – – – – – – –– – – – – – – – – – – – – – –– – – – – – – – – – – – – −0.49 −9.58– – – – – – – – – – – – – −1.60 −11.83– – – – – – – – – – – – – −0.97 −11.42– – – – – – – – – – – – – −0.88 −10.98

273E.Playà

etal.

/Sedim

entaryGeology

201(2007)

267–285


Fig. 4. Geochemical profiles of the evaporite-micrite level (MD-32; 853–874 cm) of the Carpentaria basin, with data shown separately for the gypsum and micrite sample types.

274E.Playà

etal.

/Sedim

entaryGeology

201(2007)

267–285


to each sample. Nd and Yb were determined by isotopedilution and the enriched isotopes were used as internalstandards for the other elements, correcting for instru-mental drift during analysis. Oxide production waslimited to less than 1% for La (139LaO/139Lab0.01) andmiddle rare-earth elements (MREE) and heavy rare-earth elements (HREE) were corrected for oxideinterferences by light rare-earth elements (LREE) andBa using oxide ratios determined at the start of each day,corrected for drift using 163DyO/163Dy which wasmonitored on standards throughout the day. Theprecision and accuracy of the method was determinedby repeat analysis of the NASS-4 seawater standard.The relative standard deviation (%RSD) of the methodranges from 2% to 6% and the concentrations are ingood agreement with literature reports of analysis ofREE in NASS-4 (Wyndham et al., 2004).

In order to guarantee that the obtained analyses closelyrepresented the composition of the gypsum crystals,several approaches were adopted: a) 1-cm-thick sampleswere sieved and N63 μm gypsum crystals were separated;thus, analyses were performed on isolated gypsumcrystals avoiding contamination from possible free orinterstitial pore waters and/or dispersed clay, micrite orpyrite. b) Potential contamination by fluid inclusions wasnot a problem given that the gypsum samples weremicrocrystalline. c) No residue was observable afterdissolution of the samples. d)Mineral identification undera binocular microscope indicated that the sieved sampleswere pure gypsum. On the basis of these considerations, itis concluded that the obtained elemental concentrations(ppm to hundreds of ppb) closely represented thecomposition of the gypsum crystals. In the case of REE-elements these are considered as a mixture of REE-

Fig. 5. A) MUQ-normalised (Kamber et al., 2005) pattern of gypsum REEs from Carpentaria; B) Gypsum; average REE pattern (×103). Rivers;average REE pattern (×106 for all 3 major rivers from the Wenlock to the Jardine river, including tributaries; the average comprises two differentsampling seasons for each river (Cendón et al., 2004a). Seawater: average REE pattern (×109) for the 5 and 49 m depths (Alibo and Nozaki, 1999).



elements derived from the brine and incorporated duringcrystallization from suspended particulates (b0.45 μm)trapped during crystal growth.

Gypsum samples (N63 μm fraction; converted toBaSO4) were also analysed for their oxygen and sulfurisotope compositions. The respective CO2 and SO2

gases produced from the sulfates were analysed on acontinuous-flow elemental-analyser Finnigan DELTAplus XP mass spectrometer, with TC/EA pyrolyser foroxygen and Finnigan MAT CHN 1108 analyser forsulfur. The values are given relative to the V-SMOW(Vienna Standard Mean Ocean Water) reference forδ18O of sulfates, and to the V-CDT (Vienna CañonDiablo Troilite) reference for δ34S of sulfates; themeasurement precisions are ±0.4‰.

The strontium isotope values of gypsum weredetermined with a Finnigan Neptune MC-ICP-MS. Theremnant of each aliquot used for trace-element analysis

was diluted with 2%NHO3 to a concentration of∼50 ppbSr. After evaporating to dryness and re-dissolving in 2 ml2 M HNO3, samples were passed through an ion-exchange micro-column packed with Sr-specific resin.This allows Sr separation while removing Rb, Ca andother matrix elements. The accuracy and precision ofresults was monitored during the analysis by intercalatingthe standard strontium carbonate SRM-987with 87Sr/86Srof 0.710290 (±63, 2σ, n=10).

Individual carbonate laminae were also analysed fortheir oxygen and carbon isotope compositions. Carbo-nates were separated from enclosing gypsum using adental microdrill. Possible inclusions of organic matterwere cleaned by sodium hypochlorite leaching. 60–70 μg of each sample were reacted with 103% H3PO4

for 2 min in vacuum at 70 °C. The CO2 was analysedusing an on-line carbonate device connected to aFinnigan MAT252 mass spectrometer. The values are

Fig. 6. Correlation among δ34S, δ18O values and Sr isotopes within individual gypsum samples from core MD-32, Carpentaria. (A) Oxygen vs.sulfur isotope compositions (B) strontium contents vs. oxygen and (C) sulfur isotope compositions, and (D) oxygen vs. strontium isotopecompositions.



expressed with respect to the V-PDB (Vienna Pee DeeBelemnite) standard with a precision of ±0.02‰ forδ13C and ±0.08‰ for δ18O.

4. Results

4.1. Facies and petrology

Evaporites in the Carpentaria cores are restricted tocore MD-32, between 853 and 873 cm in depth; Reeveset al. (2007) also described the occurrence of individualgypsum laminae between 1120 and 1070 cm (∼90 ka)and sparse disseminated gypsum crystals above 70 m,the latter in the upper marine unit, post-dating thetransgression at ∼12 ka. Only minor amounts ofdisseminated gypsum crystals have been detected inthe other five cores and relating to similar time frames(MD-31 and MD-33). These three cores form a transectfrom the depocenter (MD-33) to more distal parts (MD-31), with core MD-32 located in a relatively centralposition of the modern Gulf of Carpentaria in 64 mwaterdepth (Fig. 1). The evaporitic sequence is enclosedwithin muddy carbonate sediments with intercalations ofsilty clay and sandy levels (Fig. 2), interpreted as mainlymarine in origin (Chivas et al., 2001).

The evaporitic level is composed of an alternation ofthinmuddy carbonate and gypsum laminae (Fig. 3A, B, C).Individual laminae are from 50 μm to 15 mm in thickness,and the thickness (and the purity) of the gypsum laminaeincreases in the central part of the evaporitic succession anddecreases towards the upper part, where the muddycarbonate laminae become more abundant. The bottomand top of each lamina are flat and continuous and displayno dissolution features; discrete crenullations and, locally,possible tractive structures are present. Internal subcycleswithin the laminae are indicated by: a) the intercalation ofvery thin laminae of evaporites in the carbonate horizons,and whose thickness is generally one-single gypsumcrystal, or, b) an increase of micrite content within thegypseous laminae.

4.1.1. Gypsum laminaeGypsum laminae are composed of fine-grained

gypsum crystals, nearly equant and euhedral to sub-euhedral, from 15 to 175 μm across (with a mean size of30–80 μm). Gypsum is primary, that is, precipitated asgypsum in the sedimentary environment and not fromhydration of previously precipitated anhydrite. In thesecrystals, twinning is not common. These laminae do notdisplay any internal arrangement, but in some cases,some laminae show imperfect fining-upwards trends(Fig. 3C, D). Enclosed micritic calcite grains may be

present among the gypsum (up to 10% of micrite,depending on the laminae). Other accompanying compo-nents are quartz, clays, micas and minor pyrite framboids.Ferroan oxides appear in some cases in the exfoliationplanes of gypsum. No celestite, dolomite, chert orbiogenic textures such as microfossils or burrowingwere detected.

4.1.2. Carbonate laminaeMicritic calcite with crystals ranging from 5 to 10 μm

in size is the main component of the carbonate laminaefound between the gypsum laminae (Fig. 3A, B, C). Themicritic laminae show a homogeneous orange lumines-cence (Fig. 3F). Other components less commonlypresent in these laminae include vegetal (carbon)fragments, framboidal pyrite and possible hematite,and euhedral–subeuhedral gypsum crystals. No porosityhas been observed. The amount of enclosed gypsumcrystals in these laminae varies from nearly 0 to 50%.Microfossils (benthic Foraminifera, Ostracoda andechinoderm spicules) are in general very scarce, butsome laminae display abundant benthic microfauna(discorbids); these levels can show irregular upper-surface contacts and have been interpreted as reworkedsediments.

4.2. Geochemistry

4.2.1. Minor, trace and REE element contents ofgypsum

Strontium concentrations have been used as apalaeosalinity and/or palaenvironmental indicators tradi-tionally (Butler, 1973; Kushnir, 1982). The Sr contents inthe studied samples range from 691 to 1353 ppm, with amean of 1118 ppm (Table 1, Fig. 4), well within theexpected contents of marine-derived lithofacies (Ortíet al., 1984; Dronkert, 1985; Michalzik et al., 1993; Luet al., 1997; Rosell et al., 1998; Playà et al., 2000; Lu et al.,2002; Playà and Rosell, 2005; among others). However,the Sr contents in gypsum cannot be used by themselvesas marine or non-marine indicators; gypsum precipitatedfrom mixed marine–non-marine brines can displaytypical marine Sr contents even if non-marine influenceis low or non-marine waters are Sr-depleted.

The Mn, Ni and Ba concentrations are mostly in the1–5 ppm range. Li, V, Co, Cu, Zn, Rb and Pb rangebetween 100 and 700 ppb with Th and U in the low ppbrange 5–25 ppb (Table 1, Fig. 4). These concentrationsare similar to those reported in Toulkeridis et al. (1998),particularly for those elements more abundant inseawater. Transition metals, show more variations asthey are relatively more affected by non-marine inputs.



Positive correlations can also be observed between some ofthe analysed elements (Ni–Co, Rb–Li, V–Li, Ba–Sr, Cu–Zn, Cu–Mn). These elemental affinities within alkali andalkaline earth minor and trace elements have also beenobserved in the rivers of the region (Cendón et al., 2004b,c)as well as in other river systems (Gaillardet et al., 2004).

The Yand REE concentrations in the gypsum samplesare all in the low ppb range with lower concentrations forthe HREE. The general knowledge of REE concentra-tions in gypsum is very limited. The concentrationsreported here (Table 1, Fig. 5) are overall, similar to thosereported by Toulkeridis et al. (1998) for gypsum grownin an organic-matter-rich mangrove soil.

All gypsum samples in core MD-32 show parallelREE patterns when plotted against MUQ (MUd fromQueensland) (Kamber et al., 2005; Fig 5A). The relativelevels of the LREE increase from La to Sm and those ofthe MREE and HREE are depleted in a decreasing orderfrom Sm to Lu. We have plotted our REE values againstMUQ as this reference composite was derived fromsediments in 25 rivers of Queensland including 5catchments in the Gulf of Carpentaria area (Fig. 5B).

4.2.2. Oxygen and sulfur isotopic composition ofgypsum

The oxygen and sulfur isotope compositions weredetermined in 10 gypsum samples (Table 1, Fig. 4). Theδ18OV-SMOW values range from +14.1 to +16.9‰ (meanof+15.6‰), while the δ34SV-CDT values vary in anarrow range between +21.8 and +22.5‰, with a meanof +22.1‰. Modern oceanic dissolved sulfate has δ34Svalues in the range +19.3 to +21.12‰ for severallaboratories and analytical techniques. Taking intoaccount the average gypsum-water 34S fractionation of+1.65 ‰ (Thode and Monster, 1965), the original brinewould have had an average δ34S value of about +20.5‰well within normal marine values. In the case of oxygen,modern values range between +8.6 and +10.1‰.However considering the fractionation due to crystalli-zation, δ18O=+3.5‰ (Lloyd, 1968), the averageoriginal brine in the core MD-32 samples would havehad a δ18O=+12.1‰, well above the expected marineδ18O value of about +9.5‰. The oxygen and sulfurisotopic compositions and Sr contents of gypsum showa correlation (Figs. 4, 6A, B, C). This suggests that bulkchanges in the composition of the brines affected thethree components (Sr, O, S) together, which cannot beexplained by evaporation alone.

4.2.3. Strontium isotopic composition of gypsumThe strontium isotopic ratios of the same 10 gypsum

samples have values ranging from 0.70932 to 0. 70977,

with a mean of 0.70951 (±1) (Table 1, Fig. 4). Thepossible error in the 87Sr/86Sr due to Rb is negligiblegiven that Rb contents of the gypsum samples are verylow (122–317 ppb) and that they are Late Pleistocenesediments. When Rb/Sr ratios are b0.1, a rubidiumcorrection is not necessary (Clauer, 1976). Moreover,the Sr-specific resin separates quite effectively Rb asmonitored during analysis. The 87Sr/86Sr ratios for theCarpentaria gypsum samples are more radiogenic thanthe value for modern seawater of 0.70920. Thestrontium isotope ratios of the ocean during the time(∼70 ka) of Carpentaria gypsum precipitation were notdifferent from that of modern seawater, and even in theEarly Pleistocene (1.78 Ma) the value was 0.70910(Farrell et al., 1995). The oxygen and strontium isotopiccompositions of gypsum display a slight positivecovariance (Fig. 6D), suggesting an increase incontinental input probably associated with dissolutionof previously deposited gypsum with rising water levelin Lake Carpentaria.

4.2.4. Oxygen and carbon isotopic composition ofcarbonates

The carbonate laminae have δ18OV-PDB values rangingfrom − 4.07 to + 0.21‰, with δ13CV-PDB values between− 17.09 and − 9.58‰ (Table 1, Fig. 4). The δ18OV-PDB

values of these carbonates are lower than the expectedvalues for carbonates precipitated from seawater. Theδ18OV-SMOWof the Quaternary seawater ranges from−0.5to +1‰, with lower values during the interglacial stages(Berger, 1979), and Foraminifera living in these watersshow δ18OV-PDB values from −2 to +5.1‰ (e.g.,Duplessy, 1978; Broecker, 1986; McManus et al., 1999;Kroon et al., 2000). The obtained δ13CV-PDB values are farlower than those of carbonates precipitated in isotopicequilibrium with Quaternary seawater (δ13CV-PDB valuesranging from +0.4 and +2.1‰, Duplessy, 1978). Theoxygen and carbon-isotope values display similar varia-tions through time as shown in their isotopic stratigraphy(Fig. 4), but a correlation with the isotopic composition ofgypsum is not observed due to the lower samplingresolution from the evaporites.

5. Discussion

5.1. The precipitation of gypsum and carbonates.Evaporitic model

The fine-grained gypsum crystals of the evaporiticlaminae can be interpreted as chemical precipitateswithin the brine body and/or close to the brine-air surfaceand accumulated at the substrate-brine interface, with a



minor clastic component and little or no reworking.Good preservation of the crystal habits, microcrystallinesize, purity and internal structure of the laminae, nearabsence of typical detrital accompanying minerals (e.g.quartz, mica), and lack of tractive structures supports thisinterpretation. Similar lithofacies composed of very fine-grained, equant and prismatic gypsum crystals inter-preted as free precipitates from a brine, have beendescribed in the modern salt-works of Santa Pola (SESpain; Ortí et al., 1984) and in the Quaternary sedimentsof the Prungle Lakes (SE Australia; Magee, 1991) and inthe evaporitic units of the Baza Basin (SE Spain; Gibertet al., 2007). Moreover, nodular-enterolithic lithofaciesand development of microbial laminites typical ofsabkha environments do not occur, and there are nosigns of mudcracks and desiccation features or extensiveinterstitial and displacive nucleation within the sediment,which preclude interstitial or pedogenic precipitation ofgypsum in intertidal or subaerial environments.

The Carpentaria material has few clastic laminaeintercalated within the chemical precipitates, which canbe interpreted as relatively high-energy events (stormsor flash-flooding from the continent); little reworking isalso evident in some mixed laminae (gypsum-carbonate)and in the rough normal grading of some gypsumlaminae, but relatively good preservation of the gypsumcrystals seems to point to an intrabasinal (and notextensive) reworking. Micrite seems to be mainlyrelated to chemical precipitation, on the basis of itsfine-grain size and absence of micritization in theinvestigated layers. Intrabasinal reworking of the micriteis also considered, as indicated by the presence ofapparently clastic gypsum crystals enclosed in thecarbonates.

Lateral continuity and fine lamination is one of thebetter indicators of evaporite deposition in deep water(Schreiber, 1978; Kendall, 1979; Warren, 1983; amongothers); nevertheless, fine lamination in primary gypsumis not always evidence of deep evaporite precipitation.Warren (1982) described shallow (b10 m depth)submillimetre to centimetre-thick gypseous laminae inSouth Australian coastal salinas. The Carpentaria gyp-sum-carbonate cycles do not have lateral continuity, thus,the evaporitic layers within the Gulf of Carpentaria can beinterpreted as having formed in a shallow water body.Evaporites would have been deposited in the marginalzone of a shallow permanent water body. This petrolog-ical evidence agrees with the presence of discorbids,which are frequent in shallow transitional environmentswith a seasonal variability, and with the depositionalfacies defined by Reeves et al. (2007) on the basis ofmicropalaeontological analysis of these sediments.

5.2. Marine vs. continental geochemical signatures in atransitional evaporitic setting

5.2.1. Minor and trace element contents of gypsumExtensive information about minor/trace element

contents in gypsum in not available; Sr is the mostlycommonly reported element in gypseous units (Lu et al.,1997, Rosell et al., 1998; Toulkeridis et al., 1998; Luet al., 2002; Matano et al., 2005; among others). Thepartition coefficients of the majority of elements havenot been established, with the exception of Sr, Na, Mgand K (Butler, 1973; Kushnir, 1982). In principle, the Srcontents (and also sulfur isotopes) of Carpentariangypsum accord with a marine-derived precipitate.However, further discussion is warranted; it is notunusual for some evaporitic formations with seleniticlithofacies displaying Sr contents and sulfate isotopiccomposition typical of marine conditions to havesignificant meteoric (continental) influence as deducedfrom their 87Sr/86Sr ratios (Playà et al., 2000; Taberneret al., 2004). The Sr concentrations in the modernnorthern Australian river waters are low (from 2 to15 ppb; Cendón et al., 2004b). The average 87Sr/86Srsignature for northern Cape York rivers being 0.7148(Cendón et al., 2004c), considerably more radiogenicthan world average riverine inputs (0.7119; Palmer andEdmond, 1989) and that of modern marine values(0.7092; Elderfield, 1986). The low Sr concentration inthe northern rivers would suggest that a large continentalfluvial input would be necessary to produce the87Sr/86Sr values observed in the gypsum (0.7093 to0.7098). However marine-derived Sr would have playeda major role during the gypsum precipitation.

Elemental ratios can provide some clues on the originof the different elements. For those elements whoseconcentrations in seawater are much higher than inprobable continental inputs (i.e.: Sr and Ba), gypsumsamples have Ba/Sr ratios close to that of seawater.However, for elements that may not behave conserva-tively during evaporation (e.g. Mn) or where continentalwaters have concentrations within the same order ofmagnitude or higher than seawater (V, Co, Ni, Cu, Zn)these have ratios similar to those found in thecontinental inputs (i.e. Ni/Cu).

As would be expected by their geochemical similar-ities, the Rb and Li contents of the Carpentaria gypsumsamples are positively correlated (Fig. 4). Rb and Lihave behaved similarly during brine evolution, accu-mulating in the marine brine during evaporation untilthe precipitation of K–Mg-bearing phases (Holser,1979). The V contents in the studied gypsum samplesare similar to those of Rb and Li (Table 1); considering



that V concentration in seawater is much lower than Rband Li and given that Rb–Li also display a similar trendwith V, this reinforces the argument for the presence ofboth marine and continental inputs during gypsumprecipitation. Nevertheless, covariance between the Sr–Ba contents in gypsum is also related to theirgeochemical affinities (ionic radii and charge); thus, itreinforces the notion that Ba2+ can substitute for Ca2+.

5.2.2. REE contents of gypsumThe distinction between what constitutes the REE

dissolved load and the REE suspended load in waters hasbeen a subject of debate over the last few decades(Goldstein and Jacobsen, 1988; Sholkovitz, 1992,Sholkovitz et al., 1999; among others). Gypsumprecipitating in an open environment would naturallyincorporate not only those elements in the brine but alsosome suspended particulate matter. While our Carpen-taria samples were filtered after dissolution of gypsum(0.45 μm pore size) this may not totally eliminatemineral particles, particularly clays, and would noteliminate colloidal contributions. However, a compari-son of the REE distribution in gypsum samples and REEin 0.45 μm filtered water samples from the surroundingrivers reveals a common distribution of REE in bothgypsum precipitates and modern river water (Fig. 5B),ultimately pointing to a common origin for the REEdissolved and suspended load in gypsum and rivers.

The behaviour of REEs during evaporative concen-tration and gypsum precipitation is not well understood.However, several factors influence REE uptake: a) theREE concentrations in water seem to increase duringevaporation. A decrease in dissolved REE content isgenerally caused by their capture within the crystallattice and compensated by the decrease in the volume ofsolution resulting from the evaporation, thereforecausing a general net increase in REE (Kagi et al.,1993); b) REE3+ cations can substitute for Ca2+

(Guichard et al., 1979; Baumer et al., 1997); and, c)chemical sediments are most likely to reflect thecomposition of the fluid from which they precipitate(Rollinson, 1993; Roy and Smykatz-Kloss, 2007).

Wyndham et al. (2004) and Akagi et al. (2004)present evidence that the REE compositions in coralsamples reflect that of the surrounding seawater and thatthe distribution coefficients do not substantially varyamong the REE. Similarity this has also been observedin the distribution coefficents among hydrothermalfluids and precipitated anhydrites (Kagi et al., 1993;Humphris and Bach, 2005). Conversely, Baumer et al.(1997) observed from their experiments that light REEcan show higher relative concentrations in gypsum but

heavy REE are less prone to enter into the lattice,suggesting a decrease of the distribution coefficientsfrom the light to the heavy REEs. Our data support thisobservation, with a similar behaviour for the heaviestREE. While the general concentration decrease from Smto Tm mimics that of river water from the northern CapeYork rivers (Table 1, Fig. 5), the Carpentaria gypsumdeviates from this pattern with depletion from possiblyTm (not analysed) to Yb and Lu. This supports afractionation of HREE during gypsum precipitation withrespect to the parent brine.

The concentration of REE in seawater is much lowerthan in rivers. Therefore REE are typically mainlycontrolled by the continental input. Open seawatershows high Yand very low Ce with a gradual enrichmentin REE concentrations from the light to heavy REE on anormalized plot (Fig. 5B; Elderfield and Greaves, 1982;Bau et al., 1997; Alibo and Nozaki, 1999). Thisdistribution has been probably maintained, at least,from the Cretaceous until present-day (Lécuyer et al.,2004). The REE distribution in the Carpentaria gypsumsamples is basically the inverse of that observed inseawater and coincides closely with that determined inriver waters from northern Cape York rivers (Fig. 5B).The normalized REE patterns for gypsum show thatcontinental water input was restricted to a specific groupof rivers from northern Cape York. Rivers south of theWenlock river such as the Archer–Coen river haveabundant igneous lithologies in their catchments andtheir REE normalized patterns are significantly different.Our study shows that only the northernmost riverscontributed water to Lake Carpentaria during gypsumprecipitation, therefore abundant rainfall only reachedthe north of Cape York. If rainfall extended further south,their rivers may have largely evaporated before coveringthe N200 km from the continent to the Lake. Mostauthors also consider that deviations from the openseawater REE profile in their studied samples probablyindicate a non-marine influence, such as riverine waters,erosion and weathering of volcanics, etc. (Toulkeridiset al., 1998; Nozaki et al., 1999; Amawaka et al., 2000;Frimmel and Jiang, 2001, among others). Moreover, forour data set we propose that small variations of thegypsum REE distributions are due to seawater influence.Seawater presence is particularly noticeable in the LREEwhere marine and continental water normalized valuesare more different. The Ce/Ce⁎ anomaly (Table 1)suggests an increase in continental input from the bottomof the evaporitic sequence up to sample 858–859 (Ce/Ce⁎=1.01) followed by a change to a more marineinfluence (Ce/Ce⁎=0.89) towards the top of the gypsumsequence (Fig. 4).



5.2.3. Oxygen and sulfur isotope composition ofgypsum

In spite of the relatively high sulfate-oxygenisotope value calculated for the Carpentaria brine(SO4–δ

18Ovsmow=+12.1‰), this is quite common inevaporitic environments, where the isotopic enrich-ment commonly pertains to a bacterially mediatedprocess. Furthermore a small proportion of internalrecycling cannot be ruled out; clastic recycling hasbeen indicated by the intercalation of a few reworkedgypsum crystals within the chemical precipitates.Clastic and chemical recycling could coexist givingthe oxygen isotopic enrichment in the gypsum. Thelow topographic gradient of the basin would favourdissolution processes with rising water levels.

The oxygen and sulfur isotopic compositionsdisplay a similar vertical evolution (Fig. 6A); Srcontents covary inversely with isotopic values (Figs.4, 6B, C). These vertical changes could be interpretedas the result of successive opening and restriction of theevaporitic basin. Increasing Sr contents could reflect anincrease of salinity due to higher evaporation and thedecreasing δ34S and δ18O values in the precipitatedgypsum would be due to a reservoir effect in the brine;compatible with a relatively closed basin. Restriction ofthe basin and disconnection from the sea need not betotal; the isotopic composition of sulfur lies within thenormal marine range, suggesting a seawater influxbefore the onset of evaporation or more plausibly alarge remnant seawater body left trapped as a lake aftersea-level fall. Furthermore, the sulfate concentrationin the northern river waters is very low, less than 1 ppm,and their major ion ratios are like those found in sea-water with their sulfate isotopic composition expectedto be similar to that of the seawater (Chivas et al.,1991).

5.2.4. Strontium isotope composition of gypsumThe 87Sr/86Sr values found in the gypsum are more

radiogenic than average seawater values in all samplessuggesting a continental influence. Cendón et al. (2004c)calculated an approximate weighted average of the87Sr/86Sr of all rivers in the Gulf of Carpentaria duringthe dry-season, obtaining an estimate value of 0.7178.The 87Sr/86Sr of northern Cape York rivers is lessradiogenic, average 0.7148 (Cendón et al., 2004c). Thesevalues rise significantly in both Sr concentration andisotopic ratio as rivers south of the Coen and Archer areprogressively included (Fig. 1). Therefore the influenceof non-marine waters that are more radiogenic thanseawater in the Carpentaria region is highlighted by thestrontium isotope compositions.

5.2.5. Oxygen and carbon isotope composition ofcarbonates

The micritic laminae display more negative oxygenand carbon isotopic values than carbonates precipitatedfrom seawater. The oxygen isotope composition ofcarbonates depends on the δ18O value and temperatureof the water from which they precipitate. In evaporiticenvironments, the isotopic enrichment of the brine byevaporation is larger than the temperature effect (whichwill decrease the δ18O values). As a result, marineevaporitic carbonates generally show δ18O valueshigher (up to +11‰; Aharon et al., 1977) than thosefrom open marine environment. Additionally, undernormal marine conditions, the total dissolved inorganiccarbon (TDIC) reservoir is linked to the atmosphericCO2 reservoir and the carbonates show positive δ13Cvalues, up to +2‰ (Duplessy, 1978). Thus, the morenegative δ18O (and δ13C) values in Carpentariacarbonates are far from those expected for evaporatingmarine brines and strongly reflect the presence ofmeteoric waters in the Carpentaria basin.

A similar oscillation trend between δ13C and δ18O incarbonates is found (Fig. 4). Carbon isotope composi-tions display the widest dispersion of values (7.5‰)whereas oxygen isotope values only vary by 4‰. Giventhat the δ13C of a calcium carbonate crystallized inisotopic equilibrium with the TDIC reservoir is barelyaffected by temperature variations, it is clear thattemperature, and evaporation, are not the dominantparameters affecting the δ13C and δ18O values. This factsuggests that bulk changes in the composition of thebrines during evaporation are significant. According toour geochemical data on gypsum, oscillations in freshwater inputs can explain the positive correlationbetween the δ13C and δ18O values. Similar isotopicevolutions have been described by Magaritz (1987) inmarine carbonates related to evaporites.

Furthermore, under open system conditions, impor-tant negative δ13C variations of TDIC have been relatedto a variation on the relative proportion of C4/C3 plantspresent (Cerling, 1984; Melo et al., 2003). Thus,although the δ13C values of the carbonate precipitatedin isotopic equilibrium with the total dissolvedinorganic carbon reservoir (TDIC) is only slightlyaffected by temperature variations, an increase oftemperature, would reflect drier conditions, dominanceof C4 grasses, and a lack of meteoric water input, withhigher δ18O and δ13C values. Conversely, temperaturedecrease would reflect more humid conditions, de-crease of C4 grasses and increased influence of C3 trees(rainforest), and a higher meteoric water input, withlower isotope values.



5.3. Cyclicity

A repetition of carbonate-gypsum laminae at a μm tomm scale occurs in the whole unit studied; couplets are afew micrometres to 15 mm thick, with a mean of 400–700 μm. The central part of the evaporitic unit reveals anincrease of the evaporitic component compared to theupper and lowermost parts: enclosedmicrite in the gypsumlevels are less and these laminae are thicker (up to 15 mm)and purer. Internal cyclicity in the thickest cycles isobserved. Each couplet of calcite and gypsum may beinterpreted as sub-Milankovitch (climatically driven)cycles, possibly reflecting an annual cycle of deposition.The thickness of these cycles is consistent with seasonal,varve-like carbonate–evaporite cycles (Anderson et al.,1972; Magaritz, 1987; Magee, 1991), and markingoscillations from more diluted to more saline conditions.Thus, in the annual cycles, salinity increases during the“dry season” due to high evaporation, producing hypersa-line conditions. During the “wet season”, salinity decreasesbecause of freshwater input. The thickness of the calcite-gypsum couplets is also consistent with that precipitatedannually in modern salt-works (Ortí et al., 1984).

The existence of small gypsum crystals indicates thatsupersaturation produces initial sudden nucleation due tothe rapid increase of water salinity during the driest andhottest seasons. Reverse gradded-bedding of fine-grained(chemically-precipitated) gypsum crystals— this can indi-cate that the equilibrium is reached at the top by the in-crease in the stability of the brine (Schreiber, 1978;Magee,1991)— is not observed; this fact can reflect the instabilityof the brines in the studied evaporitic sequence and therapid environmental change from evaporitic to carbonaticsedimentation. This interpretation is supported by the totalabsence of “grass-like” or selenitic gypsum lithofacies,which represent the most stable Ca-sulfate evaporitic litho-facies, with vertically radiating growth from the surface ofthe substrate. The lamina boundaries provide no evidenceof exposure nor dissolution andprecipitation of halite is notobserved, even though post-precipitation dissolutioncannot be excluded; thus, the high salinity stage is notreached in the brine and salinity changes in the water bodyseem to be quick, reverting to more dilute conditions (andsedimentation of the micritic laminae). This interpretationof the relatively low salinity of the brine is also supportedby the fact that fine-grained gypsum precipitation is com-monly related to the lower salinity stage in coastal salinasand modern marine salt-works, preceding selenitic pre-cipitation (Ortí et al., 1984; Geisler-Cussey, 1985; Utrilla,1985; Rosell et al., 1998; Ortí and Salvany, 2004).

Higher order of fluctuation (0.5 to 1.5 cm thick) isrevealed by the isotopic composition of micritic laminae

(Fig. 4) and contain between 10 and 25 lower ordercycles. This fluctuation curve could be related topluriannual climatic oscillations and could have beencaused by short-term weather variations which stronglyinfluenced rainfall, producing decreased salinity. De-cadal monsoonal cyclicity, as reported by Charles et al.(1997) and Cole et al. (2000), is a possibility.

6. Palaeogeographic implications and conclusions

The evaporitic level of core MD-32 is stratigraphicallyvery close to the transition betweenmarine and non-marineconditions (Fig. 2), on the basis of Foraminifera andOstracoda assemblages (Reeves et al., 2007). Theevaporitic level, from available age constraints, and faciescoupled to sea-level changes corresponds toMarine IsotopeStage (MIS) 4 (Waelbroeck et al., 2002; Reeves et al.,2007). The ages are in good agreement with thegeochemical data from the gypsum and carbonates,which points to a strong influence of meteoric waters inthe Carpentaria basin during evaporite precipitation butwith residual evidence of marine solutes.

Throughout evaporite deposition, sea-level remainedbelow the level of the Arafura Sill (− 53 m). However, alocalized increase in marine influence suggests that thesea-level was close to the Sill with storms, peak high tidesor other sporadic causes (i.e. tsunamis) possibly beingable to breach the sill. When evaporitic conditions started,the initial Carpentaria Lakewater would have beenmostlymarine and extended up to the present −53 m water depthcontour (Fig. 1; Chivas et al., 2001). The prevalent dryconditions would have driven evaporation, causing a vastland surface to be exposed, with water receding to thedeepest part of the basin around the position of coresMD-32 and MD-33 (the latter core did not penetrate to asediment depth sufficient to encounter the evaporitelayer). The evaporitic conditions (dry-climate) reached itsmaximum during precipitation of the central laminae,after which continental inputs start to be more noticeable.This could be due to; a) an increase in rainfall, and for, b)the lake reaching hydrological equilibrium after gradualcontraction from the moment of ocean disconnection (seebathymetry in Fig. 1). The lake reached its equilibriumonce the volume of total evaporation losses were approx-imately balanced for by riverine inflows. During theprocess leading to a “sustainable” lake, evaporite depo-sition took place. Three subunits can be differentiated inthe studied evaporitic interval of the core MD-32 basedmainly on our geochemical information (Fig. 4):1) From 873–868 cm. Micritic deposition intercalated

with very thin gypsum laminae. The isotopiccompositions of carbonate display small oscillations



with a general trend to lower values indicating thatLake Carpentaria was already isolated from theoriginal marine seawater with evidence of fresh waterinput towards the top of this unit.

2) From 868–862 cm. Maximum development ofgypsum laminae alternatingwith thinmicritic laminae.Drier conditions prevail causing increased evapora-tion. Sr concentrations increase while sulfate-δ18Ovalues decrease, probably due to a reservoir effectduring chemical precipitation of gypsumcrystals in thebrine with increasing salinity. Strontium isotope ratiosand REE contents of gypsum and isotope composi-tions of micrite show the influence of continentalinputs. Carbonate-δ18O values show a marked oscil-lation, in contrast with the more stable curve insubunits 1 and 3; this suggests periodic dilution of thebrine due to sporadic rainfall increase. The reservoireffect is not observed in the isotopic compositions ofcarbonates given that they are mainly controlled by thecontinental isotope signals in this subunit; conversely,sulfate is mainly residual marine in origin.

3) From862–853 cm.Gypsum laminae are progressivelythinner. An increase in continental run-off and rainfallcause exposed evaporites to redissolve and precipitatein deeper areas, shifting sulfate-δ18O to higher values,Sr to lower concentrations and 87Sr/86Sr to moreradiogenic (continental) values. The uppermost 3 cen-timetres within this unit contain evidence for the entryof seawater, not enough to overprint the lake signaturebut enough to shift Ce/Ce⁎ anomalies and carbonate-δ18O values toward marine values.

The thickness of the calcite-gypsum couplets is, ingeneral, similar to those precipitated annually in modernenvironments. Furthermore, the oscillation of gypsum-calcite indicates variations fromhumid to drier conditions.The lack of dissolution features ensures that the record isfairly complete at least during gypsum precipitation.Multi-annual climatic oscillations are also inferred,supported by the cyclicity in the carbonate-δ18O andδ13C values (Fig. 4). It is postulated that a reducedmonsoon-like rainfall pattern operated in the north ofCarpentaria and CapeYork during evaporite precipitation,which was more intense in the 868–862 cm interval.

The multiple approach taken in this study highlightsthe importance of avoiding single-parameter palaeoen-vironmental reconstructions in complex settings such asevaporative basins. Although Sr concentrations and thesulfate-δ34S values show an initial marine origin for thebrines, sulfate-δ18O, carbonate-δ13C and δ18O values,87Sr/86Sr and REE show the influence of continentalinputs varying throughout evaporite deposition. The

combination of these geochemical markers togetherwith detailed petrography has allowed a soundpalaeoenvionmental reconstruction.

Acknowledgments

The present work was supported by the SpanishGovernment Projects CGL2005-05337/BTE, andCGL2006-04860/BTE and the Grup Consolidat deRecerca Geologia Sedimentària 2005SGR-00890. Ini-tial sedimentological studies, trace-element and REEanalyses were supported by the Australian ResearchCouncil grants A39600498 and DP0208605. Radiocar-bon dates were supported by AINSE (AustralianInstitute of Nuclear Science and Engineering) grants98/155R, 01/032, 02/025 and 05/027. Laboratory workwas greatly facilitated by the technical assistance of TimWyndham and Graham Mortimer at the ResearchSchool of Earth Sciences (ANU, Canberra). The δ18O,δ34S and δ13C isotopic analyses were carried out at theServeis Cientifico-Tècnics, SCT of the Universitat deBarcelona. The authors are indebted to Drs. F. Ortí, L.Rosell, J.J. Pueyo and the two anonymous reviewers fortheir helpful comments, and to Drs. J. Serra and C.Ferrández for identification of biogenic componentswithin the gypsum-micrite layers. We also thank StuartHankin from ANSTO for drafting Fig. 1.

References

Aharon, P., Kolodny, Y., Sass, E., 1977. Recent hot brine dolomi-tization in the “Solar Lake”, Gulf of Elat, isotopic, chemical andmineralogical study. J. Geol. 85, 27–48.

Akagi, T., Hashimoto, Y., Fu, F.F., Tsuno, H., Tao, H., Nakano, Y.,2004. Variation of the distribution coefficients of rare earthelements in modern coral-lattices: species and site dependencies.Geochim. Cosmochim. Acta 68, 2265–2273.

Alibo, D.S, Nozaki, Y., 1999. Rare earth elements in seawater: particleassociation, shale-normalization and Ce oxidation. Geochim.Cosmochim. Acta 63 (3/4), 363–372.

Amawaka, H., Alibo, D.S., Nozaki, Y., 2000. Nd isotopic compositionandREE pattern in the surfacewaters of the eastern IndianOcean andits adjacent seas. Geochim. Cosmochim. Acta 64, 1715–1727.

Anderson, R.Y., Dean Jr., W.E., Kirkland, D.W., Snider, H.I., 1972.Permian Castile varved evaporite sequence, West Texas and NewMexico. Geol. Soc. Amer. Bull. 83, 59–86.

Bau, M., Möller, P., Dulski, P., 1997. Yttrium and lanthanides ineastern Mediterranean seawater and their fractionation duringredox-cycling. Mar. Chem. 56, 123–131.

Baumer, A., Blanc, Ph., Cesbron, F., Ohnenstetter, D., 1997.Cathodoluminescence of synthetic (doped with rare-earth ele-ments) and natural anhydrites. Chem. Geol. 138, 73–80.

Berger,W.H., 1979. Stable isotopes in Foraminifera. In: Lipps, J.H., Berger,W.H., Buzas, M.A., Douglas, R.G., Ross, C.A. (Eds.), ForaminiferalEcology and Paleoecology. SEPM Short Course, vol. 6. Soc. Econ.Mineral. Paleontol. Miner., Houston, USA, pp. 156–197.



Bowler, J.M., Wyrwoll, K.H., Lu, Y., 2001. Variations of the northwestAustralian summer monsoon over the last 300,000 years: thepaleohydrological record of the Gregory (Mulan) Lakes System.Quat. Int. 83–85, 63–80.

Broecker, W.S., 1986. Oxygen isotope constraints on surface oceantemperatures. Quat. Res. 26, 121–134.

Butler, G.P., 1973. Strontium geochemistry of modern and ancientcalcium sulphate minerals. In: Purser, B.H. (Ed.), The Persian Gulf.Holocene Carbonate Sedimentation and Diagenesis in a ShallowEpicontinental Sea. Springer–Verlag, Germany, pp. 423–452.

Cendón, D.I., Peryt, T.M., Ayora, C., Pueyo, J.J., Taberner, C., 2004a.The importance of recycling processes in the Middle MioceneBadenian evaporite basin (Carpathian foredeep): paleoenviron-mental implications. Palaeogeogr. Palaeoclimatol. Palaeoecol. 212,141–158.

Cendón, D.I., Chivas, A.R., García, A., 2004b. Chemistry of the riversin the Gulf of Carpentaria drainage division and possiblecorrelations with the sedimentary record during lake phases. 17thAustralian Geological Convention. Dynamic Earth: Past, Presentand Future. Hobart, Australia, vol. 73, p. 228. Abstracts.

Cendón, D.I., Chivas, A.R., Wyndham, T., García, A., 2004c.Chemistry of the rivers in the Gulf of Carpentaria drainagedivision (N-Australia): palaeoclimatic implications. 32nd Interna-tional Geological Congress. CD Publication, Florence, Italy.

Cerling, T.E., 1984. The stable isotopic composition of modern soilcarbonate and its relationship to climate. Earth Planet. Sci. Lett. 71,229–240.

Chappell, J., Omura, A., Esat, T., McCulloch, M., Pandolfi, J., Ota, Y.,Pillans, B., 1996. Reconciliation of late Quaternary sea levelsderived from coral terraces at Huon Peninsula with deep seaoxygen isotope records. Earth Planet. Sci. Lett. 141, 227–236.

Charles, C.D., Hunter, D.E., Fairbanks, R.G., 1997. Interactionbetween the ENSO and the Asian Monsoon in a coral record oftropical climate. Science 277, 925–928.

Chivas, A.R., 2007. Terrestrial evaporites. In: Nash, D.J.,McLaren, S.J.(Eds.), Geochemical Sediments and Landscapes. BlackwellScience, pp. 330–364.

Chivas, A.R., Andrew, A.S., Lyons, W.B., Bird, M.I., Donnelly, T.H.,1991. Isotopic constraints on the origin of salts in Australianplayas. 1. Sulphur. Palaeogeogr. Palaeoclimatol. Palaeoecol. 84,309–332.

Chivas, A.R., García, A., van der Kaars, S., Couapel, M.J.J., Holt, S.,Reeves, J.M., Wheeler, D.J., Switzer, A.D., Murray-Wallace, C.V.,Banerjee, D., Price, D.M., Wang, S.X., Pearson, G., Edgar, N.T.,Beaufort, L., De Deckker, P., Lawson, E., Cecil, C.B., 2001. Sea-level and environmental changes since the last interglacial in the Gulfof Carpentaria, Australia: an overview. Quat. Int. 83–85, 19–46.

Clauer, N., 1976. 87Sr/86Sr composition of evaporitic carbonates andsulphates from Miocene sediment cores in the Mediterranean Sea(D.S.D.P., Leg 13). Sedimentology 23, 133–140.

Cole, J.E., Dunbar, R.B., McClanahan, T.R., Muthiga, N.A., 2000.Tropical Pacific forcing decadal SST variability in theWestern IndianOcean over the past two centuries. Science 287, 617–619.

Dronkert, H., 1985. Evaporite models and sedimentology of Messinianand recent evaporites. PhD Thesis, University of Amsterdam.GUA Papers of Geology, series 1, n. 24, 283 pp.

Duplessy, J.C., 1978. Isotope studies. In: Gribbin, J. (Ed.), ClimaticChange. Cambridge University Press, Cambridge, UK, pp. 46–67.

Edgar, N.T., Cecil, C.B., Mattick, R.E., Chivas, A.R., De Deckker, P.,Djajadihardja, Y.S., 2003. A modern analogue for tectonic,eustatic, and climatic processes in cratonic basins: Gulf ofCarpentaria, Northern Australia. In: Cecil, C.B., Edgar, N.T.

(Eds.), Climate Controls on Stratigraphy. SEPM Spec. Publ.,vol. 77, pp. 193–205.

Eggins, S.M., Woodhead, J.D., Kinsley, L.P.J., Mortimer, G.E.,Sylvester, P., McCulloch, M.T., Hergt, J.M., Handler, M.R.,1997. A simple method for the precise determination of N40 traceelements in geological samples by ICPMS using enriched isotopeinternal standardisation. Chem. Geol. 134, 311–326.

Elderfield, H., 1986. Strontium isotope stratigraphy. Palaeogeogr.Palaeoclimatol. Palaeoecol. 57, 71–90.

Elderfield, H., Greaves, M.J., 1982. The rare earth elements inseawater. Nature 296, 214–219.

Farrell, J.W., Clemens, S.C., Gromet, L.P., 1995. Improved chronos-tratigraphic reference curve of late Neogene seawater 87Sr/86Sr.Geology 23, 403–406.

Frimmel, H.E., Jiang, S.-Y., 2001. Marine evaporites from an oceanicisland in the Neoproterozoic Adamastor ocean. Precambrian Res.105, 57–71.

Gaillardet, J., Viers, J., Dupré, B., 2004. Trace elements in river waters.In: Turekian, K.K., Holland, H.D. (Eds.), Treatise on Geochem-istry, vol. 5. Elsevier, pp. 225–272.

Geisler-Cussey, D.,1985. Aproche sédimentologique et géochimiquedes mécanismes générateurs des formations évaporitiques actuelleset fossilles. PhD Thesis, Université Nancy I, France.

Gibert, Ll., Ortí, F., Rosell, L., 2007. Plio-Pleistocene lacustrine evaporitesof the Baza Basin (Betic Chain, SE Spain). Sediment. Geol. 200,89–116.

Goldstein, S.J., Jacobsen, S.B., 1988. Rare earth elements in riverwaters. Earth Planet. Sci. Lett. 89, 35–47.

Guichard, F., Church, T.M., Treuil, M., Jaffrezic, H.H., 1979. Rareearths in barite: distribution and effects on aqueous partitioning.Geochim. Cosmochim. Acta 43, 983–997.

Holser, W.T., 1979. Trace elements and isotopes in evaporites. In:Burns, R.G. (Ed.), Marine Minerals. . Reviews in Mineralogy, vol.6. Min. Soc. Amer., Chantilly, USA, pp. 295–346.

Humphris, S.E., Bach, W., 2005. On the Sr isotope and REEcompositions of anhydrites from the TAG seafloor hydrothermalsystem. Geochim. Cosmochim. Acta 69, 1511–1525.

Jones, M.R., Torgersen, T., 1988. Late Quaternary evolution of LakeCarpentaria on the Australia–New Guinea continental shelf. Aust.J. Earth Sci. 35, 313–324.

Kagi, H., Dohmoto, Y., Takano, S., Masuda, A., 1993. Tetrad effect inlanthanide partitioning between calcium sulphate crystal and itssaturated solution. Chem. Geol. 107, 71–82.

Kamber, B.S., Greig, A., Collerson, K.D., 2005. A new estimate forthe composition of weathered young upper continental crust fromalluvial sediments, Queensland, Australia. Geochim. Cosmochim.Acta 69, 1041–1058.

Kendall, A.C., 1979. Subaqueous evaporites. Geosci. Can. 5, 124–139.Kroon,D., Reijmer, J.J.G., Rendle, R.H., 2000.Mid-to Late-Quaternary

variations in the oxygen isotope signature ofGlobigerinoides ruberat Site 1006 in the western subtropical Atlantic. In: Swart, P.K.,Eberli, G.P., Malone, M.J., Sarg, J.F. (Eds.), Proceedings of theOcean Drilling Program. Scientific Results, vol. 166, pp. 13–22.

Kushnir, J., 1982. The composition and origin of brines during theMessinian desiccation event in the Mediterranean basin as deducedfrom concentrations of ions coprecipitated with gypsum andanhydrite. Chem. Geol. 35, 333–350.

Lécuyer, Ch., Reynard, B., Grandjean, P., 2004. Rare earth elementevolution of Phanerozoic seawater recorded in biogenic apatites.Chem. Geol. 204, 63–102.

Lloyd, R.M., 1968. Oxygen isotope behavior in the sulphate watersystem. J. Geophys. Res. 73, 6099–6110.



Lu, F.H., Meyers, W.J., Schoonen, M.A.A., 1997. Minor and traceelement analyses on gypsum: an experimental study. Chem. Geol.142, 1–10.

Lu, F.H., Meyers, W.J., Hanson, G.N., 2002. Trace elements andenvironmental significance of Messinian gypsum deposits, theNijar Basin, southeastern Spain. Chem. Geol. 192, 149–161.

Magaritz, M., 1987. A new explanation for cyclic deposition in marineevaporite basins: meteoric water input. Chem. Geol. 62, 239–250.

Magee, J.W., 1991. Late Quaternary lacustrine, groundwater, aeolianand pedogenic gypsum in the Prungle Lakes, southeasternAustralia. Palaeogeogr. Palaeoclimatol. Palaeoecol. 84, 3–42.

Matano, F., Barbieri, M., Di Nocera, S., Torre, M., 2005. Stratigraphyand strontium geochemistry of Messinian evaporite-bearingsuccessions of the Apennines foredeep, Italy: implications forthe Mediterranean “salinity crisis” and regional palaeogeography.Palaeogeogr. Palaeoclimatol. Palaeoecol. 217, 87–114.

McCulloch, M.T., De Deckker, P., Chivas, A.R., 1989. Strontiumisotope variations in single ostracod valves from the Gulf ofCarpentaria, Australia: A paleoenvironmental indicator. Geochim.Cosmochim. Acta vol. 53, 1703–1710.

McManus, J.F., Oppo, D.W., Cullen, J.L., 1999. A 0.5-million-yearrecord of millennial-scale climate variability in the North Atlantic.Science 283, 971–974.

Melo, M.S., Giannini, P.C.F., Pessenda, L.C.R., Brandt Neto, M.,2003. Holocene paleoclimatic reconstruction based on the LagoaDourada deposits, southern Brazil. Geologica Acta 1, 289–302.

Michalzik, D., Elbracht, J., Mauthe, F., Reinhold, C., Scheider, B.,1993. Messinian facies relations in the San Miguel de SalinasBasin, SE-Spain. Dtsch. Geol. Ges. 144, 352–369.

Nozaki, Y., Alibo, D.S., Amakawa, H., Gamo, T., Hasumoto, H., 1999.Dissolved rare earth elements and hydrography in the Sulu Sea.Geochim. Cosmochim. Acta 63, 2171–2181.

Ortí, F., Salvany, J.M., 2004. Coastal salina evaporites of the Triassic–Liassic boundary in the Iberian Peninsula: the Alacón borehole.Geologica Acta 2, 291–304.

Ortí, F., Pueyo, J.J., Geisler, D., Dulau, N., 1984. Evaporiticsedimentation in the coastal salinas of Santa Pola (Alicante,Spain). Rev. Inst. Invest. Geol. 38–39, 169–220.

Palmer, M.R., Edmond, J.M., 1989. The strontium isotope budget ofthe modern ocean. Earth Planet. Sci. Lett. 92, 11–26.

Passmore, V.L., Williamson, P.E., Gray, A.R.G., Wellman, P., 1993a.The Bamaga Basin—a new exploration target. In: Carman, G.J.,Carman, Z. (Eds.), Petroleoum Exploration and Development inPapua New Guinea. Proceed. Second Papua New Guinea Petrol.Convention. Port Moresby, New Guinea, pp. 233–240.

Passmore, V.L., Williamson, P.E., Maung, T.U., Gray, A.R.G., 1993b.The Gulf of Carpentaria — a new basin exploration targets.APEA (Australian Petroleum Exploration Association) J. 33,297–314.

Playà, E., Rosell, L., 2005. The celestite problem in gypsum Srgeochemistry: an evaluation of the purifying methods ofgypsiferous samples. Chem. Geol. 221, 102–116.

Playà, E., Ortí, F., Rosell, L., 2000. Marine to non-marine sedimentationin the upper Miocene evaporites of the Eastern Betics, SE Spain:sedimentological and geochemical evidence. Sediment. Geol. 133,135–166.

Reeves, J.M., Chivas, A.R., García, A., De Deckker, P., 2007.Palaeoenvironmental change in the Gulf of Carpentaria (Australia)since the last interglacial based on Ostracoda. Palaeogeogr.Palaeoclimatol. Palaeoecol. 246, 163–187.

Reeves, J.M., Chivas, A.R., García, A., Holt, S., Couapel, M.J.J.,Jones, B.G., Cendón, D.I., and Fink, D., in press. The sedimentaryrecord of palaeoenvironments and sea-level change in the Gulf ofCarpentaria, Australia, through the last glacial cycle. Quat. Int.

Rollinson, H., 1993. Using Geochemical Data. Prentice Hall, Essex,UK, p. 352.

Rosell, L., Ortí, F., Kasprzyk, A., Playà, E., Peryt, T.M., 1998.Strontium geochemistry of Miocene primary gypsum: Messinianof southeastern Spain and Sicily and Badenian of Poland.J. Sediment. Res. 68, 63–79.

Roy, P.D., Smykatz-Kloss, W., 2007. REE geochemistry of the recentplaya sediments from the Thar Desert, India: An implication toplaya sediment provenance. Chemie der Erde 67, 55–68.

Schreiber, B.Ch., 1978. Environments of subaqueous deposition: agenetic model. In: Dean, W.E., Schreiber, B.Ch. (Eds.), MarineEvaporites. SEPM Short Course, vol. 4. Soc. Econ. Mineral.Paleontol. Miner., Houston, USA, pp. 43–73.

Sholkovitz, E.R., 1992. Chemical evolution of rare earth elements:fractionation between colloidal and solution phases of filtered riverwater. Earth Planet. Sci. Lett. 114, 77–84.

Sholkovitz, E.R., Elderfield, H., Szymczak, R., Casey, K., 1999. Islandweathering: river sources of rare earth elements to the WesternPacific Ocean. Mar. Chem. 68, 39–57.

Smart, J., Grimes, K.G., Doutch, H.F., Pinchin, J., 1980. The Carpentariaand Karumba basins, North Queensland. Australia, Bureau ofMineral Resources, Geology and Geophysics, Bull., vol. 202. 73 pp.

Taberner, C., Rouchy, J.M., Pueyo, J.J., Thirlwall, M., 2004. Origin ofsolutes and evaporite deposition at the end of the Messiniansalinity crisis. The onset of “Lago Mare” sedimentation. Abstracts4th International Congress Environment and Identity in theMediterranean. TheMessinian Salinity Crisis Revisited. Universitàdi Corsica, p. 83.

Thode, H.G., Monster, J., 1965. Sulfur-isotope geochemistry ofpetroleum, evaporites and ancient seas. Mem. Am. Assoc. Petrol.Geol. 4, 367–377.

Torgersen, T., Luly, J., De Deckker, P., Jones, M.R., Searle, D.E.,Chivas, A.R., Ullman, W.J., 1988. Late Quaternary environmentsof the Carpentaria basin, Australia. Palaeogeogr. Palaeoclimatol.Palaeoecol. 67, 245–261.

Toulkeridis, T., Podwojewski, P., Clauer, N., 1998. Tracing the sourceof gypsum in New Caledonian soils by REE contents and S–Srisotopic compositions. Chem. Geol. 145, 61–71.

Utrilla, R., 1985. Estudi sedimentològic i geoquímic de les salines dela Trinitat (Delta de l’Ebre) i San Pedro del Pinatar (Mar Menor).MSc Thesis, Universitat de Barcelona, Spain.

Waelbroeck, C., Labeyrie, L., Michel, E., Duplessy, J.C., McManus, J.F.,Lambeck, K., Balbon, E., Labracherie, M., 2002. Sea-level and deepwater temperature changes derived from benthic foraminiferaisotopic records. Quat. Sci. Rev. 21, 295–305.

Warren, J.K., 1982. The hydrological setting, occurrence andsignificance of gypsum in late Quaternary salt lakes in SouthAustralia. Sedimentology 29, 609–637.

Warren, J.K, 1983. On the significance of evaporite lamination. Proc.Sixth Int. Symposium on Salt. Toronto, Canada, vol. 1, pp. 161–170.

Wyrwoll, K.H., Miller, G.H., 2001. Initiation of the Australian summermonsoon 14,000 years ago. Quat. Int. 83–85, 119–128.

Wyndham, T., McCulloch, M., Fallon, S., Alibert, Ch., 2004. High-resolution coral records of rare earth elements in coastal seawater:biogeochemical cycling and a new environmental proxy. Geochim.Cosmochim. Acta. 68, 2067–2080.


Non-marine evaporites with both inherited marine and continental signatures: The Gulf of Carpentaria, Australia, at ∼ 70 ka

Documents